1 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
3 * asan.c (create_cond_insert_point): Maintain profile.
4 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
6 * basic-block.h (struct basic_block_def): Remove frequency.
7 (EDGE_FREQUENCY): Use to_frequency
8 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
10 (find_traces): Update to use to_frequency.
11 (find_traces_1_round): Likewise; use only IPA counts.
12 (bb_to_key): Likewise.
13 (connect_traces): Use IPA counts only.
14 (copy_bb_p): Update to use to_frequency.
15 (fix_up_crossing_landing_pad): Likewise.
16 (sanitize_hot_paths): Likewise.
17 * bt-load.c (basic_block_freq): Likewise.
18 * cfg.c (init_flow): Set count_max to uninitialized.
19 (check_bb_profile): Remove frequencies; check counts.
20 (dump_bb_info): Do not dump frequencies.
21 (update_bb_profile_for_threading): Update counts only.
22 (scale_bbs_frequencies_int): Likewise.
23 (MAX_SAFE_MULTIPLIER): Remove.
24 (scale_bbs_frequencies_gcov_type): Update counts only.
25 (scale_bbs_frequencies_profile_count): Update counts only.
26 (scale_bbs_frequencies): Update counts only.
27 * cfg.h (struct control_flow_graph): Add count-max.
28 (update_bb_profile_for_threading): Update prototype.
29 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
30 (find_many_sub_basic_blocks): Likewise.
31 * cfgcleanup.c (try_forward_edges): Likewise.
32 (try_crossjump_to_edge): Likewise.
33 * cfgexpand.c (expand_gimple_cond): Likewise.
34 (expand_gimple_tailcall): Likewise.
35 (construct_init_block): Likewise.
36 (construct_exit_block): Likewise.
37 * cfghooks.c (verify_flow_info): Check consistency of counts.
38 (dump_bb_for_graph): Do not dump frequencies.
39 (split_block_1): Do not update frequencies.
40 (split_edge): Do not update frequencies.
41 (make_forwarder_block): Do not update frequencies.
42 (duplicate_block): Do not update frequencies.
43 (account_profile_record): Do not update frequencies.
44 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
45 for global heuristics.
46 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
47 (expected_loop_iterations_unbounded): Use counts only.
48 * cfgloopmanip.c (scale_loop_profile): Simplify.
49 (create_empty_loop_on_edge): Simplify
51 (duplicate_loop_to_header_edge): Simplify
52 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
53 (update_br_prob_note): Take care of removing note when profile
55 (relink_block_chain): Do not dump frequency.
56 (rtl_account_profile_record): Use to_frequency.
57 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
58 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
59 (cgraph_update_edges_for_call_stmt_node): Likewise.
60 (cgraph_edge::verify_count_and_frequency): Update.
61 (cgraph_node::verify_node): Temporarily disable frequency verification.
62 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
64 (cgraph_edge::rebuild_edges): Convert to ipa counts.
65 * cgraphunit.c (init_lowered_empty_function): Do not initialize
67 (cgraph_node::expand_thunk): Update profile.
68 * except.c (dw2_build_landing_pads): Do not update frequency.
69 * final.c (compute_alignments): Use to_frequency.
70 (dump_basic_block_info): Do not dump frequency.
71 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
72 (dump_gimple_bb_header): Do not dump frequency.
73 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
75 * gimple-streamer-in.c (input_bb): Do not stream frequency.
76 * gimple-streamer-out.c (output_bb): Do not stream frequency.
77 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
78 (init_before_recovery): Do not update frequency.
79 (sched_create_recovery_edges): Do not update frequency.
80 * hsa-gen.c (convert_switch_statements): Do not update frequency.
81 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
82 (ipa_cp_c_finalize): Set max_count to uninitialized.
83 * ipa-fnsummary.c (get_minimal_bb): Use counts.
84 (param_change_prob): Use counts.
85 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
87 * ipa-split.c (consider_split): Use to_frequency.
88 (split_function): Use to_frequency.
89 * ira-build.c (loop_compare_func): Likewise.
90 (mark_loops_for_removal): Likewise.
91 (mark_all_loops_for_removal): Likewise.
92 * loop-doloop.c (doloop_modify): Do not update frequency.
93 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
95 * lto-streamer-in.c (input_function): Update count_max.
96 * omp-expand.c (expand_omp_taskreg): Update count_max.
97 * omp-simd-clone.c (simd_clone_adjust): Update profile.
98 * predict.c (maybe_hot_frequency_p): Use to_frequency.
99 (maybe_hot_count_p): Use ipa counts only.
100 (maybe_hot_bb_p): Simplify.
101 (maybe_hot_edge_p): Simplify.
102 (probably_never_executed): Do not take frequency argument.
103 (probably_never_executed_bb_p): Do not pass frequency.
104 (probably_never_executed_edge_p): Likewise.
105 (combine_predictions_for_bb): Check that profile is nonzero.
106 (propagate_freq): Do not set frequency.
107 (drop_profile): Simplify.
108 (counts_to_freqs): Simplify.
109 (expensive_function_p): Use to_frequency.
110 (propagate_unlikely_bbs_forward): Simplify.
111 (determine_unlikely_bbs): Simplify.
112 (estimate_bb_frequencies): Add hack to silence graphite issues.
113 (compute_function_frequency): Use ipa counts.
114 (pass_profile::execute): Update.
115 (rebuild_frequencies): Use counts only.
116 (force_edge_cold): Use counts only.
117 * profile-count.c (profile_count::dump): Dump new count types.
118 (profile_count::differs_from_p): Check compatiblity.
119 (profile_count::to_frequency): New function.
120 (profile_count::to_cgraph_frequency): New function.
121 * profile-count.h (struct function): Declare.
122 (enum profile_quality): Add profile_guessed_local and
123 profile_guessed_global0.
124 (class profile_proability): Decrease number of bits to 29;
125 update from_reg_br_prob_note and to_reg_br_prob_note.
126 (class profile_count: Update comment; decrease number of bits
127 to 61. Check compatibility.
128 (profile_count::compatible_p): New private member function.
129 (profile_count::ipa_p): New member function.
130 (profile_count::operator<): Handle global zero correctly.
131 (profile_count::operator>): Handle global zero correctly.
132 (profile_count::operator<=): Handle global zero correctly.
133 (profile_count::operator>=): Handle global zero correctly.
134 (profile_count::nonzero_p): New member function.
135 (profile_count::force_nonzero): New member function.
136 (profile_count::max): New member function.
137 (profile_count::apply_scale): Handle IPA scalling.
138 (profile_count::guessed_local): New member function.
139 (profile_count::global0): New member function.
140 (profile_count::ipa): New member function.
141 (profile_count::to_frequency): Declare.
142 (profile_count::to_cgraph_frequency): Declare.
143 * profile.c (OVERLAP_BASE): Delete.
144 (compute_frequency_overlap): Delete.
145 (compute_branch_probabilities): Do not use compute_frequency_overlap.
146 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
147 * sched-ebb.c (rank): Use counts only.
148 * shrink-wrap.c (handle_simple_exit): Use counts only.
149 (try_shrink_wrapping): Use counts only.
150 (place_prologue_for_one_component): Use counts only.
151 * tracer.c (find_best_predecessor): Use to_frequency.
152 (find_trace): Use to_frequency.
153 (tail_duplicate): Use to_frequency.
154 * trans-mem.c (expand_transaction): Do not update frequency.
155 * tree-call-cdce.c: Do not update frequency.
156 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
157 (gimple_merge_blocks): Likewise.
158 (gimple_split_edge): Likewise.
159 (gimple_duplicate_sese_region): Likewise.
160 (gimple_duplicate_sese_tail): Likewise.
161 (move_sese_region_to_fn): Likewise.
162 (gimple_account_profile_record): Likewise.
163 (insert_cond_bb): Likewise.
164 * tree-complex.c (expand_complex_div_wide): Likewise.
165 * tree-eh.c (lower_resx): Update profile.
166 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
168 (initialize_cfun): Do not initialize frequencies
169 (freqs_to_counts): Delete.
170 (copy_cfg_body): Ignore count parameter.
172 (expand_call_inline): Update count_max.
173 (optimize_inline_calls): Update count_max.
174 (tree_function_versioning): Update count_max.
175 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
176 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
178 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
179 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
180 (try_peel_loop): Likewise.
181 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
183 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
184 (tree_transform_and_unroll_loop): Do not use frequencies
185 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
186 Use reliable prediction only.
187 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
188 * tree-ssa-sink.c (select_best_block): Use to_frequency.
189 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
191 * tree-ssa-threadupdate.c (create_block_for_threading): Do
193 (any_remaining_duplicated_blocks): Likewise.
194 (update_profile): Likewise.
195 (estimated_freqs_path): Delete.
196 (freqs_to_counts_path): Delete.
197 (clear_counts_path): Delete.
198 (ssa_fix_duplicate_block_edges): Likewise.
199 (duplicate_thread_path): Likewise.
200 * tree-switch-conversion.c (gen_inbound_check): Use counts.
201 * tree-tailcall.c (decrease_profile): Do not update frequency.
202 (eliminate_tail_call): Likewise.
203 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
204 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
205 (optimize_mask_stores): Likewise.
206 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
207 * ubsan.c (ubsan_expand_null_ifn): Update profile.
208 (ubsan_expand_ptr_ifn): Update profile.
209 * value-prof.c (gimple_ic): Simplify.
210 * value-prof.h (gimple_ic): Update prototype.
211 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
212 * ipa-inline.c (compute_uninlined_call_time): Be sure that
214 (want_inline_self_recursive_call_p): Likewise.
215 (resolve_noninline_speculation): Only cummulate defined counts.
216 (inline_small_functions): Use nonzero_p.
217 (ipa_inline): Do not access freed node.
219 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
221 * config/aarch64/aarch64.c (aarch64_override_options_internal):
222 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
224 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
226 * config/riscv/riscv.c (riscv_legitimize_move): Handle
227 non-legitimate address.
229 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
231 * config/rs6000/rs6000.md (*lt0_disi): Delete.
232 (*lt0_<mode>di, *lt0_<mode>si): New.
234 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
236 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
239 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
240 Jim Wilson <jim.wilson@linaro.org>
242 * config/aarch64/aarch64-cores.def (saphira): New CPU.
243 * config/aarch64/aarch64-tune.md: Regenerated.
244 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
245 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
247 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
249 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
250 (arc_expand_prologue): Restore blink for millicode.
251 * config/arc/linux.h (LINK_EH_SPEC): Defined.
253 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
256 * config/i386/i386.c (ix86_vector_duplicate_value): Use
257 gen_vec_duplicate after forcing the scalar into a register.
259 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
261 * combine (try_combine): Print the insns input to try_combine to the
264 2017-11-02 Steve Ellcey <sellcey@cavium.com>
267 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
268 Remove second argument from aarch64_process_target_attr call.
269 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
271 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
273 (aarch64_handle_attr_arch): Remove second argument.
274 (aarch64_handle_attr_cpu): Ditto.
275 (aarch64_handle_attr_tune): Ditto.
276 (aarch64_handle_attr_isa_flags): Ditto.
277 (aarch64_process_one_target_attr): Ditto.
278 (aarch64_process_target_attr): Ditto.
279 (aarch64_option_valid_attribute_p): Remove second argument.
280 on aarch64_process_target_attr call.
282 2017-11-02 David Malcolm <dmalcolm@redhat.com>
284 * diagnostic.c: Include "selftest-diagnostic.h".
285 (selftest::assert_location_text): New function.
286 (selftest::test_diagnostic_get_location_text): New function.
287 (selftest::diagnostic_c_tests): Call it.
289 2017-11-02 David Malcolm <dmalcolm@redhat.com>
291 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
292 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
293 (class selftest::test_diagnostic_context): Move to...
294 * selftest-diagnostic.c: New file.
295 * selftest-diagnostic.h: New file.
297 2017-11-02 James Bowman <james.bowman@ftdichip.com>
299 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
300 offset range for FT32B.
301 * config/ft32/ft32.h: option "mcompress" enables relaxation.
302 * config/ft32/ft32.md: Add TARGET_NOPM.
303 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
304 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
306 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
308 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
310 2017-11-02 Jeff Law <law@redhat.com>
312 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
313 virtual keyword on FINAL OVERRIDE members.
315 * tree-ssa-propagate.h (ssa_propagation_engine): Group
316 virtuals together. Add virtual destructor.
317 (substitute_and_fold_engine): Similarly.
319 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
321 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
323 2017-11-02 Richard Biener <rguenther@suse.de>
325 PR tree-optimization/82795
326 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
328 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
330 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
332 (gcc_GAS_CHECK_FEATURE): Remove.
333 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
334 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
335 * configure: Regenerate.
337 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
339 * config/arc/arc.c (hwloop_optimize): Account for empty
342 2017-11-02 Richard Biener <rguenther@suse.de>
345 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
346 Truncate ARRAY_REF index and element size.
348 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
350 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
352 2017-11-01 Jeff Law <law@redhat.com>
354 * tree-ssa-ccp.c (ccp_folder): New class derived from
355 substitute_and_fold_engine.
356 (ccp_folder::get_value): New member function.
357 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
358 (ccp_fold_stmt): Remove prototype.
359 (ccp_finalize): Call substitute_and_fold from the ccp_class.
360 * tree-ssa-copy.c (copy_folder): New class derived from
361 substitute_and_fold_engine.
362 (copy_folder::get_value): Renamed from get_value.
363 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
364 * tree-vrp.c (vrp_folder): New class derived from
365 substitute_and_fold_engine.
366 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
367 (vrp_folder::get_value): New member function.
368 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
369 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
370 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
371 provide a class interface to folder/substitute routines.
372 (ssa_prop_fold_stmt_fn): Remove typedef.
373 (ssa_prop_get_value_fn): Likewise.
374 (subsitute_and_fold): Remove prototype.
375 (replace_uses_in): Likewise.
376 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
377 Renamed from replace_uses_in. Call the virtual member function
378 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
379 (substitute_and_fold_dom_walker): Remove initialization of
380 data member entries for calbacks. Add substitute_and_fold_engine
381 member and initialize it.
382 (substitute_and_fold_dom_walker::before_dom_children0: Use the
383 member functions for get_value, replace_phi_args_in c
384 replace_uses_in, and fold_stmt calls.
385 (substitute_and_fold_engine::substitute_and_fold): Renamed from
386 substitute_and_fold. Remove assert. Update ctor call.
388 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
389 (ssa_prop_visit_phi_fn): Likewise.
390 (class ssa_propagation_engine): New class to provide an interface
392 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
394 (ssa_prop_visit_phi): Likewise.
395 (ssa_propagation_engine::simulate_stmt): Moved into class.
396 Call visit_phi/visit_stmt from the class rather than via
397 file scoped static variables.
398 (ssa_propagation_engine::simulate_block): Moved into class.
399 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
400 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
401 set file scoped statics for the visit_stmt/visit_phi callbacks.
402 * tree-complex.c (complex_propagate): New class derived from
403 ssa_propagation_engine.
404 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
405 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
406 (tree_lower_complex): Call ssa_propagate via the complex_propagate
408 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
409 ssa_propagation_engine.
410 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
411 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
412 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
413 * tree-ssa-copy.c (copy_prop): New class derived from
414 ssa_propagation_engine.
415 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
416 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
417 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
418 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
419 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
420 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
421 (execute_vrp): Call ssa_propagate from the vrp_prop class.
423 2017-11-01 Jakub Jelinek <jakub@redhat.com>
425 PR rtl-optimization/82778
426 PR rtl-optimization/82597
427 * compare-elim.c (struct comparison): Add in_a_setter field.
428 (find_comparison_dom_walker::before_dom_children): Remove killed
429 bitmap and df_simulate_find_defs call, instead walk the defs.
430 Compute last_setter and initialize in_a_setter. Merge definitions
431 with first initialization for a few variables.
432 (try_validate_parallel): Use insn_invalid_p instead of
433 recog_memoized. Return insn rather than just the pattern.
434 (try_merge_compare): Fix up comment. Don't uselessly test if
435 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
437 (execute_compare_elim_after_reload): Remove df_chain_add_problem
440 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
441 Alan Hayward <alan.hayward@arm.com>
442 David Sherwood <david.sherwood@arm.com>
444 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
445 aarch64_hard_regno_nregs to get the number of registers
448 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
449 Alan Hayward <alan.hayward@arm.com>
450 David Sherwood <david.sherwood@arm.com>
452 * config/aarch64/constraints.md (Upl): Rename to...
454 * config/aarch64/aarch64.md
455 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
458 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
459 Alan Hayward <alan.hayward@arm.com>
460 David Sherwood <david.sherwood@arm.com>
462 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
463 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
466 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
467 Alan Hayward <alan.hayward@arm.com>
468 David Sherwood <david.sherwood@arm.com>
470 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
471 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
472 (aarch64_evpc_dup): Generate rtl direcly, rather than using
474 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
476 * config/aarch64/iterators.md: Add a comment above the permute
477 unspecs to say that they are generated directly by
478 aarch64_expand_vec_perm_const.
479 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
481 2017-11-01 Nathan Sidwell <nathan@acm.org>
483 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
485 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
487 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
488 and medany code models, and describe what they do.
490 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
492 Revert accidental duplicate:
494 * combine.c (can_change_dest_mode): Reject changes in
495 REGMODE_NATURAL_SIZE.
497 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
499 PR rtl-optimization/64682
500 PR rtl-optimization/69567
501 PR rtl-optimization/69737
502 PR rtl-optimization/82683
503 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
504 register mentioned in the note, drop the note, unless it came from I3,
505 in which case it should go to I3 again.
507 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
509 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
510 and return false if not.
511 (clear_bytes_written_by, live_bytes_read): Update accordingly.
513 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
515 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
516 range is known to be empty.
518 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
519 Alan Hayward <alan.hayward@arm.com>
520 David Sherwood <david.sherwood@arm.com>
522 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
523 and CONST_VECTOR_NUNITS instead of computing the number of units from
524 the byte sizes of the vector and element.
525 (simplify_binary_operation_1): Likewise.
526 (simplify_const_binary_operation): Likewise.
527 (simplify_ternary_operation): Likewise.
529 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
530 Alan Hayward <alan.hayward@arm.com>
531 David Sherwood <david.sherwood@arm.com>
533 * var-tracking.c (INT_MEM_OFFSET): Replace with...
534 (int_mem_offset): ...this new function.
535 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
536 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
537 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
540 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
541 Alan Hayward <alan.hayward@arm.com>
542 David Sherwood <david.sherwood@arm.com>
544 * lower-subreg.c (interesting_mode_p): New function.
545 (compute_costs, find_decomposable_subregs, decompose_register)
546 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
547 (resolve_clobber, dump_choices): Use it.
549 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
550 Alan Hayward <alan.hayward@arm.com>
551 David Sherwood <david.sherwood@arm.com>
553 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
555 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
556 Alan Hayward <alan.hayward@arm.com>
557 David Sherwood <david.sherwood@arm.com>
559 * alias.c (find_base_value, find_base_term): Only process integer
560 truncations. Check the precision rather than the size.
562 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
563 Alan Hayward <alan.hayward@arm.com>
564 David Sherwood <david.sherwood@arm.com>
566 * machmode.h (is_narrower_int_mode): New function
567 * optabs.c (expand_float, expand_fix): Use it.
568 * dwarf2out.c (rotate_loc_descriptor): Likewise.
570 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
571 Alan Hayward <alan.hayward@arm.com>
572 David Sherwood <david.sherwood@arm.com>
574 * rtl.h (narrower_subreg_mode): New function.
575 * ira-color.c (update_costs_from_allocno): Use it.
577 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
578 Alan Hayward <alan.hayward@arm.com>
579 David Sherwood <david.sherwood@arm.com>
581 * optabs-query.h (convert_optab_p): New function, split out from...
582 (convert_optab_handler): ...here.
583 (widening_optab_handler): Delete.
584 (find_widening_optab_handler): Remove permit_non_widening parameter.
585 (find_widening_optab_handler_and_mode): Likewise. Provide an
586 override that operates on mode class wrappers.
587 * optabs-query.c (widening_optab_handler): Delete.
588 (find_widening_optab_handler_and_mode): Remove permit_non_widening
589 parameter. Assert that the two modes are the same class and that
590 the "from" mode is narrower than the "to" mode. Use
591 convert_optab_handler instead of widening_optab_handler.
592 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
593 instead of widening_optab_handler.
594 * expr.c (expand_expr_real_2): Update calls to
595 find_widening_optab_handler.
596 * optabs.c (expand_widen_pattern_expr): Likewise.
597 (expand_binop_directly): Take the insn_code as a parameter.
598 (expand_binop): Only call find_widening_optab_handler for
599 conversion optabs; use optab_handler otherwise. Update calls
600 to find_widening_optab_handler and expand_binop_directly.
601 Use convert_optab_handler instead of widening_optab_handler.
602 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
603 find_widening_optab_handler and use scalar_mode rather than
605 (convert_plusminus_to_widen): Likewise.
607 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
608 Alan Hayward <alan.hayward@arm.com>
609 David Sherwood <david.sherwood@arm.com>
611 * machmode.h (fixed_size_mode): New class.
612 * rtl.h (get_pool_mode): Return fixed_size_mode.
613 * gengtype.c (main): Add fixed_size_mode.
614 * target.def (get_raw_result_mode): Return a fixed_size_mode.
615 (get_raw_arg_mode): Likewise.
616 * doc/tm.texi: Regenerate.
617 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
618 * targhooks.c (default_get_reg_raw_mode): Likewise.
619 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
620 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
621 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
622 (msp430_get_raw_result_mode): Likewise.
623 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
624 * dbxout.c (dbxout_parms): Require fixed-size modes.
625 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
626 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
627 * omp-low.c (lower_oacc_reductions): Likewise.
628 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
629 (simplify_subreg): Update accordingly.
630 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
631 (force_const_mem): Update accordingly. Return NULL_RTX for modes
632 that aren't fixed-size.
633 (get_pool_mode): Return a fixed_size_mode.
634 (output_constant_pool_2): Take a fixed_size_mode.
636 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
637 Alan Hayward <alan.hayward@arm.com>
638 David Sherwood <david.sherwood@arm.com>
640 * doc/rtl.texi (vec_series): Document.
641 (const): Say that the operand can be a vec_series.
642 * rtl.def (VEC_SERIES): New rtx code.
643 * rtl.h (const_vec_series_p_1): Declare.
644 (const_vec_series_p): New function.
645 * emit-rtl.h (gen_const_vec_series): Declare.
646 (gen_vec_series): Likewise.
647 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
648 (gen_vec_series): Likewise.
649 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
650 * simplify-rtx.c (simplify_unary_operation): Handle negations
652 (simplify_binary_operation_series): New function.
653 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
654 (test_vector_ops_series): New function.
655 (test_vector_ops): Call it.
656 * config/powerpcspe/altivec.md (altivec_lvsl): Use
657 gen_const_vec_series.
658 (altivec_lvsr): Likewise.
659 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
661 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
662 Alan Hayward <alan.hayward@arm.com>
663 David Sherwood <david.sherwood@arm.com>
665 * doc/rtl.texi (const): Update description of address constants.
666 Say that vector constants are allowed too.
667 * common.md (E, F): Use CONSTANT_P instead of checking for
669 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
670 checking for CONST_VECTOR.
671 * expmed.c (make_tree): Use build_vector_from_val for a CONST
673 * expr.c (expand_expr_real_2): Check for vector modes instead
674 of checking for CONST_VECTOR.
675 * rtl.h (const_vec_p): New function.
676 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
677 (unwrap_const_vec_duplicate): Handle them here too.
679 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
680 David Malcolm <dmalcolm@redhat.com>
681 Alan Hayward <alan.hayward@arm.com>
682 David Sherwood <david.sherwood@arm.com>
684 * rtl.h (vec_duplicate_p): New function.
685 * selftest-rtl.c (assert_rtx_eq_at): New function.
686 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
687 (assert_rtx_eq_at): Declare.
688 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
689 * selftest-run-tests.c (selftest::run_tests): Call it.
690 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
691 (simplify_unary_operation_1): Recursively handle vector duplicates.
692 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
694 (simplify_subreg): Handle subregs of vector duplicates.
695 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
696 (selftest::simplify_rtx_c_tests): New functions.
698 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
699 Alan Hayward <alan.hayward@arm.com>
700 David Sherwood <david.sherwood@arm.com>
702 * emit-rtl.h (gen_const_vec_duplicate): Declare.
703 (gen_vec_duplicate): Likewise.
704 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
706 (gen_const_vector): ...here.
707 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
708 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
709 whose elements are all equal.
710 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
711 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
712 (simplify_relational_operation): Likewise.
713 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
715 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
716 (aarch64_expand_vector_init): Likewise.
717 * config/arm/arm.c (neon_vdup_constant): Likewise.
718 (neon_expand_vector_init): Likewise.
719 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
720 (arm_block_set_unaligned_vect): Likewise.
721 (arm_block_set_aligned_vect): Likewise.
722 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
723 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
724 (expand_vec_perm_even_odd_pack): Likewise.
725 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
726 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
727 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
728 gen_const_vec_duplicate.
729 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
730 * config/mips/mips.c (mips_gen_const_int_vector): Use
731 gen_const_vec_duplicate.
732 (mips_expand_vector_init): Use CONST0_RTX.
733 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
734 (define_split): Use gen_const_vec_duplicate.
735 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
736 (define_split): Use gen_const_vec_duplicate.
737 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
738 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
739 * config/spu/spu.c (spu_const): Likewise.
741 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
742 Alan Hayward <alan.hayward@arm.com>
743 David Sherwood <david.sherwood@arm.com>
745 * combine.c (can_change_dest_mode): Reject changes in
746 REGMODE_NATURAL_SIZE.
748 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
750 * configure.ac (--enable-libssp): New.
751 (gcc_cv_libc_provides_ssp): Check for explicit setting before
752 trying to determine target-specific default. Adjust indentation.
753 * configure: Regenerated.
754 * doc/install.texi (Configuration): Expand --disable-libssp
757 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
759 config/i386/i386.c (ix86_expand_epilogue): Correct stack
762 2017-10-31 Martin Jambor <mjambor@suse.cz>
765 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
767 2017-10-31 David Malcolm <dmalcolm@redhat.com>
769 * auto-profile.c (autofdo_source_profile::read): Use
770 UNKNOWN_LOCATION rather than 0.
771 * diagnostic-core.h (warning_at_rich_loc): Rename to...
772 (warning_at): ...this overload.
773 (warning_at_rich_loc_n): Rename to...
774 (warning_n): ...this overload.
775 (error_at_rich_loc): Rename to...
776 (error_at): ...this overload.
777 (pedwarn_at_rich_loc): Rename to...
778 (pedwarn): ...this overload.
779 (permerror_at_rich_loc): Rename to...
780 (permerror): ...this overload.
781 (inform_at_rich_loc): Rename to...
782 (inform): ...this overload.
783 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
784 (diagnostic_n_impl_richloc): Rename to...
785 (diagnostic_n_impl): ...this rich_location *-based decl.
786 (inform_at_rich_loc): Rename to...
787 (inform): ...this, and add an assertion.
788 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
789 (warning_at_rich_loc): Rename to...
790 (warning_at): ...this, and add an assertion.
791 (warning_at_rich_loc_n): Rename to...
792 (warning_n): ...this, and add an assertion.
793 (warning_n): Update location_t-based implementation for removal of
794 location_t-based diagnostic_n_impl.
795 (pedwarn_at_rich_loc): Rename to...
796 (pedwarn): ...this, and add an assertion.
797 (permerror_at_rich_loc): Rename to...
798 (permerror): ...this, and add an assertion.
799 (error_n): Update for removal of location_t-based diagnostic_n_impl.
800 (error_at_rich_loc): Rename to...
801 (error_at): ...this, and add an assertion.
802 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
803 (driver::do_spec_on_infiles): Likewise.
804 * substring-locations.c (format_warning_va): Update for renaming
805 of inform_at_rich_loc.
807 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
809 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
810 _Float<N>X built-in functions so that the variant without the
811 "__builtin_" prefix is only enabled for the GNU C and Objective C
812 languages when they are in non-strict ANSI/ISO mode.
813 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
814 * target.def (floatn_builtin_p): Add a target hook to control
815 whether _Float<N> and _Float<N>X built-in functions without the
816 "__builtin_" prefix are enabled, and return true for C and
817 Objective C in the default hook. Include langhooks.h in
819 * targhooks.h (default_floatn_builtin_p): Likewise.
820 * targhooks.c (default_floatn_builtin_p): Likewise.
821 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
822 floatn_builtin_p target hook.
823 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
825 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
826 Eric Botcazou <ebotcazou@adacore.com>
828 PR rtl-optimization/81803
829 * lra-constraints.c (curr_insn_transform): Also reload the whole
830 register for a strict subreg no wider than a word if this is for
831 a WORD_REGISTER_OPERATIONS target.
833 2017-10-31 Jason Merrill <jason@redhat.com>
835 * gdbinit.in: Skip over inlines from timevar.h.
837 2017-10-31 Martin Liska <mliska@suse.cz>
839 * doc/gcov.texi: Document new option.
840 * gcov.c (print_usage): Likewise print it.
841 (process_args): Support the argument.
842 (format_count): New function.
843 (format_gcov): Use the function.
845 2017-10-31 Martin Liska <mliska@suse.cz>
847 * gcov.c (struct name_map): do not use typedef.
848 Define operator== and operator<.
849 (name_search): Remove.
851 (main): Do not allocate names.
852 (process_file): Add vertical space.
853 (generate_results): Use std::find.
854 (release_structures): Do not release memory.
855 (find_source): Use std::find.
857 2017-10-31 Martin Liska <mliska@suse.cz>
859 * gcov.c (struct line_info): Remove it's typedef.
860 (line_info::line_info): Add proper ctor.
861 (line_info::has_block): Do not use a typedef.
862 (struct source_info): Do not use typedef.
864 (get_cycles_count): Likewise.
865 (output_intermediate_file): Iterate via vector iterator.
866 (add_line_counts): Use std::vector methods.
867 (accumulate_line_counts): Likewise.
868 (output_lines): Likewise.
870 2017-10-31 Martin Liska <mliska@suse.cz>
872 * gcov.c (struct source_info): Remove typedef.
873 (source_info::source_info): Add proper ctor.
874 (accumulate_line_counts): Use struct, not it's typedef.
875 (output_gcov_file): Likewise.
876 (output_lines): Likewise.
877 (main): Do not allocate an array.
878 (output_intermediate_file): Use size of vector container.
879 (process_file): Resize the vector.
880 (generate_results): Do not preallocate, use newly added vector
882 (release_structures): Do not release sources.
883 (find_source): Use vector methods.
884 (add_line_counts): Do not use typedef.
886 2017-10-31 Martin Liska <mliska@suse.cz>
888 * doc/gcov.texi: Document that.
889 * gcov.c (add_line_counts): Mark lines with a non-executed
891 (output_line_beginning): Handle such lines.
892 (output_lines): Pass new argument.
893 (output_intermediate_file): Print it in intermediate format.
895 2017-10-31 Martin Liska <mliska@suse.cz>
897 * color-macros.h: New file.
898 * diagnostic-color.c: Factor out color related to macros to
900 * doc/gcov.texi: Document -k option.
901 * gcov.c (INCLUDE_STRING): Include string.h.
902 (print_usage): Add -k option.
903 (process_args): Parse it.
904 (pad_count_string): New function.
905 (output_line_beginning): Likewise.
906 (DEFAULT_LINE_START): New macro.
907 (output_lines): Support color output.
909 2017-10-31 Martin Liska <mliska@suse.cz>
911 PR gcov-profile/82633
912 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
913 their interaction with GCOV infrastructure.
914 * configure.ac: Add -fkeep-{inline,static}-functions to
916 * configure: Regenerate.
918 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
921 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
923 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
926 * config/rs6000/rs6000.md (allocate_stack): Force update interval
927 into a register if it does not fit into an immediate offset field.
929 2017-10-31 Olivier Hainque <hainque@adacore.com>
931 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
933 2017-10-31 Julia Koval <julia.koval@intel.com>
935 * config.gcc: Add gfniintrin.h.
936 * config/i386/gfniintrin.h: New.
937 * config/i386/i386-builtin-types.def
938 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
939 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
940 __builtin_ia32_vgf2p8affineinvqb_v32qi,
941 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
942 __builtin_ia32_vgf2p8affineinvqb_v16qi,
943 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
944 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
945 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
946 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
947 V64QI_FTYPE_V64QI_V64QI_INT): New types.
948 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
949 * config/i386/immintrin.h: Include gfniintrin.h.
950 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
952 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
954 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
956 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
958 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
959 (arm_ashldi3_1bit): Remove pattern.
960 (ashrdi3): Remove shift by 1 expansion.
961 (arm_ashrdi3_1bit): Remove pattern.
962 (lshrdi3): Remove shift by 1 expansion.
963 (arm_lshrdi3_1bit): Remove pattern.
964 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
965 cost of ashldi3 by 1.
966 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
967 (<shift>di3_neon): Likewise.
969 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
971 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
972 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
973 and (*aarch64_simd_mov<VQ:mode>).
974 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
976 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
977 attributes to match pattern alternative.
979 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
981 * config.gcc (powerpc*-*-*): Add emmintrin.h.
982 * config/rs6000/emmintrin.h: New file.
983 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
985 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
987 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
988 * (movdi_vfp_cortexa8): Remove pattern.
990 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
992 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
994 (Specific, alpha*-dec-osf5.1): Remove.
995 (Specific, mips-sgi-irix5): Remove.
996 (Specific, mips-sgi-irix6): Remove.
998 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1001 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
1002 arguments to clear_bit_region_be.
1004 2017-10-30 Jim Wilson <wilson@tuliptree.org>
1006 * gimplify.c: Include memmodel.h.
1008 2017-10-30 Martin Jambor <mjambor@suse.cz>
1010 * omp-grid.c (grid_attempt_target_gridification): Also insert a
1011 condition whether loop should be executed at all.
1013 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
1015 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1016 gimple folding of vec_madd() intrinsics.
1017 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
1018 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
1019 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
1021 2017-10-30 Richard Biener <rguenther@suse.de>
1023 PR tree-optimization/82762
1025 2017-10-23 Richard Biener <rguenther@suse.de>
1027 PR tree-optimization/82129
1029 2017-08-01 Richard Biener <rguenther@suse.de>
1031 PR tree-optimization/81181
1032 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
1033 (compute_antic): ... end of iteration here.
1035 2017-10-30 Joseph Myers <joseph@codesourcery.com>
1037 * doc/invoke.texi (C Dialect Options): Document -std=c17,
1038 -std=iso9899:2017 and -std=gnu17.
1039 * doc/standards.texi (C Language): Document C17 support.
1040 * doc/cpp.texi (Overview): Mention -std=c17.
1041 (Standard Predefined Macros): Document C11 and C17 values of
1042 __STDC_VERSION__. Do not refer to C99 support as incomplete.
1043 * doc/extend.texi (Inline): Do not list individual options for
1044 standards newer than C99.
1045 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
1047 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
1050 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
1052 * asan.c (asan_finish_file): Align asan globals array by shadow
1055 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1058 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
1059 (struct store_immediate_info): Add bitregion_start and bitregion_end
1061 (store_immediate_info::store_immediate_info): Add brs and bre
1062 arguments and initialize bitregion_{start,end} from those.
1063 (struct merged_store_group): Add bitregion_start, bitregion_end,
1064 align_base and mask fields. Drop unnecessary struct keyword from
1065 struct store_immediate_info. Add do_merge method.
1066 (clear_bit_region_be): Use memset instead of loop storing zeros.
1067 (merged_store_group::do_merge): New method.
1068 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
1069 stores as long as the surrounding bitregions have no gaps.
1070 (merged_store_group::merge_overlapping): Use do_merge.
1071 (merged_store_group::apply_stores): Test that bitregion_{start,end}
1072 is byte aligned, rather than requiring that start and width are
1073 byte aligned. Drop unnecessary struct keyword from
1074 struct store_immediate_info. Allocate and populate also mask array.
1075 Make start of the arrays relative to bitregion_start rather than
1076 start and size them according to bitregion_{end,start} difference.
1077 (struct imm_store_chain_info): Drop unnecessary struct keyword from
1078 struct store_immediate_info.
1079 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
1080 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
1081 struct keyword from struct store_immediate_info.
1082 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
1083 between stores as long as the surrounding bitregions have no gaps.
1085 (struct split_store): Add orig non-static data member.
1086 (split_store::split_store): Initialize orig to false.
1087 (find_constituent_stmts): Return store_immediate_info *, non-NULL
1088 if there is exactly a single original stmt. Change stmts argument
1089 to pointer from reference, if NULL, don't push anything to it. Add
1090 first argument, use it to optimize skipping over orig stmts that
1091 are known to be before bitpos already. Simplify.
1092 (split_group): Return unsigned int count how many stores are or
1093 would be needed rather than a bool. Add allow_unaligned argument.
1094 Change split_stores argument from reference to pointer, if NULL,
1095 only do a dry run computing how many stores would be produced.
1096 Rewritten algorithm to use both alignment and misalign if
1097 !allow_unaligned and handle bitfield stores with gaps.
1098 (imm_store_chain_info::output_merged_store): Set start_byte_pos
1099 from bitregion_start instead of start. Compute allow_unaligned
1100 here, if true, do 2 split_group dry runs to compute which one
1101 produces fewer stores and prefer aligned if equal. Punt if
1102 new count is bigger or equal than original before emitting any
1103 statements, rather than during that. Remove no longer needed
1104 new_ssa_names tracking. Replace num_stmts with
1105 split_stores.length (). Use 32-bit stack allocated entries
1106 in split_stores auto_vec. Try to reuse original store lhs/rhs1
1107 if possible. Handle bitfields with gaps.
1108 (pass_store_merging::execute): Ignore bitsize == 0 stores.
1109 Compute bitregion_{start,end} for the stores and construct
1110 store_immediate_info with that. Formatting fixes.
1112 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
1115 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
1116 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
1118 2017-10-29 Jim Wilson <wilson@tuliptree.org>
1120 * gimplify.c: Include tm_p.h.
1122 * common.opt (gcoff): Re-add as ignored option.
1123 (gcoff1, gcoff2, gcoff3): Likewise.
1125 * Makefile.in (OBJS): Delete sdbout.o.
1126 (GTFILES): Delete $(srcdir)/sdbout.c.
1127 * debug.h: Delete sdb_debug_hooks.
1128 * final.c: Delete sdbout.h include.
1129 (final_scan_insn): Delete SDB_DEBUG check.
1130 (rest_of_clean_state): Likewise.
1131 * output.h: Delete sdb_begin_function_line.
1134 * toplev.c: Delete sdbout.h include.
1135 (process_options): Delete SDB_DEBUG check.
1136 * tree-core.h (tree_type_common): Delete pointer field of
1138 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
1139 TYPE_SYMTAB_POINTER.
1140 * tree.h (TYPE_SYMTAB_POINTER): Delete.
1141 (TYPE_SYMTAB_IS_POINTER): Delete.
1142 (TYPE_SYMTAB_IS_DIE): Renumber.
1143 * xcoffout.c: Refer to former sdbout.c file.
1144 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
1146 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
1147 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
1148 (Debugging Options): Delete -gcoff.
1149 (-gstabs): Delete SDB reference.
1151 (-gcoff@var{level}): Delete.
1152 * doc/passes.texi (Debugging information output): Delete SDB and
1153 sdbout.c references.
1154 * doc/tm.texi: Regenerate.
1155 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
1156 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
1158 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
1159 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
1160 (SDB_DEBUGGING_INFO): Delete.
1161 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
1162 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
1163 * target.def (output_source_filename): Delete COFF reference.
1165 * common.opt (gcoff): Delete.
1166 (gxcoff+): Update Negative chain.
1167 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
1169 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
1170 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
1171 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
1173 (expand_function_start): Change sdb reference to past tense.
1174 (expand_function_end): Change sdb reference to past tense.
1175 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
1176 * opts.c (debug_type_names): Delete coff entry.
1177 (common_handle_option): Delete OPT_gcoff case.
1178 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
1180 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
1181 * config/cris/cris.h: Delete SDB reference in comment.
1182 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
1183 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
1184 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
1185 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
1187 (ix86_expand_prologue): Likewise.
1188 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
1189 * config/ia64/ia64.h: Likewise.
1190 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
1191 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
1193 * config/mmix/mmix.h: Likewise.
1194 * config/nds32/nds32.c: Likewise.
1195 * config/stormy/storym16.h: Likewise.
1196 * config/visium/visium.h: Likewise.
1197 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
1199 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
1201 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
1202 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
1203 FRAME_POINTER_REGNUM point at high end of local var area.
1205 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
1207 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
1208 Move comment around. Do not reset best_edge for a copiable
1209 destination if the copy would cause a partition change.
1210 (better_edge_p): Remove redundant check.
1212 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
1214 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
1216 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
1218 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
1219 for math functions that have _Float<N> and _Float<N>X variants.
1220 (mathfn_built_in_2): Add support for math functions that have
1221 _Float<N> and _Float<N>X variants.
1222 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
1223 (expand_builtin_mathfn_ternary): Add support for fma with
1224 _Float<N> and _Float<N>X variants.
1225 (expand_builtin): Likewise.
1226 (fold_builtin_3): Likewise.
1227 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
1228 create math function _Float<N> and _Float<N>X variants as external
1230 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
1231 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
1232 the __builtin_ prefix and if not strict ansi, without the prefix.
1233 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
1234 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
1235 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
1236 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
1237 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
1238 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
1239 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
1240 function signatures for fma _Float<N> and _Float<N>X variants.
1241 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
1242 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
1243 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
1244 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
1245 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
1246 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
1247 * gencfn-macros.c (print_case_cfn): Add support for math functions
1248 that have _Float<N> and _Float<N>X variants.
1249 (print_define_operator_list): Likewise.
1250 (fltfn_suffixes): Likewise.
1252 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
1253 for math functions that have _Float<N> and _Float<N>X variants.
1254 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
1255 and _Float<N>X variants.
1256 (COPYSIGN): Likewise.
1259 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
1260 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
1262 (integer_valued_read_call_p): Likewise.
1263 * fold-const-call.c (fold_const_call_ss): Likewise.
1264 (fold_const_call_sss): Add support for copysign, fmin, and fmax
1265 _Float<N> and _Float<N>X variants.
1266 (fold_const_call_ssss): Add support for fma _Float<N> and
1267 _Float<N>X variants.
1268 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
1269 support for copysign and fma _Float<N> and _Float<N>X variants.
1270 (backprop::process_builtin_call_use): Likewise.
1271 * tree-call-cdce.c (can_test_argument_range); Add support for
1272 sqrt _Float<N> and _Float<N>X variants.
1273 (edom_only_function): Likewise.
1274 (get_no_error_domain): Likewise.
1275 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
1276 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
1277 copysign _Float<N> and _Float<N>X variants.
1278 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
1279 handled by machine independent code.
1280 (FMAF128): Likewise.
1281 * doc/cpp.texi (Common Predefined Macros): Document defining
1282 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
1283 fma _Float<N> and _Float<N>X variants.
1285 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
1288 * config/i386/i386-modes.def (CCFPU): Remove definition.
1289 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
1290 (ix86_cc_modes_compatible): Ditto.
1291 (ix86_expand_carry_flag_compare): Ditto.
1292 (ix86_expand_int_movcc): Ditto.
1293 (ix86_expand_int_addcc): Ditto.
1294 (ix86_reverse_condition): Ditto.
1295 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
1296 Return true/false for unordered/ordered fp comparisons.
1297 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
1298 (ix86_prepare_fp_compare_args): Update for rename.
1299 (ix86_expand_fp_compare): Update for rename. Generate unordered
1300 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
1301 (ix86_expand_sse_compare_and_jump): Ditto.
1302 * config/i386/predicates.md (fcmov_comparison_operator):
1303 Remove CCFPU mode handling.
1304 (ix86_comparison_operator): Ditto.
1305 (ix86_carry_flag_operator): Ditto.
1306 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
1307 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
1308 (*cmpu<mode>_cc_i387): Ditto.
1309 (FPCMP): Remove mode iterator.
1310 (unord): Remove mode attribute.
1311 (unord_subst): New define_subst transformation
1312 (unord): New define_subst attribute.
1314 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
1315 (*cmpi<unord>xf_i387): Ditto.
1316 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
1317 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
1318 using unord_subst transformation.
1319 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
1320 (round_saeonly): Also handle CCFP mode.
1321 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
1322 Remove UNSPEC_SAHF unspec handling.
1324 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
1326 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
1328 2017-10-27 Jeff Law <law@redhat.com>
1330 * gimple-ssa-sprintf.c: Include domwalk.h.
1331 (class sprintf_dom_walker): New class, derived from dom_walker.
1332 (sprintf_dom_walker::before_dom_children): New function.
1333 (struct call_info): Moved into sprintf_dom_walker class
1334 (compute_formath_length, handle_gimple_call): Likewise.
1335 (sprintf_length::execute): Call the dominator walker rather
1336 than walking the statements.
1338 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
1339 gimple statement locations.
1340 (check_array_bounds): Corresponding changes. Get the statement's
1341 location directly from wi->stmt.
1343 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
1346 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
1348 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
1350 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
1351 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
1353 2017-10-27 Jakub Jelinek <jakub@redhat.com>
1356 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
1357 * config/i386/i386.c (maybe_get_pool_constant): Removed.
1358 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
1359 maybe_get_pool_constant.
1360 * config/i386/predicates.md (zero_extended_scalar_load_operand):
1363 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1365 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
1366 2.26 caveat. Update gas and gld versions.
1367 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
1370 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1372 * cgraph.h (set_malloc_flag): Declare.
1373 * cgraph.c (set_malloc_flag_1): New function.
1374 (set_malloc_flag): Likewise.
1375 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
1376 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
1378 (read_ipa_call_summary): Add support for reading is_return_callee.
1379 (write_ipa_call_summary): Stream is_return_callee.
1380 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
1381 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
1382 ipa-prop.h, ipa-fnsummary.h.
1383 (pure_const_names): Change to static.
1384 (malloc_state_e): Define.
1385 (malloc_state_names): Define.
1386 (funct_state_d): Add field malloc_state.
1387 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
1388 (check_retval_uses): New function.
1389 (malloc_candidate_p): Likewise.
1390 (analyze_function): Add support for malloc attribute.
1391 (pure_const_write_summary): Stream malloc_state.
1392 (pure_const_read_summary): Add support for reading malloc_state.
1393 (dump_malloc_lattice): New function.
1394 (propagate_malloc): New function.
1395 (warn_function_malloc): New function.
1396 (ipa_pure_const::execute): Call propagate_malloc and
1397 ipa_free_fn_summary.
1398 (pass_local_pure_const::execute): Add support for malloc attribute.
1399 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
1400 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
1402 2017-10-27 Martin Liska <mliska@suse.cz>
1404 PR gcov-profile/82457
1405 * doc/invoke.texi: Document that one needs a non-strict ISO mode
1406 for fork-like functions to be properly instrumented.
1408 2017-10-27 Richard Biener <rguenther@suse.de>
1411 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
1412 info when we redirected EH.
1414 2017-10-26 Michael Collison <michael.collison@arm.com>
1416 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
1418 (<optab>_trunchf<GPI:mode>2: New pattern.
1419 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
1420 * config/aarch64/iterators.md (wv): New mode attribute.
1421 (vf, VF): New mode attributes.
1422 (vgp, VGP): New mode attributes.
1423 (s): Update attribute with SImode and DImode prefixes.
1425 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
1427 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
1428 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
1429 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
1430 (nios2_option_overide): Initialize it. Don't allow R0-relative
1431 addressing with PIC.
1432 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
1433 (nios2_symbolic_constant_p): Likewise.
1434 (nios2_legitimate_address_p): Likewise.
1435 (nios2_r0rel_section_name_p): New.
1436 (nios2_symbol_ref_in_r0rel_data_p): New.
1437 (nios2_emit_move_sequence): Handle r0rel_constant_p.
1438 (r0rel_constant_p): New.
1439 (nios2_print_operand_address): Handle r0rel_constant_p.
1440 (nios2_cdx_narrow_form_p): Likewise.
1441 * config/nios2/nios2.opt (mr0rel-sec=): New option.
1442 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
1443 (Nios II Options): Document -mr0rel-sec.
1445 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
1447 * config/nios2/nios2.c: Include xregex.h.
1448 (nios2_gprel_sec_regex): New.
1449 (nios2_option_overide): Initialize it. Don't allow GP-relative
1450 addressing with PIC.
1451 (nios2_small_section_name_p): Check for regex match.
1452 * config/nios2/nios2.opt (mgprel-sec=): New option.
1453 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
1454 (Nios II Options): Document -mgprel-sec.
1456 2017-10-26 Jim Wilson <wilson@tuliptree.org>
1458 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
1460 2017-10-26 Tom de Vries <tom@codesourcery.com>
1462 PR tree-optimization/82707
1463 * gimple.c (gimple_copy): Fix unsharing of
1464 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
1466 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
1468 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
1469 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
1470 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
1471 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
1472 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
1473 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
1474 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
1475 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
1476 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
1477 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
1478 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
1479 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
1480 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
1481 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
1482 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
1483 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
1484 _mm512_mask_cmpunord_ps_mask): New intrinsics.
1486 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
1488 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
1490 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
1491 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
1492 warning to rs6000.c. Remove the Undocumented flag, since it has
1494 (-mabi=ibmlongdouble): Likewise.
1495 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
1496 already set, set the default format for long double.
1497 (rs6000_debug_reg_global): Print whether long double is IBM or
1499 (rs6000_option_override_internal): Rework setting long double
1500 format. Only warn if the user is changing the long double default
1501 and they did not use -Wno-psabi.
1502 * doc/invoke.texi (PowerPC options): Update the documentation for
1503 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
1505 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
1506 Alan Hayward <alan.hayward@arm.com>
1507 David Sherwood <david.sherwood@arm.com>
1509 * rtl.h (wider_subreg_mode): New function.
1510 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
1511 rather than an unsigned int *.
1512 * ira-color.c (regno_max_ref_width): Replace with...
1513 (regno_max_ref_mode): ...this new variable.
1514 (coalesced_pseudo_reg_slot_compare): Update accordingly.
1515 Use wider_subreg_mode.
1516 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
1517 rather than an unsigned int *.
1518 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
1519 (process_alt_operands): Likewise.
1520 (invariant_p): Likewise.
1521 * lra-spills.c (assign_mem_slot): Likewise.
1522 (add_pseudo_to_slot): Likewise.
1523 * lra.c (collect_non_operand_hard_regs): Likewise.
1524 (add_regs_to_insn_regno_info): Likewise.
1525 * reload1.c (regno_max_ref_width): Replace with...
1526 (regno_max_ref_mode): ...this new variable.
1527 (reload): Update accordingly. Update call to
1528 ira_sort_regnos_for_alter_reg.
1529 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
1530 (init_eliminable_invariants): Update to use regno_max_ref_mode.
1531 (scan_paradoxical_subregs): Likewise.
1533 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
1535 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
1536 (aarch64_frame): Add emit_frame_chain boolean.
1537 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1538 Move eh_return case to aarch64_layout_frame.
1539 (aarch64_layout_frame): Initialize emit_frame_chain.
1540 (aarch64_expand_prologue): Use emit_frame_chain.
1542 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
1544 * config/aarch64/aarch64.c (aarch64_layout_frame):
1545 Ensure LR is always stored at the bottom of the callee-saves.
1546 Remove rarely used frame layout which saves callee-saves at top of
1547 frame, so the store of LR can be used as a valid probe in all cases.
1549 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
1551 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
1552 Improve unaligned TImode/TFmode base/offset split.
1554 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
1555 Alan Hayward <alan.hayward@arm.com>
1556 David Sherwood <david.sherwood@arm.com>
1558 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
1559 * combine.c (find_single_use_1): Likewise.
1560 (expand_field_assignment): Likewise.
1561 (move_deaths): Likewise.
1562 * lra-constraints.c (simplify_operand_subreg): Likewise.
1563 (curr_insn_transform): Likewise.
1564 * lra.c (collect_non_operand_hard_regs): Likewise.
1565 (add_regs_to_insn_regno_info): Likewise.
1566 * rtlanal.c (reg_referenced_p): Likewise.
1567 (covers_regno_no_parallel_p): Likewise.
1569 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
1571 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
1572 Don't print any bits outside the precision of the value.
1573 * wide-int.cc (test_printing): Add some new tests.
1575 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1577 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
1578 supports -xbrace_comment option.
1579 * configure: Regenerate.
1580 * config.in: Regenerate.
1581 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
1582 (ASM_CPU_SPEC): Use it.
1584 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
1586 * target.def (static_rtx_alignment): New hook.
1587 * targhooks.h (default_static_rtx_alignment): Declare.
1588 * targhooks.c (default_static_rtx_alignment): New function.
1589 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
1590 * doc/tm.texi: Regenerate.
1591 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
1592 instead of targetm.constant_alignment. Remove call to
1594 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
1595 (cris_preferred_mininum_alignment): New function, split out from...
1596 (cris_constant_alignment): ...here.
1597 (cris_static_rtx_alignment): New function.
1598 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
1600 (ix86_constant_alignment): ...here.
1601 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
1602 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
1603 (mmix_static_rtx_alignment): New function.
1604 * config/spu/spu.c (spu_static_rtx_alignment): New function.
1605 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
1607 2017-10-26 Tamar Christina <tamar.christina@arm.com>
1610 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
1611 Add flag_trapping_math and flag_fp_int_builtin_inexact.
1613 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
1615 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
1616 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
1617 mark as a sign-extending load.
1618 (local_pic_load_u): Define.
1620 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
1623 * fold-const.c (operand_equal_for_comparison_p): Also return true
1624 if ARG0 is a simple variant of ARG1 with narrower precision.
1625 (fold_ternary_loc): Always pass unstripped operands to the predicate.
1627 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
1629 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
1631 * i386.h (processor_costs): Add gather_static, gather_per_elt,
1632 scatter_static, scatter_per_elt.
1633 * x86-tune-costs.h: Add new cost entries.
1635 2017-10-25 Richard Biener <rguenther@suse.de>
1637 * tree-ssa-sccvn.h (vn_eliminate): Declare.
1638 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
1639 class pass_fre): Move to ...
1640 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
1641 class pass_fre): ... here and adjust for statistics.
1643 2017-10-25 Jakub Jelinek <jakub@redhat.com>
1646 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
1647 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
1648 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
1649 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
1652 2017-10-25 Richard Biener <rguenther@suse.de>
1654 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
1655 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
1656 eliminate_push_avail, eliminate_insert): Move inside...
1657 (class eliminate_dom_walker): ... this class in preparation
1659 (fini_eliminate): Remove by merging with ...
1660 (eliminate): ... this function. Adjust for class changes.
1661 (pass_pre::execute): Remove fini_eliminate call.
1662 (pass_fre::execute): Likewise.
1664 2017-10-24 Jakub Jelinek <jakub@redhat.com>
1667 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
1668 (VPERMI2, VPERMI2I): New mode iterators.
1669 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
1670 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
1672 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
1673 mode iterator. Remove 3 old define_insn patterns.
1674 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
1675 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
1676 VPERMI2 mode iterator, remove the other two expanders.
1677 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
1678 to use VPERMI2 mode iterator, add another alternative for vpermi2*
1679 instructions, remove the other two patterns.
1680 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
1681 mode iterator, remove the other two patterns.
1682 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
1683 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
1684 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
1685 and adjust argument order accordingly.
1686 (ix86_expand_vec_perm): Adjust caller.
1687 (expand_vec_perm_1): Likewise.
1688 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
1689 (expand_vec_perm_vpermt2_vpshub2): ... this.
1690 (ix86_expand_vec_perm_const_1): Adjust caller.
1691 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
1694 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
1695 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
1696 (vec_shl_<mode>): Remove unused expander.
1697 (avx512bw_<shift_insn><mode>3): New define_insn.
1698 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
1699 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
1701 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
1704 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
1707 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
1709 PR rtl-optimization/82396
1710 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
1711 (autopref_multipass_init): Simplify initialization.
1712 (autopref_rank_data): Simplify sort order.
1713 * gcc/sched-int.h (autopref_multipass_data_): Remove
1714 multi_mem_insn_p, min_offset and max_offset.
1716 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
1719 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1720 Check special value of flag_omit_frame_pointer.
1721 (aarch64_can_eliminate): Likewise.
1722 (aarch64_override_options_after_change_1): Simplify handling of
1723 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
1725 2017-10-24 Richard Biener <rguenther@suse.de>
1727 PR tree-optimization/82697
1728 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
1729 zero for conditional load and unconditional store.
1731 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
1733 * doc/install.texi: Document bootstrap-cet.
1735 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
1738 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
1739 ENDBR instruction at function entrance if function is only
1742 2017-10-24 Jakub Jelinek <jakub@redhat.com>
1745 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
1746 patterns to better describe from which operation the CF is computed.
1747 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
1748 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
1749 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
1750 is 0, use _0 suffixed expanders instead of emitting a comparison
1753 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
1755 * config/i386/i386.md(*movsf_internal, *movdf_internal):
1756 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
1758 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
1761 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
1762 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
1763 * loop-iv.c (iv_get_reaching_def): Likewise.
1764 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
1765 variable is promoted and the partition contains undefined values.
1767 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
1769 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
1771 (nios2_address_cost): Define.
1772 (nios2_legitimize_address): Recognize (exp + constant) directly.
1773 (TARGET_ADDRESS_COST): Define.
1775 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
1777 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
1778 (nios2_symbolic_memory_operand_p): Declare.
1779 (nios2_split_large_constant): Declare.
1780 (nios2_split_symbolic_memory_operand): Declare.
1781 * config/nios2/nios2.c: Adjust includes.
1782 (nios2_symbolic_constant_allowed): New.
1783 (nios2_symbolic_constant_p): New.
1784 (nios2_plus_symbolic_constant_p): New.
1785 (nios2_valid_addr_expr_p): Recognize addresses involving
1787 (nios2_legitimate_address_p): Likewise, also LO_SUM.
1788 (nios2_symbolic_memory_operand_p): New.
1789 (nios2_large_constant_p): New.
1790 (nios2_split_large_constant): New.
1791 (nios2_split_plus_large_constant): New.
1792 (nios2_split_symbolic_memory_operand): New.
1793 (nios2_legitimize_address): Code refactoring. Handle addresses
1794 involving symbolic constants.
1795 (nios2_emit_move_sequence): Likewise.
1796 (nios2_print_operand): Improve error output.
1797 (nios2_print_operand_address): Handle LO_SUM.
1798 (nios2_cdx_narrow_form_p): Likewise.
1799 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
1800 operands involving symbolic constants.
1801 (movhi_internal, movsi_internal): Likewise.
1802 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
1803 (extendhisi2, extendqi<mode>2): Likewise.
1805 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
1807 * tree-pass.h (PROP_rtl_split_insns): Define.
1808 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
1810 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
1812 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
1814 2017-10-23 Jakub Jelinek <jakub@redhat.com>
1817 * target.def (const_not_ok_for_debug_p): Default to
1818 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
1819 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
1820 * targhooks.c (default_const_not_ok_for_debug_p): New function.
1821 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
1822 which targetm.const_not_ok_for_debug_p returned true.
1823 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
1825 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
1827 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
1828 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
1829 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
1831 (ix86_const_not_ok_for_debug_p): New function.
1832 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
1833 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
1835 2017-10-23 David Malcolm <dmalcolm@redhat.com>
1838 * system.h: Conditionally include "unique-ptr.h" if
1839 INCLUDE_UNIQUE_PTR is defined.
1840 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
1841 of defining INCLUDE_UNIQUE_PTR before including "system.h".
1843 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
1845 * config/rl78/rl78.md: New define_expand "subdi3".
1847 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
1850 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
1851 DF_REF_INSN if DF_REF_INSN_INFO is false.
1853 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
1855 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
1856 xmm_move instead of sse_move.
1857 (sse_store_index): New function.
1858 (ix86_register_move_cost): Be more sensible about mismatch stall;
1859 model AVX moves correctly; make difference between sse->integer and
1861 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
1862 moves; make difference between SSE and AVX.
1863 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
1864 and zmm_move. Increase size of sse load and store tables;
1865 add unaligned load and store tables; add ssemmx_to_integer.
1866 * x86-tune-costs.h: Update all entries according to real
1867 move latencies from Agner Fog's manual and chip documentation.
1869 2017-10-23 Jakub Jelinek <jakub@redhat.com>
1872 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
1873 * config/i386/constraints.md (Wf): New constraint.
1874 * config/i386/i386.md (UNSPEC_SBB): New unspec.
1875 (cmp<dwi>_doubleword): Removed.
1876 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
1877 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
1878 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
1879 expand with cmp<dwi>_doubleword. For LTU and GEU use
1880 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
1882 * common.opt (gcolumn-info): Enable by default.
1883 * doc/invoke.texi (gcolumn-info): Document new default.
1885 2017-10-23 Richard Biener <rguenther@suse.de>
1887 PR tree-optimization/82672
1888 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
1889 Fold the stmt if we propagated into it.
1891 2017-10-23 Richard Biener <rguenther@suse.de>
1893 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
1894 (bitmap_remove_expr_from_set): ... this. All callers call this
1895 for non-constant values.
1896 (bitmap_set_subtract): Rename to...
1897 (bitmap_set_subtract_expressions): ... this. Adjust and
1899 (bitmap_set_contains_value): Remove superfluous check.
1900 (bitmap_set_replace_value): Inline into single caller ...
1901 (bitmap_value_replace_in_set): ... here and simplify.
1902 (dependent_clean): Merge into ...
1903 (clean): ... this using an overload. Adjust.
1904 (prune_clobbered_mems): Adjust.
1905 (compute_antic_aux): Likewise.
1906 (compute_partial_antic_aux): Likewise.
1908 2017-10-23 Richard Biener <rguenther@suse.de>
1910 PR tree-optimization/82129
1912 2017-08-01 Richard Biener <rguenther@suse.de>
1914 PR tree-optimization/81181
1915 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
1916 (compute_antic): ... end of iteration here.
1918 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
1920 * target.def (starting_frame_offset): New hook.
1921 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
1922 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
1923 * doc/tm.texi.in: Regenerate.
1924 * hooks.h (hook_hwi_void_0): Declare.
1925 * hooks.c (hook_hwi_void_0): New function.
1926 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
1927 STARTING_FRAME_OFFSET.
1928 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
1929 * reload1.c (reload): Likewise.
1930 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
1931 instead of STARTING_FRAME_OFFSET.
1932 * function.c (try_fit_stack_local): Likewise.
1933 (assign_stack_local_1): Likewise
1934 (instantiate_virtual_regs): Likewise.
1935 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
1936 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
1937 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
1938 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
1939 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
1940 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
1941 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
1942 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
1943 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
1944 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
1945 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
1946 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
1947 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
1948 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
1949 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
1950 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
1951 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
1952 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
1953 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
1954 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
1955 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
1956 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
1957 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
1958 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
1959 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
1960 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
1961 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
1962 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
1963 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
1964 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
1965 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
1966 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
1967 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
1968 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
1969 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
1970 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
1971 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
1972 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
1973 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
1974 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
1975 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
1976 * config/avr/avr.c (avr_starting_frame_offset): Make static and
1977 return a HOST_WIDE_INT.
1978 (avr_builtin_setjmp_frame_value): Use it instead of
1979 STARTING_FRAME_OFFSET.
1980 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1981 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
1982 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
1984 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1985 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
1986 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
1987 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1988 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
1989 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
1990 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1991 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
1992 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
1993 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1994 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
1995 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
1997 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1998 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
1999 * config/mips/mips.c (mips_compute_frame_info): Refer to
2000 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2001 (mips_starting_frame_offset): New function.
2002 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2003 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
2004 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
2005 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
2006 and return a HOST_WIDE_INT.
2007 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2008 (mmix_initial_elimination_offset): Refer to
2009 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2010 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
2011 * config/pa/pa.c (pa_starting_frame_offset): New function.
2012 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
2013 (pa_expand_prologue): Likewise.
2014 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2015 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
2016 !FRAME_GROWS_DOWNWARD handling to...
2017 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2018 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
2019 !FRAME_GROWS_DOWNWARD handling to...
2020 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2021 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
2022 !FRAME_GROWS_DOWNWARD handling to...
2023 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2024 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
2026 (rs6000_starting_frame_offset): New function.
2027 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
2028 !FRAME_GROWS_DOWNWARD handling to...
2029 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2030 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
2031 !FRAME_GROWS_DOWNWARD handling to...
2032 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2033 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
2034 !FRAME_GROWS_DOWNWARD handling to...
2035 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2036 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
2037 (rs6000_starting_frame_offset): New function.
2038 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
2039 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
2040 * config/vax/vax.c (vax_starting_frame_offset): New function.
2041 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
2042 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2043 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
2044 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
2045 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2046 * system.h (STARTING_FRAME_OFFSET): Poison.
2048 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2050 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
2051 SCALAR_TYPE_MODE instead of TYPE_MODE.
2053 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2054 Alan Hayward <alan.hayward@arm.com>
2055 David Sherwood <david.sherwood@arm.com>
2057 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
2059 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2060 Alan Hayward <alan.hayward@arm.com>
2061 David Sherwood <david.sherwood@arm.com>
2063 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
2065 2017-10-23 Richard Biener <rguenther@suse.de>
2067 PR tree-optimization/82129
2068 * tree-ssa-pre.c (bitmap_set_and): Remove.
2069 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
2070 canonicalizing expressions in the set to those with lowest
2071 ID rather than taking that from the first edge.
2073 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2075 * combine.c (rtx_equal_for_field_assignment_p): Use
2076 byte_lowpart_offset.
2078 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2079 Alan Hayward <alan.hayward@arm.com>
2080 David Sherwood <david.sherwood@arm.com>
2082 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
2083 to a SUBREG_PROMOTED_VAR.
2085 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2086 Alan Hayward <alan.hayward@arm.com>
2087 David Sherwood <david.sherwood@arm.com>
2089 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
2090 (expand_debug_source_expr): Likewise.
2091 * combine.c (combine_simplify_rtx): Likewise.
2092 * cse.c (fold_rtx): Likewise.
2093 * optabs.c (expand_float): Likewise.
2094 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2095 (simplify_binary_operation_1): Likewise.
2097 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2098 Alan Hayward <alan.hayward@arm.com>
2099 David Sherwood <david.sherwood@arm.com>
2101 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
2102 (record_promoted_value): Likewise.
2103 * expr.c (expand_expr_real_2): Likewise.
2104 * ree.c (update_reg_equal_equiv_notes): Likewise.
2105 (combine_set_extension): Likewise.
2106 * rtlanal.c (low_bitmask_len): Likewise.
2107 * simplify-rtx.c (neg_const_int): Likewise.
2108 (simplify_binary_operation_1): Likewise.
2110 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2111 Alan Hayward <alan.hayward@arm.com>
2112 David Sherwood <david.sherwood@arm.com>
2114 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
2115 * regcprop.c (maybe_mode_change): Likewise.
2116 * reload1.c (alter_reg): Likewise.
2118 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2120 * inchash.h (inchash::hash::add_wide_int): New function.
2121 * lto-streamer-out.c (hash_tree): Use it.
2123 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2125 * inchash.h (inchash::hash::add_wide_int): Rename to...
2126 (inchash::hash::add_hwi): ...this.
2127 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
2128 (polymorphic_call_target_hasher::hash): Likewise.
2129 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
2130 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
2131 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
2132 * lto-streamer-out.c (hash_tree): Likewise.
2133 * optc-save-gen.awk: Likewise.
2134 * tree.c (add_expr): Likewise.
2136 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
2139 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
2140 for ordered inequality comparisons even with TARGET_IEEE_FP.
2142 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
2145 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
2146 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
2147 Expand with cmp<dwi>_doubleword.
2149 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2151 * extend.texi: Add x86 specific to 'nocf_check' attribute.
2152 List CET intrinsics.
2153 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
2154 specific to -fcf-protection option.
2156 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2158 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
2159 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
2160 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
2161 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
2162 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
2163 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
2164 (extra_objs): Add cet.o for Linux/x86 targets.
2165 (tmake_file): Add i386/t-cet for Linux/x86 targets.
2166 * config/i386/cet.c: New file.
2167 * config/i386/cetintrin.h: Likewise.
2168 * config/i386/t-cet: Likewise.
2169 * config/i386/cpuid.h (bit_SHSTK): New.
2170 (bit_IBT): Likewise.
2171 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
2172 pass IBT and SHSTK bits.
2173 * config/i386/i386-builtin-types.def
2174 (VOID_FTYPE_UNSIGNED_PVOID): New.
2175 (VOID_FTYPE_UINT64_PVOID): Likewise.
2176 * config/i386/i386-builtin.def: Add CET intrinsics.
2177 * config/i386/i386-c.c (ix86_target_macros_internal): Add
2178 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
2179 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
2180 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
2182 * config/i386/i386.c (rest_of_insert_endbranch): New.
2183 (pass_data_insert_endbranch): Likewise.
2184 (pass_insert_endbranch): Likewise.
2185 (make_pass_insert_endbranch): Likewise.
2186 (ix86_notrack_prefixed_insn_p): Likewise.
2187 (ix86_target_string): Add -mibt, -mshstk flags.
2188 (ix86_option_override_internal): Add flag_cf_protection
2190 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
2191 (ix86_print_operand): Add 'notrack' prefix output.
2192 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
2193 (ix86_expand_builtin): Expand CET intrinsics.
2194 (x86_output_mi_thunk): Add 'endbranch' instruction.
2195 * config/i386/i386.h (TARGET_IBT): New.
2196 (TARGET_IBT_P): Likewise.
2197 (TARGET_SHSTK): Likewise.
2198 (TARGET_SHSTK_P): Likewise.
2199 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
2200 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
2201 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
2202 (builtin_setjmp_setup): New pattern.
2203 (builtin_longjmp): Likewise.
2204 (rdssp<mode>): Likewise.
2205 (incssp<mode>): Likewise.
2206 (saveprevssp): Likewise.
2207 (rstorssp): Likewise.
2208 (wrss<mode>): Likewise.
2209 (wruss<mode>): Likewise.
2210 (setssbsy): Likewise.
2211 (clrssbsy): Likewise.
2212 (nop_endbr): Likewise.
2213 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
2215 * config/i386/immintrin.h: Include <cetintrin.h>.
2216 * config/i386/linux-common.h
2217 (file_end_indicate_exec_stack_and_cet): New prototype.
2218 (TARGET_ASM_FILE_END): New.
2220 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
2222 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
2223 latencies instead of having separate table; make difference between
2224 integer and float costs.
2225 * i386.h (processor_costs): Remove scalar_stmt_cost,
2226 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
2227 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
2229 * x86-tune-costs.h: Remove entries which has been removed in
2230 procesor_costs from all tables; make cond_taken_branch_cost
2231 and cond_not_taken_branch_cost COST_N_INSNS based.
2233 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
2235 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
2237 2017-10-20 Jakub Jelinek <jakub@redhat.com>
2239 * config/i386/i386.md (isa): Remove fma_avx512f.
2240 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
2241 <avx512>_fmadd_<mode>_mask3<round_name>,
2242 <avx512>_fmsub_<mode>_mask<round_name>,
2243 <avx512>_fmsub_<mode>_mask3<round_name>,
2244 <avx512>_fnmadd_<mode>_mask<round_name>,
2245 <avx512>_fnmadd_<mode>_mask3<round_name>,
2246 <avx512>_fnmsub_<mode>_mask<round_name>,
2247 <avx512>_fnmsub_<mode>_mask3<round_name>,
2248 <avx512>_fmaddsub_<mode>_mask<round_name>,
2249 <avx512>_fmaddsub_<mode>_mask3<round_name>,
2250 <avx512>_fmsubadd_<mode>_mask<round_name>,
2251 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
2252 (*vec_widen_umult_even_v16si<mask_name>,
2253 *vec_widen_smult_even_v16si<mask_name>): Likewise.
2254 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
2256 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2258 * extend.texi: Add 'nocf_check' documentation.
2259 * gimple.texi: Add second parameter to
2260 gimple_build_call_from_tree.
2261 * invoke.texi: Add -fcf-protection documentation.
2262 * rtl.texi: Add REG_CALL_NOTRACK documenation.
2264 2017-10-20 Richard Biener <rguenther@suse.de>
2266 PR tree-optimization/82473
2267 * tree-vect-loop.c (vectorizable_reduction): Properly get at
2268 the largest input type.
2270 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2272 * c-attribs.c (handle_nocf_check_attribute): New function.
2273 (c_common_attribute_table): Add 'nocf_check' handling.
2274 * gimple-parser.c: Add second argument NULL to
2275 gimple_build_call_from_tree.
2276 * attrib.c (comp_type_attributes): Check nocf_check attribute.
2277 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
2279 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
2280 * common.opt: Add fcf-protection flag.
2281 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
2282 * flag-types.h: Add enum cf_protection_level.
2283 * gimple.c (gimple_build_call_from_tree): Add second parameter.
2284 Add 'nocf_check' attribute propagation to gimple call.
2285 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
2286 (gimple_build_call_from_tree): Update prototype.
2287 (gimple_call_nocf_check_p): New function.
2288 (gimple_call_set_nocf_check): Likewise.
2289 * gimplify.c: Add second argument to gimple_build_call_from_tree.
2290 * ipa-icf.c: Add nocf_check attribute in statement hash.
2291 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
2292 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
2293 * toplev.c (process_options): Add flag_cf_protection handling.
2295 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
2297 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
2299 2017-10-20 Richard Biener <rguenther@suse.de>
2301 PR tree-optimization/82603
2302 * tree-if-conv.c (predicate_mem_writes): Make sure to only
2303 remove false predicated stores.
2305 2017-10-20 Richard Biener <rguenther@suse.de>
2307 * graphite-isl-ast-to-gimple.c
2308 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
2309 Remove return value and simplify, dump copied stmt after lhs
2311 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
2312 Reduce dump verbosity.
2313 (gsi_insert_earliest): Likewise.
2314 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
2315 * graphite.c (print_global_statistics): Adjust dumping.
2316 (print_graphite_scop_statistics): Likewise.
2317 (print_graphite_statistics): Do not dump loops here.
2318 (graphite_transform_loops): But here.
2320 2017-10-20 Nicolas Roche <roche@adacore.com>
2322 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
2323 * configure: Regenerate.
2325 2017-10-20 Jakub Jelinek <jakub@redhat.com>
2328 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
2329 functions when optimizing replace GIMPLE_RETURN stmts with
2330 calls to __builtin_unreachable ().
2333 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
2334 for -fsanitize=thread link of executables.
2335 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
2336 link of executables.
2339 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
2341 (<shift_insn><mode>3<mask_name>): Change the last of the 3
2342 define_insns for logical vector shifts to use VI248_AVX512BW
2343 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
2344 condition, useless isa and prefix attributes. Change the first
2345 2 of these define_insns to ...
2346 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
2347 define_insn for avx512vl.
2348 (<shift_insn><mode>3): ... and this, new define_insn without
2349 masking for non-avx512vl.
2352 * config/i386/sse.md (*andnot<mode>3,
2353 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
2354 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
2355 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
2356 not applied use empty suffix even for TARGET_AVX512VL.
2357 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
2358 is applied, supply evex,evex or evex,evex,evex instead of just
2361 2017-10-20 Julia Koval <julia.koval@intel.com>
2363 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
2364 (OPTION_MASK_ISA_GFNI_UNSET): New.
2365 (ix86_handle_option): Handle OPT_mgfni.
2366 * config/i386/cpuid.h (bit_GFNI): New.
2367 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
2368 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
2369 * config/i386/i386.c (ix86_target_string): Add -mgfni.
2370 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
2371 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
2372 * config/i386/i386.opt: Add mgfni.
2374 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
2376 * config/msp430/msp430.c (msp430_option_override): Disable
2377 -fdelete-null-pointer-checks.
2378 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
2380 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
2382 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
2383 of x87 and SSE instructions.
2385 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
2387 * asan.c (create_cond_insert_point): Do not update edge count.
2388 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
2389 (afdo_propagate_circuit): Likewise.
2390 (afdo_calculate_branch_prob): Likewise.
2391 (afdo_annotate_cfg): Likewise.
2392 * basic-block.h (struct edge_def): Remove count.
2393 (edge_def::count): New accessor.
2394 * bb-reorder.c (rotate_loop): Update.
2395 (find_traces_1_round): Update.
2396 (connect_traces): Update.
2397 (sanitize_hot_paths): Update.
2398 * cfg.c (unchecked_make_edge): Update.
2399 (make_single_succ_edge): Update.
2400 (check_bb_profile): Update.
2401 (dump_edge_info): Update.
2402 (update_bb_profile_for_threading): Update.
2403 (scale_bbs_frequencies_int): Update.
2404 (scale_bbs_frequencies_gcov_type): Update.
2405 (scale_bbs_frequencies_profile_count): Update.
2406 (scale_bbs_frequencies): Update.
2407 * cfganal.c (connect_infinite_loops_to_exit): Update.
2408 * cfgbuild.c (compute_outgoing_frequencies): Update.
2409 (find_many_sub_basic_blocks): Update.
2410 * cfgcleanup.c (try_forward_edges): Update.
2411 (try_crossjump_to_edge): Update
2412 * cfgexpand.c (expand_gimple_cond): Update
2413 (expand_gimple_tailcall): Update
2414 (construct_exit_block): Update
2415 * cfghooks.c (verify_flow_info): Update
2416 (redirect_edge_succ_nodup): Update
2417 (split_edge): Update
2418 (make_forwarder_block): Update
2419 (duplicate_block): Update
2420 (account_profile_record): Update
2421 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
2422 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
2423 * cfgloopmanip.c (scale_loop_profile): Update.
2425 (lv_adjust_loop_entry_edge): Update.
2426 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
2427 (force_nonfallthru_and_redirect): Update.
2428 (purge_dead_edges): Update.
2429 (rtl_flow_call_edges_add): Update.
2430 * cgraphunit.c (init_lowered_empty_function): Update.
2431 (cgraph_node::expand_thunk): Update.
2432 * gimple-pretty-print.c (dump_probability): Update.
2433 (dump_edge_probability): Update.
2434 * gimple-ssa-isolate-paths.c (isolate_path): Update.
2435 * haifa-sched.c (sched_create_recovery_edges): Update.
2436 * hsa-gen.c (convert_switch_statements): Update.
2437 * ifcvt.c (dead_or_predicable): Update.
2438 * ipa-inline-transform.c (inline_transform): Update.
2439 * ipa-split.c (split_function): Update.
2440 * ipa-utils.c (ipa_merge_profiles): Update.
2441 * loop-doloop.c (add_test): Update.
2442 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
2443 * lto-streamer-in.c (input_cfg): Update.
2444 (input_function): Update.
2445 * lto-streamer-out.c (output_cfg): Update.
2446 * modulo-sched.c (sms_schedule): Update.
2447 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
2448 * predict.c (maybe_hot_edge_p): Update.
2449 (unlikely_executed_edge_p): Update.
2450 (probably_never_executed_edge_p): Update.
2451 (dump_prediction): Update.
2452 (drop_profile): Update.
2453 (propagate_unlikely_bbs_forward): Update.
2454 (determine_unlikely_bbs): Update.
2455 (force_edge_cold): Update.
2456 * profile.c (compute_branch_probabilities): Update.
2457 * reg-stack.c (better_edge): Update.
2458 * shrink-wrap.c (handle_simple_exit): Update.
2459 * tracer.c (better_p): Update.
2460 * trans-mem.c (expand_transaction): Update.
2461 (split_bb_make_tm_edge): Update.
2462 * tree-call-cdce.c: Update.
2463 * tree-cfg.c (gimple_find_sub_bbs): Update.
2464 (gimple_split_edge): Update.
2465 (gimple_duplicate_sese_region): Update.
2466 (gimple_duplicate_sese_tail): Update.
2467 (gimple_flow_call_edges_add): Update.
2468 (insert_cond_bb): Update.
2469 (execute_fixup_cfg): Update.
2470 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
2471 * tree-complex.c (expand_complex_div_wide): Update.
2472 * tree-eh.c (lower_resx): Update.
2473 (unsplit_eh): Update.
2474 (cleanup_empty_eh_move_lp): Update.
2475 * tree-inline.c (copy_edges_for_bb): Update.
2476 (freqs_to_counts): Update.
2477 (copy_cfg_body): Update.
2478 * tree-ssa-dce.c (remove_dead_stmt): Update.
2479 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
2480 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
2481 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
2482 (unloop_loops): Update.
2483 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
2484 * tree-ssa-loop-split.c (connect_loops): Update.
2485 (split_loop): Update.
2486 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
2487 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
2488 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
2489 * tree-ssa-reassoc.c (branch_fixup): Update.
2490 * tree-ssa-tail-merge.c (replace_block_by): Update.
2491 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
2492 (compute_path_counts): Update.
2493 (update_profile): Update.
2494 (recompute_probabilities): Update.
2495 (update_joiner_offpath_counts): Update.
2496 (estimated_freqs_path): Update.
2497 (freqs_to_counts_path): Update.
2498 (clear_counts_path): Update.
2499 (ssa_fix_duplicate_block_edges): Update.
2500 (duplicate_thread_path): Update.
2501 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
2502 (case_bit_test_cmp): Update.
2503 (collect_switch_conv_info): Update.
2504 (gen_inbound_check): Update.
2505 (do_jump_if_equal): Update.
2506 (emit_cmp_and_jump_insns): Update.
2507 * tree-tailcall.c (decrease_profile): Update.
2508 (eliminate_tail_call): Update.
2509 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
2510 (vect_do_peeling): Update.
2511 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
2512 * ubsan.c (ubsan_expand_null_ifn): Update.
2513 (ubsan_expand_ptr_ifn): Update.
2514 * value-prof.c (gimple_divmod_fixed_value): Update.
2515 (gimple_mod_pow2): Update.
2516 (gimple_mod_subtract): Update.
2517 (gimple_ic): Update.
2518 (gimple_stringop_fixed_value): Update.
2520 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
2523 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
2525 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
2527 PR rtl-optimization/82395
2528 * ira-color.c (allocno_priority_compare_func): Fix comparison step
2529 based on non_spilled_static_chain_regno_p.
2531 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
2533 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
2534 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
2535 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
2537 2017-10-19 Martin Sebor <msebor@redhat.com>
2539 PR tree-optimization/82596
2540 * tree.c (array_at_struct_end_p): Handle STRING_CST.
2542 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
2544 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
2545 (get_mem_refs_of_builtin_call): Likewise.
2546 * builtins.c (expand_builtin_apply): Adjust call to
2547 allocate_dynamic_stack_space.
2548 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
2549 the third argument to allocate_dynamic_stack_space, otherwise -1.
2550 (expand_builtin): Deal with all alloca variants.
2551 (is_inexpensive_builtin): Likewise.
2552 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
2553 * calls.c (special_function_p): Deal with all alloca variants.
2554 (initialize_argument_information): Adjust call to
2555 allocate_dynamic_stack_space.
2556 (expand_call): Likewise.
2557 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
2558 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
2559 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
2560 use it for the stack usage computation.
2561 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
2562 * function.c (gimplify_parameters): Call build_alloca_call_expr.
2563 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
2564 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
2565 (in_loop_p): Remove first argument and useless check.
2566 (pass_walloca::execute): Remove useless test and adjust call to above.
2567 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
2568 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
2569 (gimplify_call_expr): Deal with all alloca variants.
2570 * hsa-gen.c (gen_hsa_alloca): Likewise.
2571 (gen_hsa_insns_for_call): Likewise.
2572 * ipa-pure-const.c (special_builtin_state): Likewise.
2573 * tree-chkp.c (chkp_build_returned_bound): Likewise.
2574 * tree-object-size.c (alloc_object_size): Likewise.
2575 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
2576 (call_may_clobber_ref_p_1): Likewise.
2577 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
2578 (ccp_fold_stmt): Likewise.
2579 (optimize_stack_restore): Likewise.
2580 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
2581 (mark_all_reaching_defs_necessary_1): Likewise.
2582 (propagate_necessity): Likewise.
2583 (eliminate_unnecessary_stmts): Likewise.
2584 * tree.c (build_common_builtin_nodes): Build
2585 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
2586 (build_alloca_call_expr): New function.
2587 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
2588 (CASE_BUILT_IN_ALLOCA): Likewise.
2589 (build_alloca_call_expr): Declare.
2590 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
2592 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
2595 * dwarf2out.c (new_die_raw): New static inline function.
2596 (new_die): Use it to create the DIE.
2597 (add_AT_external_die_ref): Likewise.
2598 (clone_die): Likewise.
2599 (clone_as_declaration): Likewise.
2600 (dwarf2out_vms_debug_main_pointer): Likewise.
2601 (base_type_die): Likewise. Remove early return for corner cases.
2602 Do not call add_pubtype on the DIE here.
2603 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
2604 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
2605 typedefs for base types with DW_AT_endianity. Make sure a DIE with
2606 native order exists for base types, attach the DIE manually and call
2607 add_pubtype on it. Do not equate a reverse order DIE to the type.
2609 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
2611 * config/arm/arm.c (align_ok_ldrd_strd): New function.
2612 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
2614 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
2616 2017-10-19 Jakub Jelinek <jakub@redhat.com>
2618 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
2619 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
2620 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
2621 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
2622 * opts.c (sanitizer_opts): Add builtin.
2623 * ubsan.c (instrument_builtin): New function.
2624 (pass_ubsan::execute): Call it.
2625 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
2626 * doc/invoke.texi: Document -fsanitize=builtin.
2628 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
2629 builtins, store max (log2 (align), 0) into uchar field instead of
2630 align into uptr field.
2631 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
2632 store uchar 0 field instead of uptr 0 field.
2633 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
2634 instead of passing one address of struct with 2 locations pass
2635 two addresses of structs with 1 location each.
2636 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
2637 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
2638 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
2639 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
2640 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
2641 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
2642 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
2643 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
2645 2017-10-19 Martin Liska <mliska@suse.cz>
2648 * file-find.c (remove_prefix): Remove.
2649 * file-find.h (remove_prefix): Likewise.
2650 * gcc-ar.c: Remove smartness of lookup.
2652 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
2654 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
2655 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
2656 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
2658 2017-10-19 Jakub Jelinek <jakub@redhat.com>
2661 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
2662 (setcc + and to xor + setcc): New peephole2.
2664 2017-10-19 Tom de Vries <tom@codesourcery.com>
2666 * doc/sourcebuild.texi (Test Directives, Variants of
2667 dg-require-support): Add dg-require-stack-size.
2669 2017-10-19 Martin Liska <mliska@suse.cz>
2672 * gimplify.c (gimplify_decl_expr): Do not instrument variables
2673 that have a large alignment.
2674 (gimplify_target_expr): Likewise.
2676 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
2678 PR rtl-optimization/82602
2679 * ira.c (rtx_moveable_p): Return false for volatile asm.
2681 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
2684 * config/i386/i386-modes.def (CCGZ): New CC mode.
2685 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
2686 * config/i386/predicates.md (ix86_comparison_operator):
2688 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
2689 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
2690 with double-word subtraction.
2691 (put_condition_code): Handle CCGZmode.
2693 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
2695 * wide-int.cc (debug (const wide_int &)): New.
2696 (debug (const wide_int *)): New.
2697 (debug (const widest_int &)): New.
2698 (debug (const widest_int *)): New.
2700 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
2703 * lra-constraints.c (curr_insn_transform): Use non-input operand
2704 instead of output one for matched reload.
2706 2017-10-18 Bin Cheng <bin.cheng@arm.com>
2708 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
2709 (tree-ssa-loop-ivopts.h): New header file.
2710 (struct builtin_info): New fields.
2711 (classify_builtin_1): Compute and record base and offset parts for
2712 memset builtin partition by calling strip_offset.
2713 (offset_cmp, fuse_memset_builtins): New functions.
2714 (finalize_partitions): Fuse adjacent memset partitions by calling
2716 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
2717 Expose the interface.
2718 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
2720 2017-10-18 Bin Cheng <bin.cheng@arm.com>
2722 PR tree-optimization/82574
2723 * tree-loop-distribution.c (find_single_drs): New parameter. Check
2724 that data reference must be executed exactly once per iteration
2725 against the outermost loop in nest.
2726 (classify_partition): Update call to above function.
2728 2017-10-18 Richard Biener <rguenther@suse.de>
2730 PR tree-optimization/82591
2731 * graphite.c (graphite_transform_loops): Move code gen message
2733 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
2734 Here. Handle scop_to_isl_ast failing.
2735 (scop_to_isl_ast): Limit the number of ISL operations.
2737 2017-10-18 Richard Biener <rguenther@suse.de>
2739 * graphite-isl-ast-to-gimple.c
2740 (translate_isl_ast_to_gimple::set_rename): Simplify.
2741 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
2742 (graphite_copy_stmts_from_block): ... here.
2743 (copy_bb_and_scalar_dependences): Simplify.
2744 (add_parameters_to_ivs_params): Canonicalize.
2745 (generate_entry_out_of_ssa_copies): Simplify.
2746 * graphite-sese-to-poly.c (extract_affine_name): Simplify
2747 by passing in ISL dimension.
2748 (parameter_index_in_region_1): Rename to ...
2749 (parameter_index_in_region): ... this.
2750 (extract_affine): Adjust assert, pass down parameter index.
2751 (add_param_constraints): Use range-info when available.
2752 (build_scop_context): Adjust.
2753 * sese.c (new_sese_info): Adjust.
2754 (free_sese_info): Likewise.
2755 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
2756 Remove unused typedefs.
2757 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
2759 2017-10-18 Martin Liska <mliska@suse.cz>
2761 * combine.c (simplify_compare_const): Add gcc_fallthrough.
2763 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
2765 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
2766 (s390_sched_init): Do not reset s390_sched_state if we entered the
2767 current basic block via a fallthru edge and all others are unlikely.
2769 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
2771 * config/s390/s390.c (NUM_SIDES): New variable.
2772 (LONGRUNNING_THRESHOLD): New variable.
2773 (LATENCY_FACTOR): New variable.
2774 (s390_sched_score): Decrease score for long-running instructions on
2776 (s390_sched_variable_issue): Perform bookkeeping for long-running
2779 2017-10-18 Richard Biener <rguenther@suse.de>
2781 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
2782 Simplify with removal of the parameter rename map.
2783 (set_rename): Likewise.
2784 (should_copy_to_new_region): Likewise.
2785 (graphite_copy_stmts_from_block): Likewise.
2786 (copy_bb_and_scalar_dependences): Remove initialization of
2787 unused copied_bb_map.
2789 (copy_internal_parameters): Likewise.
2790 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
2791 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
2792 Use INTEGRAL_TYPE_P.
2793 (parameter_index_in_region_1): Rename to ...
2794 (assign_parameter_index_in_region): ... this. Assert we have
2795 a parameter we handle.
2796 (scan_tree_for_params): Adjust.
2797 * sese.h (parameter_rename_map_t): Remove.
2798 (struct sese_info_t): Remove unused parameter_rename_map and
2799 copied_bb_map members.
2800 * sese.c (new_sese_info): Adjust.
2801 (free_sese_info): Likewise.
2803 2017-10-18 Martin Liska <mliska@suse.cz>
2806 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
2807 on an abnormal edge.
2809 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
2811 * doc/invoke.texi (ffunction-sections and fdata-sections):
2814 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
2816 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
2817 the use statement can throw internally.
2819 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
2821 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
2822 any RTX present on the RHS of a SET.
2823 * compare-elim.c (try_eliminate_compare): Restore comment.
2825 2017-10-17 Jakub Jelinek <jakub@redhat.com>
2827 * langhooks.h (struct lang_hooks): Document that tree_size langhook
2828 may be also called on tcc_type nodes.
2829 * langhooks.c (lhd_tree_size): Likewise.
2831 2017-10-17 David Malcolm <dmalcolm@redhat.com>
2833 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
2834 format_warning_at_substring.
2835 (maybe_warn): Convert source_range * param to a location_t. Pass
2836 UNKNOWN_LOCATION rather than NULL to fmtwarn.
2837 (format_directive): Remove code to extract source_ranges and
2838 source_range * in favor of just a location_t.
2839 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
2841 * substring-locations.c (format_warning_va): Convert
2842 source_range * param to a location_t.
2843 (format_warning_at_substring): Likewise.
2844 * substring-locations.h (format_warning_va): Likewise.
2845 (format_warning_at_substring): Likewise.
2847 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
2849 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
2851 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
2852 and scatter/gather ops.
2854 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
2855 vec_gather_load and vec_scatter_store.
2856 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
2857 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
2858 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
2859 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
2860 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
2861 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
2863 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
2865 * reg-stack.c (compare_for_stack_reg): Add bool argument.
2866 Detect FTST instruction and handle its register pops. Only pop
2867 second operand if can_pop_second_op is true.
2868 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
2869 set can_pop_second_op to false in the compare_for_stack_reg call.
2871 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
2872 output_fp_compare for stack register operands.
2873 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
2874 instructions here. Do not emit stack register pops here. Assert
2875 that FCOMPP pops next to top stack register. Rewrite function.
2877 2017-10-17 Nathan Sidwell <nathan@acm.org>
2880 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
2881 use DECL_ASSEMBLER_NAME_RAW.
2884 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
2887 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
2888 Wilco Dijkstra <wilco.dijkstra@arm.com>
2890 * builtins.c (expand_builtin_update_setjmp_buf): Add a
2891 converstion to Pmode from the buf_addr.
2893 2017-10-17 Richard Biener <rguenther@suse.de>
2895 * graphite-dependences.c (scop_get_reads_and_writes): Change
2896 output parameters to references.
2898 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
2900 PR 71026/tree-optimization
2901 * fold-const.c (distribute_real_division): Removed.
2902 (fold_binary_loc): Remove calls to distribute_real_divison.
2904 2017-10-17 Richard Biener <rguenther@suse.de>
2906 * graphite-scop-detection.c
2907 (scop_detection::stmt_has_simple_data_refs_p): Always use
2908 the full nest as region.
2909 (try_generate_gimple_bb): Likewise.
2910 * sese.c (scalar_evolution_in_region): Simplify now that
2911 SCEV can handle instantiation in regions.
2912 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
2913 in the non-loop part of a function if requested.
2915 2017-10-17 Richard Biener <rguenther@suse.de>
2917 PR tree-optimization/82563
2918 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
2920 (graphite_regenerate_ast_isl): Call it.
2921 * graphite-scop-detection.c (build_scops): Remove entry edge split.
2923 2017-10-17 Jakub Jelinek <jakub@redhat.com>
2925 PR tree-optimization/82549
2926 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
2927 Formatting fixes. Instead of calling make_bit_field_ref with negative
2930 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
2932 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
2933 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
2934 _mm_maskz_reduce_ss): New.
2935 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
2936 __builtin_ia32_reducess_mask): Ditto..
2937 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
2938 * config/i386/sse.md (reduces<mode>): Renamed to ...
2939 (reduces<mode><mask_scalar_name>): ... this.
2940 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
2942 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
2943 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
2945 2017-10-16 David Malcolm <dmalcolm@redhat.com>
2947 * Makefile.in (OBJS): Add unique-ptr-tests.o.
2948 * selftest-run-tests.c (selftest::run_tests): Call
2949 selftest::unique_ptr_tests_cc_tests.
2950 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
2951 * unique-ptr-tests.cc: New file.
2953 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
2956 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
2958 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
2959 (make_hard_regno_born, make_hard_regno_dead): Update
2960 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
2962 2017-10-16 Jeff Law <law@redhat.com>
2964 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
2966 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
2968 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
2970 2017-10-16 Olivier Hainque <hainque@adacore.com>
2972 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
2973 with_cpu if we were configured for an e500v2 target cpu name.
2975 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
2977 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
2978 * doc/invoke.texi: Document +nodsp as a valid extension for
2981 2017-10-16 Martin Liska <mliska@suse.cz>
2983 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
2984 (test_set_range): Likewise.
2985 (test_range_functions): Rename to ...
2986 (test_bit_in_range): ... this.
2987 (sbitmap_c_tests): Add new test.
2989 2017-10-16 Tamar Christina <tamar.christina@arm.com>
2991 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
2993 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
2994 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
2996 2017-10-16 Tamar Christina <tamar.christina@arm.com>
2998 * config/aarch64/aarch64-builtins.c
2999 (aarch64_types_quadopu_lane_qualifiers): New.
3000 (TYPES_QUADOPU_LANE): New.
3001 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
3002 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
3003 (aarch64_<sur>dot_laneq<vsi2qi>): New.
3004 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
3005 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
3006 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
3007 (Vdottype, DOTPROD): New.
3008 (sur): Add SDOT and UDOT.
3010 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3012 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
3013 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
3014 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
3016 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
3017 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
3018 Enable TARGET_DOTPROD.
3019 (cortex-a75.cortex-a55): Likewise.
3020 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
3022 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3024 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
3025 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
3027 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
3029 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
3030 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
3031 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
3032 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
3033 * config/arm/types.md (neon_dot, neon_dot_q): New.
3034 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
3036 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3038 * config/arm/arm.h (TARGET_DOTPROD): New.
3039 * config/arm/arm.c (arm_arch_dotprod): New.
3040 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
3041 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
3042 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
3043 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
3044 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
3045 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
3046 * doc/invoke.texi (armv8.2-a): Document dotprod
3048 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3050 * i386.c (ix86_vec_cost): New function.
3051 (ix86_rtx_costs): Handle vector operations better.
3052 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
3053 * x86-tune-costs.h: Add new costs to all tables.
3055 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3057 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
3059 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
3060 divsd, sqrtss and sqrtsd
3061 * x86-tune-costs.h: Add new entries to all costs.
3062 (znver1_cost): Fix to match real instruction latencies.
3064 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3065 Michael Collison <michael.collison@arm.com>
3067 * compare-elim.c: Include emit-rtl.h.
3068 (can_merge_compare_into_arith): New function.
3069 (try_validate_parallel): Likewise.
3070 (try_merge_compare): Likewise.
3071 (try_eliminate_compare): Call the above when no previous clobber
3073 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
3076 2017-10-14 Jakub Jelinek <jakub@redhat.com>
3080 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
3081 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
3085 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
3086 to be any operand_equal_p operands. For & (B - 1) require
3087 B to be power of 2. Recognize
3088 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
3090 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3093 * optabs.c (expand_memory_blockage): Fix call of
3094 targetm.have_memory_blockage.
3096 2017-10-14 Jakub Jelinek <jakub@redhat.com>
3099 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
3100 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
3101 objects to extra_objs instead of overwriting it.
3103 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3105 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
3106 Use any_fp_register_operand as operand[3] predicate. Simplify
3107 equality test for operands[2] and operands[4] memory location.
3108 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
3109 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
3110 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
3111 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
3112 any_fp_register_operand as operand[1] predicate. Simplify
3113 equality test for operands[0] and operands[3] memory location.
3114 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
3115 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
3116 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
3118 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3120 * target-insns.def: Add memory_blockage.
3121 * optabs.c (expand_memory_blockage): New function.
3122 (expand_asm_memory_barrier): Rename ...
3123 (expand_asm_memory_blockage): ... to this.
3124 (expand_mem_thread_fence): Call expand_memory_blockage
3125 instead of expand_asm_memory_barrier.
3126 (expand_mem_singnal_fence): Ditto.
3127 (expand_atomic_load): Ditto.
3128 (expand_atomic_store): Ditto.
3129 * doc/md.texi (Standard Pattern Names For Generation):
3130 Document memory_blockage instruction pattern.
3132 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
3134 * config/rl78/rl78.c (rl78_emit_libcall): New function.
3135 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
3136 * config/rl78/rl78.md: New define_expand "adddi3".
3138 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
3140 * cfghooks.c (verify_flow_info): Disable check that all probabilities
3143 2017-10-13 Jeff Law <law@redhat.com>
3145 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
3147 2017-10-13 Jakub Jelinek <jakub@redhat.com>
3150 * internal-fn.c (expand_mul_overflow): If both operands have
3151 the same highpart of -1 or 0 and the topmost bit of lowpart
3152 is different, overflow is if res <= 0 rather than res < 0.
3154 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
3156 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
3157 TARGET_P9_VECTOR code for unaligned_load case.
3159 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
3161 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
3163 2017-10-13 Nathan Sidwell <nathan@acm.org>
3165 * tree-core.h (tree_contains_struct): Make bool.
3166 * tree.c (tree_contains_struct): Likewise.
3167 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
3168 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
3169 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
3170 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
3171 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
3173 2017-10-13 Richard Biener <rguenther@suse.de>
3175 * graphite-isl-ast-to-gimple.c
3176 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
3177 parameters and dominance check.
3178 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
3179 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
3180 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
3181 Do not update SSA form here or do intermediate IL verification.
3182 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
3183 (graphite_initialize): Remove check on the number of loops in
3184 the function and inline into graphite_transform_loops.
3185 (graphite_finalize): Inline into graphite_transform_loops.
3186 (graphite_transform_loops): Perform SSA update and IL verification
3188 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
3190 2017-10-13 Richard Biener <rguenther@suse.de>
3192 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
3193 graphite_expression_type_precision): Avoid global constructor
3195 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
3196 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
3197 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
3198 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
3199 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
3200 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
3202 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
3205 * config/i386/i386.h (ix86_red_zone_size): New.
3206 * config/i386/i386.md (push peephole2s): Replace
3207 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
3209 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3210 Alan Hayward <alan.hayward@arm.com>
3211 David Sherwood <david.sherwood@arm.com>
3213 * combine.c (can_change_dest_mode): Reject changes in
3214 REGMODE_NATURAL_SIZE.
3216 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3217 Alan Hayward <alan.hayward@arm.com>
3218 David Sherwood <david.sherwood@arm.com>
3220 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
3221 (expand_debug_source_expr): Likewise.
3222 * combine.c (combine_simplify_rtx): Likewise.
3223 * cse.c (fold_rtx): Likewise.
3224 * fwprop.c (canonicalize_address): Likewise.
3225 * targhooks.c (default_shift_truncation_mask): Likewise.
3227 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3228 Alan Hayward <alan.hayward@arm.com>
3229 David Sherwood <david.sherwood@arm.com>
3231 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
3232 (widened_mode): Likewise.
3233 (expand_unop): Likewise.
3234 * ree.c (transform_ifelse): Likewise.
3235 (merge_def_and_ext): Likewise.
3236 (combine_reaching_defs): Likewise.
3237 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3239 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3240 Alan Hayward <alan.hayward@arm.com>
3241 David Sherwood <david.sherwood@arm.com>
3243 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
3244 * combine.c (gen_lowpart_for_combine): Likewise.
3245 * dwarf2out.c (rtl_for_decl_location): Likewise.
3246 * final.c (alter_subreg): Likewise.
3247 * rtlhooks.c (gen_lowpart_general): Likewise.
3248 (gen_lowpart_if_possible): Likewise.
3250 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3251 Alan Hayward <alan.hayward@arm.com>
3252 David Sherwood <david.sherwood@arm.com>
3254 * calls.c (expand_call): Use subreg_lowpart_offset.
3255 * cse.c (cse_insn): Likewise.
3256 * regcprop.c (copy_value): Likewise.
3257 (copyprop_hardreg_forward_1): Likewise.
3259 2017-10-13 Jakub Jelinek <jakub@redhat.com>
3262 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
3263 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
3264 =Q constraints to +Q and into insn condition add check
3265 that operands[0] and operands[1] are equal.
3266 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
3267 =Q constraints to +Q and into insn condition add check
3268 that operands[0] is equal to either operands[1] or operands[2].
3271 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
3272 instead of handling MINUS_EXPR twice (once for each argument),
3273 canonicalize operand order and handle just once, use rtype where
3274 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
3277 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
3278 any values of __C while still being pattern recognizable as a simple
3281 2017-10-13 Richard Biener <rguenther@suse.de>
3283 PR tree-optimization/82451
3285 2017-10-02 Richard Biener <rguenther@suse.de>
3287 PR tree-optimization/82355
3288 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
3289 a mapping for the enclosing loop but avoid generating one for
3291 (copy_bb_and_scalar_dependences): Remove premature codegen
3292 error on PHIs in blocks duplicated into multiple places.
3293 * graphite-scop-detection.c
3294 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
3295 in the region use it as loop and nest to analyze the DR in.
3296 (try_generate_gimple_bb): Likewise.
3297 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
3298 (add_loop_constraints): For blocks in a loop not in the region
3299 create a dimension with a single iteration.
3300 * sese.h (gbb_loop_at_index): Remove assert.
3302 * cfgloop.c (loop_preheader_edge): For the loop tree root
3303 return the single successor of the entry block.
3304 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3305 Reset the SCEV hashtable and niters.
3306 * graphite-scop-detection.c
3307 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
3308 assert that we only have POLYNOMIAL_CHREC that vary in loops
3309 contained in the region.
3310 (scop_detection::graphite_can_represent_expr): Adjust.
3311 (scop_detection::stmt_has_simple_data_refs_p): For loops
3312 not in the region set loop to NULL. The nest is now the
3313 entry edge to the region.
3314 (try_generate_gimple_bb): Likewise.
3315 * sese.c (scalar_evolution_in_region): Adjust for
3316 instantiate_scev change.
3317 * tree-data-ref.h (graphite_find_data_references_in_stmt):
3318 Make nest parameter the edge into the region.
3319 (create_data_ref): Likewise.
3320 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
3321 entry edge into a region and adjust instantiate_scev calls.
3322 (create_data_ref): Likewise.
3323 (graphite_find_data_references_in_stmt): Likewise.
3324 (find_data_references_in_stmt): Pass the loop preheader edge
3325 from the nest argument.
3326 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
3327 parameter the edge into the region.
3328 (instantiate_parameters): Use the loop preheader edge as entry.
3329 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
3331 (get_instantiated_value_entry): Make instantiate_below parameter
3332 the edge into the region.
3333 (instantiate_scev_name): Likewise. Adjust dominance checks,
3334 when we cannot use loop-based instantiation instantiate by
3335 walking use-def chains.
3336 (instantiate_scev_poly): Adjust.
3337 (instantiate_scev_binary): Likewise.
3338 (instantiate_scev_convert): Likewise.
3339 (instantiate_scev_not): Likewise.
3340 (instantiate_array_ref): Remove.
3341 (instantiate_scev_3): Likewise.
3342 (instantiate_scev_2): Likewise.
3343 (instantiate_scev_1): Likewise.
3344 (instantiate_scev_r): Do not blindly handle N-operand trees.
3345 Do not instantiate array-refs. Handle all constants and invariants.
3346 (instantiate_scev): Make instantiate_below parameter
3347 the edge into the region.
3348 (resolve_mixers): Use the loop preheader edge for the region
3349 parameter to instantiate_scev_r.
3350 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
3352 2017-10-13 Richard Biener <rguenther@suse.de>
3354 PR tree-optimization/82525
3355 * graphite-isl-ast-to-gimple.c
3356 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
3358 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
3359 Fail code generation when we cannot represent the isl integer.
3360 (binary_op_to_tree): Elide modulo operations that are no-ops
3361 in the type we code generate. Remove now superfluous code
3364 2017-10-13 Richard Biener <rguenther@suse.de>
3366 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
3367 (scop_detection::harmful_loop_in_region): Remove premature
3368 IV type restriction.
3369 (scop_detection::graphite_can_represent_scev): We can handle
3370 pointer IVs just fine.
3372 2017-10-13 Alan Modra <amodra@gmail.com>
3374 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
3375 "Clobbers and Scratch Registers". Add paragraph on
3376 alternative to clobbers for scratch registers and OpenBLAS
3379 2017-10-13 Alan Modra <amodra@gmail.com>
3381 * doc/extend.texi (Clobbers): Correct vax example. Delete old
3382 example of a memory input for a string of known length. Move
3383 commentary out of table. Add a number of new examples
3384 covering array memory inputs.
3386 2017-10-12 Martin Liska <mliska@suse.cz>
3388 PR tree-optimization/82493
3389 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
3390 (test_range_functions): New function.
3391 (sbitmap_c_tests): Likewise.
3392 * selftest-run-tests.c (selftest::run_tests): Run new tests.
3393 * selftest.h (sbitmap_c_tests): New function.
3395 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
3397 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
3399 * config/rs6000/amo.h: Fix spacing issue.
3401 2017-10-12 Jakub Jelinek <jakub@redhat.com>
3404 * config/i386/i386.md (*ashl<mode>3_mask_1,
3405 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
3406 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
3409 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
3411 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
3412 (profile_probability): Set max_probability
3413 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
3415 * profile-count.c (profile_probability::differs_from_p): Do not
3416 rely on max_probaiblity == 10000
3418 2017-10-12 Jeff Law <law@redhat.com>
3420 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
3423 2017-10-12 Martin Sebor <msebor@redhat.com>
3427 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
3428 (handle_alias_pairs): Call it.
3429 * common.opt (-Wattribute-alias): New option.
3430 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
3431 * doc/invoke.texi (-Wattribute-alias): Document.
3433 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
3436 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
3438 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3440 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3441 (make_hard_regno_born, make_hard_regno_dead): Update
3442 bb_killed_pseudos and bb_gen_pseudos.
3444 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
3446 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
3448 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
3450 * config/alpha/alpha.c (alpha_split_conditional_move):
3451 Use std::swap instead of manually swapping.
3452 (alpha_stdarg_optimize_hook): Ditto.
3453 (alpha_canonicalize_comparison): Ditto.
3455 2017-10-12 Bin Cheng <bin.cheng@arm.com>
3457 * tree-loop-distribution.c (struct builtin_info): New struct.
3458 (struct partition): Refactor fields into struct builtin_info.
3459 (partition_free): Free struct builtin_info.
3460 (build_size_arg_loc, build_addr_arg_loc): Delete.
3461 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
3462 information from struct builtin_info.
3463 (find_single_drs): New function refactored from classify_partition.
3464 Also moved builtin validity checks to this function.
3465 (compute_access_range, alloc_builtin): New functions.
3466 (classify_builtin_st, classify_builtin_ldst): New functions.
3467 (classify_partition): Refactor code into functions find_single_drs,
3468 classify_builtin_st and classify_builtin_ldst.
3469 (distribute_loop): Don't do runtime alias check when distributing
3471 (find_seed_stmts_for_distribution): New function.
3472 (pass_loop_distribution::execute): Refactor code finding seed
3473 stmts into above function. Support distribution for the innermost
3474 two-level loop nest. Adjust dump information.
3476 2017-10-12 Bin Cheng <bin.cheng@arm.com>
3478 * tree-loop-distribution.c: Adjust the general comment.
3479 (NUM_PARTITION_THRESHOLD): New macro.
3480 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
3481 (classify_partition): Skip builtin pattern of loop nest's inner loop.
3482 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
3483 in call to build_partition_graph.
3484 (finalize_partitions): New parameter. Make loop distribution more
3485 conservative by fusing more partitions.
3486 (distribute_loop): Don't do runtime alias check in case of loop nest
3488 (find_seed_stmts_for_distribution): New function.
3489 (prepare_perfect_loop_nest): New function.
3490 (pass_loop_distribution::execute): Refactor code finding seed stmts
3491 and loop nest into above functions. Support loop nest distribution.
3492 Adjust dump information accordingly.
3494 2017-10-12 Bin Cheng <bin.cheng@arm.com>
3496 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
3497 and set PTYPE_SEQUENTIAL for merged partition.
3499 2017-10-12 Richard Biener <rguenther@suse.de>
3501 PR tree-optimization/69728
3503 2017-09-19 Richard Biener <rguenther@suse.de>
3505 PR tree-optimization/69728
3506 * graphite-sese-to-poly.c (schedule_error): New global.
3507 (add_loop_schedule): Handle empty domain by failing the
3509 (build_original_schedule): Handle schedule_error.
3511 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
3512 domain by returning an unchanged schedule.
3514 2017-10-12 Jakub Jelinek <jakub@redhat.com>
3516 * genrecog.c (validate_pattern): For VEC_SELECT verify that
3517 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
3519 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
3521 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
3524 2017-10-12 Jakub Jelinek <jakub@redhat.com>
3527 * expr.c (store_field): Don't optimize away bitsize == 0 store
3528 from CALL_EXPR with addressable return type.
3530 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
3532 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
3533 * config/rs6000/rs6000.md (sel): Delete mode attribute.
3534 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
3535 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
3536 TARGET_ISEL instead of TARGET_ISEL<sel>.
3538 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
3540 * config/rs6000/rs6000.c
3541 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
3543 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
3545 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
3547 (reg_or_cint_operand): Fix comment.
3548 (reg_or_zero_operand): New predicate.
3549 * config/rs6000/rs6000-protos.h (output_isel): Delete.
3550 * config/rs6000/rs6000.c (output_isel): Delete.
3551 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
3552 instead of reg_or_cint_operand. Output instruction directly (not via
3554 (isel_unsigned_<mode>): Ditto.
3555 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
3556 gpc_reg_operand. Add an instruction alternative for this. Output
3557 instruction directly.
3558 (*isel_reversed_unsigned_<mode>): Ditto.
3560 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
3562 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
3563 (TARGET_CANONICALIZE_COMPARISON): Define.
3565 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
3568 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
3569 Check whether the dest is REG before adding REG_EQUIV note.
3571 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
3574 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3576 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3577 (make_hard_regno_born, make_hard_regno_dead): Update
3578 bb_killed_pseudos and bb_gen_pseudos.
3580 2017-10-11 Nathan Sidwell <nathan@acm.org>
3582 * incpath.h (enum incpath_kind): Name enum, prefix values.
3583 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
3584 * incpath.c (heads, tails): Use INC_MAX.
3585 (add_env_var_paths, add_standard_paths): Use incpath_kind.
3586 (merge_include_chains, split_quote_chain,
3587 register_include_chains): Update incpath_kind names.
3588 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
3589 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
3591 (add_framework_path, darwin_register_objc_includes): Likewise.
3592 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
3594 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
3596 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
3597 Do not use float_operator operator predicate.
3598 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
3599 * config/i386/predicates.md (float_operator): Remove predicate.
3601 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
3603 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
3604 (*jccxf_i387): Ditto.
3605 (*jcc<mode>_i387): Ditto.
3606 (*jccu<mode>_i387): Ditto.
3607 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
3608 (*jcc_*_i387 splitters): Remove.
3609 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
3610 * config/i386/i386.c (ix86_split_fp_branch): Remove.
3611 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
3614 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
3616 * profile-count.h (slow_safe_scale_64bit): New function.
3617 (safe_scale_64bit): New inline.
3618 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
3619 * profile-count.c: Include wide-int.h
3620 (slow_safe_scale_64bit): New.
3622 2017-10-11 Nathan Sidwell <nathan@acm.org>
3624 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
3625 HAS_DECL_ASSEMBLER_NAME_P.
3626 * gimple-expr.c (gimple_decl_printable_name: Check
3627 HAS_DECL_ASSEMBLER_NAME_P too.
3628 * ipa-utils.h (type_in_anonymous_namespace_p): Check
3629 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
3630 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
3631 * passes.c (rest_of_decl_compilation): Check
3632 HAS_DECL_ASSEMBLER_NAME_P too.
3633 * recog.c (verify_changes): Likewise.
3634 * tree-pretty-print.c (dump_decl_name): Likewise.
3635 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
3637 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
3638 (SET_DECL_ASSEMBLER_NAME): Use it.
3639 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
3640 (COPY_DECL_ASSEMBLER_NAME): Likewise.
3641 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
3643 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
3645 * config.gcc (i386, x86_64): Add extra objects.
3646 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
3647 (ix86_min_insn_size): Declare.
3648 (ix86_issue_rate): Declare.
3649 (ix86_adjust_cost): Declare.
3650 (ia32_multipass_dfa_lookahead): Declare.
3651 (ix86_macro_fusion_p): Declare.
3652 (ix86_macro_fusion_pair_p): Declare.
3653 (ix86_bd_has_dispatch): Declare.
3654 (ix86_bd_do_dispatch): Declare.
3655 (ix86_core2i7_init_hooks): Declare.
3656 (ix86_atom_sched_reorder): Declare.
3657 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
3658 (COSTS_N_BYTES): Move to x86-tune-costs.h.
3659 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
3660 (rip_relative_addr_p): Rename to ...
3661 (ix86_rip_relative_addr_p): ... this one; export.
3662 (memory_address_length): Update.
3663 (ix86_issue_rate): Move to x86-tune-sched.c.
3664 (ix86_flags_dependent): Move to x86-tune-sched.c.
3665 (ix86_agi_dependent): Move to x86-tune-sched.c.
3666 (exact_dependency_1): Move to x86-tune-sched.c.
3667 (exact_store_load_dependency): Move to x86-tune-sched.c.
3668 (ix86_adjust_cost): Move to x86-tune-sched.c.
3669 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
3670 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
3671 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
3672 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
3673 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
3674 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
3675 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
3676 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
3677 (min_insn_size): Rename to ...
3678 (ix86_min_insn_size): ... this one; export.
3679 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
3680 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
3681 (core2i7_first_cycle_multipass_backtrack): Move to
3682 x86-tune-sched-core.c.
3683 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
3684 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
3685 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
3686 (ix86_avoid_jump_mispredicts): Update.
3687 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
3688 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
3689 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
3690 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
3691 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
3692 (MAX_INSN): Move to ix86-tune-sched-bd.c.
3693 (MAX_IMM): Move to ix86-tune-sched-bd.c.
3694 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
3695 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
3696 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
3697 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
3698 (MAX_STORE): Move to ix86-tune-sched-bd.c.
3699 (BIG): Move to ix86-tune-sched-bd.c.
3700 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
3701 (enum insn_path): Move to ix86-tune-sched-bd.c.
3702 (get_mem_group): Move to ix86-tune-sched-bd.c.
3703 (is_cmp): Move to ix86-tune-sched-bd.c.
3704 (dispatch_violation): Move to ix86-tune-sched-bd.c.
3705 (is_branch): Move to ix86-tune-sched-bd.c.
3706 (is_prefetch): Move to ix86-tune-sched-bd.c.
3707 (init_window): Move to ix86-tune-sched-bd.c.
3708 (allocate_window): Move to ix86-tune-sched-bd.c.
3709 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
3710 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
3711 (process_end_window): Move to ix86-tune-sched-bd.c.
3712 (allocate_next_window): Move to ix86-tune-sched-bd.c.
3713 (find_constant): Move to ix86-tune-sched-bd.c.
3714 (get_num_immediates): Move to ix86-tune-sched-bd.c.
3715 (has_immediate): Move to ix86-tune-sched-bd.c.
3716 (get_insn_path): Move to ix86-tune-sched-bd.c.
3717 (get_insn_group): Move to ix86-tune-sched-bd.c.
3718 (count_num_restricted): Move to ix86-tune-sched-bd.c.
3719 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
3720 (add_insn_window): Move to ix86-tune-sched-bd.c.
3721 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
3722 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
3723 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
3724 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
3725 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
3726 (do_dispatch): Move to ix86-tune-sched-bd.c.
3727 (has_dispatch): Move to ix86-tune-sched-bd.c.
3728 * config/i386/t-i386: Add new object files.
3729 * config/i386/x86-tune-costs.h: New file.
3730 * config/i386/x86-tune-sched-atom.c: New file.
3731 * config/i386/x86-tune-sched-bd.c: New file.
3732 * config/i386/x86-tune-sched-core.c: New file.
3733 * config/i386/x86-tune-sched.c: New file.
3735 2017-10-11 Liu Hao <lh_mouse@126.com>
3737 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
3738 the generic version below instead.
3739 (should_colorize): Recognize Windows consoles as terminals
3741 * pretty-print.c [__MINGW32__] (write_all): New function.
3742 [__MINGW32__] (find_esc_head): Likewise.
3743 [__MINGW32__] (find_esc_terminator): Likewise.
3744 [__MINGW32__] (eat_esc_sequence): Likewise.
3745 [__MINGW32__] (mingw_ansi_fputs): New function that handles
3747 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
3750 2017-10-11 Richard Biener <rguenther@suse.de>
3752 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
3753 Properly call analyze_scalar_evolution with the loop of the stmt.
3755 2017-10-11 Richard Biener <rguenther@suse.de>
3757 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
3758 * tree-core.h (tree_base): Add chrec_var union member.
3759 * tree.h (CHREC_VAR): Remove.
3760 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
3761 * tree-chrec.h (build_polynomial_chrec): Adjust.
3762 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
3763 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
3765 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
3767 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
3768 * match.pd: ... here.
3769 ((T) X == (T) Y): Relax condition.
3771 2017-10-11 Bin Cheng <bin.cheng@arm.com>
3773 PR tree-optimization/82472
3774 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
3776 (break_alias_scc_partitions): Update postorder number.
3778 2017-10-11 Martin Liska <mliska@suse.cz>
3781 * opts.c (parse_no_sanitize_attribute): Do not use error_value
3783 * opts.h (parse_no_sanitize_attribute): Remove last argument.
3785 2017-10-11 Martin Liska <mliska@suse.cz>
3787 * print-rtl.c (print_insn): Move declaration of idbuf
3788 to same scope as name.
3790 2017-10-11 Martin Liska <mliska@suse.cz>
3795 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
3798 2017-10-11 Martin Liska <mliska@suse.cz>
3801 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
3804 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
3806 * config.gcc (default_gnu_indirect_function): Default to yes for
3807 arm*-*-linux* with glibc.
3809 2017-10-11 Richard Biener <rguenther@suse.de>
3811 * tree-scalar-evolution.c (get_scalar_evolution): Handle
3812 default-defs and types we do not want to analyze.
3813 (interpret_loop_phi): Replace unreachable code with an assert.
3814 (compute_scalar_evolution_in_loop): Remove and inline ...
3815 (analyze_scalar_evolution_1): ... here, replacing condition with
3816 what makes the intent clearer. Remove handling of cases
3817 get_scalar_evolution now handles.
3819 2017-10-10 Jim Wilson <wilson@tuliptree.org>
3821 PR rtl-optimization/81434
3822 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
3823 comment for main loop. In sched_group_found if, also add checks for
3824 pass and min_cost_group.
3826 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
3828 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
3829 (rs6000_insn_cost): New function.
3830 * config/rs6000/rs6000.md (cost): New attribute.
3832 2017-10-10 Jakub Jelinek <jakub@redhat.com>
3833 H.J. Lu <hongjiu.lu@intel.com>
3837 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
3838 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
3839 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
3840 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
3841 to OPTION_MASK_ISA_AVX512VL - builtins that have both
3842 OPTION_MASK_ISA_MMX and some other bit set require both
3843 mmx and the ISAs without the mmx bit.
3844 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
3845 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
3846 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
3847 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
3848 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
3849 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
3850 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
3851 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
3852 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
3853 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
3854 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
3855 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
3856 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
3857 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
3858 Add OPTION_MASK_ISA_MMX.
3860 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
3862 * config.gcc (armv7*-*-freebsd*): New target.
3863 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
3865 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
3867 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
3868 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
3871 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
3873 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
3874 a template parameter.
3875 (WIDE_INT_REF_FOR): Update accordingly.
3876 * tree.h (wi::int_traits <const_tree>): Delete.
3877 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
3878 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
3879 (wi::tree_to_wide_ref): New typedef.
3880 (wi::to_wide): New function.
3881 * calls.c (get_size_range): Use wi::to_wide when operating on
3883 * cgraph.c (cgraph_node::create_thunk): Likewise.
3884 * config/i386/i386.c (ix86_data_alignment): Likewise.
3885 (ix86_local_alignment): Likewise.
3886 * dbxout.c (stabstr_O): Likewise.
3887 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
3888 * expr.c (const_vector_from_tree): Likewise.
3889 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
3890 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
3891 (fold_negate_expr_1, int_const_binop_1, const_binop)
3892 (fold_convert_const_int_from_real, optimize_bit_field_compare)
3893 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
3894 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
3895 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
3896 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
3897 (fold_not_const, round_up_loc): Likewise.
3898 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
3899 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
3900 (alloca_call_type): Likewise.
3901 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
3902 * godump.c (go_output_typedef): Likewise.
3903 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
3904 * internal-fn.c (get_min_precision): Likewise.
3905 * ipa-cp.c (ipcp_store_vr_results): Likewise.
3906 * ipa-polymorphic-call.c
3907 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
3908 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
3909 (ipa_modify_call_arguments): Likewise.
3910 * match.pd: Likewise.
3911 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
3912 * print-tree.c (print_node_brief, print_node): Likewise.
3913 * stmt.c (expand_case): Likewise.
3914 * stor-layout.c (layout_type): Likewise.
3915 * tree-affine.c (tree_to_aff_combination): Likewise.
3916 * tree-cfg.c (group_case_labels_stmt): Likewise.
3917 * tree-data-ref.c (dr_analyze_indices): Likewise.
3918 (prune_runtime_alias_test_list): Likewise.
3919 * tree-dump.c (dequeue_and_dump): Likewise.
3920 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
3921 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
3922 * tree-pretty-print.c (dump_generic_node): Likewise.
3923 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
3924 (simple_iv_with_niters): Likewise.
3925 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
3926 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
3927 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
3928 * tree-ssa-loop-niter.c (split_to_var_and_offset)
3929 (refine_value_range_using_guard, number_of_iterations_ne_max)
3930 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
3931 (get_cst_init_from_scev, record_nonwrapping_iv)
3932 (scev_var_range_cant_overflow): Likewise.
3933 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
3934 * tree-ssa-pre.c (compute_avail): Likewise.
3935 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
3936 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
3937 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
3938 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
3939 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
3940 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
3941 (dump_case_nodes, try_switch_expansion): Likewise.
3942 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
3943 (vect_do_peeling): Likewise.
3944 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
3945 * tree-vect-stmts.c (vectorizable_load): Likewise.
3946 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
3947 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
3948 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
3949 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
3950 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
3951 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
3952 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
3953 (evrp_dom_walker::before_dom_children): Likewise.
3954 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
3955 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
3956 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
3957 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
3958 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
3959 (get_range_pos_neg): Likewise.
3960 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
3961 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
3962 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
3963 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
3964 * config/avr/avr.c (avr_fold_builtin): Likewise.
3965 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
3966 * config/msp430/msp430.c (msp430_attr): Likewise.
3967 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
3968 * config/powerpcspe/powerpcspe-c.c
3969 (altivec_resolve_overloaded_builtin): Likewise.
3970 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
3971 (rs6000_expand_ternop_builtin): Likewise.
3972 * config/rs6000/rs6000-c.c
3973 (altivec_resolve_overloaded_builtin): Likewise.
3974 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
3975 (rs6000_expand_ternop_builtin): Likewise.
3976 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
3978 2017-10-10 Bin Cheng <bin.cheng@arm.com>
3980 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
3981 when copying loop nest with only one inner loop.
3983 2017-10-10 Richard Biener <rguenther@suse.de>
3985 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
3986 blocks if SCEV is active.
3987 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
3989 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
3990 (scev_initialize): Assert we are not yet initialized.
3992 2017-10-10 Bin Cheng <bin.cheng@arm.com>
3994 * tree-loop-distribution.c (generate_loops_for_partition): Remove
3995 inner loop's exit stmt by making it always exit the loop, otherwise
3996 we would generate an infinite empty loop.
3998 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4000 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
4001 renaming variables in new preheader if it's deleted.
4003 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4005 * tree-loop-distribution.c (struct partition): Remove unused field
4006 loops of the structure.
4007 (partition_alloc, partition_free): Ditto.
4008 (build_rdg_partition_for_vertex): Ditto.
4010 2017-10-09 Jeff Law <law@redhat.com>
4012 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
4013 return type to match prototype and documentation.
4015 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4017 * config/rs6000/rs6000.c (processor_costs): Move to ...
4018 * config/rs6000/rs6000.h: ... here.
4019 (rs6000_cost): Declare.
4021 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
4023 * except.c (setjmp_fn): New global variable.
4024 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
4025 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
4026 if DONT_USE_BUILTIN_SETJMP is defined.
4028 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4030 * target.def (insn_cost): New hook.
4031 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
4032 * doc/tm.texi: Regenerate.
4033 * rtlanal.c (insn_cost): Use the new hook.
4035 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4037 * combine.c (combine_validate_cost): Compute the new insn_cost,
4038 not just pattern_cost.
4039 (try_combine): Adjust comment.
4041 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4043 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
4045 * combine.c (uid_insn_cost): Adjust comment.
4046 (combine_validate_cost): Adjust comment. Use pattern_cost instead
4048 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
4049 * dse.c (find_shift_sequence): Ditto.
4050 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
4051 (bb_valid_for_noce_process_p): Use pattern_cost.
4052 * rtl.h (insn_rtx_cost): Delete.
4053 (pattern_cost): New prototype.
4054 (insn_cost): New prototype.
4055 * rtlanal.c (insn_rtx_cost): Rename to...
4056 (pattern_cost): ... this.
4059 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
4061 * config/i386/i386.md (*jcc_2): Remove insn pattern.
4062 (*jcc<mode>_0_r_i387): Ditto.
4063 (*jccxf_r_i387): Ditto.
4064 (*jcc<mode>_r_i387): Ditto.
4065 (*jccu<mode>_r_i387): Ditto.
4066 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
4067 (*jcc): Rename from *jcc_1.
4069 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4071 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
4072 deferred rescans after the lvx/stvx recombination pre-pass.
4074 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
4076 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
4077 memory operation instruction support.
4078 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
4079 (rs6000-ibm-aix[789]*): Likewise.
4080 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
4081 Document new functions.
4083 2017-10-09 Richard Biener <rguenther@suse.de>
4085 PR tree-optimization/82397
4086 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
4087 equality only for semantically equal trees.
4089 2017-10-09 Richard Biener <rguenther@suse.de>
4091 PR tree-optimization/82449
4092 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
4093 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
4094 allow constant addresses.
4095 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
4098 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4100 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
4103 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4106 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
4109 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4112 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
4114 2017-10-09 Jakub Jelinek <jakub@redhat.com>
4117 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
4118 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
4120 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
4122 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
4123 (WI_BINARY_PREDICATE_RESULT): Likewise.
4124 (wi::binary_traits::operator_result): New type.
4125 (wi::binary_traits::predicate_result): Likewise.
4126 (generic_wide_int::operator~, unary generic_wide_int::operator-)
4127 (generic_wide_int::operator==, generic_wide_int::operator!=)
4128 (generic_wide_int::operator&, generic_wide_int::and_not)
4129 (generic_wide_int::operator|, generic_wide_int::or_not)
4130 (generic_wide_int::operator^, generic_wide_int::operator+
4131 (binary generic_wide_int::operator-, generic_wide_int::operator*):
4133 (operator~, unary operator-, operator==, operator!=, operator&)
4134 (operator|, operator^, operator+, binary operator-, operator*): New
4136 * expr.c (get_inner_reference): Use wi::bit_and_not.
4137 * fold-const.c (fold_binary_loc): Likewise.
4138 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
4139 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
4140 (bit_value_binop): Likewise.
4141 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
4142 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
4143 (extract_range_from_binary_expr_1): Likewise.
4144 (masked_increment): Likewise.
4145 (simplify_bit_ops_using_ranges): Likewise.
4147 2017-10-09 Martin Jambor <mjambor@suse.cz>
4150 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
4151 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
4152 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
4154 (hsa_fixup_mov_insn_type): New function.
4155 (hsa_op_with_type::get_in_type): Use it.
4156 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
4157 immediates in an assert.
4158 (hsa_op_with_type::extend_int_to_32bit): New method.
4159 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
4160 types. Convert to dest type if necessary.
4161 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
4162 (reg_for_gimple_ssa): Pass false as min32int to
4163 hsa_type_for_scalar_tree_type.
4164 (gen_hsa_addr): Fixup type when creating addresable temporary.
4165 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
4166 (gen_hsa_unary_operation): Extend operands and convert to dest type if
4167 necessary. Call hsa_fixup_mov_insn_type.
4168 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
4169 extend operands and convert to dest type if necessary.
4170 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
4171 to dest type if necessary.
4172 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
4173 if conversion nt necessary and size matches.
4174 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
4175 to dest type if necessary.
4176 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
4177 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
4179 (gen_hsa_clrsb): Likewise.
4180 (gen_hsa_ffs): Likewise.
4181 (gen_hsa_divmod): Extend operands and convert to dest type if
4183 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
4185 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
4187 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
4188 Remove empty default arguments. Use a brace block as output
4190 (conditional return): Ditto.
4192 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
4193 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
4195 (group_ending_nop): Ditto.
4196 (doloop_end): Ditto.
4197 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
4198 (splitters for those): Ditto.
4200 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
4202 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
4203 a conditional jump (and the compare for it) so that pc_rtx is the
4205 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
4206 for the deleted and renamed ctr<mode>_internal[234] patterns.
4207 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
4208 Delete second conditional return pattern.
4209 (ctr<mode>_internal2): Delete this second bdnz pattern.
4210 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
4211 (ctr<mode>_internal4): Delete this second bdz pattern.
4213 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
4215 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
4216 (always_initialized_rtx_for_ssa_name_p): New predicate.
4217 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
4218 (finish_out_of_ssa): Free new field of SA.
4219 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
4220 * tree-ssa-coalesce.c: Include tree-ssa.h.
4221 (get_parm_default_def_partitions): Remove extern keyword.
4222 (get_undefined_value_partitions): New function.
4223 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
4224 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
4226 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
4228 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
4230 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
4232 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
4234 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
4235 for targets that preffer 128bit.
4237 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
4239 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
4241 2017-10-08 Olivier Hainque <hainque@adacore.com>
4243 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
4244 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
4246 (thumb_set_return_address): Likewise.
4248 2017-10-08 Olivier Hainque <hainque@adacore.com>
4250 * common/config/arm/arm-common.c (arm_except_unwind_info):
4251 Handle DWARF2_UNWIND_INFO.
4253 2017-10-07 Michael Collison <michael.collison@arm.com>
4255 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
4258 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
4260 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
4261 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
4262 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
4263 defined, force the creation of a new block for a dispatch label.
4265 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
4267 * invoke.texi (Wsuggest-attribute=cold): Document.
4268 * common.opt (Wsuggest-attribute=cold): New
4269 * ipa-pure-const.c (warn_function_cold): New function.
4270 * predict.c (compute_function_frequency): Use it.
4271 * predict.h (warn_function_cold): Declare.
4273 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
4275 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
4278 2017-10-06 Martin Liska <mliska@suse.cz>
4280 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
4281 keyword for member functions.
4282 (struct sanopt_tree_couple): New struct.
4283 (struct sanopt_tree_couple_hash): New function.
4284 (struct sanopt_ctx): Add new hash_map.
4285 (has_dominating_ubsan_ptr_check): New function.
4286 (record_ubsan_ptr_check_stmt): Likewise.
4287 (maybe_optimize_ubsan_ptr_ifn): Likewise.
4288 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
4289 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
4291 2017-10-06 Sudakshina Das <sudi.das@arm.com>
4294 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
4295 aarch64_simd_valid_immediate on CONST_VECTORs.
4296 (aarch64_reg_or_bic_imm): Likewise.
4298 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
4300 PR rtl-optimization/82396
4301 * haifa-sched.c (ready_sort_real): Disable qsort checking.
4303 2017-10-06 Sebastian Pop <sebpop@gmail.com>
4305 * graphite-dependences.c (scop_get_reads): Move code to...
4306 (scop_get_must_writes): Move code to...
4307 (scop_get_may_writes): Move code to...
4308 (scop_get_reads_and_writes): ... here.
4309 (scop_get_dependences): Call scop_get_reads_and_writes.
4311 2017-10-06 Jakub Jelinek <jakub@redhat.com>
4313 PR tree-optimization/82434
4314 * fold-const.h (can_native_encode_type_p,
4315 can_native_encode_string_p): Remove.
4316 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
4317 don't encode anything, just return what would be otherwise returned.
4318 (native_encode_fixed, native_encode_complex, native_encode_vector):
4320 (native_encode_string): Likewise. Inline by hand
4321 can_native_encode_string_p.
4322 (can_native_encode_type_p): Remove.
4323 (can_native_encode_string_p): Remove.
4324 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
4325 STRING_CSTs using can_native_encode_string_p, test all
4326 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
4327 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
4328 argument from native_encode_expr.
4329 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
4330 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
4333 2017-10-06 Richard Biener <rguenther@suse.de>
4335 PR tree-optimization/82397
4336 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
4337 operand_equal_p but rely on data_ref_compare_tree for detecting
4339 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
4340 to match up with dr_group_sort_cmp.
4342 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4345 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
4347 * config/s390/s390-builtin-types.def: Regenerate.
4349 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4352 * config/s390/s390-builtin-types.def: Regenerate.
4353 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
4354 Change flag from B_VXE to B_VX.
4355 (s390_vec_min_dbl): Remove B_VXE flag.
4357 2017-10-06 Richard Biener <rguenther@suse.de>
4359 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
4360 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
4361 translate_isl_ast_to_gimple::is_valid_rename,
4362 translate_isl_ast_to_gimple::get_rename,
4363 translate_isl_ast_to_gimple::get_def_bb_for_const,
4364 translate_isl_ast_to_gimple::get_new_name,
4365 translate_isl_ast_to_gimple::collect_all_ssa_names,
4366 translate_isl_ast_to_gimple::copy_loop_phi_args,
4367 translate_isl_ast_to_gimple::collect_all_ssa_names,
4368 translate_isl_ast_to_gimple::copy_loop_phi_args,
4369 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
4370 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
4371 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
4372 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
4373 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
4374 translate_isl_ast_to_gimple::copy_cond_phi_args,
4375 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
4376 translate_isl_ast_to_gimple::edge_for_new_close_phis,
4377 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
4378 translate_isl_ast_to_gimple::rename_uses,
4379 translate_isl_ast_to_gimple::rename_all_uses): Remove.
4380 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
4381 (set_rename_for_each_def): Likewise.
4382 (graphite_copy_stmts_from_block): Handle debug stmt resetting
4383 here. Handle rewriting SCEV analyzable uses here.
4384 (copy_bb_and_scalar_dependences): Generate code for PHI
4386 (graphite_regenerate_ast_isl): Adjust.
4387 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
4388 (add_write, add_read): New functions.
4389 (build_cross_bb_scalars_def): Use it and simplify.
4390 (build_cross_bb_scalars_use): Likewise.
4391 (graphite_find_cross_bb_scalar_vars): Inline into...
4392 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
4393 simulating out-of-SSA. Compute liveout and add dependencies.
4394 (build_scops): Force an empty entry block.
4395 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
4397 (sese_build_liveouts): Declare.
4398 (sese_trivially_empty_bb_p): Likewise.
4399 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
4400 compute liveout and debug_liveout.
4401 (sese_bad_liveouts_use): Remove.
4402 (sese_reset_debug_liveouts_bb): Likewise.
4403 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
4404 (sese_build_liveouts): Build liveout and debug_liveout and store
4406 (new_sese_info): Adjust.
4407 (free_sese_info): Likewise.
4408 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
4409 do not build liveout here.
4410 (move_sese_in_condition): Adjust region entry.
4411 (scev_analyzable_p): Match up with chrec_apply requirements.
4412 (sese_trivially_empty_bb_p): New.
4413 * tree-into-ssa.c (get_reaching_def): Properly support generating
4414 default-defs for incremental rewrite of anonymous names.
4416 2017-10-06 Richard Biener <rguenther@suse.de>
4418 * graphite-sese-to-poly.c (extract_affine): For casts increasing
4419 precision do not perform modulo reduction.
4421 2017-10-06 Richard Biener <rguenther@suse.de>
4423 PR tree-optimization/82436
4424 * tree-vect-slp.c (vect_supported_load_permutation_p): More
4425 conservatively choose the vectorization factor when checking
4426 whether we can perform the required load permutation.
4427 (vect_transform_slp_perm_load): Assert when we may not fail.
4429 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
4431 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
4432 message for incompatible -msdata=* and -mcall-* options.
4434 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
4436 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
4437 rate for post-reload scheduling.
4439 2017-10-05 Tamar Christina <tamar.christina@arm.com>
4441 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
4443 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
4445 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
4446 to improve monte carlo in scimark.
4448 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
4450 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
4451 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
4452 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
4453 pentium4_cost, nocona_cost): Set reassociation width to 1.
4454 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
4455 width to 2 for fp operations and 1 otherwise.
4456 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
4458 (atom_cost): Set reassociation width to 2.
4459 (slm_cost, generic_cost): Set fp reassociation width
4460 to 2 and 1 otherwise.
4461 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
4462 (core_cost): Set fp reassociation width to 4 and vector to 2.
4463 (ix86_reassociation_width): Rewrite using cost table; special case
4464 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
4465 and TARGET_AVX128_OPTIMAL.
4466 * config/i386/i386.h (processor_costs): Add
4467 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
4468 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
4469 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
4470 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
4471 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
4472 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
4474 2017-10-05 Nathan Sidwell <nathan@acm.org>
4476 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
4478 2017-10-05 Tamar Christina <tamar.christina@arm.com>
4480 * config/arm/arm.c (arm_test_fpu_data): New.
4481 (arm_run_selftests): Call arm_test_fpu_data.
4483 2017-10-04 Nathan Sidwell <nathan@acm.org>
4485 * toplev.c (toplev::main): Remove excess parens on pretty_printer
4487 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
4489 2017-10-04 Sudakshina Das <sudi.das@arm.com>
4491 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
4492 check type for aarch64_simd_valid_immediate.
4493 (aarch64_output_simd_mov_immediate): Update prototype.
4494 (aarch64_simd_valid_immediate): Update prototype.
4495 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
4496 support for ORR-immediate.
4497 (and<mode>3): modified pattern to add support for BIC-immediate.
4498 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
4499 now checks for valid immediate for BIC and ORR based on new enum
4501 (aarch64_output_simd_mov_immediate): Function now used to output
4502 BIC/ORR imm as well based on new enum argument.
4503 * config/aarch64/constraints.md (Do): New vector immediate constraint.
4505 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
4506 (aarch64_reg_or_bic_imm): Likewise.
4508 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4510 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
4511 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
4513 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
4517 PR rtl-optimization/82396
4518 * haifa-sched.c (autopref_multipass_init): Simplify
4520 (autopref_rank_data): Simplify sort order.
4521 * sched-int.h (autopref_multipass_data_): Remove
4522 multi_mem_insn_p, min_offset and max_offset.
4524 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4526 * doc/sourcebuild.texi: Document vect_peeling_profitable.
4528 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4530 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
4533 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4535 * doc/sourcebuild.texi: Document vect_long_mult.
4537 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
4539 PR tree-optimization/82413
4540 * fold-const.c (build_range_check): Use widest_int when comparing
4541 the maximum ETYPE value with HIGH.
4543 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
4545 PR rtl-optimization/82396
4546 * haifa-sched.c (autopref_multipass_init): Simplify
4548 (autopref_rank_data): Simplify sort order.
4549 * sched-int.h (autopref_multipass_data_): Remove
4550 multi_mem_insn_p, min_offset and max_offset.
4552 2017-10-04 Jakub Jelinek <jakub@redhat.com>
4554 PR tree-optimization/82381
4555 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
4556 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
4559 PR tree-optimization/82374
4560 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
4561 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
4562 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
4563 current_function_decl to the new decl.
4565 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
4567 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
4568 helper macro for IEEE float128 hardware built-in functions.
4569 (SQRTF128_ODD): Add built-in functions with the round-to-odd
4571 (TRUNCF128_ODD): Likewise.
4572 (ADDF128_ODD): Likewise.
4573 (SUBF128_ODD): Likewise.
4574 (MULF128_ODD): Likewise.
4575 (DIVF128_ODD): Likewise.
4576 (FMAF128_ODD): Likewise.
4577 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
4578 UNSPEC_TRUNC_ROUND_TO_ODD.
4579 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
4580 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
4581 floating point round to odd instructions.
4582 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
4583 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
4584 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
4585 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
4586 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
4587 (trunc<mode>sf2_hw): Change the truncate with round to odd
4588 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
4589 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
4590 to odd hardware instructions.
4591 (sub<mode>3_odd): Likewise.
4592 (mul<mode>3_odd): Likewise.
4593 (div<mode>3_odd): Likewise.
4594 (sqrt<mode>2_odd): Likewise.
4595 (fma<mode>4_odd): Likewise.
4596 (fms<mode>4_odd): Likewise.
4597 (nfma<mode>4_odd): Likewise.
4598 (nfms<mode>4_odd): Likewise.
4599 (trunc<mode>df2_odd): Change the truncate with round to odd
4600 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
4602 * doc/extend.texi (PowerPC built-in functions): Update documentation
4603 for existing IEEE float128-bit built-in functions. Add built-in
4604 functions that generate the IEEE 128-bit floating point round to
4607 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
4609 PR rtl-optimization/77729
4610 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
4611 to (X&(C1&~C2))|C2 transformations.
4613 2017-10-03 Martin Jambor <mjambor@suse.cz>
4615 PR tree-optimization/82363
4616 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
4617 mismatch, mark lacc written regardless of racc.
4619 2017-10-03 Jakub Jelinek <jakub@redhat.com>
4621 PR tree-optimization/82381
4622 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
4623 stmt_to_insert nor wheather SSA_NAMEs are default defs.
4624 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
4625 fallthrough into reassoc_stmt_dominates_stmt_p.
4628 * combine.c (combine_instructions): Don't combine in unreachable
4631 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
4634 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
4635 function to not use the have_cpu variable. Do not set cpu_index,
4636 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
4638 (rs6000_valid_attribute_p): Remove duplicate initializations of
4639 old_optimize and func_optimize.
4640 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
4641 (rs6000_activate_target_options): Make global.
4642 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
4645 2017-10-02 Jakub Jelinek <jakub@redhat.com>
4647 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
4648 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
4649 Set *poffset to 0 and *psize and *pmax_size to -1 if
4650 *poffset + *psize overflows in HOST_WIDE_INT.
4652 PR tree-optimization/82387
4653 PR tree-optimization/82388
4654 PR tree-optimization/82389
4655 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
4656 instead of live_bytes non-NULL.
4658 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
4661 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
4664 2017-10-02 Richard Biener <rguenther@suse.de>
4666 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
4667 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
4668 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
4670 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
4672 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
4673 requested precision matches the type's.
4674 * calls.c (alloc_max_size): Calculate the new candidate size as
4675 a widest_int and use wi::to_widest when comparing it with the
4676 current candidate size.
4677 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
4678 zero rather than integer_zero_node.
4679 * match.pd: Check for a no-op conversion before using wi::add
4680 rather than after. Use tree_to_uhwi when summing small shift
4681 counts into an unsigned int.
4683 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
4684 Alan Hayward <alan.hayward@arm.com>
4685 David Sherwood <david.sherwood@arm.com>
4688 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
4689 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
4690 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
4691 POINTER_AND_FP_REGS.
4693 2017-10-02 Richard Biener <rguenther@suse.de>
4695 PR tree-optimization/82355
4696 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
4697 a mapping for the enclosing loop but avoid generating one for
4699 (copy_bb_and_scalar_dependences): Remove premature codegen
4700 error on PHIs in blocks duplicated into multiple places.
4701 * graphite-scop-detection.c
4702 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
4703 in the region use it as loop and nest to analyze the DR in.
4704 (try_generate_gimple_bb): Likewise.
4705 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
4706 (add_loop_constraints): For blocks in a loop not in the region
4707 create a dimension with a single iteration.
4708 * sese.h (gbb_loop_at_index): Remove assert.
4710 2017-10-01 Kevin Buettner <kevinb@redhat.com>
4712 * omp-expand.c (adjust_context_scope): New function.
4713 (expand_parallel_call): Call adjust_context_scope.
4715 2017-10-01 Jeff Law <law@redhat.com>
4717 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
4718 dom_opt_dom_walker class with direct access to private members.
4719 Add comments. Call test_for_singularity.
4720 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
4721 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
4722 m_dummy_cond anymore.
4723 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
4725 (pass_dominator:execute): Build the dummy_cond here and pass it
4726 to the dom_opt_dom_walker ctor.
4727 (test_for_singularity): New function.
4729 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
4730 Maya Rashish <coypu@sdf.org>
4732 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
4733 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
4734 (alpha*-*-netbsd*) Use nbsd_tm_file.
4735 (arm*-*-netbsdelf*) Likewise.
4736 (i[34567]86-*-netbsdelf*) Likewise.
4737 (x86_64-*-netbsd*) Likewise.
4738 (mips*-*-netbsd*) Likewise.
4739 (powerpc-*-netbsd*) Likewise.
4740 (sh*-*-netbsd*) Likewise.
4741 (sparc-*-netbsdelf*) Likewise.
4742 (sparc64-*-netbsd*) Likewise.
4743 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
4745 (vax-*-netbsdelf*) Likewise.
4746 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
4747 (UINT_FAST8_TYPE) Likewise.
4748 (INT_FAST16_TYPE) Check CHAR_FAST16.
4749 (UINT_FAST16_TYPE) Likewise.
4751 2017-09-30 Jakub Jelinek <jakub@redhat.com>
4754 * config/i386/i386.md
4755 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
4756 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
4757 *divmodsi4_zext_2): New define_insn_and_split.
4758 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
4759 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
4760 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
4761 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
4762 New define_insn_and_split.
4763 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
4764 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
4765 operands[1] having DImode when mode is SImode.
4767 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
4768 always SImode for DIV and MOD in REG_EQUAL notes.
4770 2017-09-29 Yury Gribov <tetra2005@gmail.com>
4773 * match.pd: Fix handling of NaNs in pattern.
4775 2017-09-29 Jeff Law <law@redhat.com>
4777 * sbitmap.c (bitmap_bit_in_range_p): New function.
4778 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
4779 * tree-ssa-dse.c (live_bytes_read): New function.
4780 (dse_classify_store): Ignore reads of dead bytes.
4782 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
4783 typos and whitespace errors.
4784 * config/i386/predicates.md (address_no_seg_operand): Likewise.
4785 * config/s390/s390.c (s390_emit_prologue): Likewise.
4787 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
4790 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
4791 with a symbol for LRA.
4793 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
4795 PR rtl-optimization/82338
4796 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
4798 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
4800 * genmodes.c (calc_wider_mode): Suppress qsort macro.
4801 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
4802 (qsort_chk): Declare.
4803 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
4804 (qsort_chk): New function.
4806 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4808 PR tree-optimization/82337
4809 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
4810 phi definition if the PHI result appears in an abnormal PHI.
4811 (find_basis_for_base_expr): Don't record a basis if the LHS of the
4812 basis appears in an abnormal PHI.
4814 2017-09-29 Richard Biener <rguenther@suse.de>
4816 * graphite-isl-ast-to-gimple.c
4817 (translate_isl_ast_to_gimple::set_codegen_error): New function.
4818 (binary_op_to_tree): Use it.
4819 (get_rename_from_scev): Likewise.
4820 (copy_loop_phi_nodes): Likewise.
4821 (copy_bb_and_scalar_dependences): Likewise.
4822 (translate_pending_phi_nodes): Likewise.
4824 2017-09-29 Jakub Jelinek <jakub@redhat.com>
4827 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
4828 for movabsq $(i32 << shift), r64.
4830 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
4832 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
4833 index when encoding %esp as %rsp to avoid 0x67 prefix.
4835 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
4837 * config/i386/i386.md (*movsf_internal, *movdf_internal):
4838 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
4840 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
4842 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
4843 Extensions with more than 16 double VFP registers.
4844 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
4845 to_clear_mask and all code related to it. Replace the remaining
4846 entry by a sbitmap and adapt code accordingly.
4848 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
4850 * brig-builtins.def: Change pure attributes to const.
4852 2017-09-28 Joseph Myers <joseph@codesourcery.com>
4854 * config.gcc (default_gnu_indirect_function): Default to yes for
4855 sparc*-*-linux* with glibc.
4857 2017-09-28 Joseph Myers <joseph@codesourcery.com>
4859 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
4860 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
4861 when creating .init_array and .fini_array sections with priority
4864 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
4867 * config/aarch64/aarch64.c
4868 (aarch64_builtin_support_vector_misalignment): Always return false
4869 when misalignment is unknown.
4871 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
4873 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
4874 this function to return false if the definition used by the swap
4875 instruction is artificial, or if the memory address from which the
4876 constant value is loaded is not represented by a base address held
4877 in a register or if the base address register is a frame or stack
4878 pointer. Additionally, return false if the base address of the
4879 loaded constant is a SYMBOL_REF but is not considered to be a
4881 (replace_swapped_load_constant): New function.
4882 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
4883 loaded constant vector with a load of a swapped constant vector.
4885 2017-09-27 Carl Love <cel@us.ibm.com>
4887 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
4888 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
4889 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
4892 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
4894 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
4895 first, always call autopref_rank_data otherwise.
4897 2017-09-27 Richard Biener <rguenther@suse.de>
4899 * graphite-scop-detection.c (find_scop_parameters): Move
4900 loop bound handling ...
4901 (gather_bbs::before_dom_children): ... here, avoiding the need
4902 to build scop_info->loop_nest.
4903 (record_loop_in_sese): Remove.
4904 * sese.h (sese_info_t::loop_nest): Remove.
4905 * sese.c (new_sese_info): Do not allocate loop_nest.
4906 (free_sese_info): Do not free loop_nest.
4908 2017-09-27 Jakub Jelinek <jakub@redhat.com>
4911 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
4912 lhs from calls if the lhs has addressable type.
4914 2017-09-27 Richard Biener <rguenther@suse.de>
4916 * graphite.h (scop::max_alias_set): New member.
4917 * graphite-scop-detection.c: Remove references to non-existing
4918 --param in comments.
4919 (build_alias_sets): Record the maximum alias set used for drs.
4920 (build_scops): Support zero as unlimited for
4921 --param graphite-max-arrays-per-scop.
4922 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
4924 (build_poly_sr_1): ... here. Compute alias set based on the
4925 maximum alias set used for drs rather than
4926 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
4928 2017-09-27 Richard Biener <rguenther@suse.de>
4930 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
4931 --param loop-block-tile-size=0 to disable tiling.
4933 2017-09-27 Richard Biener <rguenther@suse.de>
4935 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
4936 (graphite-max-nb-scop-params): Document special value zero.
4937 * domwalk.h (dom_walker::STOP): New symbolical constant.
4938 (dom_walker::dom_walker): Add optional parameter for bb to
4940 (dom_walker::~dom_walker): Declare.
4941 (dom_walker::before_dom_children): Document STOP return value.
4942 (dom_walker::m_user_bb_to_rpo): New member.
4943 (dom_walker::m_bb_to_rpo): Likewise.
4944 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
4945 mapping here if not provided by the user.
4946 (dom_walker::~dom_walker): Free bb to RPO mapping if not
4947 provided by the user.
4948 (dom_walker::STOP): Define.
4949 (dom_walker::walk): Do not compute bb to RPO mapping here.
4950 Support STOP return value from before_dom_children to stop
4952 * graphite-optimize-isl.c (optimize_isl): If the schedule
4953 is the same still generate code if -fgraphite-identity
4954 or -floop-parallelize-all are given.
4955 * graphite-scop-detection.c: Include cfganal.h.
4956 (gather_bbs::gather_bbs): Get and pass through bb to RPO
4958 (gather_bbs::before_dom_children): Return STOP for BBs
4960 (build_scops): Compute bb to RPO mapping and pass it to
4961 the domwalk. Treat --param graphite-max-nb-scop-params=0
4962 as not limiting the number of params.
4963 * graphite.c (graphite_initialize): Remove limit on the
4964 number of basic-blocks in a function.
4965 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
4966 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
4967 default value of 10.
4969 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
4971 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
4972 Adjust code to eliminate needing to do the shift right 32-bits
4973 operation after XSCVDPSPN.
4975 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4977 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
4978 ((X / Y) != 0 -> X >= Y): Likewise.
4980 2017-09-26 Carl Love <cel@us.ibm.com>
4982 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
4983 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
4984 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
4985 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
4986 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
4987 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
4988 definitions and overloading.
4989 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
4990 statement for P9V_BUILTIN_XST_LEN_R.
4991 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
4992 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
4993 define_expand and define_insn for the instructions and builtins.
4994 * doc/extend.texi: Update the built-in documentation file for the new
4996 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
4997 define_insn for the instructions
4999 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
5002 * gcc/config/netbsd-protos.h: New file.
5003 * gcc/config/netbsd.c: New file.
5004 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
5005 * gcc/config/t-netbsd: New file.
5006 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
5007 (tmake_file) Add t-netbsd.
5008 (extra_objs) Add netbsd.o.
5010 2017-09-26 Janus Weil <janus@gcc.gnu.org>
5014 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
5016 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5018 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
5019 sign extension from a vector register to a GPR by doing a 32-bit
5020 direct move and then an EXTSW.
5021 (extendsi<mode>2 splitter): Likewise.
5022 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
5023 right or vector extract after doing XSCVDPSPN. Use
5024 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
5026 (movdi_from_sf_zero_ext): Likewise.
5027 (reload_gpr_from_vsxsf): Likewise.
5028 (p8_mfvsrd_4_disf): Delete, no longer used.
5029 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
5030 then needing to move the SFmode to a GPR to use the XSCVDPSP
5031 instruction instead of FRSP and XSCVDPSPN.
5032 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
5033 it is adjacent to the other XSCVSPDP insns.
5034 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
5035 SFmode to be in traditional Altivec registers.
5036 (vsx_xscvdpspn): Eliminate useless alternative constraint.
5037 (vsx_xscvspdpn): Likewise.
5038 (vsx_xscvspdpn_scalar): Likewise.
5040 2017-09-26 Martin Jambor <mjambor@suse.cz>
5042 * tree-sra.c (compare_access_positions): Put integral types first,
5043 stabilize sorting of integral types, remove conditions putting
5044 non-full-precision integers last.
5045 (sort_and_splice_var_accesses): Disable scalarization if a
5046 non-integert would be represented by a non-full-precision integer.
5048 2017-09-26 Joseph Myers <joseph@codesourcery.com>
5050 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
5051 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
5052 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
5053 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
5054 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
5055 conditionals inside the function instead of around it. Call
5056 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
5057 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
5059 2017-09-26 Richard Biener <rguenther@suse.de>
5061 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
5063 (scop_detection::build_scop_breadth): ... this. Removed.
5064 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
5065 (scop_detection::harmful_stmt_in_bb): Likewise.
5066 (scop_detection::graphite_can_represent_stmt): Likewise.
5067 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
5068 (scop_detection::can_represent_loop): Remove recursion, fold in ...
5069 (scop_detection::can_represent_loop_1): ... this. Removed.
5070 (scop_detection::harmful_loop_in_region): Simplify after inlining
5071 the above and remove more quadraticness.
5072 (build_scops): Adjust.
5073 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
5076 2017-09-26 Jakub Jelinek <jakub@redhat.com>
5079 * config/i386/i386.c (ix86_print_operand_address_as): Only test
5080 REGNO (base) == SP_REG if base is a REG.
5083 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
5084 if it is different SSA_NAME.
5085 (optimize_range_tests_cmp_bitwise): New function.
5086 (optimize_range_tests): Call it.
5088 2017-09-26 Richard Biener <rguenther@suse.de>
5090 PR tree-optimization/82321
5091 * graphite.c (canonicalize_loop_closed_ssa): Properly check
5092 for the def being inside the loop.
5094 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5096 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
5098 * config/s390/s390-builtins.def: Fix constraint on op4.
5100 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5102 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
5103 independent expanders.
5104 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
5105 ("vec_ordered", "vec_unordered"): New expanders.
5107 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5109 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
5112 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5114 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
5115 vec_unpacks_lo_v16qi.
5116 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
5118 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5120 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
5121 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
5122 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
5124 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5126 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
5128 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
5130 ("*vec_slb<mode>"): New insn pattern.
5131 ("vec_shr_<mode>"): New expander.
5132 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
5133 and force the shift count operand to V16QImode.
5134 ("vec_srb<mode>"): Set shift count mode to V16QI.
5136 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5138 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
5139 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
5140 ("vec_widen_smult_hi_<mode>"): New expander definitions.
5142 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
5145 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
5147 2017-09-26 Richard Biener <rguenther@suse.de>
5149 PR tree-optimization/82320
5150 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
5153 2017-09-25 Jeff Law <law@redhat.com>
5155 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
5156 prototype for new argument.
5157 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
5158 mostly extracted from rs6000_emit_allocate_stack.
5159 (rs6000_emit_probe_stack_range_stack_clash): New function.
5160 (rs6000_emit_allocate_stack): Call
5161 rs6000_emit_probe_stack_range_stack_clash as needed.
5162 (rs6000_emit_probe_stack_range): Add additional argument
5163 to call to gen_probe_stack_range{si,di}.
5164 (output_probe_stack_range): New.
5165 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
5166 (output_probe_stack_range_stack_clash): New.
5167 (rs6000_emit_prologue): Emit notes into dump file as requested.
5168 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
5169 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
5170 Add additional operand and pass it to output_probe_stack_range.
5172 2017-09-25 Bin Cheng <bin.cheng@arm.com>
5174 PR tree-optimization/82163
5175 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
5176 (checking_verify_loop_closed_ssa): New parameter.
5177 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
5178 (check_loop_closed_ssa_stmt): Delete.
5179 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
5180 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
5181 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
5184 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
5186 * brig-builtins.def: Treat HSAIL barrier builtins as
5187 setjmp/longjump style functions.
5189 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
5191 * target.def (constant_alignment): New hook.
5192 * defaults.h (CONSTANT_ALIGNMENT): Delete.
5193 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
5194 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
5195 * doc/tm.texi: Regenerate.
5196 * targhooks.h (default_constant_alignment): Declare.
5197 (constant_alignment_word_strings): Likewise.
5198 * targhooks.c (default_constant_alignment): New function.
5199 (constant_alignment_word_strings): Likewise.
5200 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
5201 instead of CONSTANT_ALIGNMENT.
5202 * varasm.c (align_variable, get_variable_align, build_constant_desc)
5203 (force_const_mem): Likewise.
5204 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
5205 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
5206 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
5207 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5208 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
5210 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
5211 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5212 constant_alignment_word_strings.
5213 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
5214 (CONSTANT_ALIGNMENT): Likewise.
5215 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5216 (arm_constant_alignment): New function.
5217 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
5218 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5219 constant_alignment_word_strings.
5220 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
5221 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5222 constant_alignment_word_strings.
5223 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
5224 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5225 (cris_constant_alignment): New function.
5226 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
5227 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5228 (epiphany_constant_alignment): New function.
5229 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
5230 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5231 constant_alignment_word_strings.
5232 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
5233 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5234 constant_alignment_word_strings.
5235 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
5236 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5237 constant_alignment_word_strings.
5238 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
5239 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
5240 * config/i386/i386.c (ix86_constant_alignment): Make static.
5241 Use the same interface as the target hook.
5242 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5243 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
5244 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5245 constant_alignment_word_strings.
5246 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
5247 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
5248 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5249 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
5250 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5251 constant_alignment_word_strings.
5252 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
5253 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5254 constant_alignment_word_strings.
5255 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
5256 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5257 constant_alignment_word_strings.
5258 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
5259 * config/microblaze/microblaze.c (microblaze_constant_alignment):
5261 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5262 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
5263 * config/mips/mips.c (mips_constant_alignment): New function.
5264 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5265 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
5266 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
5267 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5268 (mmix_constant_alignment): Make static. Use the same interface
5270 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
5271 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5272 constant_alignment_word_strings.
5273 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
5274 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5275 constant_alignment_word_strings.
5276 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
5277 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5278 constant_alignment_word_strings.
5279 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
5280 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5281 (rs6000_constant_alignment): New function.
5282 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
5283 * config/riscv/riscv.c (riscv_constant_alignment): New function.
5284 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5285 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
5286 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5287 (rs6000_constant_alignment): New function.
5288 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
5289 * config/s390/s390.c (s390_constant_alignment): New function.
5290 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5291 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
5292 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5293 constant_alignment_word_strings.
5294 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
5295 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5296 (sparc_constant_alignment): New function.
5297 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
5298 * config/spu/spu.c (spu_constant_alignment): New function.
5299 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5300 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
5301 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5302 constant_alignment_word_strings.
5303 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
5304 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5305 constant_alignment_word_strings.
5306 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
5307 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5308 constant_alignment_word_strings.
5309 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
5310 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5311 (visium_constant_alignment): New function.
5312 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
5313 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5314 (xtensa_constant_alignment): New function.
5315 * system.h (CONSTANT_ALIGNMENT): Poison.
5317 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
5319 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
5320 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
5321 (rs6000_builtin_valid_without_lhs): New helper function.
5322 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5323 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
5325 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
5327 * target.h (vec_perm_indices): Use unsigned short rather than
5329 (auto_vec_perm_indices): Likewise.
5330 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
5331 Use unsigned int rather than unsigned char.
5332 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
5334 2017-09-25 Richard Biener <rguenther@suse.de>
5336 * cfgloop.h (sort_sibling_loops): Declare.
5337 * cfgloop.c (sort_sibling_loops_cmp): New helper.
5338 (sort_sibling_loops): New function sorting the sibling loop list
5340 * graphite.c (graphite_transform_loops): Sort sibling loops.
5342 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
5344 * target.def (vec_perm_const_ok): Change sel parameter to
5346 * optabs-query.c (can_vec_perm_p): Update accordingly.
5347 * doc/tm.texi: Regenerate.
5348 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
5349 auto_vec_perm_indices and remove separate nelt field.
5350 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
5351 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
5352 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
5353 (aarch64_expand_vec_perm_const): Update accordingly.
5354 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
5355 to vec_perm_indices.
5356 * config/arm/arm.c (expand_vec_perm_d): Change perm to
5357 auto_vec_perm_indices and remove separate nelt field.
5358 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
5359 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
5360 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
5362 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
5363 to vec_perm_indices.
5364 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
5365 sel to vec_perm_indices.
5366 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
5367 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
5368 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
5370 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
5373 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
5376 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
5377 on the FUNCTION_DECL function context if it has a DIE that is a
5380 2017-09-25 Richard Biener <rguenther@suse.de>
5382 PR tree-optimization/82285
5383 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
5386 2017-09-25 Tom de Vries <tom@codesourcery.com>
5390 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
5393 2017-09-25 Richard Biener <rguenther@suse.de>
5395 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
5396 ISL errors other than isl_error_quota happen. Dump if the
5397 schedule is the same.
5398 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
5399 errors instead of aborting inside ISL.
5401 2017-09-25 Iain Sandoe <iain@codesourcery.com>
5404 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
5405 of libgcc_eh for m64.
5406 * config/i386/darwin64.h: Likewise.
5408 2017-09-25 Richard Biener <rguenther@suse.de>
5411 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
5412 attribute for incomplete types nor twice for complete ones.
5414 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
5417 * config/i386/i386.c (ix86_print_operand_address_as): Encode
5418 %esp as %rsp to avoid 0x67 prefix if there is no index or base
5421 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
5424 * config/i386/i386.opt (mprefer-avx256): Use
5425 ix86_target_flags variable.
5426 * config/i386/i386.c (ix86_target_string): Move
5427 -mprefer-avx256 to flag2_opts.
5429 2017-09-22 Jakub Jelinek <jakub@redhat.com>
5432 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
5433 and x != -1 | y != -1 into (x & y) != -1.
5435 2017-09-22 Steve Ellcey <sellcey@cavium.com>
5437 * config.gcc: Add new case statement to set
5438 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
5439 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
5440 s390x-*-linux* case statements. Added aarch64 to the list of
5441 supported architectures.
5443 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
5445 PR tree-optimization/82289
5446 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
5447 STMT_VINFO_RELEVANT_P.
5449 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
5450 Alan Hayward <alan.hayward@arm.com>
5451 David Sherwood <david.sherwood@arm.com>
5453 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
5454 for VR_RANGE only; don't allow VR_ANTI_RANGE.
5455 (extract_range_from_binary_expr_1): Don't call
5456 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
5458 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
5459 Alan Hayward <alan.hayward@arm.com>
5460 David Sherwood <david.sherwood@arm.com>
5462 * target.def (preferred_vector_alignment): New hook.
5463 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
5465 * doc/tm.texi: Regenerate.
5466 * targhooks.h (default_preferred_vector_alignment): Declare.
5467 * targhooks.c (default_preferred_vector_alignment): New function.
5468 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
5470 (DR_TARGET_ALIGNMENT): New macro.
5471 (aligned_access_p): Update commentary.
5472 (vect_known_alignment_in_bytes): New function.
5473 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
5475 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
5476 Calculate the misalignment based on the target alignment rather than
5478 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
5479 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
5480 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
5481 the target alignment, rather than masking the element misalignment
5482 with the number of elements in a vector. Also use the target
5483 alignment when calculating the maximum number of peels.
5484 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
5485 instead of TYPE_ALIGN_UNIT.
5486 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
5487 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
5488 (vect_create_addr_base_for_vector_ref): Update call accordingly.
5489 (vect_create_data_ref_ptr): Likewise.
5490 (vect_setup_realignment): Realign by ANDing with
5491 -DR_TARGET_MISALIGNMENT.
5492 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
5493 the number of peels based on DR_TARGET_ALIGNMENT.
5494 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
5495 with the guaranteed alignment boundary when deciding whether
5497 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
5498 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
5499 (ensure_base_align): Remove stmt_info parameter. Get the
5500 target base alignment from DR_TARGET_ALIGNMENT.
5501 (vectorizable_store): Update call accordingly. Interpret
5502 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
5504 (vectorizable_load): Likewise.
5506 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
5507 Alan Hayward <alan.hayward@arm.com>
5508 David Sherwood <david.sherwood@arm.com>
5510 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
5511 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
5512 (vect_enhance_data_refs_alignment): Likewise.
5514 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
5516 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
5517 error. Only quit immediately if parsing is complete.
5518 (BEGIN): Initialize fatal_err and parse_done.
5519 (begin fpu, end fpu): Check number of arguments.
5520 (begin arch, end arch): Likewise.
5521 (begin cpu, end cpu): Likewise.
5522 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
5523 (optalias): Likewise.
5525 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
5527 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
5528 * config/arm/arm-isa.h: Delete. Move definitions to ...
5529 * arm-cpus.in: ... here. Use new feature and fgroup values.
5530 * config/arm/arm.c (arm_option_override): Use lower case for feature
5532 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
5533 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
5534 * config/arm/parsecpu.awk (END): Add new command 'isa'.
5536 (print_isa_bits_for): New function.
5537 (gen_isa): New function.
5538 (gen_comm_data): Use print_isa_bits_for.
5539 (define feature): New keyword.
5540 (define fgroup): New keyword.
5541 * config/arm/t-arm (TM_H): Remove.
5542 (GTM_H): Add arm-isa.h.
5543 (arm-isa.h): Add rule to generate file.
5544 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
5545 case for feature bit names.
5547 2017-09-22 Richard Biener <rguenther@suse.de>
5549 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
5551 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
5552 print of no dependency loops ...
5553 * graphite.c (graphite_transform_loops): ... here.
5554 (canonicalize_loop_closed_ssa_form): Work from inner to outer
5556 (same_close_phi_node, remove_duplicate_close_phi,
5557 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
5558 (canonicalize_loop_closed_ssa): ... here and simplify.
5559 * graphite-optimize-isl.c: Include tree-vectorizer.h.
5560 (optimize_isl): Use dump_printf_loc to tell when we stopped
5561 optimizing because of an ISL timeout.
5563 2017-09-22 Richard Biener <rguenther@suse.de>
5565 PR tree-optimization/82291
5566 * tree-if-conv.c (predicate_mem_writes): Make sure to
5567 remove writes in blocks predicated with false.
5569 2017-09-22 Richard Biener <rguenther@suse.de>
5571 * sese.c: Include cfganal.h.
5572 (if_region_set_false_region): Remove.
5573 (create_if_region_on_edge): Likewise.
5574 (move_sese_in_condition): Re-implement without destroying
5577 2017-09-22 Richard Biener <rguenther@suse.de>
5579 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
5580 Verify both BBs contain loop PHI nodes before dispatching to
5582 (graphite_regenerate_ast_isl): Do not recompute dominators,
5583 do not verify three times. Restructure for clarity.
5584 * graphite-scop-detection.c (same_close_phi_node,
5585 remove_duplicate_close_phi, make_close_phi_nodes_unique,
5586 defined_in_loop_p, canonicalize_loop_closed_ssa,
5587 canonicalize_loop_closed_ssa_form): Simplify, remove excess
5588 checking and SSA rewrite, move to ...
5589 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
5590 (graphite_initialize): Do not pass in ctx, do not reset the
5591 SCEV cache, compute only dominators.
5592 (graphite_transform_loops): Allocate ISL ctx after
5593 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
5594 Maintain post-dominators only around build_scops.
5595 * sese.c (if_region_set_false_region): Make static. Free
5596 and recompute dominators.
5597 (move_sese_in_condition): Assert we don't get called with
5598 post-dominators computed.
5599 * sese.h (if_region_set_false_region): Remove.
5601 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
5603 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
5604 mode attribute for TARGET_AVX512VL.
5606 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
5608 * config/i386/i386.opt (mprefer-avx256): New option.
5609 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
5611 (ix86_preferred_simd_mode): Return 256-bit AVX modes
5612 for TARGET_PREFER_AVX256.
5613 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
5615 2017-09-21 Jeff Law <law@redhat.com>
5617 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
5618 Fix dump output if the only stack space is for pushed registers.
5620 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5622 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
5625 2017-09-21 Martin Sebor <msebor@redhat.com>
5628 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
5629 code (in C++) or code that triggers warnings.
5631 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
5633 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
5635 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
5637 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
5638 * sched-rgn.c: Ditto.
5639 * sel-sched-ir.c: Ditto.
5641 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
5643 * toplev.h (set_random_seed): Adjust return type.
5644 * toplev.c (init_local_tick): Move eager initialization of random_seed
5645 to get_random_seed. Adjust comment.
5646 (init_random_seed): Inline to get_random_seed, delete.
5647 (get_random_seed): Initialize random_seed lazily.
5648 (set_random_seed): Do not return previous value.
5649 (print_switch_value): Do not call get_random_seed.
5651 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
5653 * cgraph.c (delete_function_version): New, broken out from...
5654 (cgraph_node::delete_function_version): ...here. Rename to
5655 cgraph_node::delete_function_version_by_decl. Update all uses.
5656 (cgraph_node::remove): Call delete_function_version.
5658 2017-09-21 Jakub Jelinek <jakub@redhat.com>
5661 * tree-inline.c (expand_call_inline): Emit clobber stmts for
5662 VAR_DECLs to which addressable non-volatile parameters are mapped
5663 and for id->retvar after the return value assignment. Clear
5664 id->retval and id->retbnd after inlining.
5666 2017-09-21 Richard Biener <rguenther@suse.de>
5668 PR tree-optimization/82276
5669 PR tree-optimization/82244
5670 * tree-vrp.c (build_assert_expr_for): Set
5671 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
5673 (remove_range_assertions): Revert earlier change.
5675 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
5678 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
5680 2017-09-21 Richard Biener <rguenther@suse.de>
5682 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5683 Restore valid IL after code generation errors.
5684 * graphite.c (graphite_transform_loops): Diagnose code
5685 generation issues as MSG_MISSED_OPTIMIZATION and continue
5686 with processing SCOPs.
5688 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5689 Alan Hayward <alan.hayward@arm.com>
5690 David Sherwood <david.sherwood@arm.com>
5692 * calls.c (compute_argument_addresses): Use simplify_gen_binary
5693 rather than choosing between plus_constant and gen_rtx_<CODE>.
5694 * expr.c (emit_push_insn): Likewise.
5695 (expand_expr_real_2): Likewise.
5697 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5698 Alan Hayward <alan.hayward@arm.com>
5699 David Sherwood <david.sherwood@arm.com>
5701 * loop-unroll.c (split_iv): Call copy_rtx on the step.
5703 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5704 Alan Hayward <alan.hayward@arm.com>
5705 David Sherwood <david.sherwood@arm.com>
5707 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
5708 calling tree_to_uhwi.
5710 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5711 Alan Hayward <alan.hayward@arm.com>
5712 David Sherwood <david.sherwood@arm.com>
5714 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
5715 INTEGER_CST rather than a negative test for ADDR_EXPR.
5717 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5718 Alan Hayward <alan.hayward@arm.com>
5719 David Sherwood <david.sherwood@arm.com>
5721 * tree-vrp.c (extract_range_from_binary_expr_1): Check
5722 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
5724 2017-09-21 Richard Biener <rguenther@suse.de>
5726 PR tree-optimization/71351
5727 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
5728 graphite_create_new_loop_guard): Remove, fold remaining parts
5730 (translate_isl_ast_node_for): ... here and simplify.
5732 2017-09-21 Jakub Jelinek <jakub@redhat.com>
5735 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
5736 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
5737 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
5738 alternative always use QI mode, for -Os imov (=R,R) alternative
5739 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
5742 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5743 Jeff Law <law@redhat.com>
5745 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
5746 (allocate_stack_space): New function, partially extracted from
5748 (s390_emit_prologue): Track offset to most recent stack probe.
5749 Code to allocate space moved into allocate_stack_space.
5750 Dump actions when no stack is allocated.
5751 (s390_prologue_plus_offset): New function.
5752 (s390_emit_stack_probe): Likewise.
5754 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
5756 * common.opt (Wa, Wl, Wp, g, gz=): Add
5758 (gno-column-info): Remove.
5759 (gcolumn-info): Drop RejectNegative.
5761 (gno-record-gcc-switches): Remove.
5762 (grecord-gcc-switches): Drop RejectNegative.
5763 (gno-split-dwarf): Remove.
5764 (gsplit-dwarf): Drop RejectNegative.
5765 (gno-strict-dwarf): Remove.
5766 (gstrict-dwarf): Drop RejectNegative.
5767 * config/darwin.opt (gfull, gused): Add RejectNegative.
5768 * dwarf2out.c (gen_producer_string): Drop
5769 gno-record-gcc-switches handler.
5770 * optc-gen.awk: Add g to prefixes with negative forms.
5771 * opts-common.c (remapping_prefix_p): New.
5772 (find_opt): Check it.
5773 (generate_canonical_option): Test g prefix.
5774 (option_map): Add -gno- mapping.
5775 (add_misspelling_candidates): Check remapping_prefix_p.
5777 2017-09-20 Jeff Law <law@redhat.com>
5779 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
5780 thinko in stack clash protection support.
5782 * explow.c (compute_stack_clash_protection_loop_data): Use
5783 CONST_INT_P instead of explicit test. Verify object is a
5784 CONST_INT_P before looking at INTVAL.
5785 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
5786 instead of explicit test.
5788 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
5791 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
5792 address instead of to r1 and r11.
5794 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
5796 * config.gcc: Support "knm".
5797 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
5798 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
5800 * config/i386/i386.c (m_KNM): Define.
5801 (processor_target_table): Add "knm".
5803 (ix86_option_override_internal): Add "knm".
5804 (ix86_issue_rate): Add PROCESSOR_KNM.
5805 (ix86_adjust_cost): Ditto.
5806 (ia32_multipass_dfa_lookahead): Ditto.
5807 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
5808 (fold_builtin_cpu): Add M_INTEL_KNM.
5809 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
5810 (processor_type): Add PROCESSOR_KNM.
5811 * config/i386/x86-tune.def: Add m_KNM.
5812 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
5814 2017-09-20 Richard Biener <rguenther@suse.de>
5816 PR tree-optimization/80213
5817 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
5818 are allowed in empty BBs as well.
5819 (canonicalize_loop_closed_ssa): Also look for other complex
5821 (scop_detection::get_sese): Include the loop-closed PHI block
5823 (scop_detection::merge_sese): Remove code adding extra blocks.
5824 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
5825 (build_scops): Assert the final returned scop is invalid.
5827 2017-09-20 Richard Biener <rguenther@suse.de>
5829 PR tree-optimization/82264
5830 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
5832 (vn_phi_lookup): Likewise.
5833 (vn_phi_insert): Likewise.
5835 2017-09-20 Jakub Jelinek <jakub@redhat.com>
5837 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
5838 that fits into uhwi or shwi, add DW_AT_const_value regardless
5839 of early_dwarf without going through RTL, using add_AT_unsigned
5842 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
5843 (DEBUG_LTO_ABBREV_SECTION): Likewise.
5844 (DEBUG_LTO_MACINFO_SECTION): Likewise.
5845 (DEBUG_MACRO_SECTION): Likewise.
5846 (DEBUG_LTO_MACRO_SECTION): Likewise.
5847 (DEBUG_STR_DWO_SECTION): Likewise.
5848 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
5849 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
5850 (DEBUG_LTO_DWO_LINE_SECTION): Define.
5851 (DEBUG_LTO_LINE_STR_SECTION): Define.
5852 (init_sections_and_labels): Initialize debug_line_str_section
5853 variable. Initialize debug_loc_section for -gdwarf-5 to
5854 DEBUG_LOCLISTS_SECTION. Formatting fixes.
5856 2017-09-20 Richard Biener <rguenther@suse.de>
5858 * graphite-sese-to-poly.c (extract_affine): Properly handle
5859 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
5861 2017-09-20 Richard Biener <rguenther@suse.de>
5863 PR tree-optimization/81373
5864 * graphite-scop-detection.c (build_cross_bb_scalars_def):
5865 Force SESE live-out defs to be handled even if they are
5868 2017-09-19 Jeff Law <law@redhat.com>
5870 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
5871 nothing for stack adjustments with REG_STACK_CHECK.
5872 * sched-deps.c (parse_add_or_inc): Reject insns with
5873 REG_STACK_CHECK from dependency breaking.
5874 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
5875 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
5876 * reg-notes.def (STACK_CHECK): New note.
5878 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
5879 (ix86_expand_prologue): Dump stack clash info as needed.
5880 Call ix86_adjust_stack_and_probe_stack_clash as needed.
5882 * function.c (dump_stack_clash_frame_info): New function.
5883 * function.h (dump_stack_clash_frame_info): Prototype.
5884 (enum stack_clash_probes): New enum.
5886 * config/alpha/alpha.c (alpha_expand_prologue): Also check
5887 flag_stack_clash_protection.
5888 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
5889 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
5890 (arm_frame_pointer_required): Likewise.
5891 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
5892 (ia64_expand_prologue): Likewise.
5893 * config/mips/mips.c (mips_expand_prologue): Likewise.
5894 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
5895 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
5896 (sparc_flat_expand_prologue): Likewise.
5897 * config/spu/spu.c (spu_expand_prologue): Likewise.
5899 * explow.c: Include "params.h".
5900 (anti_adjust_stack_and_probe_stack_clash): New function.
5901 (get_stack_check_protect): Likewise.
5902 (compute_stack_clash_protection_loop_data): Likewise.
5903 (emit_stack_clash_protection_loop_start): Likewise.
5904 (emit_stack_clash_protection_loop_end): Likewise.
5905 (allocate_dynamic_stack_space): Use get_stack_check_protect.
5906 Use anti_adjust_stack_and_probe_stack_clash.
5907 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
5908 (emit_stack_clash_protection_loop_start): Likewise.
5909 (emit_stack_clash_protection_loop_end): Likewise.
5910 * rtl.h (get_stack_check_protect): Prototype.
5911 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
5912 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
5913 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
5915 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
5917 * doc/tm.texi: Rebuilt.
5918 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
5919 get_stack_check_protect.
5920 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
5921 * config/arm/arm.c (arm_expand_prologue): Likewise.
5922 (arm_frame_pointer_required): Likewise.
5923 * config/i386/i386.c (ix86_expand_prologue): Likewise.
5924 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
5925 * config/mips/mips.c (mips_expand_prologue): Likewise.
5926 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
5927 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
5928 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
5929 (sparc_flat_expand_prologue): Likewise.
5931 * common.opt (-fstack-clash-protection): New option.
5932 * flag-types.h (enum stack_check_type): Note difference between
5933 -fstack-check= and -fstack-clash-protection.
5934 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
5935 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
5936 * toplev.c (process_options): Issue warnings/errors for cases
5937 not handled with -fstack-clash-protection.
5938 * doc/invoke.texi (-fstack-clash-protection): Document new option.
5939 (-fstack-check): Note additional problem with -fstack-check=generic.
5940 Note that -fstack-check is primarily for Ada and refer users
5941 to -fstack-clash-protection for stack-clash-protection.
5942 Document new params for stack clash protection.
5944 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
5946 * config/i386/i386.c (ix86_split_long_move): Do not handle
5947 address used for LEA in a special way.
5949 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
5951 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
5953 2017-09-19 Martin Sebor <msebor@redhat.com>
5956 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
5957 of incompatible types.
5959 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
5961 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
5962 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
5963 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5964 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
5966 2017-09-19 Richard Biener <rguenther@suse.de>
5968 PR tree-optimization/82244
5969 * tree-vrp.c (remove_range_assertions): Do not propagate
5970 a constant to abnormals but replace the assert with a copy.
5972 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
5974 PR rtl-optimization/57878
5975 PR rtl-optimization/68988
5976 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
5977 avoidance test involving non_reload_pseudos. Move frequency test
5978 below the general fragmentation avoidance test.
5980 2017-09-19 Richard Biener <rguenther@suse.de>
5982 PR tree-optimization/69728
5983 * graphite-sese-to-poly.c (schedule_error): New global.
5984 (add_loop_schedule): Handle empty domain by failing the
5986 (build_original_schedule): Handle schedule_error.
5988 2017-09-19 Richard Biener <rguenther@suse.de>
5990 * graphite-scop-detection.c (scop_detection::can_represent_loop):
5991 Do not iterate to sibling loops but only to siblings of inner
5994 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
5997 * config/m68k/m68k.md (moveq feeding equality comparison): Check
5998 that the registers are different.
6000 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
6002 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
6003 to processor_model and "amdfam17h" to arch_names_table.
6004 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
6006 2017-09-18 Jakub Jelinek <jakub@redhat.com>
6009 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
6011 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6012 Alan Hayward <alan.hayward@arm.com>
6013 David Sherwood <david.sherwood@arm.com>
6015 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
6017 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
6019 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
6020 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
6021 vect_schedule_slp_instance.
6022 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
6023 Update call to vect_slp_analyze_node_operations. Simplify return
6025 (vect_slp_analyze_bb_1): Update call accordingly.
6026 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
6027 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
6028 (vect_schedule_slp): Update call accordingly.
6030 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6031 Alan Hayward <alan.hayward@arm.com>
6032 David Sherwood <david.sherwood@arm.com>
6034 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
6035 with types that aren't in fact scalar.
6037 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6039 * tree-vect-slp.c (vect_record_max_nunits): New function,
6041 (vect_build_slp_tree_1): ...here.
6042 (vect_build_slp_tree_2): Call it for phis too.
6044 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6046 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
6047 to vect_get_vec_def_for_operand when getting the mask operand.
6049 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6050 Alan Hayward <alan.hayward@arm.com>
6051 David Sherwood <david.sherwood@arm.com>
6053 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
6056 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6057 Alan Hayward <alan.hayward@arm.com>
6058 David Sherwood <david.sherwood@arm.com>
6060 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
6061 calculation for vector booleans.
6063 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6064 Alan Hayward <alan.hayward@arm.com>
6065 David Sherwood <david.sherwood@arm.com>
6067 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
6069 (vect_transform_stmt): ...here.
6070 (vect_analyze_stmt): Use it instead of calling
6071 vectorizable_live_operation directly.
6073 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
6075 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
6076 non-SIMT targets in acc vector loops.
6078 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
6080 * configure.ac: Add arc and check if assembler supports gdwarf2.
6081 * configure: Regenerate.
6083 2017-09-18 Richard Biener <rguenther@suse.de>
6085 PR tree-optimization/82220
6086 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
6087 epilogue niters from the min_profitable_iters compute.
6089 2017-09-18 Jakub Jelinek <jakub@redhat.com>
6092 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
6093 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
6094 (ix86_init_pic_reg): Revert 2017-09-01 changes.
6096 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
6099 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
6100 switching to a new text section.
6102 2017-09-18 Richard Biener <rguenther@suse.de>
6104 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
6106 (build_alias_set): Reject aliases with no access function.
6108 2017-09-18 Richard Biener <rguenther@suse.de>
6110 PR tree-optimization/79622
6111 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
6113 (build_cross_bb_scalars_use): Likewise.
6115 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
6117 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
6119 2017-09-18 Alan Modra <amodra@gmail.com>
6122 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
6123 stack_pointer_rtx for count 0. Update comments. Break up
6124 large rtl expression.
6126 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
6128 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
6129 Increase to 20 bytes.
6130 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
6131 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
6132 or avx version of the stub.
6134 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
6137 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
6138 compute the minimum stack alignment. Also update preferred stack
6139 boundary for leaf functions.
6141 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
6143 PR tree-optimization/82228
6144 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
6147 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
6149 * common/config/nds32/nds32-common.c
6150 (nds32_option_optimization_table): Refine formatting.
6151 (nds32_option_optimization_table): Use -fsched-pressure and
6152 -fomit-frame-pointer for specific optimization level.
6154 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
6156 * config/nds32/nds32.c: Refine formatting and comments.
6157 * config/nds32/nds32.h: Likewise.
6158 * config/nds32/nds32.md: Likewise.
6159 * config/nds32/nds32-cost.c: Likewise.
6160 * config/nds32/nds32-isr.c: Likewise.
6161 * config/nds32/nds32-md-auxiliary.c: Likewise.
6162 * config/nds32/nds32-multiple.md: Likewise.
6163 * config/nds32/nds32-predicates.c: Likewise.
6165 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
6166 Jakub Jelinek <jakub@redhat.com>
6168 Add support for -std=c++2a.
6169 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
6171 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
6173 2017-09-15 Steve Ellcey <sellcey@cavium.com>
6176 * doc/extend.texi (Common Function Attributes): Add
6177 references to ARM, AArch64, and S/390 specific attributes.
6178 (Function Specific Option Pragmas): Add AArch64 and S/390
6179 to list of back ends that support the target pragma.
6181 2017-09-15 Nathan Sidwell <nathan@acm.org>
6183 * doc/standards.texi: Fix C++17 description. Update URLs for
6186 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
6188 * common.opt (Wcast-align=strict): New warning option.
6189 * doc/invoke.texi: Document -Wcast-align=strict.
6191 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
6193 * cgraph.h (cgraph_thunk_info): Add comments.
6194 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
6195 assert for VIRTUAL_* arguments stricter.
6197 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
6199 PR tree-optimization/71026
6200 * match.pd: Move RDIV patterns from fold-const.c
6201 * fold-const.c (distribute_real_division): Removed.
6202 (fold_binary_loc): Remove calls to distribute_real_divison.
6204 2017-09-15 Jakub Jelinek <jakub@redhat.com>
6206 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
6207 c++1z and gnu++1z as deprecated. Change other references to
6208 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
6209 Change -Wc++1z-compat to -Wc++17-compat.
6210 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
6211 * dwarf2out.c (highest_c_language): Handle C++17.
6212 (gen_compile_unit_die): Likewise.
6214 2017-09-15 Jakub Jelinek <jakub@redhat.com>
6216 PR rtl-optimization/82192
6217 * combine.c (make_extraction): Don't look through non-paradoxical
6218 SUBREGs or TRUNCATE if pos + len is or might be bigger than
6221 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6222 Alan Hayward <alan.hayward@arm.com>
6223 David Sherwood <david.sherwood@arm.com>
6225 * target.def (function_arg_offset): New hook.
6226 * targhooks.h (default_function_arg_offset): Declare.
6227 * targhooks.c (default_function_arg_offset): New function.
6228 * function.c (locate_and_pad_parm): Use
6229 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
6230 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
6231 (TARGET_FUNCTION_ARG_OFFSET): ...this.
6232 * doc/tm.texi: Regenerate.
6233 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
6234 * config/spu/spu.c (spu_function_arg_offset): New function.
6235 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
6236 * system.h (FUNCTION_ARG_OFFSET): Poison.
6238 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6239 Alan Hayard <alan.hayward@arm.com>
6240 David Sherwood <david.sherwood@arm.com>
6242 * target.def (truly_noop_truncation): New hook.
6243 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
6244 than TRULY_NOOP_TRUNCATION.
6245 * hooks.h (hook_bool_uint_uint_true): Declare.
6246 * hooks.c (hook_bool_uint_uint_true): New function.
6247 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
6248 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
6249 * doc/tm.texi: Regenerate.
6250 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
6251 rather than TRULY_NOOP_TRUNCATION in comments.
6252 (simplify_comparison): Likewise.
6253 (record_truncated_value): Likewise.
6254 * expmed.c (extract_bit_field_1): Likewise.
6255 (extract_split_bit_field): Likewise.
6256 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
6257 instead of TRULY_NOOP_TRUNCATION.
6258 * function.c (assign_parm_setup_block): Likewise.
6259 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
6260 * rtlhooks.c: Include target.h.
6261 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
6262 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
6263 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
6264 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
6265 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
6266 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
6267 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
6268 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
6269 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
6270 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
6271 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
6272 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
6273 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
6274 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
6275 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
6276 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
6277 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
6278 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
6279 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
6280 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
6281 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
6282 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
6283 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
6284 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
6285 * config/mips/mips.c (mips_truly_noop_truncation): New function.
6286 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
6287 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
6288 rather than TRULY_NOOP_TRUNCATION in comments.
6289 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
6290 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
6291 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
6292 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
6293 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
6294 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
6295 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
6296 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
6297 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
6298 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
6299 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
6300 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
6301 rather than TRULY_NOOP_TRUNCATION in comments.
6302 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
6303 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
6304 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
6305 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
6306 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
6307 TRULY_NOOP_TRUNCATION condition.
6308 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
6309 (TRULY_NOOP_TRUNCATION): Delete.
6310 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
6311 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
6312 * config/spu/spu.c (spu_truly_noop_truncation): New function.
6313 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
6314 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
6315 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
6316 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
6317 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
6318 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
6319 rather than TRULY_NOOP_TRUNCATION in comments.
6320 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
6321 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
6322 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
6323 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
6324 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
6325 * system.h (TRULY_NOOP_TRUNCATION): Poison.
6327 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
6330 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
6331 (*cmp_ior): Likewise.
6332 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
6333 (*ior_scc_scc_cmp): Likewise.
6334 (*and_scc_scc): Likewise.
6335 (*and_scc_scc_cmp): Likewise.
6337 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6338 Alan Hayard <alan.hayward@arm.com>
6339 David Sherwood <david.sherwood@arm.com>
6341 * target.def (can_change_mode_class): New hook.
6342 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
6343 (hard_regno_nregs): Likewise.
6344 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
6345 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
6346 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
6347 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
6348 (LOAD_EXTEND_OP): Update accordingly.
6349 * doc/tm.texi: Regenerate.
6350 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
6351 CANNOT_CHANGE_MODE_CLASS.
6352 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
6353 (REG_CAN_CHANGE_MODE_P): ...this new macro.
6354 * combine.c (simplify_set): Update accordingly.
6355 * emit-rtl.c (validate_subreg): Likewise.
6356 * recog.c (general_operand): Likewise.
6357 * regcprop.c (mode_change_ok): Likewise.
6358 * reload1.c (choose_reload_regs): Likewise.
6359 (inherit_piecemeal_p): Likewise.
6360 * rtlanal.c (simplify_subreg_regno): Likewise.
6361 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
6362 instead of CANNOT_CHANGE_MODE_CLASS.
6363 (reload_cse_simplify_operands): Likewise.
6364 * reload.c (push_reload): Use targetm.can_change_mode_class
6365 instead of CANNOT_CHANGE_MODE_CLASS.
6366 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
6367 REG_CANNOT_CHANGE_MODE_P.
6368 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6369 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
6370 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6371 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6372 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6373 (arm_can_change_mode_class): New function.
6374 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
6375 than CANNOT_CHANGE_MODE_CLASS in comments.
6376 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6377 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
6378 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
6379 (ix86_can_change_mode_class): ...this new function, inverting the
6380 sense of the return value.
6381 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6382 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6383 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6384 (ia64_can_change_mode_class): New function.
6385 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6386 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
6387 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
6388 (m32c_can_change_mode_class): ...this new function, inverting the
6389 sense of the return value.
6390 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6391 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6392 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
6393 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
6394 (mips_can_change_mode_class): ...this new function, inverting the
6395 sense of the return value.
6396 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6397 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6398 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6399 (msp430_can_change_mode_class): New function.
6400 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6401 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
6402 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6403 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6404 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6405 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
6406 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6407 (pa_cannot_change_mode_class): Replace with...
6408 (pa_can_change_mode_class): ...this new function, inverting the
6409 sense of the return value.
6410 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
6411 than CANNOT_CHANGE_MODE_CLASS in comments.
6412 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6413 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
6414 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6415 (pdp11_cannot_change_mode_class): Replace with...
6416 (pdp11_can_change_mode_class): ...this new function, inverting the
6417 sense of the return value.
6418 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6419 * config/powerpcspe/powerpcspe-protos.h
6420 (rs6000_cannot_change_mode_class_ptr): Delete.
6421 * config/powerpcspe/powerpcspe.c
6422 (rs6000_cannot_change_mode_class_ptr): Delete.
6423 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6424 (rs6000_option_override_internal): Assign to
6425 targetm.can_change_mode_class instead of
6426 rs6000_cannot_change_mode_class_ptr.
6427 (rs6000_cannot_change_mode_class): Replace with...
6428 (rs6000_can_change_mode_class): ...this new function, inverting the
6429 sense of the return value.
6430 (rs6000_debug_cannot_change_mode_class): Replace with...
6431 (rs6000_debug_can_change_mode_class): ...this new function.
6432 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6433 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
6434 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6435 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6436 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
6438 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
6439 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6440 (rs6000_option_override_internal): Assign to
6441 targetm.can_change_mode_class instead of
6442 rs6000_cannot_change_mode_class_ptr.
6443 (rs6000_cannot_change_mode_class): Replace with...
6444 (rs6000_can_change_mode_class): ...this new function, inverting the
6445 sense of the return value.
6446 (rs6000_debug_cannot_change_mode_class): Replace with...
6447 (rs6000_debug_can_change_mode_class): ...this new function.
6448 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6449 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
6450 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
6451 (s390_can_change_mode_class): ...this new function, inverting the
6452 sense of the return value.
6453 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6454 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6455 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
6456 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6457 (sh_cannot_change_mode_class): Replace with...
6458 (sh_can_change_mode_class): ...this new function, inverting the
6459 sense of the return value.
6460 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6461 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6462 (sparc_can_change_mode_class): New function.
6463 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6464 * config/spu/spu.c (spu_can_change_mode_class): New function.
6465 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6466 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6467 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6468 (visium_can_change_mode_class): New function.
6469 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
6471 2017-09-15 Richard Biener <rguenther@suse.de>
6473 PR tree-optimization/82217
6474 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
6475 but not undefined case.
6477 2017-09-15 Jakub Jelinek <jakub@redhat.com>
6480 * postreload.c (reload_cse_simplify_operands): Skip
6481 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
6483 2017-09-15 Richard Biener <rguenther@suse.de>
6485 PR tree-optimization/68823
6486 * graphite-scop-detection.c (build_alias_set): If we have a
6487 possible dependence check whether we can handle them by just
6488 looking at the DRs DR_ACCESS_FNs.
6489 (build_scops): If build_alias_set fails, fail the SCOP.
6491 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
6493 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
6494 to support float128 built-in functions that require the ISA 3.0
6496 (BU_FLOAT128_3_HW): Likewise.
6497 (SQRTF128): Add support for the IEEE 128-bit square root and fma
6499 (FMAF128): Likewise.
6501 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
6502 support for built-in functions that need the ISA 3.0 IEEE 128-bit
6503 floating point instructions.
6504 (rs6000_invalid_builtin): Likewise.
6505 (rs6000_builtin_mask_names): Likewise.
6506 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
6507 (RS6000_BTM_FLOAT128_HW): Likewise.
6508 (RS6000_BTM_COMMON): Likewise.
6509 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
6511 * doc/extend.texi (RS/6000 built-in functions): Document the
6512 IEEE 128-bit floating point square root and fused multiply-add
6515 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
6517 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
6518 reg (r2) isn't in the set of registers defined in the prologue.
6520 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6521 Alan Hayward <alan.hayward@arm.com>
6522 David Sherwood <david.sherwood@arm.com>
6524 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
6525 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
6526 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
6527 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
6528 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
6530 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
6531 max_vectorization_factor.
6532 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
6534 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6535 Alan Hayward <alan.hayward@arm.com>
6536 David Sherwood <david.sherwood@arm.com>
6538 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
6539 (vect_worthwhile_without_simd_p): Declare.
6540 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
6541 (vectorizable_reduction): Use it.
6542 * tree-vect-stmts.c (vectorizable_shift): Likewise.
6543 (vectorizable_operation): Likewise.
6545 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6546 Alan Hayward <alan.hayward@arm.com>
6547 David Sherwood <david.sherwood@arm.com>
6549 * tree-vectorizer.h (vect_get_num_copies): New function.
6550 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
6551 * tree-vect-loop.c (vectorizable_reduction): Likewise.
6552 (vectorizable_induction): Likewise.
6553 (vectorizable_live_operation): Likewise.
6554 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
6555 (vectorizable_bswap): Likewise.
6556 (vectorizable_call): Likewise.
6557 (vectorizable_conversion): Likewise.
6558 (vectorizable_assignment): Likewise.
6559 (vectorizable_shift): Likewise.
6560 (vectorizable_operation): Likewise.
6561 (vectorizable_store): Likewise.
6562 (vectorizable_load): Likewise.
6563 (vectorizable_condition): Likewise.
6564 (vectorizable_comparison): Likewise.
6565 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
6567 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6568 Alan Hayward <alan.hayward@arm.com>
6569 David Sherwood <david.sherwood@arm.com>
6571 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
6572 of vect_init_vector.
6574 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6575 Alan Hayward <alan.hayward@arm.com>
6576 David Sherwood <david.sherwood@arm.com>
6578 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
6579 an inline wrapper that provides a location.
6580 (gimple_build_vector): Likewise.
6581 * gimple-fold.c (gimple_build_vector_from_val): New function.
6582 (gimple_build_vector): Likewise.
6583 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
6584 functions to build the initial value. Always return a gimple value.
6585 (get_initial_defs_for_reduction): Likewise. Only compute
6587 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
6588 vect_init_vector on the results from get_initial_def(s)_for_reduction.
6589 (vectorizable_induction): Use gimple_build_vector rather than
6592 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6593 Alan Hayward <alan.hayward@arm.com>
6594 David Sherwood <david.sherwood@arm.com>
6596 * target.h (vec_perm_indices): New typedef.
6597 (auto_vec_perm_indices): Likewise.
6598 * optabs-query.h: Include target.h
6599 (can_vec_perm_p): Take a vec_perm_indices *.
6600 * optabs-query.c (can_vec_perm_p): Likewise.
6601 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
6602 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
6603 * tree-vect-generic.c (lower_vec_perm): Likewise.
6604 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
6605 (vect_grouped_load_supported): Likewise.
6606 (vect_shift_permute_load_chain): Likewise.
6607 (vect_permute_store_chain): Use auto_vec_perm_indices.
6608 (vect_permute_load_chain): Likewise.
6609 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
6610 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
6611 Update uses of can_vec_perm_p.
6612 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
6613 mode with a number of elements. Take a vec_perm_indices *.
6614 (vect_create_epilog_for_reduction): Update accordingly.
6615 Use auto_vec_perm_indices.
6616 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
6617 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
6618 (vect_transform_slp_perm_load): Likewise.
6619 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
6620 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
6621 (vect_gen_perm_mask_checked): Likewise.
6622 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
6623 (vect_gen_perm_mask_checked): Likewise.
6624 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
6625 (vectorizable_store): Likewise.
6626 (vectorizable_load): Likewise.
6627 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
6628 (vectorizable_bswap): Likewise.
6630 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6631 Alan Hayward <alan.hayward@arm.com>
6632 David Sherwood <david.sherwood@arm.com>
6634 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
6635 * tree.c (build_vector): Likewise.
6636 (build_vector_from_ctor): Update accordingly.
6637 (build_vector_from_val): Likewise.
6638 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
6639 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
6640 * tree-vect-generic.c (add_rshift): Likewise.
6641 (expand_vector_divmod): Likewise.
6642 (optimize_vector_constructor): Likewise.
6643 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
6644 (vect_transform_slp_perm_load): Likewise.
6645 (vect_schedule_slp_instance): Likewise.
6646 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
6647 (vectorizable_call): Likewise.
6648 (vect_gen_perm_mask_any): Likewise. Add elements in order.
6649 * expmed.c (make_tree): Likewise.
6650 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
6651 a vector passed to build_vector.
6652 (fold_convert_const): Likewise.
6653 (exact_inverse): Likewise.
6654 (fold_ternary_loc): Likewise.
6655 (fold_relational_const): Likewise.
6656 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
6657 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
6658 (const_unop): Likewise. Store the reduction accumulator in a
6659 variable rather than an array.
6660 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
6661 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
6662 the new vector, rather than constructing it after the input arrays.
6663 (native_interpret_vector): Use auto_vec<tree> when building
6664 a vector passed to build_vector. Add elements in order.
6665 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
6666 auto_vec<tree> when building a vector passed to build_vector.
6667 (vect_create_epilog_for_reduction): Likewise.
6668 (vectorizable_induction): Likewise.
6669 (get_initial_def_for_reduction): Likewise. Fix indentation of
6671 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
6673 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
6674 passed to build_vector.
6676 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6677 Alan Hayward <alan.hayward@arm.com>
6678 David Sherwood <david.sherwood@arm.com>
6680 * tree-core.h (tree_base::u): Add an "nelts" field.
6681 (tree_vector): Use VECTOR_CST_NELTS as the length.
6682 * tree.c (tree_size): Likewise.
6683 (make_vector): Initialize VECTOR_CST_NELTS.
6684 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
6685 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
6686 TYPE_VECTOR_SUBPARTS.
6687 * expr.c (const_vector_mask_from_tree): Consistently use "units"
6688 as the number of units, setting it from VECTOR_CST_NELTS.
6689 (const_vector_from_tree): Likewise.
6690 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
6691 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
6692 (fold_negate_expr_1): Likewise.
6693 (fold_convert_const): Likewise.
6694 (const_binop): Likewise. Differentiate the number of output and
6696 (const_unop): Likewise.
6697 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
6698 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
6699 in cases that did the opposite.
6701 2017-09-14 Richard Biener <rguenther@suse.de>
6703 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
6706 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
6708 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
6710 2017-09-14 Jakub Jelinek <jakub@redhat.com>
6713 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
6714 if and where to split a bb, except for splitting before debug insn
6715 sequences followed by non-label real insn. Delete debug insns
6716 in between basic blocks.
6718 * combine.c (make_compound_operation_int): Formatting fixes.
6720 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
6721 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6722 * config/netbsd.h (LINK_EH_SPEC): Likewise.
6723 * config/sol2.h (LINK_EH_SPEC): Likewise.
6724 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6725 * config/s390/linux.h (LINK_SPEC): Likewise.
6726 * config/freebsd.h (LINK_EH_SPEC): Likewise.
6727 * config/openbsd.h (LINK_EH_SPEC): Likewise.
6728 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6729 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
6730 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
6731 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6732 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
6733 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
6735 2017-09-13 Jakub Jelinek <jakub@redhat.com>
6737 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
6739 (ENDFILE_LINUX_SPEC): Likewise.
6740 (LINK_EH_SPEC): Likewise.
6741 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
6742 (LINK_OS_LINUX_SPEC32): Likewise.
6743 (LINK_OS_LINUX_SPEC64): Likewise.
6744 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
6745 (LINK_OS_LINUX_SPEC): Likewise.
6747 2017-09-13 Martin Liska <mliska@suse.cz>
6750 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
6751 CASE_HIGH is NULL_TREE.
6753 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
6754 Alan Hayward <alan.hayward@arm.com>
6755 David Sherwood <david.sherwood@arm.com>
6757 * target.def (secondary_memory_needed): New hook.
6758 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
6759 instead of SECONDARY_MEMORY_NEEDED.
6760 (secondary_memory_needed_mode): Likewise.
6761 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
6762 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
6763 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
6764 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
6765 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
6766 * doc/tm.texi: Regenerate.
6767 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
6768 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
6769 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6770 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
6771 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
6772 * config/i386/i386.c (inline_secondary_memory_needed): Put the
6773 mode argument first and change the reg_class arguments to reg_class_t.
6774 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
6775 Make static. Update the call to inline_secondary_memory_needed.
6776 (ix86_register_move_cost): Update the call to
6777 inline_secondary_memory_needed.
6778 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6779 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
6781 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
6782 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
6784 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
6785 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
6786 * config/mips/mips.c (mips_secondary_memory_needed): Make static
6787 and match hook interface. Add comment from mips.h.
6788 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6789 * config/mmix/mmix.md (truncdfsf2): Refer to
6790 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
6792 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
6793 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
6794 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6795 (pa_secondary_memory_needed): New function.
6796 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
6797 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
6798 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6799 (pdp11_secondary_memory_needed): Make static and match hook interface.
6800 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
6801 * config/powerpcspe/powerpcspe-protos.h
6802 (rs6000_secondary_memory_needed_ptr): Delete.
6803 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
6805 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6806 (rs6000_option_override_internal): Assign to
6807 targetm.secondary_memory_needed rather than
6808 rs6000_secondary_memory_needed_ptr.
6809 (rs6000_secondary_memory_needed): Match hook interface.
6810 (rs6000_debug_secondary_memory_needed): Likewise.
6811 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
6812 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
6813 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
6814 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6815 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
6816 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
6818 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
6819 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6820 (rs6000_option_override_internal): Assign to
6821 targetm.secondary_memory_needed rather than
6822 rs6000_secondary_memory_needed_ptr.
6823 (rs6000_secondary_memory_needed): Match hook interface.
6824 (rs6000_debug_secondary_memory_needed): Likewise.
6825 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
6826 * config/s390/s390.c (s390_secondary_memory_needed): New function.
6827 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6828 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
6829 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6830 (sparc_secondary_memory_needed): New function.
6831 * lra-constraints.c (check_and_process_move): Refer to
6832 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
6834 (curr_insn_transform): Likewise.
6835 (process_alt_operands): Use targetm.secondary_memory_needed
6836 instead of TARGET_SECONDARY_MEMORY_NEEDED.
6837 (check_secondary_memory_needed_p): Likewise.
6838 (choose_split_class): Likewise.
6839 * reload.c: Unconditionally include code that was previously
6840 conditional on SECONDARY_MEMORY_NEEDED.
6841 (push_secondary_reload): Use targetm.secondary_memory_needed
6842 instead of TARGET_SECONDARY_MEMORY_NEEDED.
6843 (push_reload): Likewise.
6844 * reload1.c: Unconditionally include code that was previously
6845 conditional on SECONDARY_MEMORY_NEEDED.
6846 (choose_reload_regs): Use targetm.secondary_memory_needed
6847 instead of TARGET_SECONDARY_MEMORY_NEEDED.
6848 (gen_reload): Likewise.
6849 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
6851 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
6852 Alan Hayward <alan.hayward@arm.com>
6853 David Sherwood <david.sherwood@arm.com>
6855 * target.def (secondary_memory_needed_mode): New hook:
6856 * targhooks.c (default_secondary_memory_needed_mode): Declare.
6857 * targhooks.h (default_secondary_memory_needed_mode): New function.
6858 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
6859 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
6860 * doc/tm.texi: Regenerate.
6861 * lra-constraints.c (check_and_process_move): Use
6862 targetm.secondary_memory_needed_mode instead of
6863 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
6864 (curr_insn_transform): Likewise.
6865 * reload.c (get_secondary_mem): Likewise.
6866 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
6867 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
6869 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
6870 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
6871 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
6872 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
6873 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
6875 * config/powerpcspe/powerpcspe-protos.h
6876 (rs6000_secondary_memory_needed_mode): Delete.
6877 * config/powerpcspe/powerpcspe.c
6878 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
6879 (rs6000_secondary_memory_needed_mode): Make static.
6880 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
6881 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
6883 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
6885 (rs6000_secondary_memory_needed_mode): Make static.
6886 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
6887 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
6888 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
6889 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
6890 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
6892 (sparc_secondary_memory_needed_mode): New function.
6893 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
6895 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
6897 * config/aarch64/constraints.md (Umq): New constraint.
6898 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
6900 (mov<mode>): Update condition.
6902 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6904 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
6905 when bitposition is the same.
6907 2017-09-13 Richard Biener <rguenther@suse.de>
6909 * dwarf2out.c (output_die_symbol): Remove.
6910 (output_die): Do not output a DIEs symbol.
6912 2017-09-13 Richard Biener <rguenther@suse.de>
6915 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
6916 default-def to avoid breaking iterator update with the weird
6917 interaction with cgraph_update_edges_for_call_stmt_node.
6919 2017-09-13 Richard Biener <rguenther@suse.de>
6921 * tree-cfg.c (verify_gimple_assign_binary): Add verification
6922 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
6923 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
6924 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
6926 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
6928 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
6929 Disable pc relative literal load irrespective of
6930 TARGET_FIX_ERR_A53_84341 for default.
6932 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
6934 * config/sparc/sparc.c (output_return): Output the source location of
6935 the insn in the delay slot, if any.
6936 (output_sibcall): Likewise.
6938 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
6941 * common.opt (-static-pie): New alias.
6942 (shared): Negate static-pie.
6943 (-no-pie): Update help text.
6945 (static-pie): New option.
6946 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
6947 -static-pie support.
6948 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
6949 (LINK_EH_SPEC): Likewise.
6950 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6951 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
6952 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
6953 * gcc.c (LINK_COMMAND_SPEC): Likewise.
6954 (init_gcc_specs): Likewise.
6955 (init_spec): Likewise.
6956 (display_help): Update help message for -pie.
6957 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
6960 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
6962 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
6963 (movdi_aarch64): Likewise.
6964 (movti_aarch64): Likewise.
6966 2017-09-12 Simon Wright <simon@pushface.org>
6969 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
6970 calculation of the minor version, always output as 0.
6972 2017-09-12 Jakub Jelinek <jakub@redhat.com>
6975 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
6976 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
6977 on it early, rather than manual conversion late. For
6978 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
6979 instead of performing manual conversion.
6981 2017-09-12 Carl Love <cel@us.ibm.com>
6983 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
6984 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
6986 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
6987 VMULOSW): Add definitions.
6988 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6989 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
6990 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
6991 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
6992 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
6994 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
6996 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
6998 (movti_aarch64): Likewise.
6999 (movdf_aarch64): Likewise.
7000 (movtf_aarch64): Likewise.
7001 (load_pairdi): Likewise.
7002 (store_pairdi): Likewise.
7003 (load_pairdf): Likewise.
7004 (store_pairdf): Likewise.
7005 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
7006 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
7007 (ldr_got_small_<mode>): Likewise.
7008 (ldr_got_small_28k_<mode>): Likewise.
7009 (ldr_got_tiny): Likewise.
7010 * config/aarch64/iterators.md (ldst_sz): New.
7011 (ldpstp_sz): Likewise.
7012 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
7014 (thunderx_load): Split load_8 to load_16.
7015 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
7017 (thunderx2t99_storepair_basic): Split store_8 to store_16.
7018 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
7019 (xgene1_store_pair): Split store_8 to store_16.
7020 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
7021 (falkor_st_0_st_sd): Split store_8 to store_16.
7023 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7025 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
7026 and store1/2/3/4 to store_4/8/12/16.
7027 * config/aarch64/aarch64.md: Update for rename.
7028 * config/arm/arm.md: Likewise.: Likewise.
7029 * config/arm/arm.c: Likewise.
7030 * config/arm/thumb1.md: Likewise.
7031 * config/arm/thumb2.md: Likewise.
7032 * config/arm/vfp.md: Likewise.
7033 * config/arm/arm-generic.md: Likewise.
7034 * config/arm/arm1020e.md: Likewise.
7035 * config/arm/arm1026ejs.md: Likewise.
7036 * config/arm/arm1136jfs.md: Likewise.
7037 * config/arm/arm926ejs.md: Likewise.
7038 * config/arm/cortex-a15.md: Likewise.
7039 * config/arm/cortex-a17.md: Likewise.
7040 * config/arm/cortex-a5.md: Likewise.
7041 * config/arm/cortex-a53.md: Likewise.
7042 * config/arm/cortex-a57.md: Likewise.
7043 * config/arm/cortex-a7.md: Likewise.
7044 * config/arm/cortex-a8.md: Likewise.
7045 * config/arm/cortex-a9.md: Likewise.
7046 * config/arm/cortex-m4.md: Likewise.
7047 * config/arm/cortex-m7.md: Likewise.
7048 * config/arm/cortex-r4.md: Likewise.
7049 * config/arm/exynos-m1.md: Likewise.
7050 * config/arm/fa526.md: Likewise.
7051 * config/arm/fa606te.md: Likewise.
7052 * config/arm/fa626te.md: Likewise.
7053 * config/arm/fa726te.md: Likewise.
7054 * config/arm/fmp626.md: Likewise.
7055 * config/arm/iwmmxt.md: Likewise.
7056 * config/arm/ldmstm.md: Likewise.
7057 * config/arm/marvell-pj4.md: Likewise.
7058 * config/arm/xgene1.md: Likewise.
7059 * config/aarch64/thunderx.md: Likewise.
7060 * config/aarch64/thunderx2t99.md: Likewise.
7061 * config/aarch64/falkor.md: Likewise.
7063 2017-09-12 Martin Liska <mliska@suse.cz>
7065 * attribs.c (private_lookup_attribute): New function.
7066 * attribs.h (private_lookup_attribute): Declared here.
7067 (lookup_attribute): Called from this place.
7069 2017-09-12 Richard Biener <rguenther@suse.de>
7071 PR tree-optimization/82157
7072 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
7073 stmts with side-effects.
7075 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7076 Alan Hayward <alan.hayward@arm.com>
7077 David Sherwood <david.sherwood@arm.com>
7079 * target.def (hard_regno_nregs): New hook.
7080 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
7081 * targhooks.h (default_hard_regno_nregs): Declare.
7082 * targhooks.c (default_hard_regno_nregs): New function.
7083 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
7084 (TARGET_HARD_REGNO_NREGS): ...this hook.
7085 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
7086 (CLASS_MAX_NREGS): Likewise.
7087 * doc/tm.texi: Regenerate.
7088 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
7089 instead of HARD_REGNO_NREGS.
7090 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
7091 HARD_REGNO_NREGS in the comment.
7092 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
7093 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
7094 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
7095 Return an unsigned int.
7096 (TARGET_HARD_REGNO_NREGS): Redefine.
7097 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
7098 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
7099 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
7100 (arc_hard_regno_nregs): New function.
7101 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
7102 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
7103 (arm_hard_regno_nregs): New function.
7104 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
7105 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
7106 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
7107 (TARGET_HARD_REGNO_NREGS): Redefine.
7108 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
7109 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
7110 (HARD_REGNO_NREGS): Delete.
7111 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
7112 (cr16_hard_regno_nregs): New function.
7113 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
7114 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
7115 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
7116 (cris_hard_regno_nregs): New function.
7117 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
7118 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
7119 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
7120 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
7121 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
7122 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
7123 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
7124 (frv_hard_regno_nregs): Make static. Take and return an
7126 (frv_class_max_nregs): Remove outdated copy of documentation.
7127 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
7128 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
7129 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
7130 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
7131 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
7132 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
7133 (TARGET_HARD_REGNO_NREGS): Redefine.
7134 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
7135 (CLASS_MAX_NREGS): Update comment.
7136 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
7137 (ia64_hard_regno_nregs): New function.
7138 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
7139 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
7140 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
7141 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
7142 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
7144 (m32c_hard_regno_nregs): Likewise. Make static.
7145 (TARGET_HARD_REGNO_NREGS): Redefine.
7146 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
7147 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
7148 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
7149 (m68k_hard_regno_nregs): New function.
7150 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
7151 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
7152 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
7153 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
7154 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
7155 Take and return an unsigned int.
7156 (TARGET_HARD_REGNO_NREGS): Redefine.
7157 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
7158 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
7159 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
7160 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
7161 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
7162 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
7163 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
7164 (msp430_hard_regno_nregs): Make static. Take and return an
7166 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
7167 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
7168 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
7169 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
7170 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
7171 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
7172 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
7173 (TARGET_HARD_REGNO_NREGS): Redefine.
7174 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
7175 (PA_HARD_REGNO_NREGS): ...this.
7176 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
7177 (PA_HARD_REGNO_NREGS): ...this.
7178 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
7179 (pa_hard_regno_nregs): New function.
7180 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
7181 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
7182 (pdp11_hard_regno_nregs): New function.
7183 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
7184 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
7185 (rs6000_hard_regno_nregs_hook): New function.
7186 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
7187 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
7188 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
7189 Take and return an unsigned int. Move earlier in file.
7190 (TARGET_HARD_REGNO_NREGS): Redefine.
7191 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
7192 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
7193 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
7194 (rl78_hard_regno_nregs): Make static. Take and return an
7196 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
7197 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
7198 (rs6000_hard_regno_nregs_hook): New function.
7199 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
7200 * config/rx/rx.c (rx_hard_regno_nregs): New function.
7201 (TARGET_HARD_REGNO_NREGS): Redefine.
7202 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
7203 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
7204 instead of HARD_REGNO_NREGS.
7205 (s390_hard_regno_nregs): New function.
7206 (s390_hard_regno_mode_ok): Add comment from s390.h.
7207 (TARGET_HARD_REGNO_NREGS): Redefine.
7208 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
7209 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
7210 (sh_hard_regno_nregs): New function.
7211 (sh_pass_in_reg_p): Use it.
7212 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
7213 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
7214 (sparc_hard_regno_nregs): New function.
7215 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
7216 * config/spu/spu.c (spu_hard_regno_nregs): New function.
7217 (spu_function_arg_advance): Use it, supplying a valid register number.
7218 (TARGET_HARD_REGNO_NREGS): Redefine.
7219 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
7220 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
7221 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
7222 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
7223 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
7224 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
7225 (CLASS_MAX_NREGS): Remove copy of old documentation.
7226 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
7227 (visium_hard_regno_nregs): New function.
7228 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
7229 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
7230 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
7231 xtensa_hard_regno_nregs): New function.
7232 * system.h (HARD_REGNO_NREGS): Poison.
7234 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7236 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
7237 hard_regno_nregs instead of HARD_REGNO_NREGS.
7238 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
7239 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
7240 (c6x_expand_epilogue): Likewise.
7241 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
7242 (frv_read_iacc_argument): Likewise.
7243 * config/sh/sh.c: Include regs.h.
7244 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
7245 (regs_used): Likewise.
7246 (output_stack_adjust): Likewise.
7247 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
7248 * expmed.c: Include regs.h.
7249 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
7250 * ree.c: Include regs.h.
7251 (combine_reaching_defs): Use hard_regno_nregs instead of
7253 (add_removable_extension): Likewise.
7255 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7257 * regs.h (hard_regno_nregs): Turn into a function.
7258 (end_hard_regno): Update accordingly.
7259 * caller-save.c (setup_save_areas): Likewise.
7260 (save_call_clobbered_regs): Likewise.
7261 (replace_reg_with_saved_mem): Likewise.
7262 (insert_restore): Likewise.
7263 (insert_save): Likewise.
7264 * combine.c (can_change_dest_mode): Likewise.
7265 (move_deaths): Likewise.
7266 (distribute_notes): Likewise.
7267 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
7268 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
7269 (rs6000_split_multireg_move): Likewise.
7270 (rs6000_register_move_cost): Likewise.
7271 (rs6000_memory_move_cost): Likewise.
7272 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
7273 (rs6000_split_multireg_move): Likewise.
7274 (rs6000_register_move_cost): Likewise.
7275 (rs6000_memory_move_cost): Likewise.
7276 * cselib.c (cselib_reset_table): Likewise.
7277 (cselib_lookup_1): Likewise.
7278 * emit-rtl.c (set_mode_and_regno): Likewise.
7279 * function.c (aggregate_value_p): Likewise.
7280 * ira-color.c (setup_profitable_hard_regs): Likewise.
7281 (check_hard_reg_p): Likewise.
7282 (calculate_saved_nregs): Likewise.
7283 (assign_hard_reg): Likewise.
7284 (improve_allocation): Likewise.
7285 (calculate_spill_cost): Likewise.
7286 * ira-emit.c (modify_move_list): Likewise.
7287 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
7288 (ira_hard_reg_in_set_p): Likewise.
7289 * ira.c (setup_reg_mode_hard_regset): Likewise.
7290 (clarify_prohibited_class_mode_regs): Likewise.
7291 (check_allocation): Likewise.
7292 * lra-assigns.c (find_hard_regno_for_1): Likewise.
7293 (lra_setup_reg_renumber): Likewise.
7294 (setup_try_hard_regno_pseudos): Likewise.
7295 (spill_for): Likewise.
7296 (assign_hard_regno): Likewise.
7297 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
7298 * lra-constraints.c (in_class_p): Likewise.
7299 (lra_constraint_offset): Likewise.
7300 (simplify_operand_subreg): Likewise.
7301 (lra_constraints): Likewise.
7302 (split_reg): Likewise.
7303 (split_if_necessary): Likewise.
7304 (invariant_p): Likewise.
7305 (inherit_in_ebb): Likewise.
7306 * lra-lives.c (process_bb_lives): Likewise.
7307 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
7308 (get_hard_regs): Likewise.
7309 (do_remat): Likewise.
7310 * lra-spills.c (assign_spill_hard_regs): Likewise.
7311 * mode-switching.c (create_pre_exit): Likewise.
7312 * postreload.c (reload_combine_recognize_pattern): Likewise.
7313 * recog.c (peep2_find_free_register): Likewise.
7314 * regcprop.c (kill_value_regno): Likewise.
7315 (set_value_regno): Likewise.
7316 (copy_value): Likewise.
7317 (maybe_mode_change): Likewise.
7318 (find_oldest_value_reg): Likewise.
7319 (copyprop_hardreg_forward_1): Likewise.
7320 * regrename.c (check_new_reg_p): Likewise.
7321 (regrename_do_replace): Likewise.
7322 * reload.c (push_reload): Likewise.
7323 (combine_reloads): Likewise.
7324 (find_dummy_reload): Likewise.
7325 (operands_match_p): Likewise.
7326 (find_reloads): Likewise.
7327 (find_equiv_reg): Likewise.
7328 (reload_adjust_reg_for_mode): Likewise.
7329 * reload1.c (count_pseudo): Likewise.
7330 (count_spilled_pseudo): Likewise.
7331 (find_reg): Likewise.
7332 (clear_reload_reg_in_use): Likewise.
7333 (free_for_value_p): Likewise.
7334 (allocate_reload_reg): Likewise.
7335 (choose_reload_regs): Likewise.
7336 (reload_adjust_reg_for_temp): Likewise.
7337 (emit_reload_insns): Likewise.
7338 (delete_output_reload): Likewise.
7339 * rtlanal.c (subreg_get_info): Likewise.
7340 * sched-deps.c (sched_analyze_reg): Likewise.
7341 * sel-sched.c (init_regs_for_mode): Likewise.
7342 (mark_unavailable_hard_regs): Likewise.
7343 (choose_best_reg_1): Likewise.
7344 (verify_target_availability): Likewise.
7345 * valtrack.c (dead_debug_insert_temp): Likewise.
7346 * var-tracking.c (track_loc_p): Likewise.
7347 (emit_note_insn_var_location): Likewise.
7348 * varasm.c (make_decl_rtl): Likewise.
7349 * reginfo.c (choose_hard_reg_mode): Likewise.
7350 (init_reg_modes_target): Refer directly to
7351 this_target_regs->x_hard_regno_nregs.
7353 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7355 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
7356 instead of hard_regno_nregs.
7358 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7360 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
7361 end_hard_regno instead of hard_regno_nregs.
7362 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
7363 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
7364 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
7365 * ira-color.c (improve_allocation): Likewise.
7366 * lra-assigns.c (find_hard_regno_for_1): Likewise.
7367 * lra-lives.c (mark_regno_live): Likewise.
7368 (mark_regno_dead): Likewise.
7369 * lra-remat.c (operand_to_remat): Likewise.
7370 * lra.c (collect_non_operand_hard_regs): Likewise.
7371 * postreload.c (reload_combine_note_store): Likewise.
7372 (move2add_valid_value_p): Likewise.
7373 * reload.c (regno_clobbered_p): Likewise.
7375 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7377 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
7379 * config/v850/v850.c (v850_reorg): Likewise.
7380 * reload.c (refers_to_regno_for_reload_p): Likewise.
7381 (find_equiv_reg): Likewise.
7382 * reload1.c (reload_reg_reaches_end_p): Likewise.
7384 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7386 * caller-save.c (add_used_regs): Use REG_NREGS instead of
7388 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
7389 * config/arm/arm.c (output_move_neon): Likewise.
7390 (arm_attr_length_move_neon): Likewise.
7391 (neon_split_vcombine): Likewise.
7392 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
7393 (c6x_mark_reg_written): Likewise.
7394 (c6x_dwarf_register_span): Likewise.
7395 * config/i386/i386.c (ix86_save_reg): Likewise.
7396 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
7397 (rws_access_reg): Likewise.
7398 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
7399 * mode-switching.c (create_pre_exit): Likewise.
7400 * ree.c (combine_reaching_defs): Likewise.
7401 (add_removable_extension): Likewise.
7402 * regcprop.c (find_oldest_value_reg): Likewise.
7403 (copyprop_hardreg_forward_1): Likewise.
7404 * reload.c (reload_inner_reg_of_subreg): Likewise.
7405 (push_reload): Likewise.
7406 (combine_reloads): Likewise.
7407 (find_dummy_reload): Likewise.
7408 (reload_adjust_reg_for_mode): Likewise.
7409 * reload1.c (find_reload_regs): Likewise.
7410 (forget_old_reloads_1): Likewise.
7411 (reload_reg_free_for_value_p): Likewise.
7412 (reload_adjust_reg_for_temp): Likewise.
7413 (emit_reload_insns): Likewise.
7414 (delete_output_reload): Likewise.
7415 * sel-sched.c (choose_best_reg_1): Likewise.
7416 (choose_best_pseudo_reg): Likewise.
7418 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7419 Alan Hayward <alan.hayward@arm.com>
7420 David Sherwood <david.sherwood@arm.com>
7422 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
7423 * target.def (slow_unaligned_access): New hook.
7424 * targhooks.h (default_slow_unaligned_access): Declare.
7425 * targhooks.c (default_slow_unaligned_access): New function.
7426 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
7427 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
7428 * doc/tm.texi: Regenerate.
7429 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
7430 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
7431 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
7433 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
7434 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
7436 (rs6000_slow_unaligned_access): New function.
7437 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
7438 (expand_block_compare): Likewise.
7439 (expand_strn_compare): Likewise.
7440 (rs6000_rtx_costs): Likewise.
7441 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
7442 (riscv_slow_unaligned_access): Likewise.
7443 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
7444 (riscv_slow_unaligned_access_p): ...this and make static.
7445 (riscv_option_override): Update accordingly.
7446 (riscv_slow_unaligned_access): New function.
7447 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
7448 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
7449 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
7450 (rs6000_slow_unaligned_access): New function.
7451 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
7452 (rs6000_rtx_costs): Likewise.
7453 * config/rs6000/rs6000-string.c (expand_block_compare)
7454 (expand_strn_compare): Use targetm.slow_unaligned_access instead
7455 of SLOW_UNALIGNED_ACCESS.
7456 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
7457 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
7458 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
7459 of SLOW_UNALIGNED_ACCESS.
7460 * expmed.c (simple_mem_bitfield_p): Likewise.
7461 * expr.c (alignment_for_piecewise_move): Likewise.
7462 (emit_group_load_1): Likewise.
7463 (emit_group_store): Likewise.
7464 (copy_blkmode_from_reg): Likewise.
7465 (emit_push_insn): Likewise.
7466 (expand_assignment): Likewise.
7467 (store_field): Likewise.
7468 (expand_expr_real_1): Likewise.
7469 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
7470 * lra-constraints.c (simplify_operand_subreg): Likewise.
7471 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
7472 * gimple-ssa-store-merging.c: Likewise in block comment at start
7474 * tree-ssa-strlen.c: Include target.h.
7475 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
7476 of SLOW_UNALIGNED_ACCESS.
7477 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
7479 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7481 PR rtl-optimization/82185
7482 * expmed.c (emit_store_flag_int): Only test tem if it has been
7485 2017-09-12 Richard Biener <rguenther@suse.de>
7488 * match.pd ((FTYPE) N CMP CST): Fix typo.
7490 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
7492 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
7494 (mips_near_type_p): Add 'short_call' attribute as a synonym
7496 * doc/extend.texi (short_call): Document new function attribute.
7498 2017-09-12 Jakub Jelinek <jakub@redhat.com>
7501 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
7502 assertion check that in the condition.
7503 (get_atomic_generic_size): Likewise. Before testing if parameter
7504 has pointer type, if it has array type, call for C++
7505 default_conversion to perform array-to-pointer conversion.
7507 2017-09-12 Richard Biener <rguenther@suse.de>
7509 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
7510 for operations we cannot scalarize.
7512 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
7514 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
7515 vectors heap vectors. Clean up comments.
7516 Make visited_bbs a reference.
7517 (profitable_jump_thread_path): Make GC
7518 vectors heap vectors. Clean up comments.
7520 (convert_and_register_jump_thread_path): Make GC vectors heap
7522 (check_subpath_and_update_thread_path): Same. Clean up comments.
7523 Make visited_bbs a reference.
7524 (handle_phi): Abstract common code to to
7525 register_jump_thread_path_if_profitable.
7526 Rename VAR_BB to DEF_BB.
7528 Make GC vectors heap vectors.
7529 Make visited_bbs a reference.
7530 (handle_assignment): Same.
7531 (register_jump_thread_path_if_profitable): New.
7532 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
7534 Make GC vectors heap vectors. Clean up comments.
7535 Make visited_bbs a reference.
7536 (find_jump_threads_backwards): Make visited_bbs live in the stack.
7537 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
7540 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
7543 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
7544 words of E_DImode object are reachable by xtensa_uimm8x4 access.
7546 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
7548 Revert r251800 and r251799.
7550 2017-09-11 Martin Jambor <mjambor@suse.cz>
7553 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
7554 arguments in advance.
7555 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
7556 use it to find predecessor edges.
7557 (naive_outof_ssa): Collect vector of predecessors.
7559 2017-09-08 Jason Merrill <jason@redhat.com>
7561 PR c++/70029 - ICE with ref-qualifier and -flto
7562 * langhooks.h (struct lang_hooks_for_types): Add
7563 copy_lang_qualifiers.
7564 * attribs.c (build_type_attribute_qual_variant): Use it.
7565 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
7567 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
7568 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
7570 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
7573 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
7574 (*mulsi3_sp64): New instruction.
7575 (mulsi3): New expander.
7577 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
7579 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
7581 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7583 * sancov.c: Include memmodel.h.
7585 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
7588 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
7591 2017-09-07 Carl Love <cel@us.ibm.com>
7593 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
7594 the sldi instruction.
7596 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
7598 * sancov.c: Include tm_p.h.
7600 2017-09-07 Jakub Jelinek <jakub@redhat.com>
7603 * output.h (switch_to_other_text_partition): New declaration.
7604 * varasm.c (switch_to_other_text_partition): New function.
7605 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
7606 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
7607 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
7608 to the other text partition before emitting LCL label and switch back
7609 after emitting the word after it.
7611 2017-09-07 Richard Biener <rguenther@suse.de>
7613 * passes.def (pass_split_crit_edges): Remove instance before PRE.
7614 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
7615 critical edges here, after loop init.
7616 (pass_data_pre): Remove PROP_no_crit_edges flags.
7617 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
7618 for valueization of call args to avoid leaking VN_TOP.
7619 (visit_use): Assert we do not visit default defs.
7620 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
7621 Use error_mark_node to more easily detect leaking VN_TOP.
7622 All default-defs are varying, not VN_TOP. Mark them visited.
7623 (run_scc_vn): Make code match comment.
7625 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
7627 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
7628 OPTION_MASK_FLOAT128_KEYWORD.
7629 (POWERPC_MASKS): Likewise.
7630 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
7631 support for the -mfloat128-type option, and make -mfloat128
7632 default on PowerPC Linux systems. Define or undefine
7633 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
7634 Define __float128 to be __ieee128 if IEEE 128-bit support is
7635 enabled, or undefine it.
7636 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
7637 Delete defining __FLOAT128_TYPE__.
7638 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
7639 -mfloat128-type option and make -mfloat128 default on PowerPC
7641 (TARGET_FLOAT128_TYPE): Likewise.
7642 (-mfloat128-type): Likewise.
7643 * config/rs6000/rs6000.c (rs6000_option_override_internal):
7644 Delete the -mfloat128-type option and make -mfloat128 default on
7645 PowerPC Linux systems. Always use __ieee128 to be the keyword for
7646 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
7647 128-bit floating point is enabled. Change tests from using
7648 -mfloat128-type to -mfloat128.
7649 (rs6000_mangle_type): Use the correct mangling for the __float128
7650 type even if normal long double is restricted to 64-bits.
7651 (floatn_mode): Enable the _Float128 type by default on VSX Linux
7653 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
7654 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
7655 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
7657 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
7658 documentation for -mfloat128.
7660 2017-09-06 Olivier Hainque <hainque@adacore.com>
7662 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
7664 2017-09-06 Wish Wu <wishwu007@gmail.com>
7665 Jakub Jelinek <jakub@redhat.com>
7667 * asan.c (initialize_sanitizer_builtins): Add
7668 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
7669 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
7670 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
7671 BT_FN_VOID_UINT64_PTR variables.
7672 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
7673 (BT_FN_VOID_UINT16_UINT16): Likewise.
7674 (BT_FN_VOID_UINT32_UINT32): Likewise.
7675 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
7676 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
7677 (BT_FN_VOID_UINT64_PTR): Likewise.
7678 * common.opt (flag_sanitize_coverage): New variable.
7679 (fsanitize-coverage=trace-pc): Remove.
7680 (fsanitize-coverage=): Add.
7681 * flag-types.h (enum sanitize_coverage_code): New enum.
7682 * fold-const.c (fold_range_test): Disable non-short-circuit
7683 optimization if flag_sanitize_coverage.
7684 (fold_truth_andor): Likewise.
7685 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
7686 * opts.c (COVERAGE_SANITIZER_OPT): Define.
7687 (coverage_sanitizer_opts): New array.
7688 (get_closest_sanitizer_option): Add OPTS argument, handle also
7689 OPT_fsanitize_coverage_.
7690 (parse_sanitizer_options): Adjusted to also handle
7691 OPT_fsanitize_coverage_.
7692 (common_handle_option): Add OPT_fsanitize_coverage_.
7693 * sancov.c (instrument_comparison, instrument_switch): New function.
7694 (sancov_pass): Add trace-cmp support.
7695 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
7696 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
7697 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
7698 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
7699 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
7700 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
7701 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
7702 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
7703 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
7704 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
7706 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
7708 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
7709 error. Only quit immediately if parsing is complete.
7710 (BEGIN): Initialize fatal_err and parse_done.
7711 (begin fpu, end fpu): Check number of arguments.
7712 (begin arch, end arch): Likewise.
7713 (begin cpu, end cpu): Likewise.
7714 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
7715 (optalias): Likewise.
7717 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
7719 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
7720 * config/arm/arm-isa.h: Delete. Move definitions to ...
7721 * arm-cpus.in: ... here. Use new feature and fgroup values.
7722 * config/arm/arm.c (arm_option_override): Use lower case for feature
7724 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
7725 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
7726 * config/arm/parsecpu.awk (END): Add new command 'isa'.
7728 (print_isa_bits_for): New function.
7729 (gen_isa): New function.
7730 (gen_comm_data): Use print_isa_bits_for.
7731 (define feature): New keyword.
7732 (define fgroup): New keyword.
7733 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
7734 (arm-isa.h): Add rule to generate file.
7735 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
7736 case for feature bit names.
7738 2017-09-06 Richard Biener <rguenther@suse.de>
7740 * tree-ssa-pre.c (NECESSARY): Remove.
7741 (create_expression_by_pieces): Do not touch pass-local flags.
7742 (insert_into_preds_of_block): Likewise.
7743 (do_pre_regular_insertion): Likewise.
7744 (eliminate_insert): Likewise.
7745 (eliminate_dom_walker::before_dom_children): Likewise.
7746 (fini_eliminate): Do not look at inserted_exprs.
7747 (mark_operand_necessary): Remove.
7748 (remove_dead_inserted_code): Replace with simple work-list
7749 algorithm based on inserted_exprs and SSA uses.
7750 (pass_pre::execute): Re-order fini_eliminate and
7751 remove_dead_inserted_code.
7753 2017-09-06 Olivier Hainque <hainque@adacore.com>
7755 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
7756 for VxWorks 7. Adjust surrounding comments.
7758 2017-09-06 Richard Biener <rguenther@suse.de>
7760 * gimple-ssa-strength-reduction.c
7761 (find_candidates_dom_walker::before_dom_children): Also allow
7764 2017-09-06 Richard Biener <rguenther@suse.de>
7766 PR tree-optimization/82108
7767 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
7768 for gap in the non-permutation SLP case.
7770 2017-09-06 Martin Jambor <mjambor@suse.cz>
7772 PR tree-optimization/82078
7773 * tree-sra.c (sort_and_splice_var_accesses): Move call to
7774 add_access_to_work_queue...
7775 (build_accesses_from_assign): ...here.
7776 (propagate_all_subaccesses): Make sure racc is the group
7777 representative, if there is one.
7779 2017-09-06 Jakub Jelinek <jakub@redhat.com>
7782 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
7785 2017-09-06 Richard Biener <rguenther@suse.de>
7787 * gimple-ssa-strength-reduction.c
7788 (find_candidates_dom_walker::before_doom_children): Use a
7789 type and not a mode check.
7791 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
7794 * config/arm/predicates.md (arm_general_adddi_operand): Create new
7796 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
7798 2017-09-05 Jeff Law <law@redhat.com>
7800 PR tree-optimization/64910
7801 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
7802 cases where we have 3 or more operands.
7804 2017-09-05 Jakub Jelinek <jakub@redhat.com>
7807 * omp-low.c (lower_omp_for): Recompute tree invariant if
7808 gimple_omp_for_initial/final is ADDR_EXPR.
7811 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
7812 into gimple val before gimplification fo the COND_EXPR.
7814 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
7816 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
7817 REGION_COPY argument.
7818 (thread_through_all_blocks): Remove unused argument to
7819 duplicate_thread_path.
7821 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7822 Alan Hayward <alan.hayward@arm.com>
7823 David Sherwood <david.sherwood@arm.com>
7825 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
7826 Take a scalar_mode rather than a machine_mode.
7827 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
7828 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
7829 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
7830 (aarch64_gen_adjusted_ldpstp): Likewise.
7831 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
7833 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7834 Alan Hayward <alan.hayward@arm.com>
7835 David Sherwood <david.sherwood@arm.com>
7837 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
7838 Take a scalar_int_mode instead of a machine_mode.
7839 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
7840 (aarch64_output_scalar_simd_mov_immediate): Likewise.
7841 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
7842 (aarch64_simd_attr_length_rglist): Delete.
7843 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
7844 a scalar_int_mode instead of a machine_mode.
7845 (aarch64_add_offset): Likewise.
7846 (aarch64_internal_mov_immediate): Likewise
7847 (aarch64_add_constant_internal): Likewise.
7848 (aarch64_add_constant): Likewise.
7849 (aarch64_movw_imm): Likewise.
7850 (aarch64_rtx_arith_op_extract_p): Likewise.
7851 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
7852 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
7853 Remove assert that the mode isn't a vector.
7854 (aarch64_output_scalar_simd_mov_immediate): Likewise.
7855 (aarch64_expand_mov_immediate): Update calls after above changes.
7856 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
7857 (aarch64_and_bitmask_imm): Check for scalar integer modes.
7858 (aarch64_move_imm): Likewise.
7859 (aarch64_can_const_movi_rtx_p): Likewise.
7860 (aarch64_strip_extend): Likewise.
7861 (aarch64_extr_rtx_p): Likewise.
7862 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
7863 a CONST_INT when the mode parameter is VOIDmode.
7864 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
7866 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7868 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
7869 * stor-layout.c (bitwise_mode_for_mode): Likewise.
7870 (bitwise_type_for_mode): Update accordingly.
7872 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7874 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
7875 * stor-layout.c (mode_for_size_tree): Likewise.
7876 (mode_for_array): Update accordingly.
7877 (layout_decl): Likewise.
7878 (compute_record_mode): Likewise. Only set the mode once.
7880 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7882 * target.def (get_mask_mode): Change return type to opt_mode.
7884 * doc/tm.texi: Regenerate.
7885 * targhooks.h (default_get_mask_mode): Return an opt_mode.
7886 * targhooks.c (default_get_mask_mode): Likewise.
7887 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
7888 * optabs-query.c (can_vec_mask_load_store_p): Update use of
7889 targetm.get_mask_mode.
7890 * tree.c (build_truth_vector_type): Likewise.
7892 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7894 * machmode.h (mode_for_vector): Return an opt_mode.
7895 * stor-layout.c (mode_for_vector): Likewise.
7896 (mode_for_int_vector): Update accordingly.
7897 (layout_type): Likewise.
7898 * config/i386/i386.c (emit_memmov): Likewise.
7899 (ix86_expand_set_or_movmem): Likewise.
7900 (ix86_expand_vector_init): Likewise.
7901 (ix86_get_mask_mode): Likewise.
7902 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
7904 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
7905 * expmed.c (extract_bit_field_1): Likewise.
7906 * expr.c (expand_expr_real_2): Likewise.
7907 * optabs-query.c (can_vec_perm_p): Likewise.
7908 (can_vec_mask_load_store_p): Likewise.
7909 * optabs.c (expand_vec_perm): Likewise.
7910 * targhooks.c (default_get_mask_mode): Likewise.
7911 * tree-vect-stmts.c (vectorizable_store): Likewise.
7912 (vectorizable_load): Likewise.
7913 (get_vectype_for_scalar_type_and_size): Likewise.
7915 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7917 * machmode.h (mode_for_int_vector): New function.
7918 * stor-layout.c (mode_for_int_vector): Likewise.
7919 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
7920 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
7921 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
7922 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
7923 (s390_expand_vcond): Likewise.
7925 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7927 * machmode.h (opt_machine_mode): New type.
7928 (opt_mode<T>): Allow construction from anything that can be
7930 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
7931 (mode_for_size): Return an opt_machine_mode.
7932 * stor-layout.c (mode_for_size): Likewise.
7933 (mode_for_size_tree): Update call accordingly.
7934 (bitwise_mode_for_mode): Likewise.
7935 (make_fract_type): Likewise.
7936 (make_accum_type): Likewise.
7937 * caller-save.c (replace_reg_with_saved_mem): Update call
7939 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
7940 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
7941 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
7942 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
7943 * expmed.c (extract_bit_field_1): Likewise.
7944 * reload.c (get_secondary_mem): Likewise.
7945 * varasm.c (assemble_integer): Likewise.
7946 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
7949 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7951 * machmode.h (decimal_float_mode_for_size): New function.
7952 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
7953 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
7954 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
7955 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
7956 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
7957 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
7959 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7961 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
7962 (get_builtin_sync_mode): Likewise.
7963 (expand_ifn_atomic_compare_exchange): Likewise.
7964 (expand_builtin_atomic_clear): Likewise.
7965 (expand_builtin_atomic_test_and_set): Likewise.
7966 (fold_builtin_atomic_always_lock_free): Likewise.
7967 * calls.c (compute_argument_addresses): Likewise.
7968 (emit_library_call_value_1): Likewise.
7969 (store_one_arg): Likewise.
7970 * combine.c (combine_instructions): Likewise.
7971 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
7972 * config/arm/arm.c (arm_function_value): Likewise.
7973 (aapcs_allocate_return_reg): Likewise.
7974 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
7975 * config/i386/i386.c (construct_container): Likewise.
7976 (ix86_gimplify_va_arg): Likewise.
7977 (ix86_expand_sse_cmp): Likewise.
7978 (emit_memmov): Likewise.
7979 (emit_memset): Likewise.
7980 (expand_small_movmem_or_setmem): Likewise.
7981 (ix86_expand_pextr): Likewise.
7982 (ix86_expand_pinsr): Likewise.
7983 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
7984 * config/microblaze/microblaze.c (microblaze_block_move_straight):
7986 * config/mips/mips.c (mips_function_value_1) Likewise.
7987 (mips_block_move_straight): Likewise.
7988 (mips_expand_ins_as_unaligned_store): Likewise.
7989 * config/powerpcspe/powerpcspe.c
7990 (rs6000_darwin64_record_arg_advance_flush): Likewise.
7991 (rs6000_darwin64_record_arg_flush): Likewise.
7992 * config/rs6000/rs6000.c
7993 (rs6000_darwin64_record_arg_advance_flush): Likewise.
7994 (rs6000_darwin64_record_arg_flush): Likewise.
7995 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
7996 (sparc_function_value_1): Likewise.
7997 * config/spu/spu.c (adjust_operand): Likewise.
7998 (spu_emit_branch_or_set): Likewise.
7999 (arith_immediate_p): Likewise.
8000 * emit-rtl.c (gen_lowpart_common): Likewise.
8001 * expr.c (expand_expr_real_1): Likewise.
8002 * function.c (assign_parm_setup_block): Likewise.
8003 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8004 * reload1.c (alter_reg): Likewise.
8005 * stor-layout.c (mode_for_vector): Likewise.
8006 (layout_type): Likewise.
8008 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8010 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
8011 (spu_convert_move): Likewise.
8012 * lower-subreg.c (resolve_simple_move): Likewise.
8014 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8017 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
8018 define_insn to a define_expand.
8019 (altivec_vsum2sws_direct): New define_insn.
8020 (altivec_vsumsws): Convert from a define_insn to a define_expand.
8022 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8024 * config/arm/arm.c (arm_option_params_internal): Improve setting of
8027 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
8032 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
8033 realign stack if stack alignment needed is less than incoming
8036 2017-09-05 Marek Polacek <polacek@redhat.com>
8039 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
8042 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8044 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
8046 2017-09-05 Richard Biener <rguenther@suse.de>
8048 PR tree-optimization/82084
8049 * fold-const.c (can_native_encode_string_p): Handle wide characters.
8051 2017-09-05 Richard Biener <rguenther@suse.de>
8053 PR tree-optimization/82102
8054 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
8056 2017-09-05 Martin Liska <mliska@suse.cz>
8058 PR tree-optimization/82032
8059 * tree-cfg.c (generate_range_test): New function.
8060 * tree-cfg.h (generate_range_test): Declared here.
8061 * tree-cfgcleanup.c (convert_single_case_switch): New function.
8062 (cleanup_control_expr_graph): Use it.
8063 * tree-switch-conversion.c (try_switch_expansion): Remove
8065 (emit_case_nodes): Use generate_range_test.
8067 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
8070 * config/i386/i386.md (*<btsc><mode>_mask): Add
8071 TARGET_USE_BT to insn constraint.
8072 (*btr<mode>_mask): Ditto.
8074 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
8076 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
8077 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
8079 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
8082 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
8083 TARGET_NEON and TARGET_IWMMXT.
8084 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
8085 TARGET_NEON and TARGET_IWMMXT.
8086 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
8088 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
8090 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
8091 (ix86_rewrite_tls_address): Ditto.
8092 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
8093 (ix86_rewrite_tls_address_1): Ditto.
8094 (ix86_rewrite_tls_address): Ditto.
8095 * config/i386/predicates.md (tls_address_pattern): New predicate.
8096 * config/i386/i386.md (TLS address splitter): New splitter.
8098 2017-09-04 Richard Biener <rguenther@suse.de>
8100 PR tree-optimization/82084
8101 * fold-const.h (can_native_encode_string_p): Declare.
8102 * fold-const.c (can_native_encode_string_p): Factor out from ...
8103 (native_encode_string): ... here.
8104 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
8105 vectorizing stores from constants we later cannot handle.
8107 2017-09-04 Marek Polacek <polacek@redhat.com>
8110 * doc/invoke.texi: Update -Wtautological-compare documentation.
8112 2017-09-04 Jeff Law <law@redhat.com>
8114 PR tree-optimization/64910
8115 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
8116 swap the first and last operand if the last is a constant.
8118 2017-09-04 Marek Polacek <polacek@redhat.com>
8121 * convert.c (do_narrow): When sanitizing signed integer overflows,
8122 bail out for signed types.
8123 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
8125 2017-09-04 Richard Biener <rguenther@suse.de>
8127 PR tree-optimization/82060
8128 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
8129 Move devirtualization after stmt folding and before EH/AB/noreturn
8130 cleanup to get the stmt refs canonicalized. Use a bool instead
8131 of gimple_modified_p since that doesn't work for NOPs. Schedule
8132 NOPs generated by folding for removal.
8134 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8135 Alan Hayward <alan.hayward@arm.com>
8136 David Sherwood <david.sherwood@arm.com>
8138 * coretypes.h (pad_direction): New enum.
8139 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
8140 (FUNCTION_ARG_PADDING): Likewise.
8141 * target.def (function_arg_padding): New hook.
8142 * targhooks.h (default_function_arg_padding): Declare.
8143 * targhooks.c (default_function_arg_padding): New function.
8144 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
8145 (TARGET_FUNCTION_ARG_PADDING): ...this.
8146 * doc/tm.texi: Regenerate.
8147 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
8148 instead of direction.
8149 (compute_argument_addresses): Likewise.
8150 (load_register_parameters): Likewise.
8151 (emit_library_call_value_1): Likewise.
8152 (store_one_arg): Use targetm.calls.function_arg_padding instead
8153 of FUNCTION_ARG_PADDING.
8154 (must_pass_in_stack_var_size_or_pad): Likewise.
8155 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
8156 (emit_group_store): Likewise.
8157 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
8158 instead of FUNCTION_ARG_PADDING.
8159 (emit_push_insn): Likewise, and propagate enum change throughout
8161 * function.h (direction): Delete.
8162 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
8164 * function.c (assign_parm_find_stack_rtl): Likewise.
8165 (assign_parm_setup_block_p): Likewise.
8166 (assign_parm_setup_block): Likewise.
8167 (gimplify_parameters): Likewise.
8168 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
8169 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
8171 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
8172 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8173 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
8174 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
8175 (aarch64_function_arg_padding): ...this new function.
8176 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
8177 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8178 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
8179 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8180 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
8181 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8182 (arm_pad_arg_upward): Replace with...
8183 (arm_function_arg_padding): ...this new function.
8184 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
8186 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
8187 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
8188 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8189 (ia64_hpux_function_arg_padding): Replace with...
8190 (ia64_function_arg_padding): ...this new function. Use pad_direction
8191 instead of direction. Check for TARGET_HPUX.
8192 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
8193 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8194 (iq2000_function_arg_padding): New function.
8195 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
8196 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
8197 (mips_function_arg_padding): ...this new function.
8198 (mips_pad_reg_upward): Update accordingly.
8199 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8200 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
8201 targetm.calls.function_arg_padding.
8202 (FUNCTION_ARG_PADDING): Delete.
8203 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8204 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
8205 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8206 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
8207 (nios2_block_reg_padding): Return pad_direction instead of direction.
8208 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
8209 instead of direction.
8210 (nios2_function_arg_padding): Likewise. Make static.
8211 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8212 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
8213 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
8214 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
8215 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8216 (pa_function_arg_padding): Make static. Return pad_direction instead
8218 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
8219 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8220 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
8221 instead of direction. Use targetm.calls.function_arg_padding.
8222 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
8223 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
8224 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
8225 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
8226 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
8228 (function_arg_padding): Rename to...
8229 (rs6000_function_arg_padding): ...this. Make static. Return
8230 pad_direction instead of direction.
8231 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
8232 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
8233 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8234 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
8235 instead of direction. Use targetm.calls.function_arg_padding.
8236 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
8237 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
8238 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
8239 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
8240 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8241 (function_arg_padding): Rename to...
8242 (rs6000_function_arg_padding): ...this. Make static. Return
8243 pad_direction instead of direction.
8244 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
8245 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
8246 * config/s390/s390.c (s390_function_arg_padding): New function.
8247 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8248 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
8249 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
8250 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8251 (function_arg_padding): Rename to...
8252 (sparc_function_arg_padding): ...this. Make static. Return
8253 pad_direction instead of direction.
8254 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
8255 * config/spu/spu.c (spu_function_arg_padding): New function.
8256 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8257 * system.h (FUNCTION_ARG_PADDING): Poison.
8259 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8260 Alan Hayward <alan.hayward@arm.com>
8261 David Sherwood <david.sherwood@arm.com>
8263 * target.def (modes_tieable_p): New hook.
8264 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
8265 (TARGET_MODES_TIEABLE_P): ...this.
8266 * doc/tm.texi.in: Regenerate.
8267 * hooks.h (hook_bool_mode_mode_true): Declare.
8268 * hooks.c (hook_bool_mode_mode_true): New function.
8269 * combine.c (subst): Use targetm.modes_tieable_p instead of
8271 * dse.c (find_shift_sequence): Likewise.
8272 * expmed.c (extract_low_bits): Likewise.
8273 * lower-subreg.c: Include target.h.
8274 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
8276 * rtlanal.c (rtx_cost): Likewise.
8277 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
8278 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
8279 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
8280 (TARGET_MODES_TIEABLE_P): Redefine.
8281 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
8282 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
8283 (TARGET_MODES_TIEABLE_P): Redefine.
8284 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
8285 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
8286 (arc_modes_tieable_p): New function.
8287 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
8288 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
8289 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
8290 (arm_modes_tieable_p): Make static.
8291 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
8292 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
8293 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
8294 (TARGET_MODES_TIEABLE_P): Redefine.
8295 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
8296 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
8297 (TARGET_MODES_TIEABLE_P): Redefine.
8298 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
8299 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
8300 (cr16_modes_tieable_p): New function.
8301 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
8302 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
8303 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
8304 (TRULY_NOOP_TRUNCATION): Update comment.
8305 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
8306 (TRULY_NOOP_TRUNCATION): Update comment.
8307 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
8308 (frv_modes_tieable_p): New function.
8309 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
8310 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
8311 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
8312 (TARGET_MODES_TIEABLE_P): Redefine.
8313 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
8314 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
8315 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
8316 (TARGET_MODES_TIEABLE_P): Redefine.
8317 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
8318 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
8319 (ia64_modes_tieable_p): New function.
8320 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
8321 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
8322 (iq2000_modes_tieable_p): New function.
8323 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
8324 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
8325 (lm32_modes_tieable_p): New function.
8326 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
8327 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
8328 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
8329 (TARGET_MODES_TIEABLE_P): Redefine.
8330 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
8331 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
8332 (m32r_modes_tieable_p): New function.
8333 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
8334 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
8335 (m68k_modes_tieable_p): New function.
8336 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
8337 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
8338 (mcore_modes_tieable_p): New function.
8339 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
8340 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
8342 (TARGET_MODES_TIEABLE_P): Redefine.
8343 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
8344 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
8345 * config/mips/mips.c (mips_modes_tieable_p): Make static.
8346 (TARGET_MODES_TIEABLE_P): Redefine.
8347 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
8348 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
8349 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
8350 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
8351 (mn10300_modes_tieable_p): ...this and make static.
8352 (TARGET_MODES_TIEABLE_P): Redefine.
8353 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
8354 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
8355 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
8356 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
8357 (msp430_modes_tieable_p): Make static.
8358 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
8359 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
8360 (TARGET_MODES_TIEABLE_P): Redefine.
8361 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
8362 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
8363 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
8364 (TARGET_MODES_TIEABLE_P): Redefine.
8365 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
8366 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
8367 * config/pa/pa.c (pa_modes_tieable_p): Make static.
8368 (TARGET_MODES_TIEABLE_P): Redefine.
8369 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
8370 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
8371 (pdp11_modes_tieable_p): New function.
8372 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
8373 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
8374 (rs6000_modes_tieable_p): New function.
8375 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
8376 * config/powerpcspe/powerpcspe.md: Update comment.
8377 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
8378 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
8379 (TARGET_MODES_TIEABLE_P): Redefine.
8380 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
8381 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
8382 (rl78_modes_tieable_p): New function.
8383 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
8384 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
8385 (rs6000_modes_tieable_p): New function.
8386 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
8387 * config/rs6000/rs6000.md: Update comment.
8388 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
8389 * config/rx/rx.c (rx_modes_tieable_p): New function.
8390 (TARGET_MODES_TIEABLE_P): Redefine.
8391 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
8392 * config/s390/s390.c (s390_modes_tieable_p): New function.
8393 (TARGET_MODES_TIEABLE_P): Redefine.
8394 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
8395 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
8396 (sh_modes_tieable_p): New function.
8397 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
8398 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
8399 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
8400 (sparc_modes_tieable_p): Make static.
8401 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
8402 * config/spu/spu.c (spu_modes_tieable_p): New function.
8403 (TARGET_MODES_TIEABLE_P): Redefine.
8404 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
8405 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
8406 (TARGET_MODES_TIEABLE_P): Redefine.
8407 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
8408 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
8409 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
8410 * config/v850/v850.c (v850_modes_tieable_p): New function.
8411 (TARGET_MODES_TIEABLE_P): Redefine.
8412 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
8413 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
8414 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
8415 (visium_modes_tieable_p): New function.
8416 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
8417 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
8418 (xtensa_modes_tieable_p): New function.
8419 * system.h (MODES_TIEABLE_P): Poison.
8421 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8422 Alan Hayward <alan.hayward@arm.com>
8423 David Sherwood <david.sherwood@arm.com>
8425 * target.def (hard_regno_mode_ok): New hook.
8426 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
8427 (TARGET_HARD_REGNO_MODE_OK): ...this.
8428 * doc/tm.texi.in: Regenerate.
8429 * hooks.h (hook_bool_uint_mode_true): Declare.
8430 * hooks.c (hook_bool_uint_mode_true): New function.
8431 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
8433 * genpreds.c (write_insn_preds_c): Add an include of target.h.
8434 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
8435 instead of HARD_REGNO_MODE_OK.
8436 * caller-save.c: Include target.h.
8437 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
8439 * combine.c (can_combine_p): Likewise.
8440 (combinable_i3pat): Likewise.
8441 (can_change_dest_mode): Likewise.
8442 * expr.c (init_expr_target): Likewise.
8443 (convert_move): Likewise.
8444 (convert_modes): Likewise.
8445 * ira.c (setup_prohibited_class_mode_regs): Likewise.
8446 (setup_prohibited_mode_move_regs): Likewise.
8447 * ira.h (target_ira): Likewise.
8448 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8449 * lra-constraints.c (process_alt_operands): Likewise.
8450 (split_reg): Likewise.
8451 * recog.c (peep2_find_free_register): Likewise.
8452 * ree.c (combine_reaching_defs): Likewise.
8453 * regcprop.c (maybe_mode_change): Likewise.
8454 * reginfo.c (init_reg_sets_1): Likewise.
8455 (choose_hard_reg_mode): Likewise.
8456 (simplifiable_subregs): Likewise.
8457 * regrename.c (check_new_reg_p): Likewise.
8458 * reload.c (find_valid_class): Likewise.
8459 (find_valid_class_1): Likewise.
8460 (reload_inner_reg_of_subreg): Likewise.
8461 (push_reload): Likewise.
8462 (combine_reloads): Likewise.
8463 (find_dummy_reload): Likewise.
8464 (find_reloads): Likewise.
8465 * reload1.c (find_reg): Likewise.
8466 (set_reload_reg): Likewise.
8467 (allocate_reload_reg): Likewise.
8468 (choose_reload_regs): Likewise.
8469 (reload_adjust_reg_for_temp): Likewise.
8470 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
8471 (simplify_subreg_regno): Likewise.
8472 * sel-sched.c (init_regs_for_mode): Likewise.
8473 * varasm.c (make_decl_rtl): Likewise.
8474 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
8475 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
8477 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
8478 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
8479 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8480 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
8481 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
8482 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8483 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
8484 (arc_mode_class): Delete.
8485 (HARD_REGNO_MODE_OK): Delete.
8486 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8487 (arc_hard_regno_mode_ok): Rename old array to...
8488 (arc_hard_regno_mode_ok_modes): ...this.
8489 (arc_conditional_register_usage): Update accordingly.
8490 (arc_mode_class): Make static.
8491 (arc_hard_regno_mode_ok): New function.
8492 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
8493 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
8494 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8495 (arm_hard_regno_mode_ok): Make static.
8496 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
8498 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
8499 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
8500 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
8502 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8503 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
8504 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
8505 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
8506 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
8507 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8508 * config/bfin/predicates.md (valid_reg_operand): Use
8509 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
8510 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
8511 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
8512 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8513 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
8514 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
8515 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8516 (cr16_hard_regno_mode_ok): Make static and return a bool.
8517 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
8518 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8519 (cris_hard_regno_mode_ok): New function.
8520 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
8521 (epiphany_mode_class): Delete.
8522 (HARD_REGNO_MODE_OK): Delete.
8523 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
8524 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8525 (hard_regno_mode_ok): Rename to...
8526 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
8527 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
8528 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
8530 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
8531 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
8532 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8533 (frv_hard_regno_mode_ok): Make static and return a bool.
8534 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
8536 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
8537 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
8538 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
8539 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
8541 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8542 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
8543 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
8544 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
8546 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8547 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
8548 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8549 (ia64_hard_regno_mode_ok): New function.
8550 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
8551 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8552 (iq2000_hard_regno_mode_ok): New function.
8553 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
8554 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8555 (lm32_hard_regno_mode_ok): New function.
8556 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
8557 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
8558 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
8559 instead of HARD_REGNO_MODE_OK.
8560 (m32c_hard_regno_ok): Rename to...
8561 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
8562 (m32c_cannot_change_mode_class): Update accordingly.
8563 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8564 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
8565 (m32r_mode_class): Delete.
8566 (HARD_REGNO_MODE_OK): Delete.
8567 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8568 (m32r_hard_regno_mode_ok): Rename to...
8569 (m32r_hard_regno_modes): ...this.
8570 (m32r_mode_class): Make static.
8571 (m32r_hard_regno_mode_ok): New function.
8572 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
8573 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
8574 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8575 (m68k_hard_regno_mode_ok): Make static.
8576 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
8577 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8578 (mcore_hard_regno_mode_ok): New function.
8579 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
8580 (HARD_REGNO_MODE_OK): Delete.
8581 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
8583 (microblaze_hard_regno_mode_ok_p): ...this and make static.
8584 (microblaze_hard_regno_mode_ok): New function.
8585 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8586 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
8587 (mips_hard_regno_mode_ok): Delete.
8588 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
8589 (mips_hard_regno_mode_ok_p): ...this and make static.
8590 (mips_hard_regno_mode_ok_p): Rename to...
8591 (mips_hard_regno_mode_ok_uncached): ...this.
8592 (mips_hard_regno_mode_ok): New function.
8593 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
8594 of HARD_REGNO_MODE_OK.
8595 (mips_option_override): Update after above name changes.
8596 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8597 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
8598 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
8599 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
8600 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
8601 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8602 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
8603 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
8604 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
8605 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8606 (msp430_hard_regno_mode_ok): Make static and return a bool.
8607 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
8608 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
8609 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
8611 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8612 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
8613 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
8614 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
8615 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
8616 (PA_HARD_REGNO_MODE_OK): ...this
8617 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
8618 (PA_HARD_REGNO_MODE_OK): ...this.
8619 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8620 (pa_hard_regno_mode_ok): New function.
8621 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
8622 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8623 (pdp11_hard_regno_mode_ok): New function.
8624 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
8625 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
8627 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
8629 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8630 (rs6000_hard_regno_mode_ok): Rename to...
8631 (rs6000_hard_regno_mode_ok_uncached): ...this.
8632 (rs6000_init_hard_regno_mode_ok): Update accordingly.
8633 (rs6000_hard_regno_mode_ok): New function.
8634 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
8635 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
8636 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
8637 (riscv_hard_regno_mode_ok): ...this and make static.
8638 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8639 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
8640 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
8641 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8642 (rl78_hard_regno_mode_ok): Make static and return bool.
8643 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
8644 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
8646 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
8647 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8648 (rs6000_hard_regno_mode_ok): Rename to...
8649 (rs6000_hard_regno_mode_ok_uncached): ...this.
8650 (rs6000_init_hard_regno_mode_ok): Update accordingly.
8651 (rs6000_hard_regno_mode_ok): New function.
8652 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
8653 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
8654 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8655 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
8656 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
8657 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
8658 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8659 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
8660 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
8661 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8662 (sh_hard_regno_mode_ok): Make static.
8663 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
8664 instead of HARD_REGNO_MODE_OK.
8665 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
8666 (sparc_mode_class): Delete.
8667 (HARD_REGNO_MODE_OK): Delete.
8668 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8669 (hard_regno_mode_classes): Make static.
8670 (sparc_mode_class): Likewise.
8671 (sparc_hard_regno_mode_ok): New function.
8672 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
8673 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
8674 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
8676 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8677 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
8678 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
8679 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
8680 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
8681 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8682 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
8683 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
8684 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8685 (visium_hard_regno_mode_ok): New function.
8686 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
8687 instead of HARD_REGNO_MODE_OK.
8688 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
8689 (HARD_REGNO_MODE_OK): Delete.
8690 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
8691 (xtensa_hard_regno_mode_ok_p): ...this and make static.
8692 (xtensa_option_override): Update accordingly.
8693 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8694 (xtensa_hard_regno_mode_ok): New function.
8695 * system.h (HARD_REGNO_MODE_OK): Poison.
8697 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8698 Alan Hayward <alan.hayward@arm.com>
8699 David Sherwood <david.sherwood@arm.com>
8701 * target.def (hard_regno_call_part_clobbered): New hook.
8702 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
8703 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
8704 * doc/tm.texi: Regenerate.
8705 * hooks.h (hook_bool_uint_mode_false): Declare.
8706 * hooks.c (hook_bool_uint_mode_false): New function.
8707 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8708 * cselib.c (cselib_process_insn): Use
8709 targetm.hard_regno_call_part_clobbered instead of
8710 HARD_REGNO_CALL_PART_CLOBBERED.
8711 * ira-conflicts.c (ira_build_conflicts): Likewise.
8712 * ira-costs.c (ira_tune_allocno_costs): Likewise.
8713 * lra-constraints.c (need_for_call_save_p): Likewise.
8714 * lra-lives.c: Include target.h.
8715 (check_pseudos_live_through_calls): Use
8716 targetm.hard_regno_call_part_clobbered instead of
8717 HARD_REGNO_CALL_PART_CLOBBERED.
8718 * regcprop.c: Include target.h.
8719 (copyprop_hardreg_forward_1): Use
8720 targetm.hard_regno_call_part_clobbered instead of
8721 HARD_REGNO_CALL_PART_CLOBBERED.
8722 * reginfo.c (choose_hard_reg_mode): Likewise.
8723 * regrename.c (check_new_reg_p): Likewise.
8724 * reload.c (find_equiv_reg): Likewise.
8725 * reload1.c (emit_reload_insns): Likewise.
8726 * sched-deps.c (deps_analyze_insn): Likewise.
8727 * sel-sched.c (init_regs_for_mode): Likewise.
8728 (mark_unavailable_hard_regs): Likewise.
8729 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
8730 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8731 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
8733 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8734 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8735 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
8737 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
8739 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8740 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8741 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
8743 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8744 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8745 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
8747 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8748 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
8750 * config/powerpcspe/powerpcspe.c
8751 (rs6000_hard_regno_call_part_clobbered): New function.
8752 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8753 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8754 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
8756 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8757 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8758 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
8760 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8761 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8762 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
8764 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8765 Alan Hayward <alan.hayward@arm.com>
8766 David Sherwood <david.sherwood@arm.com>
8768 * rtl.h (subreg_memory_offset): Declare.
8769 * emit-rtl.c (subreg_memory_offset): New function.
8770 * expmed.c (store_bit_field_1): Use it.
8771 * expr.c (undefined_operand_subword_p): Likewise.
8772 * simplify-rtx.c (simplify_subreg): Likewise.
8774 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
8776 PR rtl-optimization/57448
8779 * optabs.c (expand_atomic_load): Place compiler memory barriers if
8780 using atomic_load pattern.
8781 (expand_atomic_store): Likewise.
8783 2017-09-04 Jakub Jelinek <jakub@redhat.com>
8786 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
8787 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
8788 handling. Use replace_call_with_value with NULL instead of
8789 gsi_replace, unlink_stmt_vdef and release_defs.
8791 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
8794 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
8796 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8799 * rtl.h (emit_library_call_value_1): Declare.
8800 (emit_library_call): Replace declaration with a series of overloads.
8801 Remove the parameter count argument.
8802 (emit_library_call_value): Likewise.
8803 * calls.c (emit_library_call_value_1): Make global. Replace varargs
8804 with an "rtx_mode_t *".
8805 (emit_library_call_value): Delete.
8806 (emit_library_call): Likewise.
8807 * asan.c (asan_emit_stack_protection): Update calls accordingly.
8808 (asan_emit_allocas_unpoison): Likewise.
8809 * builtins.c (expand_builtin_powi): Likewise.
8810 (expand_asan_emit_allocas_unpoison): Likewise.
8811 * cfgexpand.c (expand_main_function): Likewise.
8812 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
8813 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
8814 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
8815 * config/arm/arm.c (arm_trampoline_init): Likewise.
8816 (arm_call_tls_get_addr): Likewise.
8817 (arm_expand_divmod_libfunc): Likewise.
8818 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
8819 (smulsi3_highpart): Likewise.
8820 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
8821 (c6x_expand_compare): Likewise.
8822 (c6x_expand_movmem): Likewise.
8823 * config/frv/frv.c (frv_trampoline_init): Likewise.
8824 * config/i386/i386.c (ix86_trampoline_init): Likewise.
8825 (ix86_expand_divmod_libfunc): Likewise.
8826 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
8827 (ia64_expand_compare): Likewise.
8828 (ia64_profile_hook): Likewise.
8829 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
8830 (nonlocal_goto): Likewise.
8831 (restore_stack_nonlocal): Likewise.
8832 * config/m32r/m32r.c (block_move_call): Likewise.
8833 (m32r_trampoline_init): Likewise.
8834 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
8835 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
8836 (m68k_call_m68k_read_tp): Likewise.
8837 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
8838 (microblaze_expand_divide): Likewise.
8839 * config/mips/mips.h (mips_args): Likewise.
8840 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
8841 (MIPS_ICACHE_SYNC): Likewise.
8842 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
8843 (nios2_trampoline_init): Likewise.
8844 * config/pa/pa.c (hppa_tls_call): Likewise.
8845 (pa_trampoline_init): Likewise.
8846 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
8847 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
8848 (expand_strn_compare): Likewise.
8849 (rs6000_generate_compare): Likewise.
8850 (rs6000_expand_float128_convert): Likewise.
8851 (output_profile_hook): Likewise.
8852 (rs6000_trampoline_init): Likewise.
8853 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
8854 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
8855 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
8856 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
8857 (rs6000_generate_compare): Likewise.
8858 (rs6000_expand_float128_convert): Likewise.
8859 (output_profile_hook): Likewise.
8860 (rs6000_trampoline_init): Likewise.
8861 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
8862 * config/sh/sh.c (sh_trampoline_init): Likewise.
8863 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
8864 (sparc_emit_float_lib_cmp): Likewise.
8865 (sparc32_initialize_trampoline): Likewise.
8866 (sparc64_initialize_trampoline): Likewise.
8867 (sparc_profile_hook): Likewise.
8868 * config/spu/spu.c (ea_load_store): Likewise.
8869 * config/spu/spu.md (floatunssidf2): Likewise.
8870 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
8871 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
8872 * config/visium/visium.c (expand_block_move_4): Likewise.
8873 (expand_block_move_2): Likewise.
8874 (expand_block_move_1): Likewise.
8875 (expand_block_set_4): Likewise.
8876 (expand_block_set_2): Likewise.
8877 (expand_block_set_1): Likewise.
8878 (visium_trampoline_init): Likewise.
8879 (visium_profile_hook): Likewise.
8880 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
8881 (xtensa_setup_frame_addresses): Likewise.
8882 (xtensa_trampoline_init): Likewise.
8883 * except.c (sjlj_emit_function_enter): Likewise.
8884 (sjlj_emit_function_exit): Likewise.
8885 * explow.c (allocate_dynamic_stack_space): Likewise.
8886 (probe_stack_range): Likewise.
8887 * expr.c (convert_mode_scalar): Likewise.
8888 * optabs.c (expand_binop): Likewise.
8889 (expand_twoval_binop_libfunc): Likewise.
8890 (expand_unop): Likewise.
8891 (prepare_cmp_insn): Likewise.
8892 (prepare_float_lib_cmp): Likewise.
8893 (expand_float): Likewise.
8894 (expand_fix): Likewise.
8895 (expand_fixed_convert): Likewise.
8896 (maybe_emit_sync_lock_test_and_set): Likewise.
8897 (expand_atomic_compare_and_swap): Likewise.
8898 (expand_mem_thread_fence): Likewise.
8899 (expand_atomic_fetch_op): Likewise.
8901 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
8903 * doc/generic.texi (OpenACC): Adjust URL.
8904 * doc/invoke.texi (C Dialect Options): Ditto.
8906 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
8908 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
8909 predicate for operand 1. Add (m,<S>) constraint.
8910 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
8911 Prevent memory operand 1 with register operand 2.
8913 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
8915 PR rtl-optimization/82024
8916 * combine.c (try_combine): If the combination result is a PARALLEL,
8917 and we only need to retain the SET in there that would be placed
8918 at I2, check that we can place that at I3 instead, before doing so.
8920 2017-09-01 Jakub Jelinek <jakub@redhat.com>
8923 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
8925 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
8926 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
8929 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
8930 Jeff Law <law@redhat.com>
8932 * varasm.c (bss_initializer_p): Do not put constants into .bss
8933 (categorize_decl_for_section): Handle bss_initializer_p returning
8934 false when DECL_INITIAL is NULL.
8936 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8939 * config/s390/s390.c (s390_can_inline_p): New function.
8941 2017-09-01 Jeff Law <law@redhat.com>
8943 PR tree-optimization/82052
8944 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
8945 Always initialize the returned slot after a hash table miss
8946 when INSERT is true.
8948 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
8950 * config/s390/s390.md (mem_signal_fence): Remove.
8951 * doc/md.texi (mem_signal_fence): Remove.
8952 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
8954 * target-insns.def (mem_signal_fence): Remove.
8956 2017-09-01 Jakub Jelinek <jakub@redhat.com>
8959 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
8962 * asan.c (create_odr_indicator): Strip name encoding from assembler
8963 name before appending it after __odr_asan_.
8965 2017-09-01 Martin Liska <mliska@suse.cz>
8967 PR tree-optimization/82059
8968 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
8969 frequency only when an edge is redirected.
8971 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
8973 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
8974 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
8975 (arc_conditional_register_usage): Remove ARC600 lp_count
8977 (arc_file_start): Emit Tag_ARC_CPU_variation.
8978 (arc_can_use_doloop_p): New conditions to use ZOLs.
8979 (hwloop_fail): New function.
8980 (hwloop_optimize): Likewise.
8981 (hwloop_pattern_reg): Likewise.
8982 (arc_doloop_hooks): New struct, to be used with reorg_loops.
8983 (arc_reorg_loops): New function, calls reorg_loops.
8984 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
8985 (arc600_corereg_hazard): Remove ZOL checking, case handled by
8987 (arc_loop_hazard): Remove function, functionality moved into
8989 (arc_hazard): Remove arc_loop_hazard call.
8990 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
8991 into hwloop_optimize.
8992 (arc_label_align): Remove ZOL handling.
8993 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
8994 * config/arc/arc.md (doloop_begin): Remove pattern.
8995 (doloop_begin_i): Likewise.
8996 (doloop_end_i): Likewise.
8997 (doloop_fallback): Likewise.
8998 (doloop_fallback_m): Likewise.
8999 (doloop_end): Reimplement expand.
9000 (arc_lp): New pattern for LP instruction.
9001 (loop_end): New pattern.
9002 (loop_fail): Likewise.
9003 (decrement_and_branch_until_zero): Likewise.
9004 * config/arc/arc.opt (mlpc-width): New option.
9005 * doc/invoke.texi (mlpc-width): Document option.
9007 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9009 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
9010 (arc_ccfsm_advance): Fix checking for delay slots.
9011 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
9013 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9015 * config/arc/arc.md (movqi_insn): Add stores to save constant long
9017 (movhi_insn): Update store instruction constraint which are saving
9018 6-bit short immediates.
9019 (movsi_insn): Consider also short scaled load operations.
9020 (zero_extendhisi2_i): Use Usd constraint instead of T.
9021 (extendhisi2_i): Add q constraint.
9022 (arc_clzsi2): Add type and length attributes.
9023 (arc_ctzsi2): Likewise.
9024 * config/arc/constraints.md (Usc): Update constraint, the
9025 assembler can parse two relocations for a single instruction.
9027 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9029 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
9030 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
9032 2017-08-31 Olivier Hainque <hainque@adacore.com>
9034 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
9035 match as powerpc-wrs-vxworks*.
9037 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
9039 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
9040 register constraint for by-element operand.
9041 (aarch64_mls_elt_merge<mode>): Likewise.
9043 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9045 * config/arc/arc.c (arc_can_follow_jump): Check for short
9048 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9050 * config.gcc: Use g.opt for arc.
9051 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
9052 functionality moved to ...
9053 (legitimate_scaled_address_p): New function, ...here.
9054 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
9055 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
9056 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
9058 (arc_override_options): Handle G option.
9059 (arc_output_pic_addr_const): Correct function definition.
9060 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
9061 (arc_decl_anon_ns_mem_p): Delete.
9062 (arc_in_small_data_p): Overhaul this function to take into
9063 consideration the value given via G option.
9064 (arc_rewrite_small_data_1): Renamed and corrected old
9065 arc_rewrite_small_data function.
9066 (arc_rewrite_small_data): New function.
9067 (small_data_pattern): Don't use pic_offset_table_rtx.
9068 * config/arc/arc.h (CC1_SPEC): Recognize G option.
9069 * config/arc/simdext.md (movmisalignv2hi): Use
9070 prepare_move_operands function.
9072 (movmisalign*): Likewise.
9073 * doc/invoke.texi (ARC options): Document -G option.
9075 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9077 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
9079 * config/arc/arc.c (arc_print_operand): Output scalled address for
9080 sdata whenever is possible.
9081 (arc_in_small_data_p): Allow sdata for 64bit datum when double
9082 load/stores are available.
9083 (compact_sda_memory_operand): Check for the alignment required by
9084 code density instructions.
9085 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
9087 * config/arc/constraints.md (Usd): Update constraint.
9088 (Us0): New constraint.
9089 (Usc): Update constraint.
9091 2017-08-31 Richard Biener <rguenther@suse.de>
9094 * dwarf2out.c (dwarf2out_early_global_decl): Process each
9097 2017-08-31 Tamar Christina <tamar.christina@arm.com>
9099 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
9100 Resize type_signature.
9102 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9103 Alan Hayward <alan.hayward@arm.com>
9104 David Sherwood <david.sherwood@arm.com>
9106 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
9107 subregs whose inner modes can be stored in GPRs.
9108 (aarch64_classify_index): Likewise.
9110 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9111 Alan Hayward <alan.hayward@arm.com>
9112 David Sherwood <david.sherwood@arm.com>
9114 * config/aarch64/iterators.md (V_cmp_result): Rename to...
9115 (V_INT_EQUIV): ...this.
9116 (v_cmp_result): Rename to...
9117 (v_int_equiv): ...this.
9118 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
9119 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
9120 (copysign<mode>3): Likewise.
9121 (aarch64_simd_bsl<mode>_internal): Likewise.
9122 (aarch64_simd_bsl<mode>): Likewise.
9123 (vec_cmp<mode><mode>): Likewise.
9124 (vcond<mode><mode>): Likewise.
9125 (vcond<v_cmp_mixed><mode>): Likewise.
9126 (vcondu<mode><v_cmp_mixed>): Likewise.
9127 (aarch64_cm<optab><mode>): Likewise.
9128 (aarch64_cmtst<mode>): Likewise.
9129 (aarch64_fac<optab><mode>): Likewise.
9130 (vec_perm_const<mode>): Likewise.
9131 (vcond_mask_<mode><v_cmp_result>): Rename to...
9132 (vcond_mask_<mode><v_int_equiv>): ...this.
9133 (vec_cmp<mode><v_cmp_result>): Rename to...
9134 (vec_cmp<mode><v_int_equiv>): ...this.
9136 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9137 Alan Hayward <alan.hayward@arm.com>
9138 David Sherwood <david.sherwood@arm.com>
9140 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
9142 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
9143 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
9144 (UNSPEC_LD4_DREG): New unspecs.
9145 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
9146 (aarch64_ld2<mode>_dreg_be): Replace with...
9147 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
9149 (aarch64_ld3<mode>_dreg_le)
9150 (aarch64_ld3<mode>_dreg_be): Replace with...
9151 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
9153 (aarch64_ld4<mode>_dreg_le)
9154 (aarch64_ld4<mode>_dreg_be): Replace with...
9155 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
9158 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9160 PR tree-optimization/81987
9161 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
9162 insert an initializer in a location not dominated by the stride
9165 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
9167 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
9168 on the entire header of the finally block in the fallthru case.
9170 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
9172 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
9174 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
9176 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
9177 rs6000_emit_move_from_cr and call renamed function.
9178 (rs6000_emit_prologue): Call renamed functions.
9179 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
9180 movesi_from_cr, remove volatile CRs.
9182 2017-08-30 Jon Beniston <jon@beniston.com>
9183 Richard Biener <rguenther@suse.de>
9185 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
9186 of VECTOR_MODE_P check.
9187 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
9188 element vector types.
9190 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9192 * df.h (df_read_modify_subreg_p): Remove in favor of...
9193 * rtl.h (read_modify_subreg_p): ...this new function. Take a
9194 const_rtx instead of an rtx.
9195 * cprop.c (local_cprop_find_used_regs): Update accordingly.
9196 * df-problems.c (df_word_lr_mark_ref): Likewise.
9197 * ira-lives.c (mark_pseudo_reg_live): Likewise.
9198 (mark_pseudo_reg_dead): Likewise.
9199 (mark_ref_dead): Likewise.
9200 * reginfo.c (init_subregs_of_mode): Likewise.
9201 * sched-deps.c (sched_analyze_1): Likewise.
9202 * df-scan.c (df_def_record_1): Likewise.
9203 (df_uses_record): Likewise.
9204 (df_read_modify_subreg_p): Remove in favor of...
9205 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
9206 const_rtx instead of an rtx.
9208 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9209 Alan Hayward <alan.hayward@arm.com>
9210 David Sherwood <david.sherwood@arm.com>
9212 * rtl.h (partial_subreg_p): New function.
9213 * caller-save.c (save_call_clobbered_regs): Use it.
9214 * calls.c (expand_call): Likewise.
9215 * combine.c (combinable_i3pat): Likewise.
9216 (simplify_set): Likewise.
9217 (make_extraction): Likewise.
9218 (make_compound_operation_int): Likewise.
9219 (gen_lowpart_or_truncate): Likewise.
9220 (force_to_mode): Likewise.
9221 (make_field_assignment): Likewise.
9222 (reg_truncated_to_mode): Likewise.
9223 (record_truncated_value): Likewise.
9224 (move_deaths): Likewise.
9225 * cse.c (record_jump_cond): Likewise.
9226 (cse_insn): Likewise.
9227 * cselib.c (cselib_lookup_1): Likewise.
9228 * expmed.c (extract_bit_field_using_extv): Likewise.
9229 * function.c (assign_parm_setup_reg): Likewise.
9230 * ifcvt.c (noce_convert_multiple_sets): Likewise.
9231 * ira-build.c (create_insn_allocnos): Likewise.
9232 * lra-coalesce.c (merge_pseudos): Likewise.
9233 * lra-constraints.c (match_reload): Likewise.
9234 (simplify_operand_subreg): Likewise.
9235 (curr_insn_transform): Likewise.
9236 * lra-lives.c (process_bb_lives): Likewise.
9237 * lra.c (new_insn_reg): Likewise.
9238 (lra_substitute_pseudo): Likewise.
9239 * regcprop.c (mode_change_ok): Likewise.
9240 (maybe_mode_change): Likewise.
9241 (copyprop_hardreg_forward_1): Likewise.
9242 * reload.c (push_reload): Likewise.
9243 (find_reloads): Likewise.
9244 (find_reloads_subreg_address): Likewise.
9245 * reload1.c (alter_reg): Likewise.
9246 (eliminate_regs_1): Likewise.
9247 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9249 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
9251 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
9252 back to if statements, including unpack.
9254 2017-08-30 Martin Liska <mliska@suse.cz>
9257 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
9259 (func_checker::compare_asm_inputs_outputs): ... this function.
9260 (func_checker::compare_gimple_asm): Use the function to compare
9261 also ASM constrains.
9262 * ipa-icf-gimple.h: Rename the function.
9264 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9265 Alan Hayward <alan.hayward@arm.com>
9266 David Sherwood <david.sherwood@arm.com>
9268 * coretypes.h (complex_mode): New type.
9269 * gdbhooks.py (build_pretty_printer): Handle it.
9270 * machmode.h (complex_mode): New class.
9271 (complex_mode::includes_p): New function.
9272 (is_complex_int_mode): Likewise.
9273 (is_complex_float_mode): Likewise.
9274 * genmodes.c (get_mode_class): Handle complex mode classes.
9275 * function.c (expand_function_end): Use is_complex_int_mode.
9277 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9278 Alan Hayward <alan.hayward@arm.com>
9279 David Sherwood <david.sherwood@arm.com>
9281 * coretypes.h (scalar_mode_pod): New typedef.
9282 * gdbhooks.py (build_pretty_printer): Handle it.
9283 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
9284 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
9285 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
9286 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
9289 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9290 Alan Hayward <alan.hayward@arm.com>
9291 David Sherwood <david.sherwood@arm.com>
9293 * machmode.h (mode_for_vector): Take a scalar_mode instead
9295 * stor-layout.c (mode_for_vector): Likewise.
9296 * explow.c (promote_mode): Use as_a <scalar_mode>.
9297 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
9299 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9300 Alan Hayward <alan.hayward@arm.com>
9301 David Sherwood <david.sherwood@arm.com>
9303 * target.def (preferred_simd_mode): Take a scalar_mode
9304 instead of a machine_mode.
9305 * targhooks.h (default_preferred_simd_mode): Likewise.
9306 * targhooks.c (default_preferred_simd_mode): Likewise.
9307 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
9308 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
9309 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
9310 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
9311 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
9312 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
9313 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
9314 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
9316 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
9317 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
9318 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
9319 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
9320 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
9321 * doc/tm.texi: Regenerate.
9322 * optabs-query.c (can_vec_mask_load_store_p): Return false for
9325 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9326 Alan Hayward <alan.hayward@arm.com>
9327 David Sherwood <david.sherwood@arm.com>
9329 * target.def (scalar_mode_supported_p): Take a scalar_mode
9330 instead of a machine_mode.
9331 * targhooks.h (default_scalar_mode_supported_p): Likewise.
9332 * targhooks.c (default_scalar_mode_supported_p): Likewise.
9333 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
9334 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
9335 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
9336 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
9337 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
9338 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
9339 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
9340 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
9341 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
9342 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
9343 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
9344 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
9346 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
9347 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
9348 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
9349 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
9350 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
9352 * doc/tm.texi: Regenerate.
9354 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9355 Alan Hayward <alan.hayward@arm.com>
9356 David Sherwood <david.sherwood@arm.com>
9358 * coretypes.h (opt_scalar_mode): New typedef.
9359 * gdbhooks.py (build_pretty_printers): Handle it.
9360 * machmode.h (mode_iterator::get_2xwider): Add overload for
9362 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
9364 * expr.c (convert_mode_scalar): Likewise.
9365 * omp-low.c (omp_clause_aligned_alignment): Likewise.
9366 * optabs.c (expand_float): Likewise.
9367 (expand_fix): Likewise.
9368 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
9370 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9371 Alan Hayward <alan.hayward@arm.com>
9372 David Sherwood <david.sherwood@arm.com>
9374 * optabs.c (expand_float): Explicitly check for scalars before
9375 using a branching expansion.
9376 (expand_fix): Likewise.
9378 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9379 Alan Hayward <alan.hayward@arm.com>
9380 David Sherwood <david.sherwood@arm.com>
9382 * expr.c (convert_mode): Split scalar handling out into...
9383 (convert_mode_scalar): ...this new function. Treat the modes
9386 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9387 Alan Hayward <alan.hayward@arm.com>
9388 David Sherwood <david.sherwood@arm.com>
9390 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
9392 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
9394 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9395 Alan Hayward <alan.hayward@arm.com>
9396 David Sherwood <david.sherwood@arm.com>
9398 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
9399 rather than a machine_mode.
9400 (fixed_from_string): Likewise.
9401 (fixed_convert): Likewise.
9402 (fixed_convert_from_int): Likewise.
9403 (fixed_convert_from_real): Likewise.
9404 (real_convert_from_fixed): Likewise.
9405 * fixed-value.c (fixed_from_double_int): Likewise.
9406 (fixed_from_string): Likewise.
9407 (fixed_convert): Likewise.
9408 (fixed_convert_from_int): Likewise.
9409 (fixed_convert_from_real): Likewise.
9410 (real_convert_from_fixed): Likewise.
9411 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
9413 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9414 Alan Hayward <alan.hayward@arm.com>
9415 David Sherwood <david.sherwood@arm.com>
9417 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
9418 of separate mode class checks. Do not allow vector modes here.
9419 (immed_wide_int_const): Use as_a <scalar_mode>.
9420 * explow.c (trunc_int_for_mode): Likewise.
9421 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
9422 (wi::shwi): Likewise.
9423 (wi::min_value): Likewise.
9424 (wi::max_value): Likewise.
9425 * dwarf2out.c (loc_descriptor): Likewise.
9426 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
9429 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9430 Alan Hayward <alan.hayward@arm.com>
9431 David Sherwood <david.sherwood@arm.com>
9433 * tree.h (SCALAR_TYPE_MODE): New macro.
9434 * expr.c (expand_expr_addr_expr_1): Use it.
9435 (expand_expr_real_2): Likewise.
9436 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
9437 (fold_convert_const_fixed_from_int): Likewise.
9438 (fold_convert_const_fixed_from_real): Likewise.
9439 (native_encode_fixed): Likewise
9440 (native_encode_complex): Likewise
9441 (native_encode_vector): Likewise.
9442 (native_interpret_fixed): Likewise.
9443 (native_interpret_real): Likewise.
9444 (native_interpret_complex): Likewise.
9445 (native_interpret_vector): Likewise.
9446 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
9447 (simd_clone_adjust_argument_types): Likewise.
9448 (simd_clone_init_simd_arrays): Likewise.
9449 (simd_clone_adjust): Likewise.
9450 * stor-layout.c (layout_type): Likewise.
9451 * tree.c (build_minus_one_cst): Likewise.
9452 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
9453 * tree-inline.c (estimate_move_cost): Likewise.
9454 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
9455 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
9456 (vectorizable_reduction): Likewise.
9457 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
9458 (vect_recog_mixed_size_cond_pattern): Likewise.
9459 (check_bool_pattern): Likewise.
9460 (adjust_bool_pattern): Likewise.
9461 (search_type_for_mask_1): Likewise.
9462 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
9463 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
9464 (vectorizable_load): Likewise.
9465 (vectorizable_store): Likewise.
9466 * ubsan.c (ubsan_encode_value): Likewise.
9467 * varasm.c (output_constant): Likewise.
9469 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9470 Alan Hayward <alan.hayward@arm.com>
9471 David Sherwood <david.sherwood@arm.com>
9473 * coretypes.h (scalar_mode): New class.
9474 * machmode.h (scalar_mode): Likewise.
9475 (scalar_mode::includes_p): New function.
9476 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
9477 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
9478 * genmodes.c (get_mode_class): Handle remaining scalar modes.
9479 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
9480 * expmed.c (store_bit_field_1): Likewise.
9481 (extract_bit_field_1): Likewise.
9482 * expr.c (write_complex_part): Likewise.
9483 (read_complex_part): Likewise.
9484 (emit_move_complex_push): Likewise.
9485 (expand_expr_real_2): Likewise.
9486 * function.c (assign_parm_setup_reg): Likewise.
9487 (assign_parms_unsplit_complex): Likewise.
9488 * optabs.c (expand_binop): Likewise.
9489 * rtlanal.c (subreg_get_info): Likewise.
9490 * simplify-rtx.c (simplify_immed_subreg): Likewise.
9491 * varasm.c (output_constant_pool_2): Likewise.
9493 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9494 Alan Hayward <alan.hayward@arm.com>
9495 David Sherwood <david.sherwood@arm.com>
9497 * expmed.c (extract_high_half): Use scalar_int_mode and remove
9499 (expmed_mult_highpart_optab): Likewise.
9500 (expmed_mult_highpart): Likewise.
9502 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9503 Alan Hayward <alan.hayward@arm.com>
9504 David Sherwood <david.sherwood@arm.com>
9506 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
9507 instead of a machine_mode.
9508 (builtin_memset_read_str): Likewise.
9509 * builtins.c (c_readstr): Likewise.
9510 (builtin_memcpy_read_str): Likewise.
9511 (builtin_strncpy_read_str): Likewise.
9512 (builtin_memset_read_str): Likewise.
9513 (builtin_memset_gen_str): Likewise.
9514 (expand_builtin_signbit): Use scalar_int_mode for local variables.
9515 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
9516 instead of a machine_mode.
9517 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
9519 (make_extraction): Likewise.
9520 (try_widen_shift_mode): Take and return scalar_int_modes instead
9522 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
9523 a scalar_int_mode instead of a machine_mode.
9524 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
9525 (avr_addr_space_pointer_mode): Likewise.
9526 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
9527 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
9528 (msp430_unwind_word_mode): Likewise.
9529 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
9530 (spu_addr_space_pointer_mode): Likewise.
9531 (spu_addr_space_address_mode): Likewise.
9532 (spu_libgcc_cmp_return_mode): Likewise.
9533 (spu_libgcc_shift_count_mode): Likewise.
9534 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
9535 (rl78_addr_space_pointer_mode): Likewise.
9536 (fl78_unwind_word_mode): Likewise.
9537 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
9539 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
9540 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
9541 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
9542 (mips_valid_pointer_mode): Likewise.
9543 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
9544 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
9545 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
9547 (ft32_addr_space_address_mode): Likewise.
9548 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
9549 scalar_int_mode instead of a machine_mode.
9550 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
9552 (m32c_addr_space_address_mode): Likewise.
9553 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
9554 (rs6000_eh_return_filter_mode): Likewise.
9555 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
9556 (rs6000_eh_return_filter_mode): Likewise.
9557 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
9558 (s390_libgcc_shift_count_mode): Likewise.
9559 (s390_unwind_word_mode): Likewise.
9560 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
9562 * target.def (mode_rep_extended): Likewise.
9563 (valid_pointer_mode): Likewise.
9564 (addr_space.valid_pointer_mode): Likewise.
9565 (eh_return_filter_mode): Return a scalar_int_mode rather than
9567 (libgcc_cmp_return_mode): Likewise.
9568 (libgcc_shift_count_mode): Likewise.
9569 (unwind_word_mode): Likewise.
9570 (addr_space.pointer_mode): Likewise.
9571 (addr_space.address_mode): Likewise.
9572 * doc/tm.texi: Regenerate.
9573 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
9575 (do_jump): Use scalar_int_mode for local variables.
9576 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
9577 rather than a machine_mode.
9578 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
9579 (scompare_loc_descriptor_wide): Likewise.
9580 (scompare_loc_descriptor_narrow): Likewise.
9581 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
9583 * except.c (sjlj_emit_dispatch_table): Likewise.
9584 (expand_builtin_eh_copy_values): Likewise.
9585 * explow.c (convert_memory_address_addr_space_1): Likewise.
9586 Take a scalar_int_mode rather than a machine_mode.
9587 (convert_memory_address_addr_space): Take a scalar_int_mode rather
9588 than a machine_mode.
9589 (memory_address_addr_space): Use scalar_int_mode for local variables.
9590 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
9591 rather than a machine_mode.
9592 * expmed.c (mask_rtx): Likewise.
9593 (init_expmed_one_conv): Likewise.
9594 (expand_mult_highpart_adjust): Likewise.
9595 (extract_high_half): Likewise.
9596 (expmed_mult_highpart_optab): Likewise.
9597 (expmed_mult_highpart): Likewise.
9598 (expand_smod_pow2): Likewise.
9599 (expand_sdiv_pow2): Likewise.
9600 (emit_store_flag_int): Likewise.
9601 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
9603 (extract_low_bits): Likewise.
9604 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
9606 * expr.c (pieces_addr::adjust): Likewise.
9607 (can_store_by_pieces): Likewise.
9608 (store_by_pieces): Likewise.
9609 (clear_by_pieces_1): Likewise.
9610 (expand_expr_addr_expr_1): Likewise.
9611 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
9612 (expand_expr_real_1): Likewise.
9613 (try_casesi): Likewise.
9614 * final.c (shorten_branches): Likewise.
9615 * fold-const.c (fold_convert_const_int_from_fixed): Change the
9616 type of "mode" to machine_mode.
9617 * internal-fn.c (expand_arith_overflow_result_store): Take a
9618 scalar_int_mode rather than a machine_mode.
9619 (expand_mul_overflow): Use scalar_int_mode for local variables.
9620 * loop-doloop.c (doloop_modify): Likewise.
9621 (doloop_optimize): Likewise.
9622 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
9623 than a machine_mode.
9624 (expand_doubleword_shift_condmove): Likewise.
9625 (expand_doubleword_shift): Likewise.
9626 (expand_doubleword_clz): Likewise.
9627 (expand_doubleword_popcount): Likewise.
9628 (expand_doubleword_parity): Likewise.
9629 (expand_absneg_bit): Use scalar_int_mode for local variables.
9630 (prepare_float_lib_cmp): Likewise.
9631 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
9632 rather than a machine_mode.
9633 (convert_memory_address_addr_space): Likewise.
9634 (get_mode_bounds): Likewise.
9635 (get_address_mode): Return a scalar_int_mode rather than a
9637 * rtlanal.c (get_address_mode): Likewise.
9638 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
9639 than a machine_mode.
9640 * targhooks.c (default_mode_rep_extended): Likewise.
9641 (default_valid_pointer_mode): Likewise.
9642 (default_addr_space_valid_pointer_mode): Likewise.
9643 (default_eh_return_filter_mode): Return a scalar_int_mode rather
9644 than a machine_mode.
9645 (default_libgcc_cmp_return_mode): Likewise.
9646 (default_libgcc_shift_count_mode): Likewise.
9647 (default_unwind_word_mode): Likewise.
9648 (default_addr_space_pointer_mode): Likewise.
9649 (default_addr_space_address_mode): Likewise.
9650 * targhooks.h (default_eh_return_filter_mode): Likewise.
9651 (default_libgcc_cmp_return_mode): Likewise.
9652 (default_libgcc_shift_count_mode): Likewise.
9653 (default_unwind_word_mode): Likewise.
9654 (default_addr_space_pointer_mode): Likewise.
9655 (default_addr_space_address_mode): Likewise.
9656 (default_mode_rep_extended): Take a scalar_int_mode rather than
9658 (default_valid_pointer_mode): Likewise.
9659 (default_addr_space_valid_pointer_mode): Likewise.
9660 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
9662 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
9663 rather than a machine_mode.
9664 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
9665 for local variables.
9666 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
9667 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
9668 than a machine_mode.
9670 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9671 Alan Hayward <alan.hayward@arm.com>
9672 David Sherwood <david.sherwood@arm.com>
9674 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
9675 the mode argument to scalar_int_mode.
9676 (do_jump_by_parts_zero_rtx): Likewise.
9677 (do_jump_by_parts_equality_rtx): Likewise.
9678 (do_jump_by_parts_greater): Take a mode argument.
9679 (do_jump_by_parts_equality): Likewise.
9680 (do_jump_1): Update calls accordingly.
9682 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9683 Alan Hayward <alan.hayward@arm.com>
9684 David Sherwood <david.sherwood@arm.com>
9686 * is-a.h (safe_dyn_cast): New function.
9687 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
9688 (jump_table_for_label): Likewise.
9689 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
9690 instead of an rtx_insn *.
9691 (shorten_branches): Use dyn_cast instead of LABEL_P and
9692 JUMP_TABLE_DATA_P. Use jump_table_for_label and
9693 rtx_jump_table_data::get_data_mode.
9694 (final_scan_insn): Likewise.
9696 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9697 Alan Hayward <alan.hayward@arm.com>
9698 David Sherwood <david.sherwood@arm.com>
9700 * combine.c (try_combine): Use is_a <scalar_int_mode> when
9701 trying to combine a full-register integer set with a subreg
9704 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9705 Alan Hayward <alan.hayward@arm.com>
9706 David Sherwood <david.sherwood@arm.com>
9708 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
9709 that is always either address_mode or pointer_mode.
9711 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9712 Alan Hayward <alan.hayward@arm.com>
9713 David Sherwood <david.sherwood@arm.com>
9715 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
9716 when the two are known to be equal.
9718 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9719 Alan Hayward <alan.hayward@arm.com>
9720 David Sherwood <david.sherwood@arm.com>
9722 * simplify-rtx.c (simplify_const_unary_operation): Use
9723 is_a <scalar_int_mode> instead of checking for a nonzero
9724 precision. Forcibly convert op_mode to a scalar_int_mode
9725 in that case. More clearly differentiate the operand and
9726 result modes and use the former when deciding what the value
9727 of a count-bits operation should be. Use is_int_mode instead
9728 of checking for a MODE_INT. Remove redundant check for whether
9729 this mode has a zero precision.
9731 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9732 Alan Hayward <alan.hayward@arm.com>
9733 David Sherwood <david.sherwood@arm.com>
9735 * optabs.c (widen_leading): Change the type of the mode argument
9736 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
9737 (widen_bswap): Likewise.
9738 (expand_parity): Likewise.
9739 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
9740 (expand_ffs): Likewise.
9741 (epand_unop): Check for scalar integer modes before calling the
9744 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9745 Alan Hayward <alan.hayward@arm.com>
9746 David Sherwood <david.sherwood@arm.com>
9748 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
9750 (expand_expr_real_1): Update call accordingly.
9752 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9753 Alan Hayward <alan.hayward@arm.com>
9754 David Sherwood <david.sherwood@arm.com>
9756 * expmed.c (store_bit_field_using_insv): Add op0_mode and
9757 value_mode arguments. Use scalar_int_mode internally.
9758 (store_bit_field_1): Rename the new integer mode from imode
9759 to op0_mode and use it instead of GET_MODE (op0). Update calls
9760 to store_split_bit_field, store_bit_field_using_insv and
9761 store_fixed_bit_field.
9762 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
9763 Use scalar_int_mode internally. Use a bit count rather than a mode
9764 when calculating the largest bit size for get_best_mode.
9765 Update calls to store_split_bit_field and store_fixed_bit_field_1.
9766 (store_fixed_bit_field_1): Add mode and value_mode arguments.
9767 Remove assertion that OP0 has a scalar integer mode.
9768 (store_split_bit_field): Add op0_mode and value_mode arguments.
9769 Update calls to extract_fixed_bit_field.
9770 (extract_bit_field_using_extv): Add an op0_mode argument.
9771 Use scalar_int_mode internally.
9772 (extract_bit_field_1): Rename the new integer mode from imode to
9773 op0_mode and use it instead of GET_MODE (op0). Update calls to
9774 extract_split_bit_field, extract_bit_field_using_extv and
9775 extract_fixed_bit_field.
9776 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
9777 to extract_split_bit_field and extract_fixed_bit_field_1.
9778 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
9779 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
9781 (extract_split_bit_field): Add an op0_mode argument. Update call
9782 to extract_fixed_bit_field.
9784 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9785 Alan Hayward <alan.hayward@arm.com>
9786 David Sherwood <david.sherwood@arm.com>
9788 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
9789 * explow.c (hard_function_value): Likewise.
9790 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
9791 convert_to_mode call outside the loop.
9792 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
9793 for the mode iterator. Require the mode specified by max_pieces
9795 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
9797 (copy_blkmode_to_reg): Likewise.
9798 (set_storage_via_setmem): Likewise.
9799 * optabs.c (prepare_cmp_insn): Likewise.
9800 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
9801 * stor-layout.c (finish_bitfield_representative): Likewise.
9803 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9804 Alan Hayward <alan.hayward@arm.com>
9805 David Sherwood <david.sherwood@arm.com>
9807 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
9808 * expr.c (convert_move): Use them.
9809 (convert_modes): Likewise.
9810 (store_expr_with_bounds): Likewise.
9812 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9813 Alan Hayward <alan.hayward@arm.com>
9814 David Sherwood <david.sherwood@arm.com>
9816 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
9817 parameter for the mode of "x". Remove the "known_x", "known_mode"
9818 and "known_ret" arguments. Change the type of the mode argument
9820 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
9821 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
9822 (reg_num_sign_bit_copies_for_combine): Likewise.
9823 * rtlanal.c (nonzero_bits1): Likewise.
9824 (num_sign_bit_copies1): Likewise.
9825 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
9826 (reg_num_sign_bit_copies_general): Likewise.
9827 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
9828 (reg_nonzero_bits_general): Likewise.
9830 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9831 Alan Hayward <alan.hayward@arm.com>
9832 David Sherwood <david.sherwood@arm.com>
9834 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
9835 than in subroutines. Return 1 for non-integer modes.
9836 (cached_num_sign_bit_copies): Change the type of the mode parameter
9838 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
9839 classes. Handle CONST_INT_P first and then check whether X also
9840 has a scalar integer mode. Check the same thing for inner registers
9841 of a SUBREG and for values that are being extended or truncated.
9843 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9844 Alan Hayward <alan.hayward@arm.com>
9845 David Sherwood <david.sherwood@arm.com>
9847 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
9848 in subroutines. Return the mode mask for non-integer modes.
9849 (cached_nonzero_bits): Change the type of the mode parameter
9851 (nonzero_bits1): Likewise. Remove early exit for other mode
9852 classes. Handle CONST_INT_P first and then check whether X
9853 also has a scalar integer mode.
9855 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9856 Alan Hayward <alan.hayward@arm.com>
9857 David Sherwood <david.sherwood@arm.com>
9859 * expr.c (widest_int_mode_for_size): Make the comment match the code.
9860 Return a scalar_int_mode and assert that the size is greater than
9862 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
9863 (op_by_pieces_d::op_by_pieces_d): Likewise.
9864 (op_by_pieces_d::run): Likewise.
9865 (can_store_by_pieces): Likewise.
9867 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9868 Alan Hayward <alan.hayward@arm.com>
9869 David Sherwood <david.sherwood@arm.com>
9871 * combine.c (extract_left_shift): Add a mode argument and update
9873 (make_compound_operation_int): Change the type of the mode parameter
9874 to scalar_int_mode and update the call to extract_left_shift.
9876 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9877 Alan Hayward <alan.hayward@arm.com>
9878 David Sherwood <david.sherwood@arm.com>
9880 * combine.c (simplify_and_const_int): Change the type of the mode
9881 parameter to scalar_int_mode.
9882 (simplify_and_const_int_1): Likewise. Update recursive call.
9884 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9885 Alan Hayward <alan.hayward@arm.com>
9886 David Sherwood <david.sherwood@arm.com>
9888 * combine.c (simplify_compare_const): Check that the mode is a
9889 scalar_int_mode (rather than VOIDmode) before testing its
9891 (simplify_comparison): Move COMPARISON_P handling out of the
9892 loop and restrict the latter part of the loop to scalar_int_modes.
9893 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
9894 and when considering SUBREG_REGs. Use is_int_mode instead of
9895 checking GET_MODE_CLASS against MODE_INT.
9897 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9898 Alan Hayward <alan.hayward@arm.com>
9899 David Sherwood <david.sherwood@arm.com>
9901 * combine.c (try_widen_shift_mode): Move check for equal modes to...
9902 (simplify_shift_const_1): ...here. Use scalar_int_mode for
9903 shift_unit_mode and for modes involved in scalar shifts.
9905 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9906 Alan Hayward <alan.hayward@arm.com>
9907 David Sherwood <david.sherwood@arm.com>
9909 * combine.c (force_int_to_mode): New function, split out from...
9910 (force_to_mode): ...here. Keep xmode up-to-date and use it
9911 instead of GET_MODE (x).
9913 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9914 Alan Hayward <alan.hayward@arm.com>
9915 David Sherwood <david.sherwood@arm.com>
9917 * optabs-query.h (extraction_insn::struct_mode): Change type to
9918 opt_scalar_int_mode and update comment.
9919 (extraction_insn::field_mode): Change type to scalar_int_mode.
9920 (extraction_insn::pos_mode): Likewise.
9921 * combine.c (make_extraction): Update accordingly.
9922 * optabs-query.c (get_traditional_extraction_insn): Likewise.
9923 (get_optab_extraction_insn): Likewise.
9924 * recog.c (simplify_while_replacing): Likewise.
9925 * expmed.c (narrow_bit_field_mem): Change the type of the mode
9926 parameter to opt_scalar_int_mode.
9928 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9929 Alan Hayward <alan.hayward@arm.com>
9930 David Sherwood <david.sherwood@arm.com>
9932 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
9933 to a scalar_int_mode instead of a machine_mode.
9934 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
9935 (get_best_mode): Return a boolean and use a pointer argument to store
9936 the selected mode. Replace the limit mode parameter with a bit limit.
9937 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
9938 for the values returned by bit_field_mode_iterator::next_mode.
9939 (store_bit_field): Update call to get_best_mode.
9940 (store_fixed_bit_field): Likewise.
9941 (extract_fixed_bit_field): Likewise.
9942 * expr.c (optimize_bitfield_assignment_op): Likewise.
9943 * fold-const.c (optimize_bit_field_compare): Likewise.
9944 (fold_truth_andor_1): Likewise.
9945 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
9946 Update for new type of m_mode.
9947 (get_best_mode): As above.
9949 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9950 Alan Hayward <alan.hayward@arm.com>
9951 David Sherwood <david.sherwood@arm.com>
9953 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
9954 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
9955 (store_bit_field): Check is_a <scalar_int_mode> before calling
9956 strict_volatile_bitfield_p.
9957 (extract_bit_field): Likewise.
9959 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9960 Alan Hayward <alan.hayward@arm.com>
9961 David Sherwood <david.sherwood@arm.com>
9963 * target.def (cstore_mode): Return a scalar_int_mode.
9964 * doc/tm.texi: Regenerate.
9965 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
9966 * targhooks.h (default_cstore_mode): Likewise.
9967 * targhooks.c (default_cstore_mode): Likewise, using a forced
9969 * expmed.c (emit_cstore): Expect the target of the cstore to be
9972 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9973 Alan Hayward <alan.hayward@arm.com>
9974 David Sherwood <david.sherwood@arm.com>
9976 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
9978 (niter_desc): Likewise mode.
9979 (iv_analyze): Add a mode parameter.
9981 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
9982 and change its type to scalar_int_mode.
9983 * loop-iv.c: Update commentary at head of file.
9984 (iv_constant): Pass the mode paraeter before the rtx it describes
9985 and change its type to scalar_int_mode. Remove VOIDmode handling.
9986 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
9987 (iv_extend): Likewise.
9988 (shorten_into_mode): Likewise.
9989 (iv_add): Use scalar_int_mode.
9990 (iv_mult): Likewise.
9991 (iv_shift): Likewise.
9992 (canonicalize_iv_subregs): Likewise.
9993 (get_biv_step_1): Pass the outer_mode parameter before the rtx
9994 it describes and change its mode to scalar_int_mode. Also change
9995 the type of the returned inner_mode to scalar_int_mode.
9996 (get_biv_step): Likewise, turning outer_mode from a pointer
9997 into a direct parameter. Update call to get_biv_step_1.
9998 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
9999 iv_constant and get_biv_step.
10000 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
10001 and change its type to scalar_int_mode. Don't initialise iv->mode
10002 to VOIDmode and remove later checks for its still being VOIDmode.
10003 Update calls to iv_analyze_op and iv_analyze_expr. Check
10004 is_a <scalar_int_mode> when changing the mode under consideration.
10005 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
10006 Update call to iv_analyze_expr.
10007 (iv_analyze_op): Add a mode parameter. Reject subregs whose
10008 inner register is not also a scalar_int_mode. Update call to
10010 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
10011 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
10012 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
10013 separate mode class checks. Update calls to iv_analyze. Remove
10014 fix-up of VOIDmodes after iv_analyze_biv.
10015 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
10016 don't have a scalar_int_mode. Update call to biv_p.
10018 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10019 Alan Hayward <alan.hayward@arm.com>
10020 David Sherwood <david.sherwood@arm.com>
10022 * cfgexpand.c (convert_debug_memory_address): Use
10023 as_a <scalar_int_mode>.
10024 * combine.c (expand_compound_operation): Likewise.
10025 (make_extraction): Likewise.
10026 (change_zero_ext): Likewise.
10027 (simplify_comparison): Likewise.
10028 * cse.c (cse_insn): Likewise.
10029 * dwarf2out.c (minmax_loc_descriptor): Likewise.
10030 (mem_loc_descriptor): Likewise.
10031 (loc_descriptor): Likewise.
10032 * expmed.c (init_expmed_one_mode): Likewise.
10033 (synth_mult): Likewise.
10034 (emit_store_flag_1): Likewise.
10035 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
10036 of a comparison with size.
10037 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
10038 (reduce_to_bit_field_precision): Likewise.
10039 * function.c (expand_function_end): Likewise.
10040 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
10041 * loop-doloop.c (doloop_modify): Likewise.
10042 * optabs.c (expand_binop): Likewise.
10043 (expand_unop): Likewise.
10044 (expand_copysign_absneg): Likewise.
10045 (prepare_cmp_insn): Likewise.
10046 (maybe_legitimize_operand): Likewise.
10047 * recog.c (const_scalar_int_operand): Likewise.
10048 * rtlanal.c (get_address_mode): Likewise.
10049 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10050 (simplify_cond_clz_ctz): Likewise.
10051 * tree-nested.c (get_nl_goto_field): Likewise.
10052 * tree.c (build_vector_type_for_mode): Likewise.
10053 * var-tracking.c (use_narrower_mode): Likewise.
10055 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10056 Alan Hayward <alan.hayward@arm.com>
10057 David Sherwood <david.sherwood@arm.com>
10059 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
10060 * builtins.c (expand_builtin_signbit): Use it.
10061 * cfgexpand.c (expand_debug_expr): Likewise.
10062 * dojump.c (do_jump): Likewise.
10063 (do_compare_and_jump): Likewise.
10064 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
10065 * expmed.c (make_tree): Likewise.
10066 * expr.c (expand_expr_real_2): Likewise.
10067 (expand_expr_real_1): Likewise.
10068 (try_casesi): Likewise.
10069 * fold-const-call.c (fold_const_call_ss): Likewise.
10070 * fold-const.c (unextend): Likewise.
10071 (extract_muldiv_1): Likewise.
10072 (fold_single_bit_test): Likewise.
10073 (native_encode_int): Likewise.
10074 (native_encode_string): Likewise.
10075 (native_interpret_int): Likewise.
10076 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
10077 * internal-fn.c (expand_addsub_overflow): Likewise.
10078 (expand_neg_overflow): Likewise.
10079 (expand_mul_overflow): Likewise.
10080 (expand_arith_overflow): Likewise.
10081 * match.pd: Likewise.
10082 * stor-layout.c (layout_type): Likewise.
10083 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10084 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
10085 * tree-ssanames.c (get_range_info): Likewise.
10086 * tree-switch-conversion.c (array_value_type) Likewise.
10087 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
10088 (vect_recog_divmod_pattern): Likewise.
10089 (vect_recog_mixed_size_cond_pattern): Likewise.
10090 * tree-vrp.c (extract_range_basic): Likewise.
10091 (simplify_float_conversion_using_ranges): Likewise.
10092 * tree.c (int_fits_type_p): Likewise.
10093 * ubsan.c (instrument_bool_enum_load): Likewise.
10094 * varasm.c (mergeable_string_section): Likewise.
10095 (narrowing_initializer_constant_valid_p): Likewise.
10096 (output_constant): Likewise.
10098 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10099 Alan Hayward <alan.hayward@arm.com>
10100 David Sherwood <david.sherwood@arm.com>
10102 * machmode.h (NARROWEST_INT_MODE): New macro.
10103 * expr.c (alignment_for_piecewise_move): Use it instead of
10104 GET_CLASS_NARROWEST_MODE (MODE_INT).
10105 (push_block): Likewise.
10106 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
10108 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10110 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10111 Alan Hayward <alan.hayward@arm.com>
10112 David Sherwood <david.sherwood@arm.com>
10114 * postreload.c (move2add_valid_value_p): Change the type of the
10115 mode parameter to scalar_int_mode.
10116 (move2add_use_add2_insn): Add a mode parameter and use it instead
10118 (move2add_use_add3_insn): Likewise.
10119 (reload_cse_move2add): Update accordingly.
10121 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10122 Alan Hayward <alan.hayward@arm.com>
10123 David Sherwood <david.sherwood@arm.com>
10125 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
10127 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
10128 * optabs.c (expand_unop): Likewise.
10130 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10131 Alan Hayward <alan.hayward@arm.com>
10132 David Sherwood <david.sherwood@arm.com>
10134 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
10135 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
10136 (popcount_loc_descriptor): Likewise.
10137 (bswap_loc_descriptor): Likewise.
10138 (rotate_loc_descriptor): Likewise.
10139 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
10140 calling the functions above.
10142 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10143 Alan Hayward <alan.hayward@arm.com>
10144 David Sherwood <david.sherwood@arm.com>
10146 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
10148 (try_combine): Likewise.
10149 (simplify_if_then_else): Likewise.
10150 * cse.c (cse_insn): Likewise.
10151 * dwarf2out.c (mem_loc_descriptor): Likewise.
10152 * emit-rtl.c (gen_lowpart_common): Likewise.
10153 * simplify-rtx.c (simplify_truncation): Likewise.
10154 (simplify_binary_operation_1): Likewise.
10155 (simplify_const_relational_operation): Likewise.
10156 (simplify_ternary_operation): Likewise.
10157 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
10159 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10160 Alan Hayward <alan.hayward@arm.com>
10161 David Sherwood <david.sherwood@arm.com>
10163 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
10164 * reload.c (push_reload): Likewise.
10165 (find_reloads): Likewise.
10167 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10168 Alan Hayward <alan.hayward@arm.com>
10169 David Sherwood <david.sherwood@arm.com>
10171 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
10172 (make_compound_operation_int): Likewise.
10173 (change_zero_ext): Likewise.
10174 * expr.c (convert_move): Likewise.
10175 (convert_modes): Likewise.
10176 * fwprop.c (forward_propagate_subreg): Likewise.
10177 * loop-iv.c (get_biv_step_1): Likewise.
10178 * optabs.c (widen_operand): Likewise.
10179 * postreload.c (move2add_valid_value_p): Likewise.
10180 * recog.c (simplify_while_replacing): Likewise.
10181 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10182 (simplify_binary_operation_1): Likewise. Remove redundant
10183 mode equality check.
10185 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10186 Alan Hayward <alan.hayward@arm.com>
10187 David Sherwood <david.sherwood@arm.com>
10189 * combine.c (combine_simplify_rtx): Add checks for
10190 is_a <scalar_int_mode>.
10191 (simplify_if_then_else): Likewise.
10192 (make_field_assignment): Likewise.
10193 (simplify_comparison): Likewise.
10194 * ifcvt.c (noce_try_bitop): Likewise.
10195 * loop-invariant.c (canonicalize_address_mult): Likewise.
10196 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10198 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10199 Alan Hayward <alan.hayward@arm.com>
10200 David Sherwood <david.sherwood@arm.com>
10202 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
10203 is_a <scalar_int_mode> instead of != BLKmode.
10205 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10206 Alan Hayward <alan.hayward@arm.com>
10207 David Sherwood <david.sherwood@arm.com>
10209 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
10210 instead of != VOIDmode.
10211 * combine.c (if_then_else_cond): Likewise.
10212 (change_zero_ext): Likewise.
10213 * dwarf2out.c (mem_loc_descriptor): Likewise.
10214 (loc_descriptor): Likewise.
10215 * rtlanal.c (canonicalize_condition): Likewise.
10216 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
10218 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10219 Alan Hayward <alan.hayward@arm.com>
10220 David Sherwood <david.sherwood@arm.com>
10222 * simplify-rtx.c (simplify_binary_operation_1): Use
10223 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
10225 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10226 Alan Hayward <alan.hayward@arm.com>
10227 David Sherwood <david.sherwood@arm.com>
10229 * wide-int.h (int_traits<unsigned char>) New class.
10230 (int_traits<unsigned short>) Likewise.
10231 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
10232 Use GET_MODE_UNIT_PRECISION and remove redundant test for
10234 * combine.c (set_nonzero_bits_and_sign_copies): Use
10235 is_a <scalar_int_mode>.
10236 (find_split_point): Likewise.
10237 (combine_simplify_rtx): Likewise.
10238 (simplify_logical): Likewise.
10239 (expand_compound_operation): Likewise.
10240 (expand_field_assignment): Likewise.
10241 (make_compound_operation): Likewise.
10242 (extended_count): Likewise.
10243 (change_zero_ext): Likewise.
10244 (simplify_comparison): Likewise.
10245 * dwarf2out.c (scompare_loc_descriptor): Likewise.
10246 (ucompare_loc_descriptor): Likewise.
10247 (minmax_loc_descriptor): Likewise.
10248 (mem_loc_descriptor): Likewise.
10249 (loc_descriptor): Likewise.
10250 * expmed.c (init_expmed_one_mode): Likewise.
10251 * lra-constraints.c (lra_constraint_offset): Likewise.
10252 * optabs.c (prepare_libcall_arg): Likewise.
10253 * postreload.c (move2add_note_store): Likewise.
10254 * reload.c (operands_match_p): Likewise.
10255 * rtl.h (load_extend_op): Likewise.
10256 * rtlhooks.c (gen_lowpart_general): Likewise.
10257 * simplify-rtx.c (simplify_truncation): Likewise.
10258 (simplify_unary_operation_1): Likewise.
10259 (simplify_binary_operation_1): Likewise.
10260 (simplify_const_binary_operation): Likewise.
10261 (simplify_const_relational_operation): Likewise.
10262 (simplify_subreg): Likewise.
10263 * stor-layout.c (bitwise_mode_for_mode): Likewise.
10264 * var-tracking.c (adjust_mems): Likewise.
10265 (prepare_call_arguments): Likewise.
10267 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10268 Alan Hayward <alan.hayward@arm.com>
10269 David Sherwood <david.sherwood@arm.com>
10271 * machmode.h (is_int_mode): New fuction.
10272 * combine.c (find_split_point): Use it.
10273 (combine_simplify_rtx): Likewise.
10274 (simplify_if_then_else): Likewise.
10275 (simplify_set): Likewise.
10276 (simplify_shift_const_1): Likewise.
10277 (simplify_comparison): Likewise.
10278 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
10279 * cse.c (notreg_cost): Likewise.
10280 (cse_insn): Likewise.
10281 * cselib.c (cselib_lookup_1): Likewise.
10282 * dojump.c (do_jump_1): Likewise.
10283 (do_compare_rtx_and_jump): Likewise.
10284 * dse.c (get_call_args): Likewise.
10285 * dwarf2out.c (rtl_for_decl_init): Likewise.
10286 (native_encode_initializer): Likewise.
10287 * expmed.c (emit_store_flag_1): Likewise.
10288 (emit_store_flag): Likewise.
10289 * expr.c (convert_modes): Likewise.
10290 (store_field): Likewise.
10291 (expand_expr_real_1): Likewise.
10292 * fold-const.c (fold_read_from_constant_string): Likewise.
10293 * gimple-ssa-sprintf.c (get_format_string): Likewise.
10294 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
10295 * optabs.c (expand_binop): Likewise.
10296 (expand_unop): Likewise.
10297 (expand_abs_nojump): Likewise.
10298 (expand_one_cmpl_abs_nojump): Likewise.
10299 * simplify-rtx.c (mode_signbit_p): Likewise.
10300 (val_signbit_p): Likewise.
10301 (val_signbit_known_set_p): Likewise.
10302 (val_signbit_known_clear_p): Likewise.
10303 (simplify_relational_operation_1): Likewise.
10304 * tree.c (vector_type_mode): Likewise.
10306 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10307 Alan Hayward <alan.hayward@arm.com>
10308 David Sherwood <david.sherwood@arm.com>
10310 * machmode.h (smallest_mode_for_size): Fix formatting.
10311 (smallest_int_mode_for_size): New function.
10312 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
10313 instead of smallest_mode_for_size.
10314 * combine.c (make_extraction): Likewise.
10315 * config/arc/arc.c (arc_expand_movmem): Likewise.
10316 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
10317 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
10318 * config/s390/s390.c (s390_expand_insv): Likewise.
10319 * config/sparc/sparc.c (assign_int_registers): Likewise.
10320 * config/spu/spu.c (spu_function_value): Likewise.
10321 (spu_function_arg): Likewise.
10322 * coverage.c (get_gcov_type): Likewise.
10323 (get_gcov_unsigned_t): Likewise.
10324 * dse.c (find_shift_sequence): Likewise.
10325 * expmed.c (store_bit_field_1): Likewise.
10326 * expr.c (convert_move): Likewise.
10327 (store_field): Likewise.
10328 * internal-fn.c (expand_arith_overflow): Likewise.
10329 * optabs-query.c (get_best_extraction_insn): Likewise.
10330 * optabs.c (expand_twoval_binop_libfunc): Likewise.
10331 * stor-layout.c (layout_type): Likewise.
10332 (initialize_sizetypes): Likewise.
10333 * targhooks.c (default_get_mask_mode): Likewise.
10334 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
10336 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10337 Alan Hayward <alan.hayward@arm.com>
10338 David Sherwood <david.sherwood@arm.com>
10340 * machmode.h (opt_mode::else_blk): New function.
10341 (int_mode_for_mode): Declare.
10342 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
10343 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
10345 * cfgexpand.c (expand_debug_expr): Likewise.
10346 * combine.c (gen_lowpart_or_truncate): Likewise.
10347 (gen_lowpart_for_combine): Likewise.
10348 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
10349 * config/avr/avr.c (avr_to_int_mode): Likewise.
10350 (avr_out_plus_1): Likewise.
10351 (avr_out_plus): Likewise.
10352 (avr_out_round): Likewise.
10353 * config/i386/i386.c (ix86_split_to_parts): Likewise.
10354 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
10355 (s390_expand_vcond): Likewise.
10356 * config/spu/spu.c (spu_split_immediate): Likewise.
10357 (spu_expand_mov): Likewise.
10358 * dse.c (get_stored_val): Likewise.
10359 * expmed.c (store_bit_field_1): Likewise.
10360 (convert_extracted_bit_field): Use int_mode_for_mode instead of
10362 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
10363 (extract_low_bits): Likewise.
10364 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
10365 handling rather than repeating the check.
10366 (emit_group_store): Likewise.
10367 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
10368 * optabs.c (expand_absneg_bit): Likewise.
10369 (expand_copysign_absneg): Likewise.
10370 (expand_copysign_bit): Likewise.
10371 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
10372 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
10373 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
10374 * var-tracking.c (prepare_call_arguments): Likewise.
10375 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
10376 int_mode_for_mode instead of mode_for_size.
10377 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
10379 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10380 Alan Hayward <alan.hayward@arm.com>
10381 David Sherwood <david.sherwood@arm.com>
10383 * machmode.h (int_mode_for_size): New function.
10384 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
10385 instead of mode_for_size.
10386 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
10388 * combine.c (expand_field_assignment): Use int_mode_for_size
10389 instead of mode_for_size.
10390 (make_extraction): Likewise.
10391 (simplify_shift_const_1): Likewise.
10392 (simplify_comparison): Likewise.
10393 * dojump.c (do_jump): Likewise.
10394 * dwarf2out.c (mem_loc_descriptor): Likewise.
10395 * emit-rtl.c (init_derived_machine_modes): Likewise.
10396 * expmed.c (flip_storage_order): Likewise.
10397 (convert_extracted_bit_field): Likewise.
10398 * expr.c (copy_blkmode_from_reg): Likewise.
10399 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
10400 * internal-fn.c (expand_mul_overflow): Likewise.
10401 * lower-subreg.c (simple_move): Likewise.
10402 * optabs-libfuncs.c (init_optabs): Likewise.
10403 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10404 * tree.c (vector_type_mode): Likewise.
10405 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
10406 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
10407 * tree-vect-generic.c (expand_vector_parallel): Likewise.
10408 * tree-vect-stmts.c (vectorizable_load): Likewise.
10409 (vectorizable_store): Likewise.
10411 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10412 Alan Hayward <alan.hayward@arm.com>
10413 David Sherwood <david.sherwood@arm.com>
10415 * coretypes.h (pod_mode): New type.
10416 (scalar_int_mode_pod): New typedef.
10417 * machmode.h (pod_mode): New class.
10418 (int_n_data_t::m): Change type to scalar_int_mode_pod.
10419 * genmodes.c (emit_mode_int_n): Update accordingly.
10420 * lower-subreg.h (target_lower_subreg): Change type to
10421 scalar_int_mode_pod.
10422 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
10423 scalar_int_mode_pod.
10425 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10426 Alan Hayward <alan.hayward@arm.com>
10427 David Sherwood <david.sherwood@arm.com>
10429 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
10430 machine_mode to scalar_int_mode.
10431 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
10432 (rs6000_option_override_internal): Remove cast to int.
10433 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
10434 machine_mode to scalar_int_mode.
10435 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
10436 (rs6000_option_override_internal): Remove cast to int.
10437 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
10438 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
10440 * config/s390/s390.c (s390_expand_builtin): Likewise.
10441 * coretypes.h (scalar_int_mode): New type.
10442 (opt_scalar_int_mode): New typedef.
10443 * machmode.h (scalar_int_mode): New class.
10444 (scalar_int_mode::includes_p): New function.
10445 (byte_mode): Change type to scalar_int_mode.
10446 (word_mode): Likewise.
10447 (ptr_mode): Likewise.
10448 * emit-rtl.c (byte_mode): Likewise.
10449 (word_mode): Likewise.
10450 (ptr_mode): Likewise.
10451 (init_derived_machine_modes): Update accordingly.
10452 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
10453 and MODE_PARTIAL_INT.
10454 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
10455 opt_scalar_int_mode.
10457 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10458 Alan Hayward <alan.hayward@arm.com>
10459 David Sherwood <david.sherwood@arm.com>
10461 * target.def (libgcc_floating_mode_supported_p): Take a
10463 * doc/tm.texi: Regenerate.
10464 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
10466 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
10467 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
10470 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10471 Alan Hayward <alan.hayward@arm.com>
10472 David Sherwood <david.sherwood@arm.com>
10474 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
10475 * doc/tm.texi: Regenerate.
10476 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
10477 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
10478 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
10479 * targhooks.h (default_floatn_mode): Likewise.
10480 * targhooks.c (default_floatn_mode): Likewise.
10481 * tree.c (build_common_tree_nodes): Update accordingly.
10483 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10484 Alan Hayward <alan.hayward@arm.com>
10485 David Sherwood <david.sherwood@arm.com>
10487 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
10488 (mode_iterator::iterate_p): Likewise.
10489 (mode_iterator::get_wider): Likewise.
10490 * expr.c (init_expr_target): Use opt_scalar_float_mode.
10492 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10493 Alan Hayward <alan.hayward@arm.com>
10494 David Sherwood <david.sherwood@arm.com>
10496 * coretypes.h (opt_scalar_float_mode): New typedef.
10497 * machmode.h (float_mode_for_size): New function.
10498 * emit-rtl.c (double_mode): Delete.
10499 (init_emit_once): Use float_mode_for_size.
10500 * stor-layout.c (layout_type): Likewise.
10501 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
10503 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10504 Alan Hayward <alan.hayward@arm.com>
10505 David Sherwood <david.sherwood@arm.com>
10507 * output.h (assemble_real): Take a scalar_float_mode.
10508 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
10509 * config/arm/arm.md (consttable_4): Likewise.
10510 (consttable_8): Likewise.
10511 (consttable_16): Likewise.
10512 * config/mips/mips.md (consttable_float): Likewise.
10513 * config/s390/s390.c (s390_output_pool_entry): Likewise.
10514 * varasm.c (assemble_real): Take a scalar_float_mode.
10515 (output_constant_pool_2): Update accordingly.
10516 (output_constant): Likewise.
10518 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10519 Alan Hayward <alan.hayward@arm.com>
10520 David Sherwood <david.sherwood@arm.com>
10522 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
10523 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
10524 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
10525 (native_encode_real): Likewise.
10526 (native_interpret_real): Likewise.
10527 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
10528 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10530 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10531 Alan Hayward <alan.hayward@arm.com>
10532 David Sherwood <david.sherwood@arm.com>
10534 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
10535 <scalar_float_mode>. Simplify.
10536 (gen_extend_conv_libfunc): Likewise.
10538 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10539 Alan Hayward <alan.hayward@arm.com>
10540 David Sherwood <david.sherwood@arm.com>
10542 * coretypes.h (scalar_float_mode): New type.
10543 * machmode.h (mode_traits::from_int): Use machine_mode if
10544 USE_ENUM_MODES is defined.
10545 (is_a): New function.
10547 (dyn_cast): Likewise.
10548 (scalar_float_mode): New class.
10549 (scalar_float_mode::includes_p): New function.
10550 (is_float_mode): Likewise.
10551 * gdbhooks.py (MachineModePrinter): New class.
10552 (build_pretty_printer): Use it for scalar_float_mode.
10553 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
10554 (format_helper::format_helper): Turn into a template.
10555 * genmodes.c (get_mode_class): New function.
10556 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
10557 or machine_mode if none.
10558 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
10559 as_a <scalar_float_mode>.
10560 * dwarf2out.c (mem_loc_descriptor): Likewise.
10561 (insert_float): Likewise.
10562 (add_const_value_attribute): Likewise.
10563 * simplify-rtx.c (simplify_immed_subreg): Likewise.
10564 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
10565 (expand_unop): Update accordingly.
10566 (expand_abs_nojump): Likewise.
10567 (expand_copysign_absneg): Take a scalar_float_mode.
10568 (expand_copysign_bit): Likewise.
10569 (expand_copysign): Update accordingly.
10571 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10572 Alan Hayward <alan.hayward@arm.com>
10573 David Sherwood <david.sherwood@arm.com>
10575 * coretypes.h (opt_mode): New class.
10576 * machmode.h (opt_mode): Likewise.
10577 (opt_mode::else_void): New function.
10578 (opt_mode::require): Likewise.
10579 (opt_mode::exists): Likewise.
10580 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
10581 (GET_MODE_2XWIDER_MODE): Likewise.
10582 (mode_iterator::get_wider): Update accordingly.
10583 (mode_iterator::get_2xwider): Likewise.
10584 (mode_iterator::get_known_wider): Likewise, turning into a template.
10585 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
10586 forcing a wider mode to exist.
10587 * config/cr16/cr16.h (LONG_REG_P): Likewise.
10588 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
10589 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
10590 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
10591 * lower-subreg.c (init_lower_subreg): Likewise.
10592 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
10593 on the final iteration.
10594 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
10595 a wider mode exists before asking for a move pattern.
10596 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
10597 forcing a wider mode to exist.
10598 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
10599 returning false if no such mode exists.
10600 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
10601 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
10602 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
10603 Avoid checking for a MODE_INT if we already know the mode is not a
10605 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
10606 forcing a wider mode to exist.
10607 (expmed_mult_highpart_optab): Likewise.
10608 (expmed_mult_highpart): Likewise.
10609 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
10611 * lto-streamer-in.c (lto_input_mode_table): Likewise.
10612 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
10613 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
10614 * internal-fn.c (expand_mul_overflow): Update use of
10615 GET_MODE_2XWIDER_MODE.
10616 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10617 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
10618 GET_MODE_WIDER_MODE.
10619 (convert_plusminus_to_widen): Likewise.
10620 * tree-switch-conversion.c (array_value_type): Likewise.
10621 * var-tracking.c (emit_note_insn_var_location): Likewise.
10622 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10623 Return false inside rather than outside the loop if no wider mode
10625 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
10626 and GET_MODE_2XWIDER_MODE
10627 (can_compare_p): Use else_void.
10628 * gdbhooks.py (OptMachineModePrinter): New class.
10629 (build_pretty_printer): Use it for opt_mode.
10631 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10632 Alan Hayward <alan.hayward@arm.com>
10633 David Sherwood <david.sherwood@arm.com>
10635 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
10636 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
10638 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10639 Alan Hayward <alan.hayward@arm.com>
10640 David Sherwood <david.sherwood@arm.com>
10642 * machmode.h (mode_traits): New structure.
10643 (get_narrowest_mode): New function.
10644 (mode_iterator::start): Likewise.
10645 (mode_iterator::iterate_p): Likewise.
10646 (mode_iterator::get_wider): Likewise.
10647 (mode_iterator::get_known_wider): Likewise.
10648 (mode_iterator::get_2xwider): Likewise.
10649 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
10650 (FOR_EACH_MODE): Likewise.
10651 (FOR_EACH_MODE_FROM): Likewise.
10652 (FOR_EACH_MODE_UNTIL): Likewise.
10653 (FOR_EACH_WIDER_MODE): Likewise.
10654 (FOR_EACH_2XWIDER_MODE): Likewise.
10655 * builtins.c (expand_builtin_strlen): Use new mode iterators.
10656 * combine.c (simplify_comparison): Likewise
10657 * config/i386/i386.c (type_natural_mode): Likewise.
10658 * cse.c (cse_insn): Likewise.
10659 * dse.c (find_shift_sequence): Likewise.
10660 * emit-rtl.c (init_derived_machine_modes): Likewise.
10661 (init_emit_once): Likewise.
10662 * explow.c (hard_function_value): Likewise.
10663 * expmed.c (extract_fixed_bit_field_1): Likewise.
10664 (extract_bit_field_1): Likewise.
10665 (expand_divmod): Likewise.
10666 (emit_store_flag_1): Likewise.
10667 * expr.c (init_expr_target): Likewise.
10668 (convert_move): Likewise.
10669 (alignment_for_piecewise_move): Likewise.
10670 (widest_int_mode_for_size): Likewise.
10671 (emit_block_move_via_movmem): Likewise.
10672 (copy_blkmode_to_reg): Likewise.
10673 (set_storage_via_setmem): Likewise.
10674 (compress_float_constant): Likewise.
10675 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10676 * optabs-query.c (get_best_extraction_insn): Likewise.
10677 * optabs.c (expand_binop): Likewise.
10678 (expand_twoval_unop): Likewise.
10679 (expand_twoval_binop): Likewise.
10680 (widen_leading): Likewise.
10681 (widen_bswap): Likewise.
10682 (expand_parity): Likewise.
10683 (expand_unop): Likewise.
10684 (prepare_cmp_insn): Likewise.
10685 (prepare_float_lib_cmp): Likewise.
10686 (expand_float): Likewise.
10687 (expand_fix): Likewise.
10688 (expand_sfix_optab): Likewise.
10689 * postreload.c (move2add_use_add2_insn): Likewise.
10690 * reg-stack.c (reg_to_stack): Likewise.
10691 * reginfo.c (choose_hard_reg_mode): Likewise.
10692 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
10693 * stor-layout.c (mode_for_size): Likewise.
10694 (smallest_mode_for_size): Likewise.
10695 (mode_for_vector): Likewise.
10696 (finish_bitfield_representative): Likewise.
10697 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
10698 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
10699 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10700 * var-tracking.c (prepare_call_arguments): Likewise.
10702 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10703 Alan Hayward <alan.hayward@arm.com>
10704 David Sherwood <david.sherwood@arm.com>
10706 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
10707 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
10708 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
10709 * machmode.h (mode_size): Move earlier in file.
10710 (mode_precision): Likewise.
10711 (mode_inner): Likewise.
10712 (mode_nunits): Likewise.
10713 (mode_unit_size): Likewise.
10714 (unit_unit_precision): Likewise.
10715 (mode_wider): Likewise.
10716 (mode_2xwider): Likewise.
10717 (machine_mode): New class.
10718 (mode_to_bytes): New function.
10719 (mode_to_bits): Likewise.
10720 (mode_to_precision): Likewise.
10721 (mode_to_inner): Likewise.
10722 (mode_to_unit_size): Likewise.
10723 (mode_to_unit_precision): Likewise.
10724 (mode_to_nunits): Likewise.
10725 (GET_MODE_SIZE): Use mode_to_bytes.
10726 (GET_MODE_BITSIZE): Use mode_to_bits.
10727 (GET_MODE_PRECISION): Use mode_to_precision.
10728 (GET_MODE_INNER): Use mode_to_inner.
10729 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
10730 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
10731 (GET_MODE_NUNITS): Use mode_to_nunits.
10732 * system.h (ALWAYS_INLINE): New macro.
10733 * config/powerpcspe/powerpcspe-c.c
10734 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
10735 int for arg1_mode and arg2_mode.
10737 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10738 Alan Hayward <alan.hayward@arm.com>
10739 David Sherwood <david.sherwood@arm.com>
10741 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
10742 Prefix mode names with E_ in case statements.
10743 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10744 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
10745 (aarch64_split_simd_move): Likewise.
10746 (aarch64_gen_storewb_pair): Likewise.
10747 (aarch64_gen_loadwb_pair): Likewise.
10748 (aarch64_gen_store_pair): Likewise.
10749 (aarch64_gen_load_pair): Likewise.
10750 (aarch64_get_condition_code_1): Likewise.
10751 (aarch64_constant_pool_reload_icode): Likewise.
10752 (get_rsqrte_type): Likewise.
10753 (get_rsqrts_type): Likewise.
10754 (get_recpe_type): Likewise.
10755 (get_recps_type): Likewise.
10756 (aarch64_gimplify_va_arg_expr): Likewise.
10757 (aarch64_simd_container_mode): Likewise.
10758 (aarch64_emit_load_exclusive): Likewise.
10759 (aarch64_emit_store_exclusive): Likewise.
10760 (aarch64_expand_compare_and_swap): Likewise.
10761 (aarch64_gen_atomic_cas): Likewise.
10762 (aarch64_emit_bic): Likewise.
10763 (aarch64_emit_atomic_swap): Likewise.
10764 (aarch64_emit_atomic_load_op): Likewise.
10765 (aarch64_evpc_trn): Likewise.
10766 (aarch64_evpc_uzp): Likewise.
10767 (aarch64_evpc_zip): Likewise.
10768 (aarch64_evpc_ext): Likewise.
10769 (aarch64_evpc_rev): Likewise.
10770 (aarch64_evpc_dup): Likewise.
10771 (aarch64_gen_ccmp_first): Likewise.
10772 (aarch64_gen_ccmp_next): Likewise.
10773 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
10774 (alpha_emit_xfloating_libcall): Likewise.
10775 (emit_insxl): Likewise.
10776 (alpha_arg_type): Likewise.
10777 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
10778 (arc_preferred_simd_mode): Likewise.
10779 (arc_secondary_reload): Likewise.
10780 (get_arc_condition_code): Likewise.
10781 (arc_print_operand): Likewise.
10782 (arc_legitimate_constant_p): Likewise.
10783 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10784 * config/arc/arc.md (casesi_load): Likewise.
10785 (casesi_compact_jump): Likewise.
10786 * config/arc/predicates.md (proper_comparison_operator): Likewise.
10787 (cc_use_register): Likewise.
10788 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10789 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
10790 (arm_init_iwmmxt_builtins): Likewise.
10791 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
10792 (neon_expand_vector_init): Likewise.
10793 (arm_attr_length_move_neon): Likewise.
10794 (maybe_get_arm_condition_code): Likewise.
10795 (arm_emit_vector_const): Likewise.
10796 (arm_preferred_simd_mode): Likewise.
10797 (arm_output_iwmmxt_tinsr): Likewise.
10798 (thumb1_output_casesi): Likewise.
10799 (thumb2_output_casesi): Likewise.
10800 (arm_emit_load_exclusive): Likewise.
10801 (arm_emit_store_exclusive): Likewise.
10802 (arm_expand_compare_and_swap): Likewise.
10803 (arm_evpc_neon_vuzp): Likewise.
10804 (arm_evpc_neon_vzip): Likewise.
10805 (arm_evpc_neon_vrev): Likewise.
10806 (arm_evpc_neon_vtrn): Likewise.
10807 (arm_evpc_neon_vext): Likewise.
10808 (arm_validize_comparison): Likewise.
10809 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
10810 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
10811 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
10812 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
10813 (c6x_preferred_simd_mode): Likewise.
10814 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
10815 (epiphany_rtx_costs): Likewise.
10816 * config/epiphany/predicates.md (proper_comparison_operator):
10818 * config/frv/frv.c (condexec_memory_operand): Likewise.
10819 (frv_emit_move): Likewise.
10820 (output_move_single): Likewise.
10821 (output_condmove_single): Likewise.
10822 (frv_hard_regno_mode_ok): Likewise.
10823 (frv_matching_accg_mode): Likewise.
10824 * config/h8300/h8300.c (split_adds_subs): Likewise.
10825 (h8300_rtx_costs): Likewise.
10826 (h8300_print_operand): Likewise.
10827 (compute_mov_length): Likewise.
10828 (output_logical_op): Likewise.
10829 (compute_logical_op_length): Likewise.
10830 (compute_logical_op_cc): Likewise.
10831 (h8300_shift_needs_scratch_p): Likewise.
10832 (output_a_shift): Likewise.
10833 (compute_a_shift_length): Likewise.
10834 (compute_a_shift_cc): Likewise.
10835 (expand_a_rotate): Likewise.
10836 (output_a_rotate): Likewise.
10837 * config/i386/i386.c (classify_argument): Likewise.
10838 (function_arg_advance_32): Likewise.
10839 (function_arg_32): Likewise.
10840 (function_arg_64): Likewise.
10841 (function_value_64): Likewise.
10842 (ix86_gimplify_va_arg): Likewise.
10843 (ix86_legitimate_constant_p): Likewise.
10844 (put_condition_code): Likewise.
10845 (split_double_mode): Likewise.
10846 (ix86_avx256_split_vector_move_misalign): Likewise.
10847 (ix86_expand_vector_logical_operator): Likewise.
10848 (ix86_split_idivmod): Likewise.
10849 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
10850 (ix86_build_const_vector): Likewise.
10851 (ix86_build_signbit_mask): Likewise.
10852 (ix86_match_ccmode): Likewise.
10853 (ix86_cc_modes_compatible): Likewise.
10854 (ix86_expand_branch): Likewise.
10855 (ix86_expand_sse_cmp): Likewise.
10856 (ix86_expand_sse_movcc): Likewise.
10857 (ix86_expand_int_sse_cmp): Likewise.
10858 (ix86_expand_vec_perm_vpermi2): Likewise.
10859 (ix86_expand_vec_perm): Likewise.
10860 (ix86_expand_sse_unpack): Likewise.
10861 (ix86_expand_int_addcc): Likewise.
10862 (ix86_split_to_parts): Likewise.
10863 (ix86_vectorize_builtin_gather): Likewise.
10864 (ix86_vectorize_builtin_scatter): Likewise.
10865 (avx_vpermilp_parallel): Likewise.
10866 (inline_memory_move_cost): Likewise.
10867 (ix86_tieable_integer_mode_p): Likewise.
10868 (x86_maybe_negate_const_int): Likewise.
10869 (ix86_expand_vector_init_duplicate): Likewise.
10870 (ix86_expand_vector_init_one_nonzero): Likewise.
10871 (ix86_expand_vector_init_one_var): Likewise.
10872 (ix86_expand_vector_init_concat): Likewise.
10873 (ix86_expand_vector_init_interleave): Likewise.
10874 (ix86_expand_vector_init_general): Likewise.
10875 (ix86_expand_vector_set): Likewise.
10876 (ix86_expand_vector_extract): Likewise.
10877 (emit_reduc_half): Likewise.
10878 (ix86_emit_i387_round): Likewise.
10879 (ix86_mangle_type): Likewise.
10880 (ix86_expand_round_sse4): Likewise.
10881 (expand_vec_perm_blend): Likewise.
10882 (canonicalize_vector_int_perm): Likewise.
10883 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
10884 (expand_vec_perm_1): Likewise.
10885 (expand_vec_perm_interleave3): Likewise.
10886 (expand_vec_perm_even_odd_pack): Likewise.
10887 (expand_vec_perm_even_odd_1): Likewise.
10888 (expand_vec_perm_broadcast_1): Likewise.
10889 (ix86_vectorize_vec_perm_const_ok): Likewise.
10890 (ix86_expand_vecop_qihi): Likewise.
10891 (ix86_expand_mul_widen_hilo): Likewise.
10892 (ix86_expand_sse2_abs): Likewise.
10893 (ix86_expand_pextr): Likewise.
10894 (ix86_expand_pinsr): Likewise.
10895 (ix86_preferred_simd_mode): Likewise.
10896 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
10897 * config/i386/sse.md (*andnot<mode>3): Likewise.
10898 (<mask_codefor><code><mode>3<mask_name>): Likewise.
10899 (*<code><mode>3): Likewise.
10900 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
10901 (ia64_expand_atomic_op): Likewise.
10902 (ia64_arg_type): Likewise.
10903 (ia64_mode_to_int): Likewise.
10904 (ia64_scalar_mode_supported_p): Likewise.
10905 (ia64_vector_mode_supported_p): Likewise.
10906 (expand_vec_perm_broadcast): Likewise.
10907 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
10908 (iq2000_function_arg_advance): Likewise.
10909 (iq2000_function_arg): Likewise.
10910 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
10911 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
10912 (m68k_libcall_value): Likewise.
10913 (m68k_function_value): Likewise.
10914 (sched_attr_op_type): Likewise.
10915 * config/mcore/mcore.c (mcore_output_move): Likewise.
10916 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
10918 (microblaze_function_arg): Likewise.
10919 * config/mips/mips.c (mips16_build_call_stub): Likewise.
10920 (mips_print_operand): Likewise.
10921 (mips_mode_ok_for_mov_fmt_p): Likewise.
10922 (mips_vector_mode_supported_p): Likewise.
10923 (mips_preferred_simd_mode): Likewise.
10924 (mips_expand_vpc_loongson_even_odd): Likewise.
10925 (mips_expand_vec_unpack): Likewise.
10926 (mips_expand_vi_broadcast): Likewise.
10927 (mips_expand_vector_init): Likewise.
10928 (mips_expand_vec_reduc): Likewise.
10929 (mips_expand_msa_cmp): Likewise.
10930 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
10931 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
10932 (cc_flags_for_mode): Likewise.
10933 * config/msp430/msp430.c (msp430_print_operand): Likewise.
10934 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
10935 (nds32_output_casesi_pc_relative): Likewise.
10936 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10937 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
10938 (nvptx_gen_unpack): Likewise.
10939 (nvptx_gen_pack): Likewise.
10940 (nvptx_gen_shuffle): Likewise.
10941 (nvptx_gen_wcast): Likewise.
10942 (nvptx_preferred_simd_mode): Likewise.
10943 * config/pa/pa.c (pa_secondary_reload): Likewise.
10944 * config/pa/predicates.md (base14_operand): Likewise.
10945 * config/powerpcspe/powerpcspe-c.c
10946 (altivec_resolve_overloaded_builtin): Likewise.
10947 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
10949 (rs6000_preferred_simd_mode): Likewise.
10950 (output_vec_const_move): Likewise.
10951 (rs6000_expand_vector_extract): Likewise.
10952 (rs6000_split_vec_extract_var): Likewise.
10953 (reg_offset_addressing_ok_p): Likewise.
10954 (rs6000_legitimate_offset_address_p): Likewise.
10955 (rs6000_legitimize_address): Likewise.
10956 (rs6000_emit_set_const): Likewise.
10957 (rs6000_const_vec): Likewise.
10958 (rs6000_emit_move): Likewise.
10959 (spe_build_register_parallel): Likewise.
10960 (rs6000_darwin64_record_arg_recurse): Likewise.
10961 (swap_selector_for_mode): Likewise.
10962 (spe_init_builtins): Likewise.
10963 (paired_init_builtins): Likewise.
10964 (altivec_init_builtins): Likewise.
10965 (do_load_for_compare): Likewise.
10966 (rs6000_generate_compare): Likewise.
10967 (rs6000_expand_float128_convert): Likewise.
10968 (emit_load_locked): Likewise.
10969 (emit_store_conditional): Likewise.
10970 (rs6000_output_function_epilogue): Likewise.
10971 (rs6000_handle_altivec_attribute): Likewise.
10972 (rs6000_function_value): Likewise.
10973 (emit_fusion_gpr_load): Likewise.
10974 (emit_fusion_p9_load): Likewise.
10975 (emit_fusion_p9_store): Likewise.
10976 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
10977 (fusion_gpr_mem_load): Likewise.
10978 (fusion_addis_mem_combo_load): Likewise.
10979 (fusion_addis_mem_combo_store): Likewise.
10980 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
10981 (fusion_gpr_mem_load): Likewise.
10982 (fusion_addis_mem_combo_load): Likewise.
10983 (fusion_addis_mem_combo_store): Likewise.
10984 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10986 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
10987 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
10988 (rs6000_preferred_simd_mode): Likewise.
10989 (output_vec_const_move): Likewise.
10990 (rs6000_expand_vector_extract): Likewise.
10991 (rs6000_split_vec_extract_var): Likewise.
10992 (reg_offset_addressing_ok_p): Likewise.
10993 (rs6000_legitimate_offset_address_p): Likewise.
10994 (rs6000_legitimize_address): Likewise.
10995 (rs6000_emit_set_const): Likewise.
10996 (rs6000_const_vec): Likewise.
10997 (rs6000_emit_move): Likewise.
10998 (rs6000_darwin64_record_arg_recurse): Likewise.
10999 (swap_selector_for_mode): Likewise.
11000 (paired_init_builtins): Likewise.
11001 (altivec_init_builtins): Likewise.
11002 (rs6000_expand_float128_convert): Likewise.
11003 (emit_load_locked): Likewise.
11004 (emit_store_conditional): Likewise.
11005 (rs6000_output_function_epilogue): Likewise.
11006 (rs6000_handle_altivec_attribute): Likewise.
11007 (rs6000_function_value): Likewise.
11008 (emit_fusion_gpr_load): Likewise.
11009 (emit_fusion_p9_load): Likewise.
11010 (emit_fusion_p9_store): Likewise.
11011 * config/rx/rx.c (rx_gen_move_template): Likewise.
11012 (flags_from_mode): Likewise.
11013 * config/s390/predicates.md (s390_alc_comparison): Likewise.
11014 (s390_slb_comparison): Likewise.
11015 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
11016 (s390_vector_mode_supported_p): Likewise.
11017 (s390_cc_modes_compatible): Likewise.
11018 (s390_match_ccmode_set): Likewise.
11019 (s390_canonicalize_comparison): Likewise.
11020 (s390_emit_compare_and_swap): Likewise.
11021 (s390_branch_condition_mask): Likewise.
11022 (s390_rtx_costs): Likewise.
11023 (s390_secondary_reload): Likewise.
11024 (__SECONDARY_RELOAD_CASE): Likewise.
11025 (s390_expand_cs): Likewise.
11026 (s390_preferred_simd_mode): Likewise.
11027 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
11028 * config/sh/sh.c (sh_print_operand): Likewise.
11029 (dump_table): Likewise.
11030 (sh_secondary_reload): Likewise.
11031 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11032 * config/sh/sh.md (casesi_worker_1): Likewise.
11033 (casesi_worker_2): Likewise.
11034 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
11035 (fcc_comparison_operator): Likewise.
11036 * config/sparc/sparc.c (sparc_expand_move): Likewise.
11037 (emit_soft_tfmode_cvt): Likewise.
11038 (sparc_preferred_simd_mode): Likewise.
11039 (output_cbranch): Likewise.
11040 (sparc_print_operand): Likewise.
11041 (sparc_expand_vec_perm_bmask): Likewise.
11042 (vector_init_bshuffle): Likewise.
11043 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11044 (spu_vector_mode_supported_p): Likewise.
11045 (spu_expand_insv): Likewise.
11046 (spu_emit_branch_or_set): Likewise.
11047 (spu_handle_vector_attribute): Likewise.
11048 (spu_builtin_splats): Likewise.
11049 (spu_builtin_extract): Likewise.
11050 (spu_builtin_promote): Likewise.
11051 (spu_expand_sign_extend): Likewise.
11052 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11053 (tilegx_simd_int): Likewise.
11054 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
11055 (tilepro_simd_int): Likewise.
11056 * config/v850/v850.c (const_double_split): Likewise.
11057 (v850_print_operand): Likewise.
11058 (ep_memory_offset): Likewise.
11059 * config/vax/vax.c (vax_rtx_costs): Likewise.
11060 (vax_output_int_move): Likewise.
11061 (vax_output_int_add): Likewise.
11062 (vax_output_int_subtract): Likewise.
11063 * config/visium/predicates.md (visium_branch_operator): Likewise.
11064 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
11065 (visium_print_operand_address): Likewise.
11066 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11067 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
11068 (xtensa_expand_conditional_branch): Likewise.
11069 (xtensa_copy_incoming_a7): Likewise.
11070 (xtensa_output_literal): Likewise.
11071 * dfp.c (decimal_real_maxval): Likewise.
11072 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11074 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11075 Alan Hayward <alan.hayward@arm.com>
11076 David Sherwood <david.sherwood@arm.com>
11078 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
11079 (mode_nunits_inline): Likewise.
11080 (mode_inner_inline): Likewise.
11081 (mode_unit_size_inline): Likewise.
11082 (mode_unit_precision_inline): Likewise.
11083 (emit_insn_modes_h): Likewise. Also emit a #define of the
11085 (emit_mode_wider): Add an E_ prefix to mode names.
11086 (emit_mode_complex): Likewise.
11087 (emit_mode_inner): Likewise.
11088 (emit_mode_adjustments): Likewise.
11089 (emit_mode_int_n): Likewise.
11090 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
11091 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
11092 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
11093 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
11094 (CRC32_BUILTIN, ENTRY): Likewise.
11095 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
11096 (aarch64_pop_regs): Likewise.
11097 (aarch64_process_components): Likewise.
11098 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
11099 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
11100 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
11101 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
11102 * config/arm/arm.c (arm_init_libfuncs): Likewise.
11103 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
11105 * config/i386/i386-builtin.def (pcmpestr): Likewise.
11106 (pcmpistr): Likewise.
11107 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
11108 * config/mmix/mmix.c (mmix_output_condition): Likewise.
11109 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
11111 * config/rl78/rl78.c (mduc_regs): Likewise.
11112 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
11113 (htm_expand_builtin): Likewise.
11114 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
11115 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
11116 * config/xtensa/xtensa.c (print_operand): Likewise.
11117 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
11118 (NUM_MODE_VECTOR_INT): Likewise.
11119 * genoutput.c (null_operand): Likewise.
11120 (output_operand_data): Likewise.
11121 * genrecog.c (print_parameter_value): Likewise.
11122 * lra.c (debug_operand_data): Likewise.
11124 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11126 * dwarf2out.c (scompare_loc_descriptor_wide)
11127 (scompare_loc_descriptor_narrow): New functions, split out from...
11128 (scompare_loc_descriptor): ...here.
11129 * expmed.c (emit_store_flag_int): New function, split out from...
11130 (emit_store_flag): ...here.
11132 2017-08-30 Richard Biener <rguenther@suse.de>
11134 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
11135 (dwarf2out_early_finish): Move setting of AT_pubnames from
11136 early debug output to early finish.
11138 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11140 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
11141 and -mdata-region to the assembler.
11143 2017-08-30 Richard Biener <rguenther@suse.de>
11145 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
11147 (gen_subprogram_die): Add DW_AT_object_pointer only early.
11148 (dwarf2out_early_global_decl): Only generate a DIE for the
11149 abstract origin if it doesn't already exist or is a declaration DIE.
11150 (resolve_addr): Do not add the linkage name twice when
11151 generating a stub DIE for the DW_TAG_GNU_call_site target.
11153 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11155 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11156 Use machine_mode rather than int for arg1_mode.
11158 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
11161 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
11162 that the second argument of the built-in functions to unpack
11163 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
11164 switch statement instead a lot of if statements.
11165 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
11166 Allow 64-bit values to be in Altivec registers as well as
11167 traditional floating point registers.
11168 (pack<mode>, FMOVE128_VSX iterator): Likewise.
11170 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
11172 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
11173 MAX_REGS_PER_ADDRESS == 1.
11175 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
11177 * config/i386/i386.opt (flag_fentry): Do not init to -1.
11178 * config/i386/i386.c (ix86_option_override_internal): Simplify
11179 setting of opts->x_flag_entry.
11181 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11182 Jakub Jelinek <jakub@redhat.com>
11183 Richard Biener <rguenther@suse.de>
11185 PR tree-optimization/81503
11186 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
11187 folded constant fits in the target type; reorder tests for clarity.
11189 2017-08-29 Martin Liska <mliska@suse.cz>
11191 * passes.def: Include pass_lower_switch.
11192 * stmt.c (dump_case_nodes): Remove and move to
11193 tree-switch-conversion.
11194 (case_values_threshold): Likewise.
11195 (expand_switch_as_decision_tree_p): Likewise.
11196 (emit_case_decision_tree): Likewise.
11197 (expand_case): Likewise.
11198 (balance_case_nodes): Likewise.
11199 (node_has_low_bound): Likewise.
11200 (node_has_high_bound): Likewise.
11201 (node_is_bounded): Likewise.
11202 (emit_case_nodes): Likewise.
11203 (struct simple_case_node): New struct.
11204 (add_case_node): Remove.
11205 (emit_case_dispatch_table): Use vector instead of case_list.
11206 (reset_out_edges_aux): Remove.
11207 (compute_cases_per_edge): Likewise.
11208 (expand_case): Build list of simple_case_node.
11209 (expand_sjlj_dispatch_table): Use it.
11210 * tree-switch-conversion.c (struct case_node): Moved from
11211 stmt.c and adjusted.
11212 (emit_case_nodes): Likewise.
11213 (node_has_low_bound): Likewise.
11214 (node_has_high_bound): Likewise.
11215 (node_is_bounded): Likewise.
11216 (case_values_threshold): Likewise.
11217 (reset_out_edges_aux): Likewise.
11218 (compute_cases_per_edge): Likewise.
11219 (add_case_node): Likewise.
11220 (dump_case_nodes): Likewise.
11221 (balance_case_nodes): Likewise.
11222 (expand_switch_as_decision_tree_p): Likewise.
11223 (emit_jump): Likewise.
11224 (emit_case_decision_tree): Likewise.
11225 (try_switch_expansion): Likewise.
11226 (do_jump_if_equal): Likewise.
11227 (emit_cmp_and_jump_insns): Likewise.
11228 (fix_phi_operands_for_edge): New function.
11229 (record_phi_operand_mapping): Likewise.
11230 (class pass_lower_switch): New pass.
11231 (pass_lower_switch::execute): New function.
11232 (make_pass_lower_switch): Likewise.
11233 (conditional_probability):
11234 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
11235 * tree-pass.h: Add make_pass_lower_switch.
11237 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11240 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
11243 2017-08-29 Richard Biener <rguenther@suse.de>
11245 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
11246 we do not add a DW_AT_inline attribute twice.
11247 (gen_subprogram_die): Remove code setting DW_AT_inline on
11248 DECL_ABSTRACT_P nodes.
11250 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
11252 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
11253 calls to internal functions.
11254 (gimplify_modify_expr): Likewise.
11255 * tree-call-cdce.c (use_internal_fn): Likewise.
11256 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
11257 (convert_to_divmod): Set the nothrow flag.
11258 * tree-if-conv.c (predicate_mem_writes): Likewise.
11259 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
11260 (vectorizable_call): Likewise.
11261 (vectorizable_store): Likewise.
11262 (vectorizable_load): Likewise.
11263 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
11264 (vect_recog_mask_conversion_pattern): Likewise.
11266 2017-08-29 Martin Liska <mliska@suse.cz>
11269 * gcc.c (driver_handle_option): Add new argument.
11270 * opts-common.c (handle_option): Pass
11271 target_option_override_hook.
11272 * opts-global.c (lang_handle_option): Add new option.
11273 (set_default_handlers): Add new argument.
11274 (decode_options): Likewise.
11275 * opts.c (target_handle_option): Likewise.
11276 (common_handle_option): Call target_option_override_hook.
11277 * opts.h (struct cl_option_handler_func): Add hook for
11278 target option override.
11279 (struct cl_option_handlers): Likewise.
11280 (set_default_handlers): Add new argument.
11281 (decode_options): Likewise.
11282 (common_handle_option): Likewise.
11283 (target_handle_option): Likewise.
11284 * toplev.c (toplev::main): Pass targetm.target_option.override
11287 2017-08-29 Richard Biener <rguenther@suse.de>
11288 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
11290 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
11291 life to the active subtree.
11293 2017-08-28 Jeff Law <law@redhat.com>
11295 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
11296 derive_equivalences.
11297 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
11298 Code moved into....
11299 (edge_info::derive_equivalences): New private member function
11301 * tree-ssa-dom.c (class edge_info): Changed from a struct
11302 to a class. Add ctor/dtor, methods and data members.
11303 (edge_info::edge_info): Renamed from allocate_edge_info.
11304 Initialize additional members.
11305 (edge_info::~edge_info): New.
11306 (free_dom_edge_info): Delete the edge info.
11307 (record_edge_info): Use new class & associated member functions.
11308 Tighten forms for testing for edge equivalences.
11309 (record_temporary_equivalences): Iterate over the simple
11310 equivalences rather than assuming there's only one per edge.
11311 (cprop_into_successor_phis): Iterate over the simple
11312 equivalences rather than assuming there's only one per edge.
11313 (optimize_stmt): Use operand_equal_p rather than pointer
11314 equality for mini-DSE code.
11316 2017-08-28 Nathan Sidwell <nathan@acm.org>
11318 * gcc.c (execute): Fold SIGPIPE handling into switch
11319 statement. Adjust internal error message.
11321 2017-08-28 Richard Biener <rguenther@suse.de>
11324 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
11325 Do nothing for removed DIEs.
11327 2017-08-28 Richard Biener <rguenther@suse.de>
11329 PR tree-optimization/81977
11330 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
11333 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
11336 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
11337 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
11338 using targetm.gen_mem_thread_fence.
11340 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
11343 * config/i386/i386.md (*<btsc><mode>): Change operand 2
11344 predicate to register_operand. Reorder operands.
11345 (*btr<mode>): Ditto.
11346 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
11347 (*btr<mode>_mask): Ditto.
11349 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
11351 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
11352 * config/rs6000/xmmintrin.h: New file.
11353 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
11355 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11358 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
11359 parameter and_insn and return it.
11360 (recombine_lvx_pattern): Insert a copy to ensure availability of
11361 the base register of the copied masking operation at the point of
11362 the instruction replacement.
11363 (recombine_stvx_pattern): Likewise.
11365 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
11367 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
11368 undocumented switches.
11369 (-mpower9-dform-vector): Likewise.
11370 (-mpower9-dform): Likewise.
11371 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
11372 comments to delete references to -mpower9-dform* switches.
11373 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
11374 Delete reference to -mpower9-dform* switches, test for
11375 -mpower9-vector instead.
11376 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
11377 (OTHER_P9_VECTOR_MASKS): Likewise.
11378 (POWERPC_MASKS): Likewise.
11379 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
11380 tests against -mpower9-dform* to -mpower9-vector. Delete code
11381 that checked for -mpower9-dform* consistancy with other options.
11382 Add test for -mpower9-misc to enable other power9 switches.
11383 (rs6000_init_hard_regno_mode_ok): Likewise.
11384 (rs6000_option_override_internal): Likewise.
11385 (rs6000_emit_prologue): Likewise.
11386 (rs6000_emit_epilogue): Likewise.
11387 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
11388 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
11389 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
11391 (emit_fusion_p9_store): Likewise.
11392 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
11393 resetting these macros if the assembler does not support ISA 3.0
11395 (TARGET_P9_DFORM_VECTOR): Likewise.
11396 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
11397 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
11399 2017-08-25 Alan Modra <amodra@gmail.com>
11401 PR rtl-optimization/81747
11402 * cse.c (cse_extended_basic_block): Don't attempt to record
11403 equivalences for degenerate conditional branches that branch
11404 to their fall-through.
11406 2017-08-24 Martin Sebor <msebor@redhat.com>
11408 PR middle-end/81908
11409 * gimple-fold.c (size_must_be_zero_p): New function.
11410 (gimple_fold_builtin_memory_op): Call it.
11412 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
11414 * config/rs6000/mm_malloc.h: New file.
11416 2017-08-24 Bin Cheng <bin.cheng@arm.com>
11418 PR tree-optimization/81913
11419 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
11420 analysis when either IVs in condition can wrap.
11422 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
11424 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
11425 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
11427 2017-08-24 Richard Biener <rguenther@suse.de>
11430 * targhooks.c (default_target_can_inline_p): Properly
11431 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
11432 is present and always compare.
11433 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
11434 infer -mfpmath=sse from TARGET_SSE_P.
11435 (ix86_can_inline_p): Properly use target_option_default_node when
11436 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
11438 2017-08-24 Richard Biener <rguenther@suse.de>
11441 * dwarf2out.c (output_die): Handle flag_generate_offload like
11443 (output_comp_unit): Likewise.
11444 (gen_array_type_die): Likewise.
11445 (dwarf2out_early_finish): Likewise.
11446 (note_variable_value_in_expr): Likewise.
11447 (dwarf2out_finish): Likewise. Adjust assert.
11448 * cgraphunit.c (symbol_table::compile): Move setting of
11449 flag_generate_offload earlier ...
11450 (symbol_table::finalize_compilation_unit): ... here, before
11451 early debug finalization.
11453 2017-08-24 Richard Biener <rguenther@suse.de>
11455 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
11456 and ipa-fnsummary.h.
11457 (ix86_can_inline_p): When ix86_fpmath flags do not match
11458 check whether the callee uses FP math at all.
11460 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
11462 PR middle-end/81931
11463 * tree-ssanames.c (get_nonzero_bits): Use element_precision
11464 instead of TYPE_PRECISION.
11466 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
11467 Alan Hayward <alan.hayward@arm.com>
11468 David Sherwood <david.sherwood@arm.com>
11470 * combine.c (make_extraction): Use subreg_offset_from_lsb.
11472 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
11474 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
11476 (ix86_frame::stack_realign_allocate): New field.
11477 (struct machine_frame_state): Modify comments.
11478 (machine_frame_state::sp_realigned_fp_end): New field.
11479 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
11480 layout calculation.
11481 (sp_valid_at): Add assertion to assure no attempt to access invalid
11482 offset of a realigned stack.
11483 (fp_valid_at): Likewise.
11484 (choose_baseaddr): Modify comments.
11485 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
11486 ix86_expand_prologue.
11487 (ix86_expand_prologue): Modify stack realignment and allocation.
11488 (ix86_expand_epilogue): Modify comments.
11489 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
11490 avx2_runtime, avx512f, and avx512f_runtime.
11492 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
11494 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
11495 (mstackrealign): Do not init to -1.
11496 * config/i386/i386.c (ix86_option_override_internal):
11497 Check opts_set, not opts when setting default value of
11498 opts->x_ix86_force_align_arg_pointer.
11500 2017-08-23 Richard Biener <rguenther@suse.de>
11502 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
11503 lang_hooks.decl_printable_name.
11504 * print-rtl-function.c (print_rtx_function): Likewise.
11505 * tree-pretty-print.c (dump_function_header): Likewise.
11507 2017-08-23 Richard Biener <rguenther@suse.de>
11510 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
11511 -g0 at compile-time.
11513 2017-08-23 Tamar Christina <tamar.christina@arm.com>
11515 PR middle-end/19706
11516 * doc/sourcebuild.texi (Other hardware attributes):
11519 2017-08-23 Tamar Christina <tamar.christina@arm.com>
11521 PR middle-end/19706
11522 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
11523 Fix single-use check.
11525 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
11527 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
11529 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
11531 * doc/install.texi: Modify to add more details on running selected
11534 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
11536 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
11537 is combined with -mabi=ms.
11538 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
11541 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11543 PR tree-optimization/81488
11544 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
11545 and cached_basis fields.
11546 (MAX_SPREAD): New constant.
11547 (alloc_cand_and_find_basis): Initialize new fields.
11548 (clear_visited): New function.
11549 (create_phi_basis_1): Rename from create_phi_basis, set visited
11550 and cached_basis fields.
11551 (create_phi_basis): New wrapper function.
11552 (phi_add_costs_1): Rename from phi_add_costs, add spread
11553 parameter, set visited field, short-circuit when limits reached.
11554 (phi_add_costs): New wrapper function.
11555 (record_phi_increments_1): Rename from record_phi_increments, set
11557 (record_phi_increments): New wrapper function.
11558 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
11559 (phi_incr_cost): New wrapper function.
11560 (all_phi_incrs_profitable_1): Rename from
11561 all_phi_incrs_profitable, set visited field.
11562 (all_phi_incrs_profitable): New wrapper function.
11564 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
11565 Alan Hayward <alan.hayward@arm.com>
11566 David Sherwood <david.sherwood@arm.com>
11568 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
11569 that takes the outer and inner modes.
11570 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
11571 comparison as the canonical test for a paradoxical subreg.
11572 * combine.c (simplify_set): Use paradoxical_subreg_p.
11573 (make_extraction): Likewise.
11574 (force_to_mode): Likewise.
11575 (rtx_equal_for_field_assignment_p): Likewise.
11576 (gen_lowpart_for_combine): Likewise.
11577 (simplify_comparison): Likewise.
11578 * cse.c (equiv_constant): Likewise.
11579 * expmed.c (store_bit_field_1): Likewise.
11580 * final.c (alter_subreg): Likewise.
11581 * fwprop.c (propagate_rtx): Likewise.
11582 (forward_propagate_subreg): Likewise.
11583 * ira-conflicts.c (ira_build_conflicts): Likewise.
11584 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
11585 * lra-constraints.c (curr_insn_transform): Likewise.
11586 (split_reg): Likewise.
11587 * lra-eliminations.c (move_plus_up): Likewise.
11588 (lra_eliminate_regs_1): Likewise.
11589 * recog.c (general_operand): Likewise.
11590 * ree.c (combine_reaching_defs): Likewise.
11591 * reload.c (push_reload): Likewise.
11592 (find_reloads): Likewise.
11593 * reload1.c (elimination_effects): Likewise.
11594 (compute_reload_subreg_offset): Likewise.
11595 (choose_reload_regs): Likewise.
11596 * rtlanal.c (subreg_lsb_1): Likewise.
11597 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11598 (simplify_subreg): Likewise.
11599 * var-tracking.c (track_loc_p): Likewise.
11600 * emit-rtl.c (byte_lowpart_offset): Likewise.
11601 (paradoxical_subreg_p): Delete out-of-line definition.
11603 2017-08-22 Jeff Law <law@redhat.com>
11605 PR tree-optimization/81741
11606 PR tree-optimization/71947
11607 * tree-ssa-dom.c: Include tree-inline.h.
11608 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
11609 equivalences if one is more expensive to compute than the other.
11610 * tree-ssa-scopedtables.h (class const_or_copies): Make
11611 record_const_or_copy_raw method private.
11612 (class avail_exprs_stack): New method simplify_binary_operation.
11613 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
11614 avail_exprs_stack::simplify_binary_operation as needed.
11615 (avail_exprs_stack::simplify_binary_operation): New function.
11617 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
11619 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
11620 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
11621 (DOT_SYMBOLS): Likewise.
11622 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
11623 (RELOCATABLE_NEEDS_FIXUP): Likewise.
11624 (RS6000_ABI_NAME): Likewise.
11625 (TARGET_CMODEL): Likewise.
11626 (TOC_SECTION_ASM_OP): Likewise.
11627 (SET_CMODEL): New macro.
11628 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
11630 2017-08-22 Richard Biener <rguenther@suse.de>
11632 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
11633 to allow for free-lang-data replacements similar to verify_type_variant.
11635 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
11636 Alan Hayward <alan.hayward@arm.com>
11637 David Sherwood <david.sherwood@arm.com>
11639 * config/aarch64/aarch64.md (casesi): Use DImode rather than
11640 VOIDmode for the LABEL_REF.
11642 2017-08-22 Richard Biener <rguenther@suse.de>
11644 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
11646 2017-08-22 Richard Biener <rguenther@suse.de>
11648 * common.opt (feliminate-dwarf2-dups): Ignore.
11649 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
11650 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
11651 same_die_p_wrap, compute_section_prefix,
11652 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
11653 (comdat_symbol_id, comdat_symbol_number): Likewise.
11654 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
11656 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
11657 (output_die): Mark unreachable path unreachable.
11658 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
11659 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
11660 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
11661 (dwarf2out_early_finish): Likewise.
11663 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
11665 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
11667 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
11670 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
11671 not VAR_P. Filter attribute warnings with OPT_Wattributes.
11672 (avr_attribute_table) <io, io_low, address>: Initialize
11673 .decl_required with true.
11675 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
11677 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
11678 undocumented debugging options.
11679 (-mvsx-scalar-double): Likewise.
11680 (-mallow-df-permute): Likewise.
11681 (-mvectorize-builtins): Likewise.
11682 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
11683 (rs6000_builtin_vectorized_function): Likewise.
11684 (rs6000_builtin_md_vectorized_function): Likewise.
11685 (rs6000_opt_vars): Likewise.
11687 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
11690 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
11691 (*btrq_imm): Rename from *btrq.
11692 (*btcq_imm): Rename from *btcq.
11693 (btsc): New code attribute.
11694 (*<btsc><mode>): New insn pattern.
11695 (*btr<mode>): Ditto.
11696 (*<btsc><mode>_mask): New insn_and_split pattern.
11697 (*btr<mode>_mask): Ditto.
11699 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11700 Alan Hayward <alan.hayward@arm.com>
11701 David Sherwood <david.sherwood@arm.com>
11703 * function.c (pad_below): Simplify padding calculation.
11705 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11706 Alan Hayward <alan.hayward@arm.com>
11707 David Sherwood <david.sherwood@arm.com>
11709 * target.def (function_prologue): Remove frame size argument.
11710 (function_epilogue): Likewise.
11711 * doc/tm.texi: Regenerate.
11712 * final.c (final_start_function): Update call to function_prologue.
11713 (final_end_function): Update call to function_epilogue.
11714 (default_function_pro_epilogue): Remove frame size argument.
11715 * output.h (default_function_pro_epilogue): Likewise.
11716 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
11717 (arm_output_function_prologue): Likewise.
11718 * config/frv/frv.c (frv_function_prologue): Likewise.
11719 (frv_function_epilogue): Likewise.
11720 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
11721 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
11722 (ia64_output_function_epilogue): Likewise.
11723 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
11724 (m32r_output_function_epilogue): Likewise.
11725 * config/microblaze/microblaze.c (microblaze_function_prologue)
11726 (microblaze_function_epilogue): Likewise.
11727 * config/mips/mips.c (mips_output_function_prologue): Likewise.
11728 (mips_output_function_epilogue): Likewise.
11729 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
11730 (mmix_target_asm_function_epilogue): Likewise.
11731 * config/msp430/msp430.c (msp430_start_function): Likewise.
11732 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
11733 (nds32_asm_function_epilogue): Likewise.
11734 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
11735 * config/pa/pa.c (pa_output_function_prologue): Likewise.
11736 (pa_output_function_epilogue): Likewise.
11737 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
11738 (rs6000_output_function_epilogue): Likewise.
11739 * config/rl78/rl78.c (rl78_start_function): Likewise.
11740 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
11741 (rs6000_output_function_epilogue): Likewise.
11742 * config/rx/rx.c (rx_output_function_prologue): Likewise.
11743 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
11744 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
11745 (sparc_asm_function_epilogue): Likewise.
11747 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11749 * tree.h (type_has_mode_precision_p): New function.
11750 * convert.c (convert_to_integer_1): Use it.
11751 * expr.c (expand_expr_real_2): Likewise.
11752 (expand_expr_real_1): Likewise.
11753 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
11754 * match.pd: Likewise.
11755 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
11756 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
11757 * tree-tailcall.c (process_assignment): Likewise.
11758 * tree-vect-loop.c (vectorizable_reduction): Likewise.
11759 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
11760 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
11761 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11762 (vectorizable_assignment): Likewise.
11763 (vectorizable_shift): Likewise.
11764 (vectorizable_operation): Likewise.
11765 * tree-vrp.c (register_edge_assert_for_2): Likewise.
11767 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
11769 * match.pd: Add pow (C, x) simplification.
11771 2017-08-21 Richard Biener <rguenther@suse.de>
11773 PR tree-optimization/81900
11774 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
11775 for blocks with abnormal predecessors.
11776 (compute_antic): Do not set visited flag prematurely.
11778 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
11781 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
11783 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11785 * stor-layout.h (vector_type_mode): Move to...
11786 * tree.h (vector_type_mode): ...here.
11787 * stor-layout.c (vector_type_mode): Move to...
11788 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
11790 2017-08-21 Richard Biener <rguenther@suse.de>
11792 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
11793 register_external_die hooks.
11794 (debug_false_tree_charstarstar_uhwistar): Declare.
11795 (debug_nothing_tree_charstar_uhwi): Likewise.
11796 * debug.c (do_nothing_debug_hooks): Adjust.
11797 (debug_false_tree_charstarstar_uhwistar): New do nothing.
11798 (debug_nothing_tree_charstar_uhwi): Likewise.
11799 * dbxout.c (dbx_debug_hooks): Adjust.
11800 (xcoff_debug_hooks): Likewise.
11801 * sdbout.c (sdb_debug_hooks): Likewise.
11802 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
11803 * dwarf2out.c (macinfo_label_base): New global.
11804 (dwarf2out_register_external_die): New function for the
11805 register_external_die hook.
11806 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
11807 (dwarf2_debug_hooks): Use them.
11808 (dwarf2_lineno_debug_hooks): Adjust.
11809 (struct die_struct): Add with_offset flag.
11810 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
11811 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
11812 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
11813 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
11814 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
11815 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
11816 defining section names for the early LTO debug variants.
11817 (reset_indirect_string): New helper.
11818 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
11819 (print_dw_val): Add support for offsetted symbol references.
11820 (get_ultimate_context): Split out from is_cxx.
11821 (is_cxx): Use get_ultimate_context.
11822 (is_fortran): Add decl overload.
11823 (compute_comp_unit_symbol): Split out worker from
11824 compute_section_prefix.
11825 (compute_section_prefix): Call compute_comp_unit_symbol and
11826 set comdat_type_p here.
11827 (output_die): Skip DIE symbol output for the LTO added one.
11828 Handle DIE symbol references with offset.
11829 (output_comp_unit): Guard section name mangling properly.
11830 For LTO debug sections emit a symbol at the section beginning
11831 which we use to refer to its DIEs.
11832 (add_abstract_origin_attribute): For DIEs registered via
11833 dwarf2out_register_external_die directly refer to the early
11834 DIE rather than indirectly through the shadow one we created.
11835 Remove obsolete call to dwarf2out_abstract_function for
11836 non-function/block origins.
11837 (gen_array_type_die): When generating early LTO debug do
11838 not emit DW_AT_string_length.
11839 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
11840 late when in LTO. As suggested place a gcc_unreachable for
11841 the DECL_ABSTRACT_P case.
11842 (gen_subprogram_die): Avoid another specification DIE
11843 for early built declarations/definitions for the late LTO case.
11844 (gen_variable_die): Add type references for late duplicated VLA dies
11846 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
11847 we have the abstract instance already.
11848 (process_scope_var): Adjust decl DIE contexts in LTO which
11849 first puts them in limbo.
11850 (gen_decl_die): Do not generate type DIEs late apart from
11851 types for VLAs or for decls we do not yet have a DIE. Do not
11852 call dwarf2out_abstract_function late.
11853 (dwarf2out_early_global_decl): Make sure to create DIEs
11854 for abstract instances of a decl first.
11855 (dwarf2out_late_global_decl): Adjust comment.
11856 (output_macinfo_op): With multiple macro sections use
11857 macinfo_label_base to distinguish labels.
11858 (output_macinfo): Likewise. Update macinfo_label_base.
11859 Pass in the line info label.
11860 (note_variable_value_in_expr): When generating LTO resolve
11861 all variable values here by generating DIEs as needed.
11862 (init_sections_and_labels): Add early LTO debug flag parameter
11863 and generate different sections and names if set. Add generation
11864 counter for the labels so we can have multiple of them.
11865 (reset_dies): Helper to allow DIEs to be output multiple times.
11866 (dwarf2out_finish): When outputting DIEs to the fat part of an
11867 LTO object first reset DIEs.
11868 (dwarf2out_early_finish): Output early DIEs when generating LTO.
11869 (modified_type_die): Check for decl_ultimate_origin being self
11871 (gen_type_die_with_usage): Likewise.
11872 (gen_typedef_die): Allow decl_ultimate_origin being self.
11873 (set_decl_abstract_flags): Remove.
11874 (set_block_abstract_flags): Likewise.
11875 (dwarf2out_abstract_function): Treat the early generated DIEs
11876 as the abstract copy and only add DW_AT_inline and
11877 DW_AT_artificial here and call set_decl_origin_self.
11878 If the DIE has an abstract origin don't do anything.
11879 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
11880 if we have none yet (Go fails to build one, PR78628).
11881 (variably_modified_type_p): Prevent endless recursion for Ada
11882 cyclic pointer types.
11883 * lto-streamer-in.c: Include debug.h.
11884 (dref_queue): New global.
11885 (lto_read_tree_1): Stream in DIE references.
11886 (lto_input_tree): Register DIE references.
11887 (input_function): Stream DECL_DEBUG_ARGS.
11888 * lto-streamer-out.c: Include debug.h.
11889 (lto_write_tree_1): Output DIE references.
11890 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
11891 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
11892 (output_function): Stream DECL_DEBUG_ARGS.
11893 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
11894 Stream DECL_ABSTRACT_ORIGIN.
11895 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
11896 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
11897 DECL_CONTEXT for file-scope decls.
11898 * lto-streamer.h (struct dref_entry): Declare.
11899 (dref_queue): Likewise.
11900 * cfgexpand.c (pass_expand::execute): Do not call the
11901 outlining_inline_function hook here.
11902 * lto-wrapper.c (debug_obj): New global.
11903 (tool_cleanup): Unlink it if required.
11904 (debug_objcopy): New function.
11905 (run_gcc): Handle early debug sections in the IL files by
11906 extracting them to separate files, partially linkin them and
11907 feeding the result back as result to the linker.
11908 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
11909 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
11910 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
11911 sections into a separate segment.
11912 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
11914 (darwin_asm_dwarf_section): Likewise.
11915 (darwin_asm_output_dwarf_offset): Likewise.
11916 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
11918 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11919 Alan Hayward <alan.hayward@arm.com>
11920 David Sherwood <david.sherwood@arm.com>
11922 * read-md.h (md_reader::record_potential_iterator_use): Replace
11923 pointer argument with an rtx and an index.
11924 * read-rtl.c (iterator_group::apply_iterator): Likewise.
11925 (apply_mode_iterator): Likewise.
11926 (apply_code_iterator): Likewise.
11927 (apply_int_iterator): Likewise.
11928 (apply_subst_iterator): Likewise.
11929 (record_iterator_use): Likewise.
11930 (record_attribute_use): Likewise.
11931 (md_reader::record_potential_iterator_use): Likewise. Update calls
11932 to record_iterator_use and apply_iterator.
11933 (iterator_use): Replace ptr with x and index.
11934 (attribute_use): Likewise.
11935 (apply_attribute_uses): Update calls to apply_iterator.
11936 (apply_iterators): Likewise. Update initialization of iterator_use.
11937 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
11938 and record_potential_iterator_use.
11939 (rtx_reader::read_rtx_operand): Likewise.
11941 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11942 Alan Hayward <alan.hayward@arm.com>
11943 David Sherwood <david.sherwood@arm.com>
11945 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
11948 2017-08-21 Richard Biener <rguenther@suse.de>
11950 PR middle-end/81884
11951 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
11952 at struct end conservatively when comparing common bases.
11954 2017-08-21 Richard Biener <rguenther@suse.de>
11956 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
11957 (mem_ref_in_stmt): Remove.
11958 (determine_max_movement): Use ref index to get at the reference.
11959 (invariantness_dom_walker::before_dom_children): Deal with
11960 lim data already initialized.
11961 (gather_mem_refs_stmt): Initialize lim data and record ref index.
11963 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
11965 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
11966 (TARGET_ISA_ROUND): Ditto.
11967 (TARGET_ROUND): Ditto.
11968 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
11969 * config/i386/i386.md: Ditto.
11970 * config/i386/sse.md: Ditto.
11971 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
11972 with OPTION_MASK_ISA_SSE4_1.
11974 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
11977 * doc/extend.texi (x86 Built-in Functions): Correct the name of
11978 __builtin_ia32_lzcnt_u16.
11980 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
11983 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
11984 (rs6000_set_current_function): Rewrite function to use it.
11986 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
11989 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
11990 and TYPE_WARN_IF_NOT_ALIGN.
11991 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
11992 (handle_warn_if_not_align): New.
11993 (place_union_field): Call handle_warn_if_not_align.
11994 (place_field): Call handle_warn_if_not_align.
11995 Copy TYPE_WARN_IF_NOT_ALIGN.
11996 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
11997 (layout_type): Likewise.
11998 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
12000 (tree_decl_common): Add warn_if_not_align.
12001 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
12002 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
12003 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
12004 (DECL_WARN_IF_NOT_ALIGN): Likewise.
12005 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
12006 * doc/extend.texi: Document warn_if_not_aligned attribute.
12007 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
12009 2017-08-17 Martin Liska <mliska@suse.cz>
12012 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
12013 (get_data_dependence): Use it as pointer type.
12014 (distribute_loop): Likewise.
12016 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12018 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
12019 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
12020 (p8_vmrgow_<mode>_direct): New define_insn.
12021 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
12022 handle endianness for vmrgew and vmrgow permute patterns.
12024 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
12026 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
12027 * config/rs6000/rs6000-cpus.def: Remove comment.
12028 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
12029 (POWERPC_MASKS): Likewise.
12030 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
12031 use of TARGET_VSX_TIMODE.
12032 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
12033 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
12034 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
12035 (rs6000_option_override_internal): Remove dead code.
12036 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
12037 (rs6000_legitimize_reload_address): Likewise.
12038 (rs6000_legitimate_address_p): Likewise.
12039 (rs6000_opt_masks): Delete "vsx-timode".
12040 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
12041 from function comment.
12042 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
12043 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
12044 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
12046 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
12047 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
12048 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
12049 (VSX_TI): Likewise.
12051 (define_peephole2): Likewise.
12053 2017-08-17 Martin Sebor <msebor@redhat.com>
12056 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
12057 past the end of an array.
12058 (test_pp_format): Add test cases.
12060 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
12062 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
12063 missing ECF_NOTHROW flags.
12065 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
12068 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
12069 operands residing in integer registers.
12070 (*vsx_le_perm_load_<mode>): Likewise.
12071 (*vsx_le_perm_store_<mode>): Likewise.
12072 (define_peephole2): Add peepholes to optimize the above.
12074 2017-08-17 Marek Polacek <polacek@redhat.com>
12076 PR middle-end/81814
12077 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
12078 to mimic what shorten_compare did. Change the return type to bool.
12079 (fold_cond_expr_with_comparison): Update call to
12080 operand_equal_for_comparison_p.
12081 (fold_ternary_loc): Likewise.
12083 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
12085 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
12087 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
12089 2017-08-17 Richard Biener <rguenther@suse.de>
12091 * tree-ssa-structalias.c (solve_graph): When propagating
12092 to successors update the graphs succ edges and avoid duplicate work.
12094 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
12097 * config/i386/i386.c (ix86_option_override_internal): Save target
12098 specific options after ix86_stack_protector_guard_reg was changed.
12100 2017-08-17 Richard Biener <rguenther@suse.de>
12102 PR tree-optimization/81827
12103 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
12104 (new_var_info): Initialize it conservatively.
12105 (get_call_vi): Mark register vars.
12106 (new_scalar_tmp_constraint_exp): Likewise.
12107 (handle_rhs_call): Likewise.
12108 (handle_const_call): Likewise.
12109 (create_function_info_for): Likewise.
12110 (solve_constraints): Sort varinfos to separate register from
12111 non-register vars to pack points-to solution bitmaps during
12114 2017-08-17 Marek Polacek <polacek@redhat.com>
12116 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
12118 2017-08-17 Richard Biener <rguenther@suse.de>
12120 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
12121 to true when overflow is undefined and we saturated the result.
12123 2017-08-17 Alan Modra <amodra@gmail.com>
12126 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
12127 Don't use store multiple if only one reg needs saving.
12128 (interesting_frame_related_regno): New function.
12129 (rs6000_frame_related): Don't emit frame info for regs that
12131 (rs6000_emit_epilogue): Likewise.
12133 2017-08-16 Nathan Sidwell <nathan@acm.org>
12135 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
12136 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
12137 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
12138 * tree.c (free_lang_data_in_type): Use else-if chain. Always
12139 clear TYPE_LANG_1. Remove obsolete member-function stripping.
12140 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
12141 (verify_type): Adjust for TYPE_BINFO move.
12142 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
12143 process TYPE_BINFO directly.
12144 (hash_tree): Likewise.
12145 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
12147 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
12150 2017-08-16 David Malcolm <dmalcolm@redhat.com>
12152 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
12154 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
12157 * config/i386/i386.md (*anddi_1_btr): Change predicates of
12158 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
12159 Add ix86_binary_operator_ok to insn constraint.
12160 (*iordi_1_bts): Ditto.
12161 (*xordi_1_btc): Ditto.
12162 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
12163 Update corresponding peephole2 pattern.
12167 2017-08-16 Bin Cheng <bin.cheng@arm.com>
12169 PR tree-optimization/81832
12170 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
12171 copy loop header which has IFN_LOOP_DIST_ALIAS call.
12173 2017-08-16 Marek Polacek <polacek@redhat.com>
12176 * fold-const.c (fold_indirect_ref_1): Restore original behavior
12177 regarding size_zero_node.
12179 2017-08-16 Martin Liska <mliska@suse.cz>
12182 * config.gcc: Respect previously set extra_objs in case
12185 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
12187 PR tree-optimization/81835
12188 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
12189 the phi SSA_NAME. Check that the condition in a COND_EXPR does
12190 not depend on the phi.
12192 2017-08-16 Alan Modra <amodra@gmail.com>
12194 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
12197 2017-08-16 Alan Modra <amodra@gmail.com>
12199 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
12200 (save_reg_p): ..into this. Update all callers.
12201 (first_reg_to_save): Simplify.
12203 2017-08-16 Alan Modra <amodra@gmail.com>
12205 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
12208 2017-08-15 Joseph Myers <joseph@codesourcery.com>
12212 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
12213 constant count if that count is less than 32.
12215 2017-08-15 Nathan Sidwell <nathan@acm.org>
12217 * gcc.c (execute): Emit friendlier message if inferior is killed
12218 by an external cause.
12220 2017-08-15 Richard Biener <rguenther@suse.de>
12222 PR tree-optimization/81790
12223 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
12224 CONSTRUCTORs from simplifying and VN.
12226 2017-08-14 Martin Sebor <msebor@redhat.com>
12228 * builtin-attrs.def: Add comments.
12230 2017-08-14 Martin Sebor <msebor@redhat.com>
12233 * doc/extend.texi (attribute nonstring): Document new attribute.
12235 2017-08-14 Martin Sebor <msebor@redhat.com>
12238 * tree-diagnostic.c (default_tree_printer): Handle %G.
12239 * gimple-pretty-print.h (percent_G_format): Declare new function.
12240 * gimple-pretty-print.c (percent_G_format): Define.
12241 * tree-pretty-print.c (percent_K_format): Add argument.
12243 2017-08-14 Martin Sebor <msebor@redhat.com>
12245 PR translation/79998
12246 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
12247 Remove a stray space.
12249 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
12252 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
12253 (*iordi_1_bts): Ditto.
12254 (*xordi_1_btc): Ditto.
12256 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12259 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
12260 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12262 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
12263 quoted strings, and make more translator-friendly.
12264 (darwin_rs6000_override_options): Likewise.
12265 (rs6000_option_override_internal): Likewise.
12266 (rs6000_return_in_memory): Fix overlong line.
12267 (init_cmulative_args): Use quoted strings, and make more
12268 translator-friendly.
12269 (rs6000_pass_by_reference): Fix overlong line.
12270 (def_builtin): Use quoted strings.
12271 (altivec_expand_predicate_builtin): Use quoted strings, and make
12272 more translator-friendly.
12273 (htm_expand_builtin): Use quoted strings.
12274 (cpu_expand_builtin): Use quoted strings, and make more
12275 translator-friendly.
12276 (altivec_expand_builtin): Likewise.
12277 (paired_expand_predicate_builtin): Likewise.
12278 (rs6000_invalid_builtin): Likewise.
12279 (builtin_function_type): Use quoted strings.
12280 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
12281 more translator-friendly.
12282 (rs6000_trampoline_init): Likewise.
12283 (rs6000_handle_altivec_attribute): Likewise.
12284 (rs6000_inner_target_options): Use quoted strings.
12285 (rs6000_disable_incompatible_switches): Likewise.
12286 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
12287 strings, and make more translator-friendly.
12288 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
12290 2017-08-14 Bin Cheng <bin.cheng@arm.com>
12292 PR tree-optimization/81799
12293 * tree-loop-distribution.c (version_loop_by_alias_check): Force
12294 cond_expr to simple gimple operand.
12296 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
12298 PR middle-end/46932
12299 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
12301 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
12305 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
12306 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
12307 TARGET_GASISR_PROLOGUES.
12308 * config/avr/avr.c (avr_option_override): Same.
12309 (avr_pass_pre_proep::execute): Same.
12311 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
12314 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
12315 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
12317 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
12319 * config/i386/i386.md (*load_tp_<mode>): Redefine as
12320 define_insn_and_split. Split to a memory load from 0 in
12321 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
12322 using PTR mode iterator.
12323 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
12324 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
12325 (*add_tp_<mode>): Redefine as define_insn_and_split.
12326 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
12327 address space. Merge with *add_tp_x32 using PTR mode iterator.
12328 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
12329 Split to an add with a memory load from 0 in
12330 DEFAULT_TLS_SEG_REG address space.
12332 2017-08-12 Andrew Pinski <apinski@cavium.com>
12334 * config/aarch64/aarch64-option-extensions.def (rdma):
12335 Fix feature string to what Linux prints out in /proc/cpuinfo.
12337 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
12340 * dwarf2out.c (modified_type_die): For C typedef types that have
12341 an ultimate origin, process the ultimate origin instead of the
12343 (gen_typedef_die): Assert that input DECLs have no ultimate
12345 (gen_type_die_with_usage): For typedef variants that have an
12346 ultimate origin, just call gen_decl_die on the original DECL.
12347 (process_scope_var): Avoid creating DIEs for local typedefs and
12348 concrete static variables.
12350 2017-08-12 Alan Modra <amodra@gmail.com>
12354 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
12355 match gnu-user.h startfile.
12356 (ENDFILE_LINUX_SPEC): Similarly.
12358 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
12361 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
12363 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
12365 2017-08-11 Tamar Christina <tamar.christina@arm.com>
12366 * config/aarch64/aarch64.md (mov<mode>): Change.
12367 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
12368 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
12369 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
12371 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
12373 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
12374 for storage order barriers.
12376 2017-08-11 Martin Liska <mliska@suse.cz>
12379 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
12380 variables of void type.
12382 2017-08-11 Martin Liska <mliska@suse.cz>
12384 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
12385 TARGET_SUPPORTS_ALIASES.
12386 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
12387 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
12388 (optimize_weakref): Likewise.
12389 * symtab.c (symtab_node::noninterposable_alias): Likewise.
12390 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
12391 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
12393 2017-08-11 Martin Liska <mliska@suse.cz>
12396 * config/i386/i386.c (make_resolver_func): Do complete
12397 refactoring of the function.
12399 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
12402 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
12403 * config/i386/i386.c (ix86_stack_protect_guard): Use
12404 ix86_stack_protect_guard_symbol_str to generate varible declaration.
12405 * doc/invoke.texi (x86 Options): Document
12406 -mstack-protector-guard-symbol= option.
12408 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
12410 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
12411 * config/i386/i386.c (ix86_split_stack_guard): New function.
12412 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
12413 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
12414 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
12415 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
12416 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
12417 (split_stack_space_check): Call ix86_split_stack_guard.
12419 2017-08-10 Martin Sebor <msebor@redhat.com>
12421 * print-tree.c (print_node): Print location using the established
12423 Replace spaces with colons.
12424 (debug_raw, debug): Ditto.
12426 2017-08-10 Martin Sebor <msebor@redhat.com>
12429 * pretty-print.c (pp_format): Correct the handling of %s precision.
12431 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
12434 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
12436 (ix86_finalize_stack_frame_flags): This. Also clear
12437 frame_pointer_needed if -fno-omit-frame-pointer is used without
12439 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
12440 with ix86_finalize_stack_frame_flags.
12441 (ix86_expand_epilogue): Likewise.
12442 (ix86_expand_split_stack_prologue): Likewise.
12443 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
12445 2017-08-10 Martin Liska <mliska@suse.cz>
12448 * c-attribs.c (handle_target_attribute):
12449 Report warning for an empty string argument of target attribute.
12451 2017-08-09 Jakub Jelinek <jakub@redhat.com>
12454 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
12456 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
12457 or DECL_NONLOCAL labels.
12458 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
12459 or DECL_NONLOCAL labels here.
12461 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
12463 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
12464 to indicate when early gimple folding has been disabled.
12465 (rs6000_gimple_fold_builtin): Add debug content.
12466 (rs6000_invalid_builtin): Fix whitespace.
12467 (rs6000_expand_builtin): Fix whitespace.
12468 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
12470 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
12473 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
12474 SAVE_MULTIPLE if not all the registers that saves, should be saved.
12476 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
12478 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
12479 (qdf24xx): Likewise.
12480 * config/aarch64/aarch64.md: Include falkor.md.
12481 * config/aarch64/falkor.md: New.
12483 2017-08-09 Marek Polacek <polacek@redhat.com>
12486 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
12487 * diagnostic.c (emit_diagnostic): Add a comment.
12488 (emit_diagnostic_valist): New function.
12490 2017-08-09 Marek Polacek <polacek@redhat.com>
12493 * input.c (make_location): New overload.
12494 * input.h (make_location): Declare.
12496 2017-08-08 Alan Modra <amodra@gmail.com>
12497 H.J. Lu <hongjiu.lu@intel.com>
12500 * gcc.c (NO_PIE_SPEC): Delete.
12501 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
12503 (LINK_PIE_SPEC): ..to here.
12504 (LINK_COMMAND_SPEC): Support -no-pie.
12505 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
12506 chain of crtbegin*.o selection, update for PIE_SPEC changes and
12508 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
12509 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
12510 (ENDFILE_CRTEND_SPEC): Similarly.
12512 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
12515 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
12516 (mstack-protector-guard-offset=): Ditto.
12517 * config/i386/i386.c (ix86_option_override): Handle
12518 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
12520 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
12521 ix86_stack_protect_guard_offset variables.
12522 (TARGET_STACK_PROTECT_GUARD): Always define.
12523 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
12524 and -mstack-protector-guard-offset= options.
12526 2017-08-08 Bin Cheng <bin.cheng@arm.com>
12528 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
12529 boundary case for the last candidate.
12531 2017-08-08 Bin Cheng <bin.cheng@arm.com>
12533 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
12534 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
12536 2017-08-08 Tamar Christina <tamar.christina@arm.com>
12538 PR middle-end/19706
12539 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
12540 * config/aarch64/aarch64-builtins.c
12541 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
12542 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
12543 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
12545 2017-08-08 Tamar Christina <tamar.christina@arm.com>
12546 Andrew Pinski <pinskia@gmail.com>
12548 PR middle-end/19706
12549 * internal-fn.def (XORSIGN): New.
12550 * optabs.def (xorsign_optab): New.
12551 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
12552 (convert_expand_mult_copysign): New.
12553 (pass_optimize_widening_mul::execute): Call
12554 convert_expand_mult_copysign.
12556 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12558 PR tree-optimization/81354
12559 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
12560 Insert on edges rather than explicitly creating landing pads.
12561 (analyze_candidates_and_replace): Commit edge inserts.
12563 2017-08-08 Richard Biener <rguenther@suse.de>
12565 PR middle-end/81719
12566 * tree-ssa-loop-niter.c: Include tree-dfa.h.
12567 (expand_simple_operations): Also look through ADDR_EXPRs with
12568 MEM_REF bases treating them as POINTER_PLUS_EXPR.
12570 2017-08-08 Richard Biener <rguenther@suse.de>
12572 PR tree-optimization/81723
12573 * tree-vect-slp.c (struct bst_traits): New hash traits.
12574 (bst_fail): New global.
12575 (vect_build_slp_tree_2): New worker, split out from ...
12576 (vect_build_slp_tree): ... this now wrapping it with using
12577 bst_fail set to cache SLP tree build fails. Properly handle
12579 (vect_analyze_slp_instance): Allocate and free bst_fail.
12581 2017-08-08 Martin Liska <mliska@suse.cz>
12584 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
12585 LABEL_DECLs that can be from a different function.
12587 2017-08-08 Bin Cheng <bin.cheng@arm.com>
12589 PR tree-optimization/81744
12590 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
12591 loop's number of iterations.
12593 2017-08-08 Martin Liska <mliska@suse.cz>
12595 * asan.c: Include header files.
12596 * attribs.c (build_decl_attribute_variant): New function moved
12598 (build_type_attribute_qual_variant): Likewise.
12599 (cmp_attrib_identifiers): Likewise.
12600 (simple_cst_list_equal): Likewise.
12601 (omp_declare_simd_clauses_equal): Likewise.
12602 (attribute_value_equal): Likewise.
12603 (comp_type_attributes): Likewise.
12604 (build_type_attribute_variant): Likewise.
12605 (lookup_ident_attribute): Likewise.
12606 (remove_attribute): Likewise.
12607 (merge_attributes): Likewise.
12608 (merge_type_attributes): Likewise.
12609 (merge_decl_attributes): Likewise.
12610 (merge_dllimport_decl_attributes): Likewise.
12611 (handle_dll_attribute): Likewise.
12612 (attribute_list_equal): Likewise.
12613 (attribute_list_contained): Likewise.
12614 * attribs.h (lookup_attribute): New function moved from tree.[ch].
12615 (lookup_attribute_by_prefix): Likewise.
12616 * bb-reorder.c: Include header files.
12617 * builtins.c: Likewise.
12618 * calls.c: Likewise.
12619 * cfgexpand.c: Likewise.
12620 * cgraph.c: Likewise.
12621 * cgraphunit.c: Likewise.
12622 * convert.c: Likewise.
12623 * dwarf2out.c: Likewise.
12624 * final.c: Likewise.
12625 * fold-const.c: Likewise.
12626 * function.c: Likewise.
12627 * gimple-expr.c: Likewise.
12628 * gimple-fold.c: Likewise.
12629 * gimple-pretty-print.c: Likewise.
12630 * gimple.c: Likewise.
12631 * gimplify.c: Likewise.
12632 * hsa-common.c: Likewise.
12633 * hsa-gen.c: Likewise.
12634 * internal-fn.c: Likewise.
12635 * ipa-chkp.c: Likewise.
12636 * ipa-cp.c: Likewise.
12637 * ipa-devirt.c: Likewise.
12638 * ipa-fnsummary.c: Likewise.
12639 * ipa-inline.c: Likewise.
12640 * ipa-visibility.c: Likewise.
12642 * lto-cgraph.c: Likewise.
12643 * omp-expand.c: Likewise.
12644 * omp-general.c: Likewise.
12645 * omp-low.c: Likewise.
12646 * omp-offload.c: Likewise.
12647 * omp-simd-clone.c: Likewise.
12648 * opts-global.c: Likewise.
12649 * passes.c: Likewise.
12650 * predict.c: Likewise.
12651 * sancov.c: Likewise.
12652 * sanopt.c: Likewise.
12653 * symtab.c: Likewise.
12654 * toplev.c: Likewise.
12655 * trans-mem.c: Likewise.
12656 * tree-chkp.c: Likewise.
12657 * tree-eh.c: Likewise.
12658 * tree-into-ssa.c: Likewise.
12659 * tree-object-size.c: Likewise.
12660 * tree-parloops.c: Likewise.
12661 * tree-profile.c: Likewise.
12662 * tree-ssa-ccp.c: Likewise.
12663 * tree-ssa-live.c: Likewise.
12664 * tree-ssa-loop.c: Likewise.
12665 * tree-ssa-sccvn.c: Likewise.
12666 * tree-ssa-structalias.c: Likewise.
12667 * tree-ssa.c: Likewise.
12668 * tree-streamer-in.c: Likewise.
12669 * tree-vectorizer.c: Likewise.
12670 * tree-vrp.c: Likewise.
12671 * tsan.c: Likewise.
12672 * ubsan.c: Likewise.
12673 * varasm.c: Likewise.
12674 * varpool.c: Likewise.
12675 * tree.c: Remove functions moved to attribs.[ch].
12676 * tree.h: Likewise.
12677 * config/aarch64/aarch64.c: Add attrs.h header file.
12678 * config/alpha/alpha.c: Likewise.
12679 * config/arc/arc.c: Likewise.
12680 * config/arm/arm.c: Likewise.
12681 * config/avr/avr.c: Likewise.
12682 * config/bfin/bfin.c: Likewise.
12683 * config/c6x/c6x.c: Likewise.
12684 * config/cr16/cr16.c: Likewise.
12685 * config/cris/cris.c: Likewise.
12686 * config/darwin.c: Likewise.
12687 * config/epiphany/epiphany.c: Likewise.
12688 * config/fr30/fr30.c: Likewise.
12689 * config/frv/frv.c: Likewise.
12690 * config/ft32/ft32.c: Likewise.
12691 * config/h8300/h8300.c: Likewise.
12692 * config/i386/winnt.c: Likewise.
12693 * config/ia64/ia64.c: Likewise.
12694 * config/iq2000/iq2000.c: Likewise.
12695 * config/lm32/lm32.c: Likewise.
12696 * config/m32c/m32c.c: Likewise.
12697 * config/m32r/m32r.c: Likewise.
12698 * config/m68k/m68k.c: Likewise.
12699 * config/mcore/mcore.c: Likewise.
12700 * config/microblaze/microblaze.c: Likewise.
12701 * config/mips/mips.c: Likewise.
12702 * config/mmix/mmix.c: Likewise.
12703 * config/mn10300/mn10300.c: Likewise.
12704 * config/moxie/moxie.c: Likewise.
12705 * config/msp430/msp430.c: Likewise.
12706 * config/nds32/nds32-isr.c: Likewise.
12707 * config/nds32/nds32.c: Likewise.
12708 * config/nios2/nios2.c: Likewise.
12709 * config/nvptx/nvptx.c: Likewise.
12710 * config/pa/pa.c: Likewise.
12711 * config/pdp11/pdp11.c: Likewise.
12712 * config/powerpcspe/powerpcspe.c: Likewise.
12713 * config/riscv/riscv.c: Likewise.
12714 * config/rl78/rl78.c: Likewise.
12715 * config/rx/rx.c: Likewise.
12716 * config/s390/s390.c: Likewise.
12717 * config/sh/sh.c: Likewise.
12718 * config/sol2.c: Likewise.
12719 * config/sparc/sparc.c: Likewise.
12720 * config/spu/spu.c: Likewise.
12721 * config/stormy16/stormy16.c: Likewise.
12722 * config/tilegx/tilegx.c: Likewise.
12723 * config/tilepro/tilepro.c: Likewise.
12724 * config/v850/v850.c: Likewise.
12725 * config/vax/vax.c: Likewise.
12726 * config/visium/visium.c: Likewise.
12727 * config/xtensa/xtensa.c: Likewise.
12729 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
12732 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
12733 constraints since the -mupper-regs-* switches have been
12735 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
12736 into a vector from a double word element that was extracted from
12737 another vector, and eliminate extra XXPERMDI instructions.
12738 (vsx_concat_<mode>_2): Likewise.
12739 (vsx_concat_<mode>_3): Likewise.
12740 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
12741 concat to allow optimizing inserts from previous extracts.
12743 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
12745 * config/i386/i386.c (ix86_stack_protect_guard): Generate
12746 memory reference to a SSP offset in TLS address space.
12747 (ix86_print_operand) <case '@'>: Remove.
12748 (ix86_print_operand_punct_valid_p): Remove '@' code.
12749 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
12750 UNSPEC_SP_TLS_TEST.
12751 (stack_tls_protect_set_<mode>): Remove.
12752 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
12753 (stack_tls_protect_test_<mode>): Remove.
12754 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
12756 2017-08-07 Olivier Hainque <hainque@adacore.com>
12759 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
12761 2017-08-07 Douglas Rupp <rupp@adacore.com>
12763 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
12764 variable was referenced as multidir in command.
12766 2017-08-07 Jakub Jelinek <jakub@redhat.com>
12769 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
12772 2017-08-07 Martin Liska <mliska@suse.cz>
12774 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
12775 * config/rl78/rl78.c: Add include of attribs.h.
12776 * config/sh/sh.c: Likewise.
12777 * config/v850/v850.c: Likewise.
12779 2017-08-07 Tom de Vries <tom@codesourcery.com>
12781 PR middle-end/78266
12782 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
12784 2017-08-07 Martin Liska <mliska@suse.cz>
12786 * config/mips/mips.c: Include attribs.h.
12788 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
12791 * doc/invoke.texi: Document change in behvaior for -Ofast for
12794 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
12796 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
12798 (aarch64_pop_regs): Likewise.
12799 (aarch64_gen_load_pair): Likewise.
12800 (aarch64_save_callee_saves): Likewise.
12801 (aarch64_restore_callee_saves): Likewise.
12803 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
12805 * config/i386/i386.c: Revert the last change.
12807 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
12810 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
12812 (ix86_finalize_stack_frame_flags): This. Also clear
12813 frame_pointer_needed if -fno-omit-frame-pointer is used without
12815 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
12816 with ix86_finalize_stack_frame_flags.
12817 (ix86_expand_epilogue): Likewise.
12818 (ix86_expand_split_stack_prologue): Likewise.
12820 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
12823 * config/i386/i386.c (get_builtin_code_for_version): Set priority
12824 to P_AES for Westmere.
12826 2017-08-07 Jonathan Yong <10walls@gmail.com>
12828 * config/i386/mingw.opt (fset-stack-executable): Removed.
12829 * config/i386/cygming.opt (fset-stack-executable): Moved
12831 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
12833 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
12835 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
12837 2017-08-07 Marek Polacek <polacek@redhat.com>
12839 PR middle-end/81737
12840 * fold-const.c (fold_indirect_ref_1): Check type_domain.
12842 2017-08-07 Martin Liska <mliska@suse.cz>
12844 * attribs.h (canonicalize_attr_name): New function.
12845 (cmp_attribs): Move from c-format.c and adjusted.
12846 (is_attribute_p): Moved from tree.h.
12847 * tree-inline.c: Add new includes.
12848 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
12849 (private_is_attribute_p): Remove.
12850 (private_lookup_attribute): Likewise.
12851 (private_lookup_attribute_by_prefix): Simplify.
12852 (remove_attribute): Use is_attribute_p.
12853 * tree.h: Remove removed declarations.
12855 2017-08-07 Jakub Jelinek <jakub@redhat.com>
12857 PR middle-end/81698
12858 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
12859 instead of computing it in the function. Formatting fix.
12860 (expand_case): Don't rely on default_edge being the first edge,
12861 clear it if removing it, pass default_edge to
12862 emit_case_dispatch_table.
12863 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
12866 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
12868 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
12869 insn in the function, emit NOP after the insn.
12871 2017-08-06 Tom de Vries <tom@codesourcery.com>
12873 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
12876 2017-08-06 Tom de Vries <tom@codesourcery.com>
12878 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
12881 2017-08-04 Yury Gribov <tetra2005@gmail.com>
12883 PR tree-optimization/57371
12884 * match.pd: New pattern.
12886 2017-08-04 Marek Polacek <polacek@redhat.com>
12888 PR middle-end/81695
12889 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
12890 perform the computation in offset_int.
12892 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
12894 PR tree-optimization/81136
12895 * tree-vectorizer.h: Include tree-hash-traits.h.
12896 (vec_base_alignments): New typedef.
12897 (vec_info): Add a base_alignments field.
12898 (vect_record_base_alignments): Declare.
12899 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
12901 (DR_IS_CONDITIONAL_IN_STMT): New macro.
12902 (create_data_ref): Add an is_conditional_in_stmt argument.
12903 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
12904 the is_conditional_in_stmt field.
12905 (data_ref_loc): Add an is_conditional_in_stmt field.
12906 (get_references_in_stmt): Set the is_conditional_in_stmt field.
12907 (find_data_references_in_stmt): Update call to create_data_ref.
12908 (graphite_find_data_references_in_stmt): Likewise.
12909 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
12910 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
12911 (vect_record_base_alignment): New function.
12912 (vect_record_base_alignments): Likewise.
12913 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
12914 for nested statements even if we fail to compute a misalignment.
12915 Use pooled base alignments for unconditional references.
12916 (vect_find_same_alignment_drs): Compare base addresses instead
12918 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
12919 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
12921 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
12923 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
12924 Add an explicit name for the enum. Use auto_vec for slp_instances
12925 and grouped_stores.
12926 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
12928 (_bb_vec_info): Add a constructor and destructor.
12929 (vinfo_for_stmt): Return NULL for uids of -1 as well.
12930 (destroy_loop_vec_info): Delete.
12931 (vect_destroy_datarefs): Likewise.
12932 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
12933 (vec_info::vec_info): New function.
12934 (vec_info::~vec_info): Likewise.
12935 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
12936 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
12937 destroy_loop_vec_info.
12938 * tree-vect-loop.c (new_loop_vec_info): Replace with...
12939 (_loop_vec_info::_loop_vec_info): ...this.
12940 (destroy_loop_vec_info): Replace with...
12941 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
12942 the stmt_vec_infos. Leave handling of vec_info information to its
12943 destructor. Remove explicit vector releases.
12944 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
12945 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
12946 * tree-vect-slp.c (new_bb_vec_info): Replace with...
12947 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
12948 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
12949 (destroy_bb_vec_info): Replace with...
12950 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
12951 information to its destructor.
12952 (vect_slp_analyze_bb_1): Use new and delete instead of
12953 new_bb_vec_info and destroy_bb_vec_info.
12954 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
12957 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
12959 * tree-data-ref.h (subscript): Add access_fn field.
12960 (data_dependence_relation): Add could_be_independent_p.
12961 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
12962 (same_access_functions): Move to tree-data-ref.c.
12963 * tree-data-ref.c (ref_contains_union_access_p): New function.
12964 (access_fn_component_p): Likewise.
12965 (access_fn_components_comparable_p): Likewise.
12966 (dr_analyze_indices): Add a reference to access_fn_component_p.
12967 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
12969 (constant_access_functions): Likewise.
12970 (add_other_self_distances): Likewise.
12971 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
12972 (initialize_data_dependence_relation): Use XCNEW and remove
12973 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
12974 of access functions that have the same type. Allow the
12975 subsequence to end with different bases in some circumstances.
12976 Record the chosen access functions in SUB_ACCESS_FN.
12977 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
12978 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
12979 (subscript_dependence_tester_1): Likewise dra and drb.
12980 (build_classic_dist_vector): Update calls accordingly.
12981 (subscript_dependence_tester): Likewise.
12982 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
12983 DDR_COULD_BE_INDEPENDENT_P.
12984 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
12985 comp_alias_ddrs instead of may_alias_ddrs.
12986 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
12988 (vect_analyze_data_ref_dependence): Use it if
12989 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
12990 distance vectors if that fails.
12991 (dependence_distance_ge_vf): New function.
12992 (vect_prune_runtime_alias_test_list): Use it. Don't clear
12993 LOOP_VINFO_MAY_ALIAS_DDRS.
12995 2017-08-04 Richard Biener <rguenther@suse.de>
12997 PR middle-end/81705
12998 * fold-const.c (fold_binary_loc): Properly restrict
12999 minus_var0 && minus_var1 case when associating undefined overflow
13002 2017-08-04 Tom de Vries <tom@codesourcery.com>
13004 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
13006 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13008 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13009 Don't start diagnostic messages with a capital letter.
13010 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13012 (rs6000_invalid_builtin): Likewise.
13013 (rs6000_trampoline_init): Likewise.
13015 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13018 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
13019 after setting changeable df flags.
13021 2017-08-03 Richard Biener <rguenther@suse.de>
13023 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
13024 up if the use is in USE - X.
13026 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
13028 * toplev.c (dumpfile.h): New include.
13029 (internal_error_reentered): New static function. Use it...
13030 (internal_error_function): ...here to handle reentered internal_error.
13032 2017-08-03 Richard Biener <rguenther@suse.de>
13034 PR middle-end/81148
13035 * fold-const.c (split_tree): Add minus_var and minus_con
13036 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
13037 here but always use minus_*.
13038 (associate_trees): Assert we never associate with MINUS_EXPR
13039 and NULL first operand. Do not recurse for PLUS_EXPR operands
13040 when associating as MINUS_EXPR either.
13041 (fold_binary_loc): Track minus_var and minus_con.
13043 2017-08-03 Tom de Vries <tom@codesourcery.com>
13046 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
13047 ACCEL_COMPILER, apply finish_options on
13048 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
13050 2017-08-03 Tom de Vries <tom@codesourcery.com>
13053 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
13054 function_entry_patch_area_size > 0.
13056 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13059 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
13060 instead of 10??LU, perform unit multiplication in wide_int,
13061 don't change alloc_object_size_limit if the limit is larger
13064 PR tree-optimization/81655
13065 PR tree-optimization/81588
13066 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
13067 the case when ranges[i].low and high are 1 for unsigned type with
13070 PR middle-end/81052
13071 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
13072 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
13074 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13076 * tree-vrp.h: Add include guard.
13078 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
13081 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
13082 (ud2): New insn pattern.
13083 * config/i386/i386.c (ix86_expand_epilogue):
13084 For naked functions, generate ud2 instead of trap insn.
13086 2017-08-02 Marek Polacek <polacek@redhat.com>
13089 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
13091 2017-08-02 Tom de Vries <tom@codesourcery.com>
13092 Cesar Philippidis <cesar@codesourcery.com>
13094 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
13095 Add missing edge probabilities.
13097 2017-08-02 Tamar Christina <tamar.christina@arm.com>
13099 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
13100 Correct endianness.
13102 2017-08-02 Jakub Jelinek <jakub@redhat.com>
13104 PR middle-end/79499
13105 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
13106 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
13107 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
13108 prologue_seq sequences - if any.
13110 2017-08-02 Richard Biener <rguenther@suse.de>
13112 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
13113 via vectors if supported, integer extracts via punning if supported
13114 or otherwise vector extracts.
13116 2017-08-02 Richard Biener <rguenther@suse.de>
13118 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
13120 (bitmap_insert_into_set): ... this.
13122 2017-08-02 Richard Biener <rguenther@suse.de>
13124 PR tree-optimization/81633
13126 2015-08-17 Alan Hayward <alan.hayward@arm.com>
13128 PR tree-optimization/71752
13129 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
13131 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
13133 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
13134 (machine_function::call_ms2sysv_pad_out): Remove field.
13135 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
13136 (ix86_compute_frame_layout): Likewise.
13138 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
13141 * config/i386/i386.c (ix86_set_func_type): Disallow naked
13142 attribute with interrupt attribute.
13144 2017-08-01 Andrew Pinski <apinski@cavium.com>
13146 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
13147 BIT_INSERT_EXPR's operand 1
13148 to see if the types precision matches.
13150 2017-08-01 Martin Liska <mliska@suse.cz>
13152 PR middle-end/70140
13153 * builtins.c (expand_builtin_memcpy_args): Remove.
13154 (expand_builtin_memcpy): Call newly added function
13155 expand_builtin_memory_copy_args.
13156 (expand_builtin_memcpy_with_bounds): Likewise.
13157 (expand_builtin_mempcpy): Remove last argument.
13158 (expand_builtin_mempcpy_with_bounds): Likewise.
13159 (expand_builtin_memory_copy_args): New function created from
13160 expand_builtin_mempcpy_args with small modifications.
13161 (expand_builtin_mempcpy_args): Remove.
13162 (expand_builtin_stpcpy): Remove unused argument.
13163 (expand_builtin): Likewise.
13164 (expand_builtin_with_bounds): Likewise.
13166 2017-08-01 Martin Liska <mliska@suse.cz>
13169 Make mempcpy more optimal (PR middle-end/70140).
13171 2017-08-01 Jakub Jelinek <jakub@redhat.com>
13174 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
13175 __builtin_vec_cmpne verify both arguments are compatible vectors
13176 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
13177 verify arg1_type is a pointer or array type. For __builtin_vec_st,
13178 move computation of aligned to after checking the argument types.
13182 * config/rs6000/vsx.md (vextract_fp_from_shorth,
13183 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
13186 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
13187 Doug Rupp <rupp@adacore.com>
13188 Olivier Hainque <hainque@adacore.com>
13190 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
13191 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
13193 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
13194 for TARGET_OS_CPP_BUILTIN.
13195 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
13196 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
13198 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
13199 passing required abi options to the assembler for EABI configurations.
13200 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
13201 of .text.hot and .text.unlikely sections for kernel modules when
13202 using ARM style exceptions.
13203 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
13204 options. Add EXTRA_CC1_SPEC.
13205 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
13207 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
13209 (ARM_TARGET2_DWARF_FORMAT): Define.
13210 * config/arm/t-vxworks: Adjust multilib control to removal of the
13211 Diab command line options.
13213 2017-08-01 Martin Liska <mliska@suse.cz>
13215 PR gcov-profile/81561
13216 * gcov.c (unblock): Make unblocking safe as we need to preserve
13217 index correspondence of blocks and block_lists.
13219 2017-08-01 Richard Biener <rguenther@suse.de>
13221 PR tree-optimization/81181
13222 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
13223 (compute_antic): ... end of iteration here.
13225 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
13227 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
13228 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
13229 (ftree-slp-vectorize): Likewise.
13230 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
13231 can no longer be set independent of flag_tree_loop_vectorize.
13232 * omp-general.c (emp_max_vf): Likewise.
13233 * opts.c (enable_fdo_optimizations): Remove references to
13234 flag_tree_vectorize, these are now implicit.
13235 (common_handle_option): Remove handling for OPT_ftree_vectorize,
13236 and leave it for the options machinery.
13238 2017-08-01 Martin Liska <mliska@suse.cz>
13240 PR middle-end/70140
13241 * builtins.c (expand_builtin_memcpy_args): Remove.
13242 (expand_builtin_memcpy): Call newly added function
13243 expand_builtin_memory_copy_args.
13244 (expand_builtin_memcpy_with_bounds): Likewise.
13245 (expand_builtin_mempcpy): Remove last argument.
13246 (expand_builtin_mempcpy_with_bounds): Likewise.
13247 (expand_builtin_memory_copy_args): New function created from
13248 expand_builtin_mempcpy_args with small modifications.
13249 (expand_builtin_mempcpy_args): Remove.
13250 (expand_builtin_stpcpy): Remove unused argument.
13251 (expand_builtin): Likewise.
13252 (expand_builtin_with_bounds): Likewise.
13254 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
13257 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
13258 print "ds:" only for immediates in generic address space.
13260 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
13263 * config/i386/i386.c (ix86_funciton_naked): New prototype.
13264 (ix86_function_ok_for_sibcall): Return false for naked functions.
13266 2017-08-01 Richard Biener <rguenther@suse.de>
13268 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
13269 (compute_antic): Seed worklist with exit block predecessors.
13270 * cfganal.c (dfs_find_deadend): For a cycle return the source
13271 of the edge closing it.
13273 2017-08-01 Tamar Christina <tamar.christina@arm.com>
13275 * config/aarch64/aarch64.c
13276 (aarch64_can_const_movi_rtx_p): Move 0 check.
13278 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13280 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
13281 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
13283 * match.pd: Ditto in address comparison pattern.
13285 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13287 PR tree-optimization/81627
13288 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
13289 closed ssa form for store-store chain.
13291 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13293 PR tree-optimization/81620
13294 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
13295 for store-store chain.
13297 2017-08-01 Jakub Jelinek <jakub@redhat.com>
13299 PR tree-optimization/81588
13300 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
13301 ranges[i].in_p, invert comparison code ccode. For >/>=,
13302 swap rhs1 and rhs2 and comparison code unconditionally,
13303 for </<= don't do that. Don't swap rhs1/rhs2 again if
13304 ranges[i].in_p, instead invert comparison code ccode if
13305 opcode or oe->rank is BIT_IOR_EXPR.
13308 * optabs.def (vec_extract_optab, vec_init_optab): Change from
13309 a direct optab to conversion optab.
13310 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
13311 with GET_MODE_INNER as last argument instead of optab_handler.
13312 * expmed.c (extract_bit_field_1): Likewise. Use vector from
13313 vector extraction if possible and optab is available.
13314 * expr.c (store_constructor): Use convert_optab_handler instead
13315 of optab_handler. Use vector initialization from smaller
13316 vectors if possible and optab is available.
13317 * tree-vect-stmts.c (vectorizable_load): Likewise.
13318 * doc/md.texi (vec_extract, vec_init): Document that the optabs
13319 now have two modes.
13320 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
13321 of vec_init from half-sized vectors with the same element mode.
13322 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
13323 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
13324 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
13325 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
13326 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
13327 after mode in gen_vec_extract* calls.
13328 (vec_extract<mode>): Renamed to ...
13329 (vec_extract<mode><ssescalarmodelower>): ... this.
13330 (vec_extract<mode><ssehalfvecmodelower>): New expander.
13331 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
13332 element mode after mode in gen_vec_init* calls.
13333 (VEC_INIT_HALF_MODE): New mode iterator.
13334 (vec_init<mode>): Renamed to ...
13335 (vec_init<mode><ssescalarmodelower>): ... this.
13336 (vec_init<mode><ssehalfvecmodelower>): New expander.
13337 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
13338 (vec_extractv2sfsf): ... this.
13339 (vec_initv2sf): Renamed to ...
13340 (vec_initv2sfsf): ... this.
13341 (vec_extractv2si): Renamed to ...
13342 (vec_extractv2sisi): ... this.
13343 (vec_initv2si): Renamed to ...
13344 (vec_initv2sisi): ... this.
13345 (vec_extractv4hi): Renamed to ...
13346 (vec_extractv4hihi): ... this.
13347 (vec_initv4hi): Renamed to ...
13348 (vec_initv4hihi): ... this.
13349 (vec_extractv8qi): Renamed to ...
13350 (vec_extractv8qiqi): ... this.
13351 (vec_initv8qi): Renamed to ...
13352 (vec_initv8qiqi): ... this.
13353 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
13354 (vec_init<mode>): Renamed to ...
13355 (vec_init<mode><VEC_base_l>): ... this.
13356 (vec_extract<mode>): Renamed to ...
13357 (vec_extract<mode><VEC_base_l>): ... this.
13358 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
13359 (vec_initv2sfsf): ... this.
13360 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
13361 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
13362 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
13363 element mode after mode in gen_vec_init* calls.
13364 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
13365 (vec_init<mode><Vel>): ... this.
13366 (vec_extract<mode>): Renamed to ...
13367 (vec_extract<mode><Vel>): ... this.
13368 * config/aarch64/iterators.md (Vel): New mode attribute.
13369 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
13370 Add element mode after mode in gen_vec_extract* calls.
13371 * config/s390/vector.md (non_vec_l): New mode attribute.
13372 (vec_extract<mode>): Renamed to ...
13373 (vec_extract<mode><non_vec_l>): ... this.
13374 (vec_init<mode>): Renamed to ...
13375 (vec_init<mode><non_vec_l>): ... this.
13376 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
13377 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
13379 * config/arm/iterators.md (V_elem_l): New mode attribute.
13380 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
13381 (vec_extract<mode><V_elem_l>): ... this.
13382 (vec_extractv2di): Renamed to ...
13383 (vec_extractv2didi): ... this.
13384 (vec_init<mode>): Renamed to ...
13385 (vec_init<mode><V_elem_l>): ... this.
13386 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
13387 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
13388 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
13389 Add element mode after gen_vec_extract* calls.
13390 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
13391 (vec_init<mode><unitmode>): ... this.
13392 (vec_extract<mode>): Renamed to ...
13393 (vec_extract<mode><unitmode>): ... this.
13394 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
13395 (vec_init<mode><unitmode>): ... this.
13396 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
13397 (vec_initv2sfsf): ... this.
13398 (vec_extractv2sf): Renamed to ...
13399 (vec_extractv2sfsf): ... this.
13400 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
13401 Add element mode after gen_vec_extract* calls.
13402 * config/mips/mips.md (unitmode): New mode iterator.
13403 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
13404 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
13405 * config/spu/spu.md (inner_l): New mode attribute.
13406 (vec_init<mode>): Renamed to ...
13407 (vec_init<mode><inner_l>): ... this.
13408 (vec_extract<mode>): Renamed to ...
13409 (vec_extract<mode><inner_l>): ... this.
13410 * config/sparc/sparc.md (veltmode): New mode iterator.
13411 (vec_init<VMALL:mode>): Renamed to ...
13412 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
13413 * config/ia64/vect.md (vec_initv2si): Renamed to ...
13414 (vec_initv2sisi): ... this.
13415 (vec_initv2sf): Renamed to ...
13416 (vec_initv2sfsf): ... this.
13417 (vec_extractv2sf): Renamed to ...
13418 (vec_extractv2sfsf): ... this.
13419 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
13420 (vec_init<mode>): Renamed to ...
13421 (vec_init<mode><VEC_base_l>): ... this.
13422 (vec_extract<mode>): Renamed to ...
13423 (vec_extract<mode><VEC_base_l>): ... this.
13424 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
13425 (vec_initv2sfsf): ... this.
13426 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
13427 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
13428 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
13429 gen_vec_init* calls.
13431 2017-08-01 Richard Biener <rguenther@suse.de>
13433 PR tree-optimization/81297
13434 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
13435 TREE_OVERFLOW from INTEGER_CSTs.
13437 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
13439 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
13441 2017-07-31 Carl Love <cel@us.ibm.com>
13443 * config/rs6000/rs6000-c: Add support for built-in functions
13444 vector signed char vec_xl_be (signed long long, signed char *);
13445 vector unsigned char vec_xl_be (signed long long, unsigned char *);
13446 vector signed int vec_xl_be (signed long long, signed int *);
13447 vector unsigned int vec_xl_be (signed long long, unsigned int *);
13448 vector signed long long vec_xl_be (signed long long, signed long long *);
13449 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
13450 vector signed short vec_xl_be (signed long long, signed short *);
13451 vector unsigned short vec_xl_be (signed long long, unsigned short *);
13452 vector double vec_xl_be (signed long long, double *);
13453 vector float vec_xl_be (signed long long, float *);
13454 * config/rs6000/altivec.h (vec_xl_be): Add #define.
13455 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
13456 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
13458 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
13459 (altivec_expand_builtin): Add switch statement to call
13460 altivec_expand_xl_be for each builtin.
13461 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
13462 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
13463 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
13464 __builtin_vsx_le_be_v16qi.
13465 * doc/extend.texi: Update the built-in documentation file for the
13466 new built-in functions.
13468 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
13471 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
13473 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
13475 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13477 * config.gcc: Add z14.
13478 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
13479 CPU model numbers for z13s and z14.
13480 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
13482 * config/s390/s390-opts.h (enum processor_type): Rename
13483 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
13484 * config/s390/s390.c (processor_table): Add field for CPU name to
13485 be passed to Binutils.
13486 (s390_asm_output_machine_for_arch): Use the new field in
13487 processor_table for Binutils.
13488 (s390_expand_builtin): Replace arch12 with z14.
13489 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
13490 (s390_get_sched_attrmask): Likewise.
13491 (s390_get_unit_mask): Likewise.
13492 * config/s390/s390.opt: Add z14 to processor_type enum.
13494 2017-07-31 Martin Jambor <mjambor@suse.cz>
13497 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
13498 regardless of optimization level.
13500 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
13501 Martin Liska <mliska@suse.cz>
13503 * predict.def: Remove old comment and adjust probability.
13504 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
13505 PREDICT statements.
13507 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
13510 * config/i386/i386.c (ix86_function_naked): New function.
13511 (ix86_can_use_return_insn_p): Return false for naked functions.
13512 (ix86_expand_prologue): Skip prologue for naked functions.
13513 (ix86_expand_epilogue): Skip epilogue for naked functions
13514 and emit trap instruction.
13515 (ix86_warn_func_return): New function.
13516 (ix86_attribute_table): Add "naked" attribute specification.
13517 (TARGET_WARN_FUNC_RETURN): Define.
13518 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
13520 2017-07-31 Martin Liska <mliska@suse.cz>
13522 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
13523 (dump_gimple_bb_header): Always dump BB info.
13524 (pp_cfg_jump): Do not append info about BB when dumping a jump.
13526 2017-07-31 Martin Liska <mliska@suse.cz>
13529 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
13530 also with current_function_decl non-null equality.
13532 2017-07-31 Jakub Jelinek <jakub@redhat.com>
13535 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
13536 change type to the element type, instead add eltype variable and
13537 use it where we are interested in the element type.
13539 PR tree-optimization/81603
13540 * ipa-polymorphic-call.c
13541 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
13542 offset arithmetic in offset_int, bail out if the resulting bit offset
13543 doesn't fit into shwi.
13545 2017-07-31 Martin Liska <mliska@suse.cz>
13547 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
13548 (gimplify_save_expr): Fix comment.
13550 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
13553 * config/i386/i386.c (ix86_function_arg): Update arguments for
13555 (ix86_compute_frame_layout): Set the initial stack offset to
13556 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
13557 INCOMING_FRAME_SP_OFFSET.
13558 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
13559 stack before exception handler returns.
13560 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
13561 the 'ERROR_CODE' for exception handler.
13563 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
13565 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
13566 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
13567 (ASM_OUTPUT_REG_POP): Ditto.
13568 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
13569 instead of asm_fprintf to output pure string.
13571 2017-07-29 Jakub Jelinek <jakub@redhat.com>
13573 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
13574 to imported_module_or_decl hook.
13575 (debug_nothing_tree_tree_tree_bool): Remove.
13576 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
13577 * debug.c (do_nothing_debug_hooks): Use
13578 debug_nothing_tree_tree_tree_bool_bool instead of
13579 debug_nothing_tree_tree_tree_bool.
13580 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
13581 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
13582 * sdbout.c (sdb_debug_hooks): Likewise.
13583 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
13584 (gen_namespace_die): Add DW_AT_export_symbols attribute if
13586 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
13587 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
13588 attribute, don't add anything.
13590 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13592 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
13593 (fold_build2_stat_loc): Likewise.
13594 (fold_build3_stat_loc): Likewise.
13595 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
13596 (fold_build1_loc): Remove macro.
13597 (fold_build2_loc): Likewise.
13598 (fold_build3_loc): Likewise.
13600 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13602 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
13603 (gimple_build_debug_bind_source_stat): Likewise.
13604 * gimple.h (gimple_build_debug_bind): Remove macro.
13605 (gimple_build_debug_bind_source): Likewise.
13607 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13609 * bitmap.c (bitmap_alloc): Adjust.
13610 (bitmap_gc_alloc): Likewise.
13611 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
13613 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13615 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
13616 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
13617 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
13618 (bitmap_gc_alloc_stat): Likewise.
13619 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
13621 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13623 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
13624 * rtl.h (shallow_copy_rtx): Remove macro.
13626 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13628 * emit-rtl.c (gen_raw_REG): Adjust.
13629 * gengenrtl.c (gendef): Likewise.
13630 * rtl.c (rtx_alloc_stat): Remove _stat from name.
13631 * rtl.h (rtx_alloc): Remove macro.
13633 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13635 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
13636 (build_tree_list_stat): Likewise.
13637 * tree.h (build_tree_list): Remove macro.
13638 (build_tree_list_vec): Likewise.
13640 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13642 * tree.c (make_vector_stat): Remove _stat from name.
13643 (build_vector_stat): Likewise.
13644 * tree.h (make_vector_stat): Remove macro.
13645 (build_vector_stat): Likewise.
13647 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13649 * tree.h (build_var_debug_value): Remove prototype.
13651 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13653 * tree.c (tree_cons_stat): Remove _stat from name.
13654 * tree.h (tree_cons): Remove macro.
13656 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13658 * tree.c (build_vl_exp_stat): Remove _stat from name.
13659 * tree.h (build_vl_exp): Remove macro.
13661 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13663 * tree.c (build_decl_stat): Remove _stat from name.
13664 * tree.h (build_decl): Remove macro.
13666 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13668 * gimple.c (gimple_build_with_ops_stat): Adjust.
13669 (gimple_alloc_stat): Remove _stat from name.
13670 * gimple.h (gimple_alloc): Remove macro.
13672 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13674 * tree.c (make_tree_vec_stat): Remove _stat from name.
13675 (grow_tree_vec_stat): Likewise.
13676 * tree.h (make_tree_vec_stat): Adjust prototype.
13677 (grow_tree_vec_stat): Likewise.
13678 (make_tree_vec): Remove macro.
13679 (grow_tree_vec): Likewise.
13681 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13683 * fold-const.c (fold_build1_stat_loc): Adjust.
13684 (fold_build2_stat_loc): Likewise.
13685 (fold_build3_stat_loc): Likewise.
13686 * tree.c (build0_stat): Remove _stat from name.
13687 (build1_stat): Likewise.
13688 (build2_stat): Likewise.
13689 (build3_stat): Likewise.
13690 (build4_stat): Likewise.
13691 (build5_stat): Likewise.
13692 * tree.h (build1_loc): Remove macro, and rename _stat function
13694 (build2_loc): Likewise.
13695 (build3_loc): Likewise.
13696 (build4_loc): Likewise.
13697 (build5_loc): Likewise.
13699 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13701 * tree.c (make_int_cst_stat): Remove _stat from name.
13702 * tree.h (make_int_cst_stat): Adjust prototype.
13703 (make_int_cst): Remove macro.
13705 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13707 * tree.c (make_tre_binfo_stat): Remove _stat from name.
13708 * tree.h (make_tree_binfo_stat): Adjust prototype.
13709 (make_tree_binfo): Remove.
13711 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13713 * tree.c (copy_node_stat): Rename to copy_node.
13714 (build_distinct_type_copy): Adjust.
13715 * tree.h (copy_node_stat): Adjust prototype.
13716 (copy_node): Remove macro.
13718 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13720 * tree.c (make_node_stat): rename to make_node.
13721 (build_tree_list_stat): Adjust.
13722 (build0_stat): Likewise.
13723 (build2_stat): Likewise.
13724 (build3_stat): Likewise.
13725 (build4_stat): Likewise.
13726 (build5_stat): Likewise.
13727 (build_decl_stat): Likewise.
13728 * tree.h (make_node_stat): Adjust prototype.
13729 (make_node): remove macro.
13731 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
13733 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
13734 (PPC_FEATURE2_SCV): Likewise.
13735 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
13737 2017-07-28 Tamar Christina <tamar.christina@arm.com>
13739 * config/aarch64/aarch64.c
13740 (aarch64_internal_mov_immediate): Add new special pattern.
13741 * config/aarch64/aarch64.md (*movdi_aarch64):
13742 Add reg/32bit const mov case.
13744 2017-07-28 Tamar Christina <tamar.christina@arm.com>
13745 Richard Sandiford <richard.sandiford@linaro.org>
13747 * config/aarch64/aarch64.md (mov<mode>): Generalize.
13748 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
13749 Add integer and movi cases.
13750 (movi-split-hf-df-sf split, fp16): New.
13751 (enabled): Added TARGET_FP_F16INST.
13752 * config/aarch64/iterators.md (GPF_HF): New.
13753 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
13755 2017-07-28 Tamar Christina <tamar.christina@arm.com>
13757 * config/aarch64/aarch64.c
13758 (aarch64_simd_container_mode): Add prototype.
13759 (aarch64_expand_mov_immediate): Add HI support.
13760 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
13761 (aarch64_can_const_movi_rtx_p): New.
13762 (aarch64_preferred_reload_class):
13763 Remove restrictions of using FP registers for certain SIMD operations.
13764 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
13765 (aarch64_valid_floating_const): Add integer move validation.
13766 (aarch64_simd_imm_scalar_p): Remove.
13767 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
13768 (aarch64_legitimate_constant_p): Expand list of supported cases.
13769 * config/aarch64/aarch64-protos.h
13770 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
13771 (aarch64_reinterpret_float_as_int): New.
13772 (aarch64_simd_imm_scalar_p): Remove.
13773 * config/aarch64/constraints.md (Uvi): New.
13774 (Dd): Split into Ds and new Dd.
13775 * config/aarch64/aarch64.md (*movsi_aarch64):
13777 (*movdi_aarch64): Add SIMD mov case.
13779 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13781 * tree-predcom.c: (struct chain): Handle store-store chain in which
13782 stores for elimination only store loop invariant values.
13783 (execute_pred_commoning_chain): Ditto.
13784 (prepare_initializers_chain_store_elim): Ditto.
13785 (prepare_finalizers): Ditto.
13786 (is_inv_store_elimination_chain): New function.
13787 (initialize_root_vars_store_elim_1): New function.
13789 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13791 * tree-predcom.c: Revise general description of the pass.
13792 (enum chain_type): New enum type for store elimination.
13793 (struct chain): New field supporting store elimination.
13794 (struct component): Ditto.
13795 (dump_chain): Dump store-stores chain.
13796 (release_chain): Release resources.
13797 (split_data_refs_to_components): Compute and create component
13798 contains only stores for elimination.
13799 (get_chain_last_ref_at): New function.
13800 (make_invariant_chain): Initialization.
13801 (make_rooted_chain): Specify chain type in parameter and record it.
13802 (add_looparound_copies): Skip for store-stores chain.
13803 (determine_roots_comp): Compute type of chain and pass it to
13805 (initialize_root_vars_store_elim_2): New function.
13806 (finalize_eliminated_stores): New function.
13807 (remove_stmt): Handle store for elimination.
13808 (execute_pred_commoning_chain): Execute predictive commoning on
13809 store-store chains.
13810 (determine_unroll_factor): Skip unroll for store-stores chain.
13811 (prepare_initializers_chain_store_elim): New function.
13812 (prepare_initializers_chain): Hanlde store-store chain.
13813 (prepare_finalizers_chain, prepare_finalizers): New function.
13814 (tree_predictive_commoning_loop): Return integer value indicating
13815 if loop is unrolled or lcssa form is corrupted.
13816 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
13818 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13820 * tree-predcom.c (initialize_root): Delete.
13821 (execute_pred_commoning_chain): Initialize root vars and replace
13822 reference of non-combined chain directly, rather than call above
13825 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13827 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
13828 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
13830 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13832 * tree-predcom.c (struct chain): New field init_seq.
13833 (release_chain): Release init_seq.
13834 (prepare_initializers_chain): Record intialization stmts in above
13836 (insert_init_seqs): New function.
13837 (tree_predictive_commoning_loop): Call insert_init_seqs.
13839 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13841 * tree-predcom.c (determine_roots_comp): Skip trivial components.
13843 2017-07-28 Richard Biener <rguenther@suse.de>
13845 * match.pd: Remove superfluous :c.
13846 * genmatch.c (simplify::id): Add member.
13847 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
13849 (current_id): New global.
13850 (dt_node::parent): Move from ...
13851 (dt_operand::parent): ... here. Add for_id member.
13852 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
13853 (decision_tree::find_node): Relax order requirement when
13854 merging DT_TRUE nodes to ones inbetween the current simplify
13855 and the one we try to merge with. Add diagnostic whenever
13856 we need to enforce pattern order by not merging.
13857 (decision_tree::insert): Set current_id.
13858 (decision_tree::print_node): Dump parent node and for_id.
13859 (parser::last_id): Add member.
13860 (parser::push_simplify): Assign unique id.
13861 (parser::parser): Initialize last_id.
13863 2017-07-28 Martin Liska <mliska@suse.cz>
13866 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
13867 gimple_build_debug_bind.
13869 2017-07-28 Richard Biener <rguenther@suse.de>
13871 PR tree-optimization/81502
13872 * match.pd: Add pattern combining BIT_INSERT_EXPR with
13874 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
13876 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
13877 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
13878 for BIT_FIELD_REF args.
13879 * fold-const.c (make_bit_field_ref): Likewise.
13880 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
13882 2017-07-28 Jakub Jelinek <jakub@redhat.com>
13885 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
13886 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
13887 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
13888 Or it into SANITIZER_UNDEFINED.
13889 * ubsan.c: Include gimple-fold.h and varasm.h.
13890 (ubsan_expand_ptr_ifn): New function.
13891 (instrument_pointer_overflow): New function.
13892 (maybe_instrument_pointer_overflow): New function.
13893 (instrument_object_size): Formatting fix.
13894 (pass_ubsan::execute): Call instrument_pointer_overflow
13895 and maybe_instrument_pointer_overflow.
13896 * internal-fn.c (expand_UBSAN_PTR): New function.
13897 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
13898 * sanitizer.def (__ubsan_handle_pointer_overflow,
13899 __ubsan_handle_pointer_overflow_abort): New builtins.
13900 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
13901 * internal-fn.def (UBSAN_PTR): New internal function.
13902 * opts.c (sanitizer_opts): Add pointer-overflow.
13903 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
13904 * fold-const.c (build_range_check): Compute pointer range check in
13905 integral type if pointer arithmetics would be needed. Formatting
13908 2017-07-28 Martin Liska <mliska@suse.cz>
13911 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
13912 parameters that are of a variable-length.
13914 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
13916 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
13918 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
13919 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
13920 (CRT_CALL_STATIC_FUNCTION): Likewise.
13921 (ASM_DEFAULT_SPEC): New define.
13922 (ASM_SPEC32): Likewise.
13923 (ASM_SPEC64): Likewise.
13924 (ASM_SPEC_COMMON): Likewise.
13925 (ASM_SPEC): Likewise.
13926 (INVALID_64BIT): Likewise.
13927 (LINK_OS_DEFAULT_SPEC): Likewise.
13928 (LINK_OS_SPEC32): Likewise.
13929 (LINK_OS_SPEC64): Likewise.
13930 (POWERPC_LINUX): Likewise.
13931 (PTRDIFF_TYPE): Likewise.
13932 (RESTORE_FP_PREFIX): Likewise.
13933 (RESTORE_FP_SUFFIX): Likewise.
13934 (SAVE_FP_PREFIX): Likewise.
13935 (SAVE_FP_SUFFIX): Likewise.
13936 (SIZE_TYPE): Likewise.
13937 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
13938 (TARGET_64BIT): Likewise.
13939 (TARGET_64BIT): Likewise.
13940 (TARGET_AIX): Likewise.
13941 (WCHAR_TYPE_SIZE): Likewise.
13942 (WCHAR_TYPE): Undefine.
13943 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
13944 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
13945 (CPP_OS_RTEMS_SPEC): Delete.
13946 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
13947 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
13949 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
13951 2017-07-28 Jakub Jelinek <jakub@redhat.com>
13953 PR tree-optimization/81578
13954 * tree-parloops.c (build_new_reduction): Bail out if
13955 reduction_code isn't one of the standard OpenMP reductions.
13956 Move the details printing after that decision.
13958 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
13960 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
13961 related to reload_in_progress.
13962 (splat_input_operand): Likewise.
13963 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
13965 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
13967 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
13968 (TARGET_INSTANTIATE_DECLS): Likewise.
13969 (legitimate_indexed_address_p): Delete reload_in_progress code.
13970 (rs6000_debug_legitimate_address_p): Likewise.
13971 (rs6000_eliminate_indexed_memrefs): Likewise.
13972 (rs6000_emit_le_vsx_store): Likewise.
13973 (rs6000_emit_move_si_sf_subreg): Likewise.
13974 (rs6000_emit_move): Likewise.
13975 (register_to_reg_type): Likewise.
13976 (rs6000_pre_atomic_barrier): Likewise.
13977 (rs6000_machopic_legitimize_pic_address): Likewise.
13978 (rs6000_allocate_stack_temp): Likewise.
13979 (rs6000_address_for_fpconvert): Likewise.
13980 (rs6000_address_for_altivec): Likewise.
13981 (rs6000_secondary_memory_needed_rtx): Delete function.
13982 (rs6000_check_sdmode): Likewise.
13983 (rs6000_alloc_sdmode_stack_slot): Likewise.
13984 (rs6000_instantiate_decls): Likewise.
13985 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
13986 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
13987 Delete reload_in_progress.
13988 (*vec_reload_and_plus_<mptrsize>): Likewise.
13989 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
13990 (vsx_div_v2di): Likewise.
13991 (vsx_udiv_v2di): Likewise.
13993 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
13995 * config/rs6000/rs6000.opt (mlra): Replace with stub.
13996 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
13997 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
13998 (rs6000_debug_reg_global): Delete print of LRA status.
13999 (rs6000_option_override_internal): Delete dead LRA related code.
14000 (rs6000_lra_p): Delete function.
14001 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
14003 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
14005 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
14006 * config/riscv/rtems.h: New file.
14008 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14009 Sudakshina Das <sudi.das@arm.com>
14011 * config/aarch64/aarch64.md
14012 (define_split for and<mode>3nr_compare): Move
14013 non aarch64_logical_operand to a register.
14014 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
14015 register immediate operand to a register.
14016 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
14018 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14020 PR middle-end/81564
14021 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
14023 2017-07-27 Richard Biener <rguenther@suse.de>
14025 PR tree-optimization/81573
14026 PR tree-optimization/81494
14027 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
14028 multi defuse cycle case.
14030 2017-07-27 Richard Biener <rguenther@suse.de>
14032 PR tree-optimization/81571
14033 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
14036 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
14038 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
14039 earlier and only if MASK_FPU is set. Adjust formatting.
14041 2017-07-27 Martin Liska <mliska@suse.cz>
14043 * opt-functions.awk: Add validation of value of Init.
14044 * optc-gen.awk: Pass new argument.
14046 2017-07-27 Martin Liska <mliska@suse.cz>
14048 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
14049 Fix wrong condition.
14051 2017-07-27 Martin Liska <mliska@suse.cz>
14053 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
14054 BBs and edges seen by autoFDO.
14056 2017-07-27 Richard Biener <rguenther@suse.de>
14058 PR tree-optimization/81502
14059 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
14060 with incompatible but same sized type.
14061 (execute_update_addresses_taken): Likewise.
14063 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
14065 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
14066 flag_tree_loop_vectorize rather than flag_tree_vectorize.
14068 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14071 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
14072 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
14073 Change s_operand to memory_operand.
14075 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
14077 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
14078 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
14079 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
14080 Emit instructions rather than returning an expression. Handle TFmode
14081 and KFmode by casting to TImode.
14082 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
14083 (rs6000_emit_le_vsx_store): Likewise.
14084 * config/rs6000/vsx.md (VSX_TI): New iterator.
14085 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
14086 (*vsx_le_undo_permute_<mode>): Likewise.
14087 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
14088 emit the split sequence.
14089 (*vsx_le_perm_store_<mode>): Likewise.
14091 2017-07-27 Jakub Jelinek <jakub@redhat.com>
14093 PR tree-optimization/81555
14094 PR tree-optimization/81556
14095 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
14096 if true, force CHANGED for the recursive invocation.
14097 (reassociate_bb): Remember original length of ops array, pass
14098 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
14100 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
14101 attributes for noipa attribute. For naked attribute use
14102 lookup_attribute first before lookup_attribute_spec.
14103 * final.c (rest_of_handle_final): Disable IPA RA for functions with
14105 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
14106 for functions with noipa attribute.
14107 (cgraph_externally_visible_p): Return true for functions with noipa
14109 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
14110 for functions with noipa attribute.
14111 * doc/extend.texi: Document noipa function attribute.
14112 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
14113 also for functions with noipa attribute.
14114 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
14116 2017-07-26 Andrew Pinski <apinski@cavium.com>
14118 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
14119 vec_unalign_load_cost and vec_unalign_store_cost.
14121 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14123 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
14124 -mvsx-small-integer option.
14125 (ISA_3_0_MASKS_IEEE): Likewise.
14126 (OTHER_VSX_VECTOR_MASKS): Likewise.
14127 (POWERPC_MASKS): Likewise.
14128 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
14129 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
14130 code, only testing for DImode being allowed in non-VSX floating
14132 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
14133 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
14135 (rs6000_option_override_internal): Delete -mvsx-small-integer.
14136 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
14137 TARGET_P8_VECTOR test.
14138 (rs6000_secondary_reload_simple_move): Likewise.
14139 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
14140 since TARGET_P9_VECTOR was already tested.
14141 (rs6000_opt_masks): Remove -mvsx-small-integer.
14142 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
14143 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14145 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
14146 test for TARGET_VEXTRACTUB was used, and that uses
14148 (p9 extract splitter): Likewise.
14149 (vsx_extract_<mode>_di_p9): Likewise.
14150 (vsx_extract_<mode>_store_p9): Likewise.
14151 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
14152 for TARGET_P9_VECTOR was used. Delete code that is now dead with
14153 the elimination of TARGET_VSX_SMALL_INTEGER.
14154 (vsx_extract_<mode>_p8): Likewise.
14155 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
14156 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
14157 (vsx_set_<mode>_p9): Likewise.
14158 (vsx_set_v4sf_p9): Likewise.
14159 (vsx_set_v4sf_p9_zero): Likewise.
14160 (vsx_insert_extract_v4sf_p9): Likewise.
14161 (vsx_insert_extract_v4sf_p9_2): Likewise.
14162 * config/rs6000/rs6000.md (sign extend splitter): Change
14163 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
14164 (floatsi<mode>2_lfiwax_mem): Likewise.
14165 (floatunssi<mode>2_lfiwzx_mem): Likewise.
14166 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
14167 since a test for TARGET_P9_VECTOR was used.
14168 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14169 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
14170 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14171 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
14172 TARGET_P8_VECTOR test.
14173 (fix_trunc<mode>si2_stfiwx): Likewise.
14174 (fix_trunc<mode>si2_internal): Likewise.
14175 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
14176 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14178 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
14179 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
14180 TARGET_P8_VECTOR test.
14181 (fixuns_trunc<mode>si2_stfiwx): Likewise.
14182 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
14183 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14185 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
14186 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
14187 since a test for TARGET_P9_VECTOR was used.
14188 (splitter for loading small constants): Likewise.
14190 2017-07-26 Andrew Pinski <apinski@cavium.com>
14192 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
14195 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
14198 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
14199 (fp_valid_at): Likewise.
14201 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
14203 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
14204 (qdf24xx_addrcost_table): Likewise.
14205 (cortexa57_tunings): Update to use generic_branch_cost.
14206 (cortexa72_tunings): Likewise.
14207 (cortexa73_tunings): Likewise.
14208 (qdf24xx_tunings): Likewise.
14210 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
14212 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
14213 (thunderx2t99_branch_cost): Likewise.
14214 (cortexa35_tunings): Update to use generic_branch_cost.
14215 (cortexa53_tunings): Likewise.
14216 (cortexa57_tunings): Likewise.
14217 (cortexa72_tunings): Likewise.
14218 (cortexa73_tunings): Likewise.
14219 (thunderx2t99_tunings): Likewise.
14221 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
14223 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
14224 (sparc_option_override): Honour MASK_FSMULD.
14225 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
14226 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
14227 * config/sparc/sparc.opt (mfsmuld): New option.
14228 * doc/invoke.texi (mfsmuld): Document option.
14230 2017-07-26 Marek Polacek <polacek@redhat.com>
14232 PR middle-end/70992
14233 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
14235 2017-07-26 Richard Biener <rguenther@suse.de>
14237 * gimple-match-head.c (do_valueize): Return OP if valueize
14239 (get_def): New helper to get at the def stmt of a SSA name
14240 if valueize allows.
14241 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
14242 do_valueize to get at the def stmt.
14243 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
14245 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
14247 PR middle-end/46932
14248 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
14250 2017-07-26 Martin Liska <mliska@suse.cz>
14253 * function.c (expand_function_start): Make expansion of
14254 nonlocal_goto_save_area after parm_birth_insn.
14256 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
14258 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
14259 from all CPU target flags enable members.
14261 2017-07-26 Richard Biener <rguenther@suse.de>
14263 * genmatch.c (dt_simplify::gen): Make iterator vars const.
14264 (decision_tree::gen): Make 'type' const.
14265 (write_predicate): Likewise.
14267 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
14269 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
14270 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
14271 (rs6000_option_override_internal): Likewise.
14272 (rs6000_expand_vector_set): Likewise.
14273 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
14274 (TARGET_UPPER_REGS_SF): Likewise.
14275 (TARGET_UPPER_REGS_DI): Likewise.
14276 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
14277 (TARGET_DIRECT_MOVE_64BIT): Likewise.
14278 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
14279 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14280 (Splitters for DI constants in Altivec registers): Likewise.
14281 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
14282 (vsx_set_v4sf_p9): Likewise.
14283 (vsx_set_v4sf_p9_zero): Likewise.
14284 (vsx_insert_extract_v4sf_p9): Likewise.
14285 (vsx_insert_extract_v4sf_p9_2): Likewise.
14287 2017-07-25 Carl Love <cel@us.ibm.com>
14289 * doc/extend.texi: Update the built-in documentation file for the
14290 existing built-in functions
14291 vector signed char vec_cnttz (vector signed char);
14292 vector unsigned char vec_cnttz (vector unsigned char);
14293 vector signed short vec_cnttz (vector signed short);
14294 vector unsigned short vec_cnttz (vector unsigned short);
14295 vector signed int vec_cnttz (vector signed int);
14296 vector unsigned int vec_cnttz (vector unsigned int);
14297 vector signed long long vec_cnttz (vector signed long long);
14298 vector unsigned long long vec_cnttz (vector unsigned long long);
14300 2017-07-25 Andrew Pinski <apinski@cavium.com>
14302 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
14303 accesses where the use is for the first operand of a BIT_INSERT.
14305 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
14308 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
14309 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
14311 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
14313 * config/i386/gstabs.h: Delete.
14314 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
14316 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
14318 * config/i386/i386.c (ix86_decompose_address): Do not check for
14319 register RTX when looking at index_reg or base_reg.
14320 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
14322 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
14324 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
14325 to update EH info here.
14327 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
14329 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
14331 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
14333 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
14335 2017-07-25 Torsten Duwe <duwe@suse.de>
14337 * common.opt: Introduce -fpatchable-function-entry
14338 command line option, and its variables function_entry_patch_area_size
14339 and function_entry_patch_area_start.
14340 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
14341 including a two-value parser.
14342 * target.def (print_patchable_function_entry): New target hook.
14343 * targhooks.h (default_print_patchable_function_entry): New function.
14344 * targhooks.c (default_print_patchable_function_entry): Likewise.
14345 * toplev.c (process_options): Switch off IPA-RA if
14346 patchable function entries are being generated.
14347 * varasm.c (assemble_start_function): Look at the
14348 patchable-function-entry command line switch and current
14349 function attributes and maybe generate NOP instructions by
14350 calling the print_patchable_function_entry hook.
14351 * doc/extend.texi: Document patchable_function_entry attribute.
14352 * doc/invoke.texi: Document -fpatchable_function_entry
14353 command line option.
14354 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
14356 * doc/tm.texi: Re-generate.
14358 2017-07-25 Jakub Jelinek <jakub@redhat.com>
14361 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
14362 TARGET_AVX512DQ rather than TARGET_AVX512BW.
14364 2017-07-25 Tamar Christina <tamar.christina@arm.com>
14366 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
14368 2017-07-25 Richard Biener <rguenther@suse.de>
14370 PR tree-optimization/81455
14371 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
14372 not walk in cycles when looking for guards.
14374 2017-07-25 Richard Biener <rguenther@suse.de>
14376 PR tree-optimization/81529
14377 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
14378 when optimizing backedge uses.
14380 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
14382 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
14384 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
14385 to dl_section_ref. On AIX, append an expression to subtract
14386 the size of the section length to dl_section_ref.
14388 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
14390 * configure.ac: If any of the config.* scripts fail, exit 1.
14391 * configure: Regenerate.
14393 2017-07-25 Richard Biener <rguenther@suse.de>
14395 PR middle-end/81546
14396 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
14397 of immediate uses, be more verbose on errors.
14399 2017-07-25 Richard Biener <rguenther@suse.de>
14401 PR tree-optimization/81510
14402 * tree-vect-loop.c (vect_is_simple_reduction): When the
14403 reduction stmt is not inside the loop bail out.
14405 2017-07-25 Richard Biener <rguenther@suse.de>
14407 PR tree-optimization/81303
14408 * tree-vect-loop-manip.c (vect_loop_versioning): Build
14409 profitability check against LOOP_VINFO_NITERSM1.
14411 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
14413 * domwalk.c (cmp_bb_postorder): Simplify.
14414 (sort_bbs_postorder): New function. Use it...
14415 (dom_walker::walk): ...here to optimize common cases.
14417 2017-07-25 Martin Liska <mliska@suse.cz>
14420 * ipa-visibility.c (function_and_variable_visibility): Make the
14421 redirection just on target that supports aliasing.
14422 Fix GNU coding style.
14424 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
14427 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
14429 * config/arm/rtems.h: Likewise.
14430 * config/bfin/rtems.h: Likewise.
14431 * config/i386/rtemself.h: Likewise.
14432 * config/lm32/rtems.h: Likewise.
14433 * config/m32c/rtems.h: Likewise.
14434 * config/m68k/rtemself.h: Likewise.
14435 * config/microblaze/rtems.h: Likewise.
14436 * config/mips/rtems.h: Likewise.
14437 * config/moxie/rtems.h: Likewise.
14438 * config/nios2/rtems.h: Likewise.
14439 * config/powerpcspe/rtems.h: Likewise.
14440 * config/rs6000/rtems.h: Likewise.
14441 * config/rtems.h: Likewise.
14442 * config/sh/rtems.h: Likewise.
14443 * config/sh/rtemself.h: Likewise.
14444 * config/sparc/rtemself.h: Likewise.
14446 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
14449 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
14450 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
14451 * tree-ssa-structalias.c (alias_get_name): Same.
14453 2017-07-25 Bin Cheng <bin.cheng@arm.com>
14456 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
14457 instructions if no du chain is found.
14459 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
14461 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
14463 2017-07-25 Richard Biener <rguenther@suse.de>
14465 PR middle-end/81505
14466 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
14469 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
14471 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
14472 upper-regs options.
14473 (ISA_2_7_MASKS_SERVER): Likewise.
14474 (ISA_3_0_MASKS_IEEE): Likewise.
14475 (OTHER_P8_VECTOR_MASKS): Likewise.
14476 (OTHER_VSX_VECTOR_MASKS): Likewise.
14477 (POWERPC_MASKS): Likewise.
14478 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
14479 duplicate list of options.
14480 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
14481 explicit -mupper-regs options.
14482 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
14483 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
14484 alias for -mupper-regs-df.
14485 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
14486 (rs6000_init_hard_regno_mode_ok): Likewise.
14487 (rs6000_option_override_internal): Likewise.
14488 (rs6000_opt_masks): Likewise.
14489 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
14490 options in terms of whether -mvsx or -mpower8-vector was used.
14491 (TARGET_UPPER_REGS_DI): Likewise.
14492 (TARGET_UPPER_REGS_SF): Likewise.
14493 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
14494 -mupper-regs-* options.
14496 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
14498 * passes.c (emergency_dump_function): Print some empty lines and a
14499 header before the RTL dump.
14501 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
14503 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
14505 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
14508 * config/aarch64/aarch64.c (aarch64_classify_symbol):
14509 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
14511 2017-07-24 Carl Love <cel@us.ibm.com>
14513 * config/rs6000/rs6000-c.c: Add support for built-in functions
14514 vector float vec_extract_fp32_from_shorth (vector unsigned short);
14515 vector float vec_extract_fp32_from_shortl (vector unsigned short);
14516 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
14517 vec_extract_fp_from_shortl): Add defines for the two builtins.
14518 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
14519 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
14521 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
14522 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
14523 * doc/extend.texi: Update the built-in documentation file for the
14524 new built-in function.
14526 2017-07-24 Jakub Jelinek <jakub@redhat.com>
14529 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
14531 * doc/generic.texi: Likewise.
14532 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
14533 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
14535 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
14537 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
14538 (aarch64_mls_elt_merge<mode>): Likewise.
14540 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
14542 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
14543 having __cxa_atexit.
14545 2017-07-23 Michael Collison <michael.collison@arm.com>
14547 * config/arm/arm.c (arm_option_override): Deprecate
14548 use of -mstructure-size-boundary.
14549 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
14550 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
14552 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14555 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
14556 Reduce cost estimate for direct moves.
14558 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
14561 * config/i386/i386.c (ix86_option_override_internal): Disable
14562 BMI, BMI2 and TBM instructions for -m16.
14564 2017-07-21 Carl Love <cel@us.ibm.com>
14566 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14567 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14568 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14569 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14570 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14571 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
14572 VMULOSW): New enum "unspec" values.
14573 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
14574 altivec_vmulosw): New patterns.
14575 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14576 VMULOSW): Add definitions.
14578 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
14580 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
14581 (qdf24xx): Likewise.
14582 * config/aarch64/aarch64-options-extensions.def (rdma); New.
14583 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
14584 (AARCH64_FL_V8_1): Renumber.
14585 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
14586 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
14587 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
14588 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
14589 rdma to feature modifiers list.
14591 2017-07-21 Yury Gribov <tetra2005@gmail.com>
14593 PR middle-end/56727
14594 * ipa-visibility (function_and_variable_visibility): Convert
14595 recursive PLT call to direct call if appropriate.
14597 2017-07-21 Andrew Pinski <apinski@cavium.com>
14599 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
14600 operand 1 to see if the types precision matches.
14601 * fold-const.c (operand_equal_p): Likewise.
14603 2017-07-21 Richard Biener <rguenther@suse.de>
14605 PR tree-optimization/81303
14606 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
14607 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
14608 (vect_peeling_hash_get_lowest_cost): Adjust.
14609 (vect_enhance_data_refs_alignment): Likewise. Use
14610 vect_get_peeling_costs_all_drs to compute the penalty for no
14611 peeling to match up costs.
14613 2017-07-21 Richard Biener <rguenther@suse.de>
14615 PR tree-optimization/81500
14616 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
14617 we didn't identify a reduction path.
14619 2017-07-21 Tom de Vries <tom@codesourcery.com>
14620 Cesar Philippidis <cesar@codesourcery.com>
14622 PR gcov-profile/81442
14623 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
14626 2017-07-21 Tom de Vries <tom@codesourcery.com>
14629 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
14631 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
14632 nvptx_override_options_after_change.
14634 2017-07-21 Ulrich Drepper <drepper@redhat.com>
14636 * dwarf2out.c (output_file_names): Avoid double testing for
14637 dwarf_version >= 5.
14639 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
14641 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
14643 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
14645 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
14647 (try_crossjump_to_edge): Do not punt on partitioned functions.
14649 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
14651 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
14652 Put all BBs reachable only via paths crossing cold region to cold
14654 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
14656 2016-07-21 Richard Biener <rguenther@suse.de>
14658 PR tree-optimization/81303
14659 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
14660 into account prologue and epilogue iterations when raising
14661 min_profitable_iters to sth at least covering one vector iteration.
14663 2017-07-21 Tamar Christina <tamar.christina@arm.com>
14665 * config/arm/arm.c (arm_test_cpu_arch_dat):
14668 2017-07-20 Nathan Sidwell <nathan@acm.org>
14670 Remove TYPE_METHODS.
14671 * tree.h (TYPE_METHODS): Delete.
14672 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
14673 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
14674 (dbxout_type_methods): Scan TYPE_FIELDS.
14675 (dbxout_type): Don't check TYPE_METHODS here.
14676 * function.c (use_register_for_decl): Always ignore register for
14677 class types when not optimizing.
14678 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
14679 * tree.c (free_lang_data_in_type): Stitch out member functions and
14680 templates from TYPE_FIELDS.
14681 (build_distinct_type_copy, verify_type_variant,
14682 verify_type): Member fns are on TYPE_FIELDS.
14683 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
14684 * tree-pretty-print.c (dump_generic_node): Likewise.
14686 2017-07-20 Jakub Jelinek <jakub@redhat.com>
14689 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
14690 V2TImode and V4TImode.
14691 (ix86_expand_vector_extract): Likewise.
14692 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
14693 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
14694 (ssescalarmode): Handle V4TImode and V2TImode.
14695 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
14696 (*vec_extractv2ti, *vec_extractv4ti): New insns.
14697 (VEXTRACTI128_MODE): New mode iterator.
14698 (splitter for *vec_extractv?ti first element): New.
14699 (VEC_INIT_MODE): New mode iterator.
14700 (vec_init<mode>): Consolidate 3 expanders into one using
14701 VEC_INIT_MODE mode iterator.
14703 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
14705 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
14706 non_spilled_static_chain_regno_p.
14708 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
14710 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
14712 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
14714 * bb-reorder.c (connect_traces): Allow copying of blocks within
14717 2017-07-20 Richard Biener <rguenther@suse.de>
14719 * gimple.h (gimple_phi_result): Add gphi * overload.
14720 (gimple_phi_result_ptr): Likewise.
14721 (gimple_phi_arg): Likewise. Adjust index assert to only
14722 allow actual argument accesses rather than all slots available
14724 (gimple_phi_arg_def): Add gphi * overload.
14725 * tree-phinodes.c (make_phi_node): Initialize only actual
14727 (resize_phi_node): Clear memory not covered by old node,
14728 do not initialize excess argument slots.
14729 (reserve_phi_args_for_new_edge): Initialize new argument slot
14732 2017-07-20 Bin Cheng <bin.cheng@arm.com>
14734 PR tree-optimization/81388
14736 2016-07-21 Bin Cheng <bin.cheng@arm.com>
14738 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
14739 by removing computation of may_be_zero.
14741 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
14742 Tom de Vries <tom@codesourcery.com>
14744 PR middle-end/81030
14745 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
14746 when gimple level profile disagrees with what RTL expander did.
14748 2017-07-20 Richard Biener <rguenther@suse.de>
14750 PR tree-optimization/61171
14751 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
14752 (vect_analyze_stmt): Add slp instance parameter.
14753 (vectorizable_reduction): Likewise.
14754 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
14755 (vect_is_simple_reduction): Deal with chains not detected
14756 as SLP reduction chain, specifically not properly associated
14757 chains containing a mix of plus/minus.
14758 (get_reduction_op): Remove.
14759 (get_initial_defs_for_reduction): Simplify, pass in whether
14760 this is a reduction chain, pass in the SLP node for the PHIs.
14761 (vect_create_epilog_for_reduction): Get the SLP instance as
14763 (vectorizable_reduction): Get the SLP instance as arg.
14764 During analysis remember the SLP node with the PHIs in the
14765 instance. Simplify getting at the vectorized reduction PHIs.
14766 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
14767 through SLP instance.
14768 (vect_slp_analyze_operations): Likewise.
14769 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
14770 (vect_transform_stmt): Likewise.
14772 2017-07-20 Tom de Vries <tom@codesourcery.com>
14774 PR tree-optimization/81489
14775 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
14776 read of phi arg location to before loop that modifies phi.
14778 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14780 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
14783 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
14785 PR middle-end/81331
14786 * except.c (execute): Fix ordering issue.
14788 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
14790 PR rtl-optimization/81423
14791 * combine.c (make_compound_operation_int): Don't try to optimize
14792 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
14794 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
14796 PR rtl-optimization/81423
14797 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
14798 with a constant that is -1 in the truncated to mode.
14800 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
14802 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
14803 (determine_unlikely_bbs): ... here.
14804 * predict.h (propagate_unlikely_bbs_forward): Declare.
14805 * cfgexpand.c (pass_expand::execute): Use it.
14806 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
14808 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
14809 propagate_unlikely_bbs_forward.
14811 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
14813 PR middle-end/81331
14814 * except.c (maybe_add_nop_after_section_switch): New function.
14817 2017-07-19 Tom de Vries <tom@codesourcery.com>
14819 * gimple.h (gimple_phi_set_arg): Make assert more strict.
14821 2017-07-19 Tom de Vries <tom@codesourcery.com>
14823 * gimple.h (gimple_phi_arg): Make assert more strict.
14825 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
14827 * config.gcc (powerpc*-*-*): Add mmintrin.h.
14828 * config/rs6000/mmintrin.h: New file.
14829 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
14831 2017-07-19 Jakub Jelinek <jakub@redhat.com>
14833 PR tree-optimization/81346
14834 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
14836 2017-07-19 Tom de Vries <tom@codesourcery.com>
14838 * config/nvptx/nvptx.md (VECIM): Add V2DI.
14840 2017-07-19 Tom de Vries <tom@codesourcery.com>
14842 * config/nvptx/nvptx-modes.def: Add V2DImode.
14843 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
14844 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
14845 (nvptx_output_mov_insn): Handle lack of mov.b128.
14846 (nvptx_print_operand): Handle 'H' and 'L' codes.
14847 (nvptx_vector_mode_supported): Allow V2DImode.
14848 (nvptx_preferred_simd_mode): New function.
14849 (nvptx_data_alignment): New function.
14850 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
14851 nvptx_preferred_simd_mode.
14852 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
14854 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
14856 2017-07-19 Tom de Vries <tom@codesourcery.com>
14858 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
14859 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
14860 (nvptx_vector_mode_supported): New function. Allow V2SImode.
14861 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
14862 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
14863 (mov<VECIM>_insn): New define_insn.
14864 (define_expand "mov<VECIM>): New define_expand.
14866 2017-07-19 Tom de Vries <tom@codesourcery.com>
14868 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
14870 2017-07-19 Jakub Jelinek <jakub@redhat.com>
14872 PR tree-optimization/81346
14873 * fold-const.h (fold_div_compare, range_check_type): Declare.
14874 * fold-const.c (range_check_type): New function.
14875 (build_range_check): Use range_check_type.
14876 (fold_div_compare): No longer static, rewritten into
14877 a match.pd helper function.
14878 (fold_comparison): Don't call fold_div_compare here.
14879 * match.pd (X / C1 op C2): New optimization using fold_div_compare
14880 as helper function.
14882 2017-07-19 Nathan Sidwell <nathan@acm.org>
14884 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
14885 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
14886 * tree.c (find_decls_types_r, verify_type): Use
14887 TYPE_{MIN,MAX}_VALUE_RAW.
14888 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
14889 (hash_tree): Likewise.
14890 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
14892 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
14895 2017-07-18 Tom de Vries <tom@codesourcery.com>
14897 PR middle-end/81464
14898 * omp-expand.c (expand_omp_for_static_chunk): Handle
14899 equal-argument loop exit phi.
14901 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
14904 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
14905 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
14906 operand 2 predicate.
14907 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
14908 operand 2 predicate.
14909 (ror,rol -> rorx splitters): Use const_int_operand as
14910 operand 2 predicate.
14912 2017-06-18 Richard Biener <rguenther@suse.de>
14914 PR tree-optimization/81410
14915 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
14916 the gap in the ! slp_perm SLP case after each group.
14918 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
14920 PR middle-end/81463
14921 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
14924 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
14926 PR middle-end/81462
14927 * predict.c (set_even_probabilities): Cleanup; do not affect
14928 probabilities that are already known.
14929 (combine_predictions_for_bb): Call even when count is set.
14931 2017-07-18 Nathan Sidwell <nathan@acm.org>
14933 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
14936 2017-07-18 Bin Cheng <bin.cheng@arm.com>
14939 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
14940 optimization for loop niter analysis.
14942 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
14945 * config/avr/avr.c (avr_optimize_casesi): Don't use
14946 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
14948 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
14950 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
14951 body_cost_vec from _vect_peel_extended_info.
14952 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
14953 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
14956 2017-07-18 Bin Cheng <bin.cheng@arm.com>
14958 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
14960 2017-07-18 Richard Biener <rguenther@suse.de>
14962 PR tree-optimization/80620
14963 PR tree-optimization/81403
14964 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
14965 info when re-using a VN table entry.
14967 2017-07-18 Richard Biener <rguenther@suse.de>
14969 PR tree-optimization/81418
14970 * tree-vect-loop.c (vectorizable_reduction): Properly compute
14971 vectype_in. Verify that with lane-reducing reduction operations
14972 we have a single def-use cycle.
14974 2017-07-17 Carl Love <cel@us.ibm.com>
14976 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
14978 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14979 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14980 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14981 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14982 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14983 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
14984 VMULOSW): New enum "unspec" values.
14985 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
14986 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
14987 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
14988 altivec_vmulosw): New patterns.
14989 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14990 VMULOSW): Add definitions.
14991 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
14993 * config/alpha/alpha.c: Include predict.h.
14995 2017-07-17 Yury Gribov <tetra2005@gmail.com>
14997 * tree-vrp.c (compare_assert_loc): Fix comparison function
14998 to return predictable results.
15000 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15002 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
15004 (subdi3): Likewise.
15005 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
15006 * doc/invoke.texi (mexpand-adddi): Update text.
15008 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15010 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
15011 that also clobbers the CC register. The old expand code is moved
15013 (*arc_clzsi2): ... here.
15014 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
15015 the CC register. The old expand code is moved to ...
15016 (arc_ctzsi2): ... here.
15018 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15020 * config/arc/arc.opt (mindexed-loads): Use initial value
15021 TARGET_INDEXED_LOADS_DEFAULT.
15022 (mauto-modify-reg): Use initial value
15023 TARGET_AUTO_MODIFY_REG_DEFAULT.
15024 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15025 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15026 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15027 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15029 2017-07-17 Martin Liska <mliska@suse.cz>
15032 * opts.c (finish_options): Do not allow -fgnu-tm
15033 w/ -fsanitize={kernel-,}address. Say sorry.
15035 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15038 * tree-loop-distribution.c (classify_partition): Only assert on
15039 numer of iterations.
15040 (merge_dep_scc_partitions): Delete prameter. Update function call.
15041 (distribute_loop): Remove code handling loop with unknown niters.
15042 (pass_loop_distribution::execute): Skip loop with unknown niters.
15044 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15047 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
15048 function sort_partitions_by_post_order.
15050 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15052 PR tree-optimization/81374
15053 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
15054 the max index of basic blocks, rather than number of basic blocks.
15056 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15058 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
15060 (arc_legitimate_pic_operand_p): Likewise.
15061 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
15063 (arc_needs_pcl_p): Likewise.
15064 (arc_legitimate_pc_offset_p): Likewise.
15065 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
15066 function is also used in constrains.md.
15067 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
15068 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
15069 PLUS. Only return true/false in known cases, otherwise assert.
15070 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
15071 is already called in arc_legitimate_constant_p.
15072 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
15074 (LEGITIMATE_PIC_OPERAND_P): Use
15075 arc_raw_symbolic_reference_mentioned_p function.
15076 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
15081 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15082 Andrew Burgess <andrew.burgess@embecosm.com>
15084 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
15085 (arc_return_address_register): New function.
15086 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
15087 (arc_handle_fndecl_attribute): Add naked attribute.
15088 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15089 (TARGET_WARN_FUNC_RETURN): Likewise.
15090 (arc_allocate_stack_slots_for_args): New function.
15091 (arc_warn_func_return): Likewise.
15092 (machine_function): Change type fn_type.
15093 (arc_compute_function_type): Consider new naked function type,
15094 change function return type.
15095 (arc_must_save_register): Adapt to handle new
15096 arc_compute_function_type's return type.
15097 (arc_expand_prologue): Likewise.
15098 (arc_expand_epilogue): Likewise.
15099 (arc_return_address_regs): Delete.
15100 (arc_return_address_register): New function.
15101 (arc_epilogue_uses): Use above function.
15102 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
15103 (arc_function_type): Change encoding, add naked type.
15104 (ARC_INTERRUPT_P): Change to handle the new encoding.
15105 (ARC_FAST_INTERRUPT_P): Likewise.
15106 (ARC_NORMAL_P): Define.
15107 (ARC_NAKED_P): Likewise.
15108 (arc_compute_function_type): Delete prototype.
15109 * config/arc/arc.md (in_ret_delay_slot): Use
15110 arc_return_address_register function.
15111 (simple_return): Likewise.
15112 (p_return_i): Likewise.
15114 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15116 PR tree-optimization/81428
15117 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
15118 can't be built for those types.
15120 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
15122 Remove stuff dead since r239246.
15124 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
15125 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
15126 (avr_inform_devices): Remove dead stuff.
15128 2017-07-17 Tamar Christina <tamar.christina@arm.com>
15130 * config/arm/arm_neon.h: Fix softp typo.
15132 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15134 PR tree-optimization/81365
15135 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
15136 aggregate moves onto bb predecessor edges, make sure there are no
15137 loads that could alias the lhs in between the start of bb and the
15140 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
15143 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
15144 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
15145 [LSHIFTRT, outer_code = TRUNCATE]: Same.
15147 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15149 PR tree-optimization/81396
15150 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
15151 (init_symbolic_number): Initialize it to 1.
15152 (perform_symbolic_merge): Add n_ops from both operands into the new
15154 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
15155 without base_addr as useless if they need more than one operation.
15156 (bswap_replace): Handle !bswap case for NULL base_addr.
15158 2017-07-17 Tom de Vries <tom@codesourcery.com>
15161 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
15164 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
15166 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
15167 conditional builtin define __FIX_LEON3FT_B2BST.
15169 2017-07-17 Daniel Cederman <cederman@gaisler.com>
15171 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
15172 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
15175 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
15177 PR rtl-optimization/81424
15178 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
15179 to remove potential trapping from operands if -fnon-call-exceptions.
15181 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15183 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
15184 profile_proability for scalling.
15185 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
15187 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15189 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
15191 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15193 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
15194 fixpoint arithmetics.
15196 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15198 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
15199 fixpoint arithmetics.
15201 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15203 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
15204 fixpoint arithmetics.
15206 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15208 * profile-count.h (profile_probability::from_reg_br_prob_note,
15209 profile_probability::to_reg_br_prob_note): New functions.
15210 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
15211 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
15212 * predict.c (probability_reliable_p): Update.
15213 (edge_probability_reliable_p): Update.
15214 (br_prob_note_reliable_p): Update.
15215 (invert_br_probabilities): Update.
15216 (add_reg_br_prob_note): New function.
15217 (combine_predictions_for_insn): Update.
15218 * asan.c (asan_clear_shadow): Update.
15219 * cfgbuild.c (compute_outgoing_frequencies): Update.
15220 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
15221 (update_br_prob_note): Update.
15222 (rtl_verify_edges): Update.
15223 (purge_dead_edges): Update.
15224 (fixup_reorder_chain): Update.
15225 * emit-rtl.c (try_split): Update.
15226 * ifcvt.c (cond_exec_process_insns): Update.
15227 (cond_exec_process_if_block): Update.
15228 (dead_or_predicable): Update.
15229 * internal-fn.c (expand_addsub_overflow): Update.
15230 (expand_neg_overflow): Update.
15231 (expand_mul_overflow): Update.
15232 * loop-doloop.c (doloop_modify): Update.
15233 * loop-unroll.c (compare_and_jump_seq): Update.
15234 * optabs.c (emit_cmp_and_jump_insn_1): Update.
15235 * predict.h: Update.
15236 * reorg.c (mostly_true_jump): Update.
15238 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
15239 * config/alpha/alpha.c (emit_unlikely_jump): Update.
15240 * config/arc/arc.c: (emit_unlikely_jump): Update.
15241 * config/arm/arm.c: (emit_unlikely_jump): Update.
15242 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
15243 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
15244 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
15245 (ix86_print_operand): Update.
15246 (ix86_split_fp_branch): Update.
15247 (predict_jump): Update.
15248 * config/ia64/ia64.c (ia64_print_operand): Update.
15249 * config/mmix/mmix.c (mmix_print_operand): Update.
15250 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
15251 (rs6000_expand_split_stack_prologue): Update.
15252 * config/rs6000/rs6000.c: Update.
15253 * config/s390/s390.c (s390_expand_vec_strlen): Update.
15254 (s390_expand_vec_movstr): Update.
15255 (s390_expand_cs_tdsi): Update.
15256 (s390_expand_split_stack_prologue): Update.
15257 * config/sh/sh.c (sh_print_operand): Update.
15258 (expand_cbranchsi4): Update.
15259 (expand_cbranchdi4): Update.
15260 * config/sparc/sparc.c (output_v9branch): Update.
15261 * config/spu/spu.c (get_branch_target): Update.
15262 (ea_load_store_inline): Update.
15263 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
15264 * config/tilepro/tilepro.c: Update.
15266 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
15268 * gimplify.c (mostly_copy_tree_r): Revert latest change.
15269 (gimplify_save_expr): Likewise.
15271 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15273 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
15275 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15277 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
15279 * timevar.def (TV_IPA_FNSUMMARY): Define.
15281 2017-07-16 Daniel Cederman <cederman@gaisler.com>
15283 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
15284 to back store errata sensitive sequence from being generated.
15285 (sqrtdf2_fix): Likewise.
15287 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15289 * tree-ssa-threadupdate.c (compute_path_counts,
15290 update_joiner_offpath_counts): Use profile_probability.
15292 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
15295 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15297 * config/arm/arm-c.c (arm_cpu_builtins): Define
15298 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
15300 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
15302 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15303 array entries to represent __ieee128 versions of the
15304 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
15305 scalar_extract_sig, and scalar_insert_exp built-in functions.
15306 (altivec_resolve_overloaded_builtin): Add special case handling
15307 for the __builtin_scalar_insert_exp function, as represented by
15308 the P9V_BUILTIN_VEC_VSIEDP constant.
15309 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
15310 exponent support for __ieee128 argument.
15311 (VSESQP): Add scalar extract signature support for __ieee128
15313 (VSTDCNQP): Add scalar test negative support for __ieee128
15315 (VSIEQP): Add scalar insert exponent support for __int128 argument
15316 with __ieee128 result.
15317 (VSIEQPF): Add scalar insert exponent support for __ieee128
15318 argument with __ieee128 result.
15319 (VSTDCQP): Add scalar test data class support for __ieee128
15321 (VSTDCNQP): Add overload support for scalar test negative with
15322 __ieee128 argument.
15323 (VSTDCQP): Add overload support for scalar test data class
15324 __ieee128 argument.
15325 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
15326 UNSPEC_VSX_SXSIGDP.
15327 (UNSPEC_VSX_SIEXPQP): New constant.
15328 (xsxexpqp): New insn for VSX scalar extract exponent quad
15330 (xsxsigqp): New insn for VSX scalar extract significand quad
15332 (xsiexpqpf): New insn for VSX scalar insert exponent quad
15333 precision with floating point argument.
15334 (xststdcqp): New expand for VSX scalar test data class quad
15336 (xststdcnegqp): New expand for VSX scalar test negative quad
15338 (xststdcqp): New insn to match expansions for VSX scalar test data
15339 class quad precision and VSX scalar test negative quad precision.
15340 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
15341 special case operand checking to enforce that second operand of
15342 VSX scalar test data class with quad precision argument is a 7-bit
15344 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
15345 prototypes and descriptions of __ieee128 versions of
15346 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
15347 scalar_test_data_class, and scalar_test_neg built-in functions.
15349 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15351 PR tree-optimization/81162
15352 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
15353 replace a negate with an add.
15355 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
15357 * doc/invoke.texi (arm/-mcpu): Document +crypto.
15359 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15361 * config/arm/arm-c.c (arm_cpu_builtins): Define
15362 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
15364 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15366 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
15367 (armv8-r): Set ARM Cortex-R52 as default CPU.
15368 * config/arm/arm-tables.opt: Regenerate.
15369 * config/arm/arm-tune.md: Regenerate.
15370 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
15372 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
15373 extension for -mcpu=cortex-r52.
15375 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15377 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
15378 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
15379 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
15380 (fp-armv8): Define it as FP_ARMv8 only.
15381 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
15382 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
15384 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
15385 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
15386 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
15387 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
15388 than TARGET_FPU_ARMV8.
15389 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
15390 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
15391 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
15393 * config/arm/neon.md (neon_vrint): Likewise.
15394 (neon_vcvt): Likewise.
15395 (neon_<fmaxmin_op><mode>): Likewise.
15396 (<fmaxmin><mode>3): Likewise.
15397 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
15398 * config/arm/predicates.md (arm_cond_move_operator): Check against
15399 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
15401 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
15403 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
15404 to top of function.
15406 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15408 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
15409 loop in comment with memset.
15411 2017-07-14 Martin Liska <mliska@suse.cz>
15413 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
15414 * dwarf2out.c (is_java): Remove the function.
15415 (output_pubname): Remove usage of the function.
15416 (lower_bound_default): Remove usage of DW_LANG_Java.
15417 (gen_compile_unit_die): Likewise.
15418 * gcc.c: Remove compiler defaults for .java and .zip files.
15419 * gimple-expr.c (remove_suffix): Change as there's no longer
15420 extension than 4-letter one.
15421 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
15422 (gimplify_save_expr): Likewise.
15423 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
15424 as it's possible even for other languages than Java.
15425 * langhooks.h (struct lang_hooks): Remove Java from a comment.
15426 * lto-opts.c (lto_write_options): Remove reference to Java.
15427 * opts.c (strip_off_ending): Update file extension handling.
15428 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
15429 * tree-eh.c (lower_resx): Likewise.
15430 * tree.c (free_lang_data_in_type): Remove dead code.
15431 (find_decls_types_r): Likewise.
15432 (build_common_builtin_nodes): Remove Java from a comment.
15433 (verify_type): Remove dead code.
15434 * varasm.c (assemble_external): Remove Java from a comment.
15436 2017-07-14 Martin Liska <mliska@suse.cz>
15438 * opts.c (finish_options): Add quotes.
15439 (common_handle_option): Likewise.
15441 2017-07-14 Martin Liska <mliska@suse.cz>
15443 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
15444 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
15445 Remove N_SO_PASCAL.
15446 * dwarf2out.c (lower_bound_default): Do not handle
15448 (gen_compile_unit_die): Likewise.
15449 * gcc.c: Remove default extension binding for GNU Pascal.
15450 * stmt.c: Remove Pascal language from a comment.
15451 * xcoffout.c: Likewise.
15453 2017-07-13 David Malcolm <dmalcolm@redhat.com>
15456 * diagnostic-show-locus.c (fixit_cmp): New function.
15457 (layout::layout): Sort m_fixit_hints.
15458 (column_range::column_range): Assert that the values are valid.
15459 (struct char_span): New struct.
15460 (correction::overwrite): New method.
15461 (struct source_line): New struct.
15462 (line_corrections::add_hint): Add assertions. Reimplement memcpy
15463 calls in terms of classes source_line and char_span, and
15464 correction::overwrite.
15465 (selftest::test_overlapped_fixit_printing_2): New function.
15466 (selftest::diagnostic_show_locus_c_tests): Call it.
15468 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
15470 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
15471 early if there is no lhs.
15473 2017-07-13 Martin Liska <mliska@suse.cz>
15475 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
15476 (gen_reference_type_die): Likewise.
15477 * stor-layout.c: Remove Pascal-related comment.
15479 2017-07-13 Martin Liska <mliska@suse.cz>
15481 * opts.c (finish_options): Add quotes to error messages.
15482 (parse_sanitizer_options): Likewise.
15484 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
15486 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
15488 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
15490 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
15492 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
15494 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
15496 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
15498 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
15501 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
15502 provides the hardware capability bits, define the macro
15503 __BUILTIN_CPU_SUPPORTS__.
15504 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
15505 if GLIBC does not provide the hardware capability bits. Add a
15506 gcc_unreachable call if the built-in cpu function is neither
15507 __builtin_cpu_is nor __builtin_cpu_supports.
15508 (rs6000_get_function_versions_dispatcher): Change the warning
15509 that an old GLIBC is used which does not export the capability
15510 bits to be an error.
15511 * doc/extend.texi (target_clones attribute): Document the
15512 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
15513 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
15514 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
15515 the macros defined by GCC if the newer GLIBC is available.
15517 2017-07-12 Jeff Law <law@redhat.com>
15519 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
15520 remaining includes slightly.
15521 * config/riscv/riscv-builtins.c: Include profile-count.h.
15523 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
15526 * config/avr/avr.c (avr_set_current_function): In diagnostic
15527 messages: Quote keywords and (parts of) identifiers.
15528 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
15531 2017-07-12 Carl Love <cel@us.ibm.com>
15533 * config/rs6000/rs6000-c.c: Add support for built-in functions
15534 vector bool char vec_revb (vector bool char);
15535 vector bool short vec_revb (vector short char);
15536 vector bool int vec_revb (vector bool int);
15537 vector bool long long vec_revb (vector bool long long);
15538 * doc/extend.texi: Update the built-in documentation file for the
15539 new built-in functions.
15541 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15543 * config/s390/s390.md: Remove movcc splitter.
15545 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15547 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
15548 load/store on condition.
15550 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
15553 * config/avr/avr.c (avr_encode_section_info)
15554 [progmem && !TREE_READONLY]: Error if progmem object needs
15557 2017-07-11 Michael Collison <michael.collison@arm.com>
15559 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
15562 2017-07-11 Carl Love <cel@us.ibm.com>
15564 * config/rs6000/rs6000-c.c: Add support for builtins
15565 vector unsigned int vec_parity_lsbb (vector signed int);
15566 vector unsigned int vec_parity_lsbb (vector unsigned int);
15567 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
15568 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
15569 vector unsigned long long vec_parity_lsbb (vector signed long long);
15570 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
15571 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
15572 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
15573 * doc/extend.texi: Update the built-in documentation file for the
15574 new built-in functions.
15576 2017-07-11 David Malcolm <dmalcolm@redhat.com>
15578 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
15579 (layout::m_primary_loc): New field.
15580 (layout::layout): Initialize new field. Move location filtering
15581 logic from here to...
15582 (layout::maybe_add_location_range): ...this new method. Add
15583 support for filtering to just the lines already specified by other
15585 (layout::will_show_line_p): New method.
15586 (gcc_rich_location::add_location_if_nearby): New method.
15587 (selftest::test_add_location_if_nearby): New test function.
15588 (selftest::diagnostic_show_locus_c_tests): Call it.
15589 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
15592 2017-07-11 Tom de Vries <tom@codesourcery.com>
15594 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
15595 (bb_first_real_insn): New function.
15596 (nvptx_single): Add extra initialization of broadcasted condition
15599 2017-07-11 Nathan Sidwell <nathan@acm.org>
15601 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
15603 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
15605 * doc/extend.texi (AVR Function Attributes): Remove weblink to
15606 Binutils doc as TEXI will mess them up.
15607 * doc/invoke.texi (AVR Options): Same here.
15609 2017-07-11 Daniel Cederman <cederman@gaisler.com>
15611 * config/sparc/sparc.opt (mfix-ut700): New option.
15612 (mfix-gr712rc): Likewise.
15613 (sparc_fix_b2bst): New variable.
15614 * doc/invoke.texi (SPARC options): Document them.
15615 (ARM options): Fix warnings.
15616 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
15617 instructions to prevent sequences that can trigger the store-store
15618 errata for certain LEON3FT processors.
15619 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
15620 (sparc_option_override): Set sparc_fix_b2bst appropriately.
15621 * config/sparc/sparc.md (fix_b2bst): New attribute.
15622 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
15624 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
15627 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
15629 (*rsqrtsf2_sse): Ditto.
15631 (div<mode>3): Macroize insn from divdf3 and divsf3
15632 using MODEF mode iterator.
15634 2017-07-10 Martin Sebor <msebor@redhat.com>
15636 PR tree-optimization/80397
15637 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
15638 instead of testing for equality to INTEGER_TYPE.
15640 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
15642 * config.gcc: Remove uclibc from arc target spec.
15644 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
15646 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
15648 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15651 * lto-wrapper.c (remove_option): New function.
15652 (merge_and_complain): Merge PIC/PIE options more realistically.
15654 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
15656 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
15660 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
15661 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
15662 * config.in: Regenerate.
15663 * configure: Regenerate.
15664 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
15665 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
15666 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
15667 (TARGET_GASISR_PROLOGUES): ...target mask.
15668 * common/config/avr/avr-common.c
15669 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
15670 Set -mgas-isr-prologues.
15671 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
15672 INSERT_PASS_BEFORE for it.
15673 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
15674 * config/avr/avr.c (avr_option_override)
15675 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
15676 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
15677 (avr_attribute_table) <no_gccisr>: Add new function attribute.
15678 (avr_set_current_function) <is_no_gccisr>: Init machine field.
15679 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
15681 (make_avr_pass_pre_proep): New function.
15682 (emit_push_sfr) <treg>: Add argument to function and use it
15683 instead of TMP_REG.
15684 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
15685 and set machine->gasisr.yes.
15686 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
15687 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
15688 __gcc_isr.n_pushed to .L__stack_usage.
15689 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
15690 (avr_asm_final_postscan_insn): ...this new static function.
15691 * config/avr/avr.h (machine_function)
15692 <is_no_gccisr, use_L__stack_usage>: New fields.
15693 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
15694 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
15695 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
15696 (gasisr, *gasisr): New expander and insn.
15697 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
15698 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
15699 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
15701 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
15703 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
15706 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
15708 Move jump-tables out of .text again.
15711 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
15712 (ASM_OUTPUT_ADDR_VEC): New function.
15713 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
15714 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
15715 INSN_ADDRESSes as asm comment.
15716 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
15717 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
15718 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
15719 * config/avr/avr.md (*tablejump): Adjust comment.
15720 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
15721 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
15723 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
15724 (avr_output_addr_vec): New proto.
15725 (avr_log_t) <insn_addresses>: New field.
15727 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
15730 * config/i386/i386.c (ix86_function_arg_advance): Set
15731 outgoing_args_on_stack to true if there are outgoing arguments
15733 (ix86_function_arg): Likewise.
15734 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
15735 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
15736 * config/i386/i386.h (machine_function): Add
15737 outgoing_args_on_stack.
15739 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
15741 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
15742 supporting pthreds.
15743 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
15745 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
15747 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
15748 (REAL_H): Remove $(MACHMODE_H).
15749 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
15751 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
15752 $(MACHMODE_H) and double-int.h.
15753 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
15755 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
15758 2017-07-07 Andrew Pinski <apinski@cavium.com>
15760 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
15761 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
15763 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15765 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
15766 Add warning if GCC was not configured to link against a GLIBC that
15767 exports the hardware capability bits.
15768 (make_resolver_func): Make resolver function private and not a
15769 COMDAT function. Create the name with clone_function_name instead
15770 of make_unique_name.
15773 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
15774 correct operand in doing the split.
15776 2017-07-07 Carl Love <cel@us.ibm.com>
15778 * config/rs6000/rs6000-c: Add support for built-in function
15779 vector unsigned short vec_pack_to_short_fp32 (vector float,
15781 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
15782 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
15783 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
15784 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
15785 (convert_4f32_8i16): Add define_expand.
15786 * doc/extend.texi: Update the built-in documentation file for the
15787 new built-in function.
15789 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15791 * config/sparc/m8.md: New file.
15792 * config/sparc/sparc.md: Include m8.md.
15794 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15796 * config/sparc/sparc.opt: New option -mvis4b.
15797 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
15798 (sparc_option_override): Handle VIS4B.
15799 (enum sparc_builtins): Define
15800 SPARC_BUILTIN_DICTUNPACK{8,16,32},
15801 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
15802 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
15803 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
15804 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
15805 (check_constant_argument): New function.
15806 (sparc_vis_init_builtins): Define builtins
15807 __builtin_vis_dictunpack{8,16,32},
15808 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
15809 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
15810 __builtin_vis_fpcmpde{8,16,32}shl and
15811 __builtin_vis_fpcmpur{8,16,32}shl.
15812 (sparc_expand_builtin): Check that the constant operands to
15813 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
15814 constant and in range.
15815 * config/sparc/sparc-c.c (sparc_target_macros): Handle
15817 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
15818 (SPARC_IMM5_P): Likewise.
15819 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
15820 (enabled): Handle vis4b.
15821 (UNSPEC_DICTUNPACK): New unspec.
15822 (UNSPEC_FPCMPSHL): Likewise.
15823 (UNSPEC_FPUCMPSHL): Likewise.
15824 (UNSPEC_FPCMPDESHL): Likewise.
15825 (UNSPEC_FPCMPURSHL): Likewise.
15826 (cpu_feature): New CPU feature `vis4b'.
15827 (dictunpack{8,16,32}): New insns.
15828 (FPCSMODE): New mode iterator.
15829 (fpcscond): New code iterator.
15830 (fpcsucond): Likewise.
15831 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
15832 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
15833 (fpcmpde{8,16,32}{si,di}shl): Likewise.
15834 (fpcmpur{8,16,32}{si,di}shl): Likewise.
15835 * config/sparc/constraints.md: Define constraints `q' for unsigned
15836 2-bit integer constants and `t' for unsigned 5-bit integer
15838 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
15840 (imm5_operand_dictunpack16): Likewise.
15841 (imm5_operand_dictunpack32): Likewise.
15842 (imm2_operand): Likewise.
15843 * doc/invoke.texi (SPARC Options): Document -mvis4b.
15844 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
15845 ditunpack* and fpcmp*shl builtins.
15847 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15849 * config.gcc: Handle m8 in --with-{cpu,tune} options.
15850 * config.in: Add HAVE_AS_SPARC6 define.
15851 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
15853 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
15855 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
15856 (CPP_CPU_SPEC): Handle m8.
15857 (ASM_CPU_SPEC): Likewise.
15858 * config/sparc/sparc-opts.h (enum processor_type): Add
15860 * config/sparc/sparc.c (m8_costs): New struct.
15861 (sparc_option_override): Handle TARGET_CPU_m8.
15862 (sparc32_initialize_trampoline): Likewise.
15863 (sparc64_initialize_trampoline): Likewise.
15864 (sparc_issue_rate): Likewise.
15865 (sparc_register_move_cost): Likewise.
15866 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
15867 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
15868 (ASM_CPU64_DEFAULT_SPEC): Likewise.
15869 (CPP_CPU_SPEC): Handle M8.
15870 (ASM_CPU_SPEC): Likewise.
15871 (AS_M8_FLAG): Define.
15872 * config/sparc/sparc.md: Add m8 to the cpu attribute.
15873 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
15874 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
15876 * configure: Regenerate.
15877 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
15880 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15882 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
15884 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
15885 ("*movdi_insn_sp32"): Do not set v3pipe.
15886 ("*movsi_insn"): Likewise.
15887 ("*movdi_insn_sp64"): Likewise.
15888 ("*movsf_insn"): Likewise.
15889 ("*movdf_insn_sp32"): Likewise.
15890 ("*movdf_insn_sp64"): Likewise.
15891 ("*zero_extendsidi2_insn_sp64"): Likewise.
15892 ("*sign_extendsidi2_insn"): Likewise.
15893 ("*mov<VM32:mode>_insn"): Likewise.
15894 ("*mov<VM64:mode>_insn_sp64"): Likewise.
15895 ("*mov<VM64:mode>_insn_sp32"): Likewise.
15896 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
15897 ("<vlop:code><VL:mode>3"): Likewise.
15898 ("*not_<vlop:code><VL:mode>3"): Likewise.
15899 ("*nand<VL:mode>_vis"): Likewise.
15900 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
15901 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
15902 ("one_cmpl<VL:mode>2"): Likewise.
15903 ("faligndata<VM64:mode>_vis"): Likewise.
15904 ("alignaddrsi_vis"): Likewise.
15905 ("alignaddrdi_vis"): Likweise.
15906 ("alignaddrlsi_vis"): Likewise.
15907 ("alignaddrldi_vis"): Likewise.
15908 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
15909 ("bmaskdi_vis"): Likewise.
15910 ("bmasksi_vis"): Likewise.
15911 ("bshuffle<VM64:mode>_vis"): Likewise.
15912 ("cmask8<P:mode>_vis"): Likewise.
15913 ("cmask16<P:mode>_vis"): Likewise.
15914 ("cmask32<P:mode>_vis"): Likewise.
15915 ("pdistn<P:mode>_vis"): Likewise.
15916 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
15918 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15920 * config/sparc/sparc.md ("subtype"): New insn attribute.
15921 ("*wrgsr_sp64"): Set insn subtype.
15922 ("*rdgsr_sp64"): Likewise.
15923 ("alignaddrsi_vis"): Likewise.
15924 ("alignaddrdi_vis"): Likewise.
15925 ("alignaddrlsi_vis"): Likewise.
15926 ("alignaddrldi_vis"): Likewise.
15927 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
15928 ("fexpand_vis"): Likewise.
15929 ("fpmerge_vis"): Likewise.
15930 ("faligndata<VM64:mode>_vis"): Likewise.
15931 ("bshuffle<VM64:mode>_vis"): Likewise.
15932 ("cmask8<P:mode>_vis"): Likewise.
15933 ("cmask16<P:mode>_vis"): Likewise.
15934 ("cmask32<P:mode>_vis"): Likewise.
15935 ("fchksm16_vis"): Likewise.
15936 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
15937 ("fmean16_vis"): Likewise.
15938 ("fp<plusminus_insn>64_vis"): Likewise.
15939 ("<plusminus_insn>v8qi3"): Likewise.
15940 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
15941 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
15942 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
15943 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
15944 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
15945 ("*movqi_insn"): Likewise.
15946 ("*movhi_insn"): Likewise.
15947 ("*movsi_insn"): Likewise.
15948 ("movsi_pic_gotdata_op"): Likewise.
15949 ("*movdi_insn_sp32"): Likewise.
15950 ("*movdi_insn_sp64"): Likewise.
15951 ("movdi_pic_gotdata_op"): Likewise.
15952 ("*movsf_insn"): Likewise.
15953 ("*movdf_insn_sp32"): Likewise.
15954 ("*movdf_insn_sp64"): Likewise.
15955 ("*zero_extendhisi2_insn"): Likewise.
15956 ("*zero_extendqihi2_insn"): Likewise.
15957 ("*zero_extendqisi2_insn"): Likewise.
15958 ("*zero_extendqidi2_insn"): Likewise.
15959 ("*zero_extendhidi2_insn"): Likewise.
15960 ("*zero_extendsidi2_insn_sp64"): Likewise.
15961 ("ldfsr"): Likewise.
15962 ("prefetch_64"): Likewise.
15963 ("prefetch_32"): Likewise.
15964 ("tie_ld32"): Likewise.
15965 ("tie_ld64"): Likewise.
15966 ("*tldo_ldub_sp32"): Likewise.
15967 ("*tldo_ldub1_sp32"): Likewise.
15968 ("*tldo_ldub2_sp32"): Likewise.
15969 ("*tldo_ldub_sp64"): Likewise.
15970 ("*tldo_ldub1_sp64"): Likewise.
15971 ("*tldo_ldub2_sp64"): Likewise.
15972 ("*tldo_ldub3_sp64"): Likewise.
15973 ("*tldo_lduh_sp32"): Likewise.
15974 ("*tldo_lduh1_sp32"): Likewise.
15975 ("*tldo_lduh_sp64"): Likewise.
15976 ("*tldo_lduh1_sp64"): Likewise.
15977 ("*tldo_lduh2_sp64"): Likewise.
15978 ("*tldo_lduw_sp32"): Likewise.
15979 ("*tldo_lduw_sp64"): Likewise.
15980 ("*tldo_lduw1_sp64"): Likewise.
15981 ("*tldo_ldx_sp64"): Likewise.
15982 ("*mov<VM32:mode>_insn"): Likewise.
15983 ("*mov<VM64:mode>_insn_sp64"): Likewise.
15984 ("*mov<VM64:mode>_insn_sp32"): Likewise.
15986 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15988 * config/sparc/sparc.md ("type"): New insn type viscmp.
15989 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
15991 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
15992 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
15993 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
15994 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
15996 ("n7_vis_logical_11cycle"): Likewise.
15997 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
15998 * config/sparc/niagara2.md ("niag3_vis": Likewise.
15999 * config/sparc/niagara.md ("niag_vis"): Likewise.
16000 * config/sparc/ultra3.md ("us3_fga"): Likewise.
16001 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
16003 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16005 * config/sparc/sparc.md: New instruction type `bmask'.
16006 (bmaskdi_vis): Use the `bmask' type.
16007 (bmasksi_vis): Likewise.
16008 * config/sparc/ultra3.md (us3_array): Likewise.
16009 * config/sparc/niagara7.md (n7_array): Likewise.
16010 * config/sparc/niagara4.md (n4_array): Likewise.
16011 * config/sparc/niagara2.md (niag2_vis): Likewise.
16012 (niag3_vis): Likewise.
16013 * config/sparc/niagara.md (niag_vis): Likewise.
16015 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16017 * ipa-comdats.c: Remove optimize check from gate.
16018 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
16019 for functions not optimized.
16020 (ipa_fn_summary_read): Skip optimize check.
16021 (ipa_fn_summary_write): Likewise.
16022 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
16024 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
16026 (can_inline_edge_p): Check flag_pcc_struct_return for match.
16027 (check_callers): Give up on caller which is not optimized.
16028 (inline_small_functions): Likewise.
16029 (ipa_inline): Do not give up when not optimizing.
16030 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
16031 away unoptimizes cdtors.
16032 (whole_program_function_and_variable_visibility): Do
16033 ipa_discover_readonly_nonaddressable_vars in LTO mode.
16034 * ipa.c (process_references): Do not check optimize.
16035 (symbol_table::remove_unreachable_nodes): Update optimize check.
16036 (set_writeonly_bit): Update optimize check.
16037 (pass_ipa_cdtor_merge::gate): Do not check optimize.
16038 (pass_ipa_single_use::gate): Remove.
16040 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16042 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
16043 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
16044 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
16045 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
16046 permute_load, permute_store, adjust_extract, adjust_splat,
16047 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
16048 replace_swap_with_copy, dump_swap_insn_table,
16049 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
16050 recombine_lvx_pattern, recombine_stvx_pattern,
16051 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
16052 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
16053 to file rs6000-p8swap.c.
16054 * config/rs6000/rs6000-p8swap.c: New file.
16055 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
16056 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
16057 and rs6000*-*-* targets.
16059 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16061 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
16063 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16065 * lto-wrapper.c (merge_and_complain): Do not merge
16066 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
16067 fsigned_zeros, ftrapping_math, fwrapv.
16068 (append_compiler_options): Do not track these options.
16069 (append_linker_options): Likewie
16071 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16073 * cgraphunit.c (cgraph_node::finalize_function): When
16074 !flag_toplevel_reorde set no_reorder flag.
16075 (varpool_node::finalize_decl): Likewise.
16076 (symbol_table::compile): Drop no toplevel reorder path.
16078 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16080 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
16081 edges; zero probability is not better than uninitialized.
16083 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
16085 * asan.h (asan_sanitize_allocas_p): Declare.
16086 * asan.c (asan_sanitize_allocas_p): New function.
16087 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
16088 (handle_builtin_alloca): Likewise.
16089 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
16090 if !asan_sanitize_allocas_p.
16091 * params.def (asan-instrument-allocas): Add new option.
16092 * params.h (ASAN_PROTECT_ALLOCAS): Define.
16093 * opts.c (common_handle_option): Disable allocas sanitization for
16096 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
16098 * asan.c: Include gimple-fold.h.
16099 (get_last_alloca_addr): New function.
16100 (handle_builtin_stackrestore): Likewise.
16101 (handle_builtin_alloca): Likewise.
16102 (asan_emit_allocas_unpoison): Likewise.
16103 (get_mem_refs_of_builtin_call): Add new parameter, remove const
16104 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
16105 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
16106 (instrument_builtin_call): Pass gimple iterator to
16107 get_mem_refs_of_builtin_call.
16108 (last_alloca_addr): New global.
16109 * asan.h (asan_emit_allocas_unpoison): Declare.
16110 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
16111 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
16112 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
16113 if function calls alloca.
16114 * gimple-fold.c (replace_call_with_value): Remove static keyword.
16115 * gimple-fold.h (replace_call_with_value): Declare.
16116 * internal-fn.c: Include asan.h.
16117 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
16118 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
16120 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16122 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
16123 (C_SELFTEST_FLAGS): New.
16124 (CPP_SELFTEST_FLAGS): New.
16125 (SELFTEST_DEPS): New, from deps of s-selftest.
16126 (C_SELFTEST_DEPS): New, from deps of s-selftest.
16127 (CPP_SELFTEST_DEPS): New.
16128 (selftest): Add dependency on s-selftest-c++.
16129 (s-selftest): Rename to...
16130 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
16131 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
16132 than SELFTEST_FLAGS.
16133 (selftest-gdb): Rename to...
16134 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
16136 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
16137 (selftest-valgrind): Rename to...
16138 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
16140 (selftest-valgrind): Reintroduce as an alias for
16141 selftest-c-valgrind.
16142 (s-selftest-c++): New.
16143 (selftest-c++-gdb): New.
16144 (selftest-c++-valgrind): New.
16146 2017-07-06 Olivier Hainque <hainque@adacore.com>
16148 * gcc.c (process_command): When deciding if undefined variables
16149 should be ignored when processing specs, accept "gcc -v" as well.
16151 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16153 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
16154 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
16156 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
16158 * config/arm/arm-cpus.in (armv8-r): Add new entry.
16159 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
16160 * config/arm/arm-tables.opt: Regenerate.
16161 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
16163 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
16165 2017-07-06 Carl Love <cel@us.ibm.com>
16167 * ChangeLog: Clean up from mid air collision
16169 2017-07-06 Carl Love <cel@us.ibm.com>
16171 * config/rs6000/rs6000-c.c: Add support for built-in functions
16172 vector signed int vec_subc (vector signed int, vector signed int);
16173 vector signed __int128 vec_subc (vector signed __int128,
16174 vector signed __int128);
16175 vector unsigned __int128 vec_subc (vector unsigned __int128,
16176 vector unsigned __int128);
16177 vector signed int vec_sube (vector signed int, vector signed int,
16178 vector signed int);
16179 vector unsigned int vec_sube (vector unsigned int,
16180 vector unsigned int,
16181 vector unsigned int);
16182 vector signed __int128 vec_sube (vector signed __int128,
16183 vector signed __int128,
16184 vector signed__int128);
16185 vector unsigned __int128 vec_sube (vector unsigned __int128,
16186 vector unsigned __int128,
16187 vector unsigned __int128);
16188 vector signed int vec_subec (vector signed int, vector signed int,
16189 vector signed int);
16190 vector unsigned int vec_subec (vector unsigned int,
16191 vector unsigned int,
16192 vector unsigned int);
16193 vector signed __int128 vec_subec (vector signed __int128,
16194 vector signed __int128,
16195 vector signed__int128);
16196 vector unsigned __int128 vec_subec (vector unsigned __int128,
16197 vector unsigned __int128,
16198 vector unsigned __int128);
16199 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
16200 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
16201 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
16202 BU_ALTIVEC_OVERLOAD_X definitions.
16203 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
16204 * doc/extend.texi: Update the built-in documentation file for the new
16205 built-in functions.
16207 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16210 * diagnostic-show-locus.c (layout::layout): Use start and finish
16211 spelling location for the start and finish of each range.
16212 * genmatch.c (linemap_client_expand_location_to_spelling_point):
16213 Add unused aspect param.
16214 * input.c (expand_location_1): Add "aspect" param, and use it
16215 to access the correct part of the location.
16216 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
16218 (expand_location_to_spelling_point): Likewise.
16219 (linemap_client_expand_location_to_spelling_point): Add "aspect"
16220 param, and pass it to expand_location_1.
16222 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
16224 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
16225 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
16226 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
16227 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
16228 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
16229 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
16230 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
16231 _mm_maskz_getmant_ss): New intrinsics.
16232 (__builtin_ia32_getexpss128_mask): Changed to ...
16233 __builtin_ia32_getexpss128_round ... this.
16234 (__builtin_ia32_getexpsd128_mask): Changed to ...
16235 __builtin_ia32_getexpsd128_round ... this.
16236 * config/i386/i386-builtin-types.def
16237 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
16238 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
16239 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
16240 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
16241 __builtin_ia32_getmantss_mask_round): New builtins.
16242 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
16243 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
16244 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
16245 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
16246 * config/i386/sse.md
16247 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
16248 avx512f_sgetexp<mode><mask_scalar_name>
16249 <round_saeonly_scalar_name> ... this.
16250 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
16251 %0, %1, %2<round_saeonly_op3>}): Changed to ...
16252 vgetexp<ssescalarmodesuffix>
16253 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16254 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
16255 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
16256 avx512f_vgetmant<mode><mask_scalar_name>
16257 <round_saeonly_scalar_name> ... this.
16258 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
16259 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
16260 vgetmant<ssescalarmodesuffix>
16261 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
16262 %0<mask_scalar_operand4>, %1, %2
16263 <round_saeonly_scalar_mask_op4>, %3} ... this.
16264 * config/i386/subst.md (mask_scalar_operand4,
16265 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
16266 round_saeonly_scalar_nimm_predicate): New subst attributes.
16268 2017-07-06 Julia Koval <julia.koval@intel.com>
16270 * config/i386/i386.c (ix86_erase_embedded_rounding):
16271 Remove code for old rounding pattern.
16273 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
16275 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
16277 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
16279 * doc/sourcebuild.texi (Test Directives, Variants of
16280 dg-require-support): Add documentation for dg-require-stack-check.
16282 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
16284 * config/i386/subst.md (mask_scalar, round_scalar,
16285 round_saeonly_scalar): New meta-templates.
16286 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
16287 round_scalar_mask_operand3, round_scalar_mask_op3,
16288 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
16289 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
16290 round_saeonly_scalar_constraint,
16291 round_saeonly_scalar_prefix): New subst attribute.
16292 * config/i386/sse.md
16293 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
16294 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
16295 <round_scalar_name> ... this.
16296 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
16297 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
16298 <round_scalar_name> ... this.
16299 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
16300 <sse>_vm<code><mode>3<mask_scalar_name>
16301 <round_saeonly_scalar_name> ... this.
16302 (v<plusminus_mnemonic><ssescalarmodesuffix>
16303 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
16304 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
16305 v<plusminus_mnemonic><ssescalarmodesuffix>
16306 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16307 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
16308 (v<multdiv_mnemonic><ssescalarmodesuffix>
16309 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
16310 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
16311 v<multdiv_mnemonic><ssescalarmodesuffix>
16312 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16313 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
16314 (v<maxmin_float><ssescalarmodesuffix>
16315 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
16316 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
16317 v<maxmin_float><ssescalarmodesuffix>
16318 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16319 %0<mask_scalar_operand3>, %1, %<iptr>2
16320 <round_saeonly_scalar_mask_op3>} ... this.
16322 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
16324 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
16325 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
16327 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
16328 Alan Hayward <alan.hayward@arm.com>
16329 David Sherwood <david.sherwood@arm.com>
16331 * combine.c (simplify_if_then_else): Remove "enum" before
16333 * compare-elim.c (can_eliminate_compare): Likewise.
16334 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
16336 (aarch64_lookup_simd_builtin_type): Likewise.
16337 (aarch64_simd_builtin_type): Likewise.
16338 (aarch64_init_simd_builtin_types): Likewise.
16339 (aarch64_simd_expand_args): Likewise.
16340 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
16342 (aarch64_reverse_mask): Likewise.
16343 (aarch64_simd_emit_reg_reg_move): Likewise.
16344 (aarch64_gen_adjusted_ldpstp): Likewise.
16345 (aarch64_ccmp_mode_to_code): Likewise.
16346 (aarch64_operands_ok_for_ldpstp): Likewise.
16347 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16348 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
16350 (aarch64_min_divisions_for_recip_mul): Likewise.
16351 (aarch64_reassociation_width): Likewise.
16352 (aarch64_get_condition_code_1): Likewise.
16353 (aarch64_simd_emit_reg_reg_move): Likewise.
16354 (aarch64_simd_attr_length_rglist): Likewise.
16355 (aarch64_reverse_mask): Likewise.
16356 (aarch64_operands_ok_for_ldpstp): Likewise.
16357 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16358 (aarch64_gen_adjusted_ldpstp): Likewise.
16359 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
16361 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
16362 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
16363 (arm_lookup_simd_builtin_type): Likewise.
16364 (arm_simd_builtin_type): Likewise.
16365 (arm_init_simd_builtin_types): Likewise.
16366 (arm_expand_builtin_args): Likewise.
16367 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
16368 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
16369 (ft32_setup_incoming_varargs): Likewise.
16370 (ft32_function_arg): Likewise.
16371 (ft32_function_arg_advance): Likewise.
16372 (ft32_pass_by_reference): Likewise.
16373 (ft32_arg_partial_bytes): Likewise.
16374 (ft32_valid_pointer_mode): Likewise.
16375 (ft32_addr_space_pointer_mode): Likewise.
16376 (ft32_addr_space_legitimate_address_p): Likewise.
16377 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
16379 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
16380 (ix86_emit_outlined_ms2sysv_restore): Likewise.
16381 (iamcu_alignment): Likewise.
16382 (canonicalize_vector_int_perm): Likewise.
16383 (ix86_noce_conversion_profitable_p): Likewise.
16384 (ix86_mpx_bound_mode): Likewise.
16385 (ix86_operands_ok_for_move_multiple): Likewise.
16386 * config/microblaze/microblaze-protos.h
16387 (microblaze_expand_conditional_branch_reg): Likewise.
16388 * config/microblaze/microblaze.c
16389 (microblaze_expand_conditional_branch_reg): Likewise.
16390 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
16392 (rs6000_reassociation_width): Likewise.
16393 (rs6000_invalid_binary_op): Likewise.
16394 (fusion_p9_p): Likewise.
16395 (emit_fusion_p9_load): Likewise.
16396 (emit_fusion_p9_store): Likewise.
16397 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
16399 (riscv_hard_regno_mode_ok_p): Likewise.
16400 (riscv_address_insns): Likewise.
16401 (riscv_split_symbol): Likewise.
16402 (riscv_legitimize_move): Likewise.
16403 (riscv_function_value): Likewise.
16404 (riscv_hard_regno_nregs): Likewise.
16405 (riscv_expand_builtin): Likewise.
16406 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
16407 (riscv_build_integer): Likewise.
16408 (riscv_split_integer): Likewise.
16409 (riscv_legitimate_constant_p): Likewise.
16410 (riscv_cannot_force_const_mem): Likewise.
16411 (riscv_regno_mode_ok_for_base_p): Likewise.
16412 (riscv_valid_base_register_p): Likewise.
16413 (riscv_valid_offset_p): Likewise.
16414 (riscv_valid_lo_sum_p): Likewise.
16415 (riscv_classify_address): Likewise.
16416 (riscv_legitimate_address_p): Likewise.
16417 (riscv_address_insns): Likewise.
16418 (riscv_load_store_insns): Likewise.
16419 (riscv_force_binary): Likewise.
16420 (riscv_split_symbol): Likewise.
16421 (riscv_force_address): Likewise.
16422 (riscv_legitimize_address): Likewise.
16423 (riscv_move_integer): Likewise.
16424 (riscv_legitimize_const_move): Likewise.
16425 (riscv_legitimize_move): Likewise.
16426 (riscv_address_cost): Likewise.
16427 (riscv_subword): Likewise.
16428 (riscv_output_move): Likewise.
16429 (riscv_canonicalize_int_order_test): Likewise.
16430 (riscv_emit_int_order_test): Likewise.
16431 (riscv_function_arg_boundary): Likewise.
16432 (riscv_pass_mode_in_fpr_p): Likewise.
16433 (riscv_pass_fpr_single): Likewise.
16434 (riscv_pass_fpr_pair): Likewise.
16435 (riscv_get_arg_info): Likewise.
16436 (riscv_function_arg): Likewise.
16437 (riscv_function_arg_advance): Likewise.
16438 (riscv_arg_partial_bytes): Likewise.
16439 (riscv_function_value): Likewise.
16440 (riscv_pass_by_reference): Likewise.
16441 (riscv_setup_incoming_varargs): Likewise.
16442 (riscv_print_operand): Likewise.
16443 (riscv_elf_select_rtx_section): Likewise.
16444 (riscv_save_restore_reg): Likewise.
16445 (riscv_for_each_saved_reg): Likewise.
16446 (riscv_register_move_cost): Likewise.
16447 (riscv_hard_regno_mode_ok_p): Likewise.
16448 (riscv_hard_regno_nregs): Likewise.
16449 (riscv_class_max_nregs): Likewise.
16450 (riscv_memory_move_cost): Likewise.
16451 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
16452 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
16453 (rl78_addr_space_address_mode): Likewise.
16454 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16456 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
16457 (rs6000_reassociation_width): Likewise.
16458 (rs6000_invalid_binary_op): Likewise.
16459 (fusion_p9_p): Likewise.
16460 (emit_fusion_p9_load): Likewise.
16461 (emit_fusion_p9_store): Likewise.
16462 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
16463 (ok_for_simple_move_operands): Likewise.
16464 (ok_for_simple_move_strict_operands): Likewise.
16465 (ok_for_simple_arith_logic_operands): Likewise.
16466 (visium_legitimize_reload_address): Likewise.
16467 (visium_select_cc_mode): Likewise.
16468 (output_cbranch): Likewise.
16469 (visium_split_double_move): Likewise.
16470 (visium_expand_copysign): Likewise.
16471 (visium_expand_int_cstore): Likewise.
16472 (visium_expand_fp_cstore): Likewise.
16473 * config/visium/visium.c (visium_pass_by_reference): Likewise.
16474 (visium_function_arg): Likewise.
16475 (visium_function_arg_advance): Likewise.
16476 (visium_libcall_value): Likewise.
16477 (visium_setup_incoming_varargs): Likewise.
16478 (visium_legitimate_constant_p): Likewise.
16479 (visium_legitimate_address_p): Likewise.
16480 (visium_legitimize_address): Likewise.
16481 (visium_secondary_reload): Likewise.
16482 (visium_register_move_cost): Likewise.
16483 (visium_memory_move_cost): Likewise.
16484 (prepare_move_operands): Likewise.
16485 (ok_for_simple_move_operands): Likewise.
16486 (ok_for_simple_move_strict_operands): Likewise.
16487 (ok_for_simple_arith_logic_operands): Likewise.
16488 (visium_function_value_1): Likewise.
16489 (rtx_ok_for_offset_p): Likewise.
16490 (visium_legitimize_reload_address): Likewise.
16491 (visium_split_double_move): Likewise.
16492 (visium_expand_copysign): Likewise.
16493 (visium_expand_int_cstore): Likewise.
16494 (visium_expand_fp_cstore): Likewise.
16495 (visium_split_cstore): Likewise.
16496 (visium_select_cc_mode): Likewise.
16497 (visium_split_cbranch): Likewise.
16498 (output_cbranch): Likewise.
16499 (visium_print_operand_address): Likewise.
16500 * expmed.c (flip_storage_order): Likewise.
16501 * expmed.h (emit_cstore): Likewise.
16502 (flip_storage_order): Likewise.
16503 * genrecog.c (validate_pattern): Likewise.
16504 * hsa-gen.c (gen_hsa_addr): Likewise.
16505 * internal-fn.c (expand_arith_overflow): Likewise.
16506 * ira-color.c (allocno_copy_cost_saving): Likewise.
16507 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16508 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
16509 (process_invariant_for_inheritance): Likewise.
16510 * lra-eliminations.c (move_plus_up): Likewise.
16511 * omp-low.c (lower_oacc_reductions): Likewise.
16512 * simplify-rtx.c (simplify_subreg): Likewise.
16513 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
16514 (TARGET_CHKP_BOUND_MODE): Likewise..
16515 * targhooks.c (default_chkp_bound_mode): Likewise.
16516 (default_setup_incoming_vararg_bounds): Likewise.
16517 * targhooks.h (default_chkp_bound_mode): Likewise.
16518 (default_setup_incoming_vararg_bounds): Likewise.
16519 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
16520 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
16521 (have_whole_vector_shift): Likewise.
16522 * tree-vect-stmts.c (vectorizable_load): Likewise.
16523 * doc/tm.texi: Regenerate.
16525 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
16527 Graceful degrade if Binutils PR21472 is not available.
16530 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
16531 .rodata in flash test fails.
16532 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
16533 * confgure: Regenerate.
16534 * config.in: Regenerate.
16535 * config/avr/avr.c (avr_asm_named_section)
16536 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
16537 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
16538 (avr_asm_init_sections): Same.
16540 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16542 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
16543 (fma<VH:mode>4_intrinsic): Likewise.
16544 (*fmsub<VCVTF:mode>4): Likewise.
16545 (*fmsub<VH:mode>4_intrinsic): Likewise.
16547 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
16550 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
16551 Don't depend on "optimize > 0".
16552 (out_movhi_r_mr, out_movqi_mr_r): Same.
16553 (out_movhi_mr_r, out_movqi_r_mr): Same.
16554 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
16555 io_address_operand on "optimize > 0".
16557 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16559 * tree-loop-distribution.c: Add general explanantion on the pass.
16560 (generate_loops_for_partition): Mark distributed loop.
16561 (pg_add_dependence_edges): New parameter. Handle alias data
16562 dependence specially and record it in the parameter if asked.
16563 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
16564 (init_partition_graph_vertices, add_partition_graph_edge): New.
16565 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
16566 (free_partition_graph_vdata, build_partition_graph): New.
16567 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
16568 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
16569 (data_ref_segment_size, latch_dominated_by_data_ref): New.
16570 (compute_alias_check_pairs, version_loop_by_alias_check): New.
16571 (version_for_distribution_p, finalize_partitions): New.
16572 (distribute_loop): Handle alias data dependence specially. Factor
16573 out loop fusion code as functions and call these functions.
16575 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16577 * tree-loop-distribution.c (classify_partition): New parameter and
16578 better handle reduction statement.
16579 (rdg_build_partitions): Revise comment.
16580 (distribute_loop): Compute statements in all partitions and pass it
16581 to classify_partition.
16583 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16585 * tree-loop-distribution.c (enum partition_type): New.
16586 (struct partition): New field type.
16587 (partition_merge_into): Add parameter. Update partition type.
16588 (data_dep_in_cycle_p, update_type_for_merge): New functions.
16589 (build_rdg_partition_for_vertex): Compute partition type.
16590 (rdg_build_partitions): Dump partition type.
16591 (distribute_loop): Update calls to partition_merge_into.
16593 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16595 * tree-loop-distribution.c (struct ddr_hasher): New.
16596 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
16598 (classify_partition): Call get_data_dependence.
16599 (pg_add_dependence_edges): Ditto.
16600 (distribute_loop): Release data dependence hash table.
16602 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16604 * tree-loop-distribution.c (ref_base_address): Delete.
16605 (similar_memory_accesses): Rename ...
16606 (share_memory_accesses): ... to this. Check if partitions access
16607 the same memory reference.
16608 (distribute_loop): Call share_memory_accesses.
16610 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16612 * tree-loop-distribution.c (struct partition): New field recording
16613 its data reference.
16614 (partition_alloc, partition_free): Init and release data refs.
16615 (partition_merge_into): Merge data refs.
16616 (build_rdg_partition_for_vertex): Collect data refs for partition.
16617 (pg_add_dependence_edges): Change parameters from vector to bitmap.
16619 (distribute_loop): Remve data refs from vertice data of partition
16622 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16624 * tree-loop-distribution.c (params.h): Include header file.
16625 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
16626 (datarefs_vec): New global var.
16627 (create_rdg_vertices): Use datarefs_vec directly.
16628 (free_rdg): Don't free data references.
16629 (build_rdg): Update use. Don't free data references.
16630 (distribute_loop): Compute global variable for data references.
16631 Bail out if there are too many data references.
16633 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16635 * tree-loop-distribution.c (loop_nest): New global var.
16636 (build_rdg): Use loop directly, rather than loop nest.
16637 (pg_add_dependence_edges): Remove loop nest parameter. Use global
16639 (distribute_loop): Compute global variable loop nest. Update use.
16641 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16643 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
16644 (partition_merge_into): New parameter. Dump reason for fusion.
16645 (distribute_loop): Update use of partition_merge_into.
16647 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16649 * tree-loop-distribution.c (bb_top_order_index): New.
16650 (bb_top_order_index_size, bb_top_order_cmp): New.
16651 (stmts_from_loop): Use topological order.
16652 (pass_loop_distribution::execute): Compute and release topological
16653 order for basic blocks.
16655 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16657 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
16660 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16662 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
16663 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
16664 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
16665 * internal-fn.def (LOOP_DIST_ALIAS): New.
16666 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
16667 (fold_loop_internal_call): ... this.
16668 (vect_loop_dist_alias_call): New function.
16669 (set_uid_loop_bbs): Call fold_loop_internal_call.
16670 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
16673 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
16676 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
16677 Require dead FLAGS_REG at the beginning of a peephole.
16679 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
16682 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
16683 arguments in the call to __builtin_ia32_sbb_u32.
16684 (_subborrow_u64): Swap _X and _Y arguments in the call to
16685 __builtin_ia32_sbb_u64.
16687 2017-07-04 Jakub Jelinek <jakub@redhat.com>
16690 * tree-vrp.c (compare_assert_loc): Turn into a function template
16691 with stable template parameter. Only test if a->e is NULL,
16692 !a->e == !b->e has been verified already. Use e == NULL or
16693 e != NULL instead of e or ! e tests. If stable is true, don't use
16694 iterative_hash_expr, on the other side allow a or b or both NULL
16695 and sort the NULLs last.
16696 (process_assert_insertions): Sort using compare_assert_loc<false>
16697 instead of compare_assert_loc, later sort using
16698 compare_assert_loc<true> before calling process_assert_insertions_for
16699 in a loop. Use break instead of continue once seen NULL pointer.
16701 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
16703 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
16704 Cortex-R7 and Cortex-R8 processors.
16706 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
16708 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
16709 uninitialized while src is not.
16711 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
16713 * common/config/arm/arm-common.c: Adjust include path for
16715 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
16716 (arm-cpu.h): Create in build directory. Adjust dependency rules.
16717 (arm-cpu-data.h): Likewise.
16718 (arm-cpu-cdata.h): Likewise.
16719 * config/arm/arm-cpu.h: Delete.
16720 * config/arm/arm-cpu-cdata.h: Delete.
16721 * config/arm/arm-cpu-data.h: Delete.
16723 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
16725 * config/arm/arm-cpus.in (cortex-a55): New.
16726 (cortex-a75): Likewise.
16727 (cortex-a75.cortex-a55): Likewise.
16728 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
16730 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
16731 * config/arm/arm-cpu-cdata.h: Regenerate.
16732 * config/arm/arm-cpu-data.h: Regenerate.
16733 * config/arm/arm-cpu.h: Regenerate.
16734 * config/arm/arm-tables.opt: Regenerate.
16735 * config/arm/arm-tune.md: Regenerate.
16737 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
16739 * haifa-sched.c (sched_create_recovery_edges): Update profile.
16741 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
16743 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
16746 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
16748 PR tree-optimization/81292
16749 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
16750 full_string_p, also call adjust_related_strinfos if the adjustment
16751 is simple, otherwise invalidate related strinfos.
16753 2017-07-04 Martin Liska <mliska@suse.cz>
16756 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
16757 newly created variable as DECL_IGNORED_P.
16759 2017-07-04 Martin Liska <mliska@suse.cz>
16762 * ipa-inline.c (inline_small_functions):
16763 Use xstrdup_for_dump.
16765 2017-07-04 Tom de Vries <tom@codesourcery.com>
16767 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
16769 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
16772 * config/darwin.c (darwin_function_switched_text_sections):
16775 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
16777 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
16779 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
16781 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
16783 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16785 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
16786 min_profitable_iters, and th as inclusive lower bounds.
16787 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
16788 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
16789 for min_profitable_iters and min_profitable_estimate.
16790 (vect_transform_loop): Treat th as an inclusive lower bound.
16791 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
16793 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
16796 * config/darwin.c (darwin_function_switched_text_sections):
16797 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
16798 in two pieces, and suppress the use of buf.
16800 2017-07-03 Nathan Sidwell <nathan@acm.org>
16802 * hash-table.h (hash_table_mod1): Fix indentation.
16804 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16806 PR middle-end/81290
16807 * predict.c (force_edge_cold): Be more careful about propagation
16809 * profile-count.h (profile_probability::guessed,
16810 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
16812 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
16814 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
16816 * doc/invoke.texi (rcpc architecture extension): Document it.
16818 2017-07-03 Richard Biener <rguenther@suse.de>
16820 PR tree-optimization/60510
16821 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
16822 the scalar reduction PHI and use it.
16823 (vectorizable_reduction): Properly guard the single_defuse_cycle
16824 path for non-SLP reduction chains where we cannot use it.
16825 Rework reduc_def/index and vector type deduction. Rework
16826 vector operand gathering during reduction op code-gen.
16827 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
16828 chains dissolve the chain and leave it to non-SLP reduction
16831 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16833 * tree-data-ref.h (dr_alignment): Declare.
16834 * tree-data-ref.c (dr_alignment): New function.
16835 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
16836 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
16838 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
16840 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16842 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
16843 and base_misalignment fields.
16844 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
16845 * tree-data-ref.c: Include builtins.h.
16846 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
16847 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
16848 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
16849 * tree-vect-data-refs.c: Include tree-cfg.h.
16850 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
16851 fields instead of calculating an alignment here.
16852 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
16853 innermost_loop_behavior fields.
16855 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16857 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
16859 (DR_STEP_ALIGNMENT): New macro.
16860 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
16861 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
16862 (create_data_ref): Print it.
16863 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
16864 to tell whether the step preserves vector (mis)alignment.
16865 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
16866 Move the check for an integer step and generalise to all INTEGER_CST.
16867 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
16868 Print the outer step alignment.
16870 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16872 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
16873 with offset_alignment.
16874 (DR_ALIGNED_TO): Delete.
16875 (DR_OFFSET_ALIGNMENT): New macro.
16876 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
16877 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
16878 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
16879 (create_data_ref): Likewise.
16880 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
16881 (vect_analyze_data_refs): Likewise.
16882 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
16883 creating dummy innermost behavior.
16885 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16887 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
16888 with a "innermost_loop_behavior *" and refeence tree.
16889 * tree-data-ref.c (dr_analyze_innermost): Likewise.
16890 (create_data_ref): Update call accordingly.
16891 * tree-predcom.c (find_looparound_phi): Likewise.
16893 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16895 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
16896 fields with dr_wrt_vec_loop.
16897 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
16898 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
16899 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
16900 (vect_dr_behavior): New function.
16901 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
16902 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
16903 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
16904 track whether the step preserves the misalignment.
16905 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
16906 Use vect_dr_behavior.
16907 (vect_setup_realignment): Update call accordingly.
16908 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
16909 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
16910 call to vect_create_addr_base_for_vector_ref.
16911 (vect_create_cond_for_align_checks): Likewise.
16912 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
16913 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
16914 (vect_recog_mask_conversion_pattern): Likewise.
16915 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
16916 (new_stmt_vec_info): Remove redundant zeroing.
16918 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
16920 * common/config/arm/arm-common.c (arm_be8_option): New function.
16921 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
16922 (ISA_ARMv6): Add isa_bit_be8.
16923 * config/arm/arm.h (arm_be8_option): Add prototype.
16924 (BE8_SPEC_FUNCTION): New define.
16925 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
16926 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
16927 (mlittle-endian): Similarly.
16928 (mbe8, mbe32): New options.
16929 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
16930 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
16932 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16934 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
16936 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16938 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
16939 (cleanup_tree_cfg_bb): Use it.
16940 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
16942 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
16944 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16947 * loop-doloop.c (add_test): Update profile.
16949 2017-07-03 Martin Liska <mliska@suse.cz>
16952 * sanopt.c (rewrite_usage_of_param): New function.
16953 (sanitize_rewrite_addressable_params): Likewise.
16954 (pass_sanopt::execute): Call rewrite_usage_of_param.
16956 2017-07-03 Richard Biener <rguenther@suse.de>
16958 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
16959 back to using VIEW_CONVERT_EXPR.
16961 2017-07-03 Martin Liska <mliska@suse.cz>
16964 * doc/extend.texi: Document when a resolver function is
16965 generated for target_clones.
16967 2017-07-03 Martin Liska <mliska@suse.cz>
16969 * asan.c (asan_emit_stack_protection): Unpoison just red zones
16970 and shadow memory of auto variables which are subject of
16971 use-after-scope sanitization.
16972 (asan_expand_mark_ifn): Add do set only when is_poison.
16974 2016-07-03 Richard Biener <rguenther@suse.de>
16976 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
16978 (vect_force_simple_reduction): Record reduction def -> phi mapping.
16979 (vectorizable_reduction): Perform reduction PHI creation when
16980 visiting a reduction PHI and adjust and simplify code generation
16981 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
16982 (vect_transform_loop): Visit reduction PHIs.
16983 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
16984 defs into the SLP tree.
16985 (vect_build_slp_tree): Reduction defs terminate the recursion.
16986 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
16988 (vect_get_vec_defs_for_stmt_copy): Export.
16989 (vect_get_vec_defs): Likewise.
16990 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
16992 (vect_get_vec_defs_for_stmt_copy): Declare.
16993 (vect_get_vec_defs): Likewise.
16995 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16997 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
16998 parameter with a "loop" parameter and use it instead of the
16999 loop containing DR_STMT. Don't check simple_iv when doing
17000 BB analysis. Describe the two analysis modes in the comment.
17002 2017-07-03 Tom de Vries <tom@codesourcery.com>
17004 PR tree-optimization/69468
17005 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
17006 (find_same_succ_bb): Handle ignore_edge_flags.
17008 2017-07-03 Tom de Vries <tom@codesourcery.com>
17010 PR tree-optimization/81192
17011 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
17013 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
17015 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
17017 2017-07-03 Tom de Vries <tom@codesourcery.com>
17019 PR tree-optimization/81192
17020 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
17021 BB_SAME_SUCC (bb) == NULL.
17023 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17025 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
17028 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17030 * dumpfile.c: Include profile-count.h
17031 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
17033 (insert_cond_bb): Update profile.
17034 * tree-cfg.h (insert_cond_bb): Update prototype.
17035 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
17036 * tree-dump.c: Do not include tree-cfg.
17038 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17040 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
17042 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17044 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
17047 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17049 * tree-complex.c (expand_complex_div_wide): update profile.
17051 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17052 Alan Hayward <alan.hayward@arm.com>
17053 David Sherwood <david.sherwood@arm.com>
17055 * Makefile.in (MACHMODE_H): Remove insn-modes.h
17056 (CORETYPES_H): New define.
17057 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
17058 (insn-modes-inline.h, s-modes-inline-h): New rules.
17059 (generated_files): Add insn-modes-inline.h.
17060 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
17061 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
17062 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
17063 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
17064 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
17065 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
17066 (build/gencodes.o, build/genconditions.o): Likewise.
17067 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
17068 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
17069 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
17070 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
17071 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
17072 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
17073 * coretypes.h: Include everything up to real.h for generators.
17074 Include insn-modes.h first. Include wide-int-print.h after
17075 wide-int.h. Include insn-modes-inline.h and then machmode.h.
17076 * machmode.h: Don't include insn-modes.h here.
17077 * function-tests.c: Remove includes of signop.h, machmode.h,
17078 double-int.h and wide-int.h.
17080 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
17082 * optc-save-gen.awk: Likewise.
17083 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
17084 * godump.c: Remove include of wide-int-print.h.
17085 * pretty-print.h: Likewise.
17086 * wide-int-print.cc: Likewise.
17087 * wide-int.cc: Likewise.
17088 * hash-map-tests.c: Remove include of signop.h.
17089 * hash-set-tests.c: Likewise.
17090 * rtl-tests.c: Likewise.
17091 * mkconfig.sh: Remove include of machmode.h.
17092 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
17094 (emit_insn_modes_inline_h): ...this new function. Emit the code
17095 into an insn-modes-inline.h header file, adding appropriate
17096 include guards and end comments.
17097 (emit_insn_modes_c_header): Remove include of machmode.h.
17098 (emit_min_insn_modes_c_header): Include coretypes.h rather than
17100 (main): Handle -i flag and call emit_insn_modes_inline_h when
17103 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17105 * tree-ssa-strlen.c (strinfo): Rename the length field to
17106 nonzero_chars. Add a full_string_p field.
17107 (compare_nonzero_chars, zero_length_string_p): New functions.
17108 (get_addr_stridx): Add an offset_out parameter.
17109 Use compare_nonzero_chars.
17110 (get_stridx): Update accordingly. Use compare_nonzero_chars.
17111 (new_strinfo): Update after above changes to strinfo.
17112 (set_endptr_and_length): Set full_string_p.
17113 (get_string_length): Update after above changes to strinfo.
17114 (unshare_strinfo): Update call to new_strinfo.
17115 (maybe_invalidate): Likewise.
17116 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
17117 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
17118 as a uhwi instead of an shwi. Update after above changes to
17119 strinfo and new_strinfo.
17120 (zero_length_string): Assert that chainsi contains full strings.
17121 Use zero_length_string_p. Update call to new_strinfo.
17122 (adjust_related_strinfos): Update after above changes to strinfo.
17123 Copy full_string_p from origsi.
17124 (adjust_last_stmt): Use zero_length_string_p.
17125 (handle_builtin_strlen): Update after above changes to strinfo and
17126 new_strinfo. Install the lhs as the string length if the previous
17127 entry didn't describe a full string.
17128 (handle_builtin_strchr): Update after above changes to strinfo
17130 (handle_builtin_strcpy): Likewise.
17131 (handle_builtin_strcat): Likewise.
17132 (handle_builtin_malloc): Likewise.
17133 (handle_pointer_plus): Likewise.
17134 (handle_builtin_memcpy): Likewise. Track nonzero characters
17135 that aren't necessarily followed by a nul terminator.
17136 (handle_char_store): Likewise.
17138 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17140 PR tree-optimization/80769
17141 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
17142 for malloc and calloc. Document the new invariant that all related
17143 strinfos have delayed lengths or none do.
17144 (verify_related_strinfos): Move earlier in file.
17145 (set_endptr_and_length): New function, split out from...
17146 (get_string_length): ...here. Also set the lengths of related
17148 (zero_length_string): Assert that chainsi has known (rather than
17150 (adjust_related_strinfos): Likewise.
17152 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17154 PR tree-optimization/81136
17155 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
17156 assert that two references with the same misalignment have the same
17157 compile-time misalignment if those compile-time misalignments
17160 2017-07-01 Andi Kleen <ak@linux.intel.com>
17162 * print-tree.c (print_node): Print all attributes.
17164 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17166 * cfg.c (scale_bbs_frequencies): New function.
17167 * cfg.h (scale_bbs_frequencies): Declare it.
17168 * cfgloopanal.c (single_likely_exit): Cleanup.
17169 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
17171 (scale_loop_profile): Likewise.
17172 (loop_version): Likewise.
17173 (create_empty_loop_on_edge): Update.
17174 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
17175 scale_loop_frequencies, scale_loop_profile, loopify,
17176 loop_version): Update prototypes.
17177 * modulo-sched.c (sms_schedule): Update.
17178 * predict.c (unlikely_executed_edge_p): Also check probability.
17179 (probably_never_executed_edge_p): Fix typo.
17180 * tree-if-conv.c (version_loop_for_if_conversion): Update.
17181 * tree-parloops.c (gen_parallel_loop): Update.
17182 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
17183 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
17184 * tree-ssa-loop-split.c (split_loop): Update.
17185 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
17186 * tree-vect-loop-manip.c (vect_do_peeling): Update.
17187 (vect_loop_versioning): Update.
17188 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
17190 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17192 * trans-mem.c (split_bb_make_tm_edge): Update profile.
17194 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17196 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
17197 to keep profile consistent.
17199 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17201 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
17202 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
17203 * profile-count.h (max_safe_multiplier): Make unsigned.
17204 (profile_count::guessed_zero): New.
17206 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17208 * bb-reorder.c (fix_up_crossing_landing_pad,
17209 fix_crossing_conditional_branches): Use make_single_succ_edge
17210 to keep profile consistent.
17212 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17214 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
17217 2017-07-01 Jakub Jelinek <jakub@redhat.com>
17220 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
17221 the right scopes, make sure cond_jump isn't preserved between multiple
17222 iterations. Search for fallthru edge whenever there are 3+ edges and
17223 use find_fallthru_edge for it.
17225 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17227 Patch by Alexander Monakov <amonakov@ispras.ru>
17228 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
17229 probabilities consistently.
17231 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17233 * pa.c (pa_expand_compare_and_swap_loop): Update call of
17234 emit_cmp_and_jump_insns.
17236 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17239 * tree-inline.c (expand_call_inline): Combine profile statuses.
17241 2017-06-30 Andrew Pinski <apinski@cavium.com>
17243 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
17244 fold_stmt returned true.
17246 2017-06-30 Nathan Sidwell <nathan@acm.org>
17248 * ggc.h (empty_string): Delete.
17249 * cfgexpand.c (expand_asm_stmt): Use plain "".
17250 * optabs.c (expand_asm_memory_barrier): Likewise.
17251 * stringpool.c (empty_string): Delete.
17252 (digit_vector, digit_string): Delete.
17253 (ggc_alloc_string): Use plain "", don't optimize single digit
17254 strings. Use ggc_alloc_atomic.
17256 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
17258 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
17259 comparison set and one other set, use the cost of the non-comparison
17262 2017-06-30 Nathan Sidwell <nathan@acm.org>
17264 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
17267 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
17269 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
17270 loops. Remove now unneeded calls to gimple_switch_set_label() that
17271 just set removed labels to NULL_TREE.
17273 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
17275 * tree-ssanames.c (set_range_info_raw): Abstract from ...
17276 (set_range_info): ...here. Only call set_range_info_raw if domain
17278 (set_nonzero_bits): Call set_range_info_raw.
17279 * tree-ssanames.h (set_range_info_raw): New.
17281 2017-06-30 Jakub Jelinek <jakub@redhat.com>
17284 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
17285 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
17286 of nonimmediate_operand and <store_mask_constraint> instead of m
17287 for the input operand. For V8FI iterator, always split if input
17288 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
17289 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
17290 <store_mask_predicate> instead of register_operand and
17291 <store_mask_constraint> instead of v for the input operand. Make
17292 sure both operands aren't MEMs for if not <mask_applied>.
17294 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
17296 * lto-wrapper.c (copy_file) Close both file descriptors before
17299 2017-06-30 Martin Liska <mliska@suse.cz>
17302 * multiple_target.c (create_dispatcher_calls): Make ifunc
17303 also for function that don't have calls or are not referenced.
17305 2017-06-30 Richard Biener <rguenther@suse.de>
17307 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
17308 analyze the first scalar stmt. Move vector type computation
17309 for the BB case here from ...
17310 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
17311 live operation processing in the SLP case properly.
17313 2017-06-30 Richard Biener <rguenther@suse.de>
17315 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
17317 2017-06-30 Martin Liska <mliska@suse.cz>
17320 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
17321 before BUILT_IN_UNWIND_RESUME when ASAN is used.
17323 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
17325 * doc/invoke.texi (AArch64): Add missing options and remove redundant
17328 2017-06-30 Richard Biener <rguenther@suse.de>
17330 PR tree-optimization/81249
17331 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
17332 condition reduction result to original scalar type.
17334 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17336 * profile-count.h (enum profile_quality): Fix typos and whitespace
17339 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17341 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
17342 type for branch probabilities.
17344 2017-06-29 Julian Brown <julian@codesourcery.com>
17345 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17347 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
17348 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
17349 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
17350 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
17352 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17354 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
17355 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
17356 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
17357 CC usage from generic code to here.
17358 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
17359 CC usage into the target macros.
17361 2017-06-29 Maya Rashish <coypu@sdf.org>
17363 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
17366 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17368 * arm/arm-builtins.c: Include profile-count.h
17369 * except.c (sjlj_emit_function_enter): Use
17370 profile_probability::unlikely.
17372 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17374 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
17375 and tocrel_offset be pointer args rather than implicitly using
17377 (legitimate_constant_pool_address_p, rs6000_emit_move,
17378 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
17379 tocrel_offset and use in toc_relative_expr_p call.
17380 (print_operand, print_operand_address): Use static tocrel_base_oac
17381 and tocrel_offset_oac.
17382 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
17385 2017-06-29 Maya Rashish <coypu@sdf.org>
17387 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
17389 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
17391 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
17392 objects, take into account only the alignment of 'op0' and 'mode1' if
17395 2017-06-29 Steve Ellcey <sellcey@cavium.com>
17397 * ccmp.c (ccmp_tree_comparison_p): New function.
17398 (ccmp_candidate_p): Update to use above function.
17399 (get_compare_parts): New function.
17400 (expand_ccmp_next): Update to use new functions.
17401 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
17403 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
17404 take mode as argument.
17405 * ccmp.h (expand_ccmp_expr): Add mode as argument.
17406 * expr.c (expand_expr_real_1): Pass mode as argument.
17408 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
17410 * combine.c (combine_instructions): Print insns to dump_file, together
17413 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17415 * asan.c (asan_emit_stack_protection): Update.
17416 (create_cond_insert_point): Update.
17417 * auto-profile.c (afdo_propagate_circuit): Update.
17418 * basic-block.h (struct edge_def): Turn probability to
17419 profile_probability.
17420 (EDGE_FREQUENCY): Update.
17421 * bb-reorder.c (find_traces_1_round): Update.
17422 (better_edge_p): Update.
17423 (sanitize_hot_paths): Update.
17424 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
17425 (make_single_succ_edge): Update.
17426 (check_bb_profile): Update.
17427 (dump_edge_info): Update.
17428 (update_bb_profile_for_threading): Update.
17429 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
17430 probabilitycount to 0.
17431 * cfgbuild.c (compute_outgoing_frequencies): Update.
17432 * cfgcleanup.c (try_forward_edges): Update.
17433 (outgoing_edges_match): Update.
17434 (try_crossjump_to_edge): Update.
17435 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
17436 (expand_gimple_tailcall): Update.
17437 (construct_init_block): Use make_single_succ_edge.
17438 (construct_exit_block): Use make_single_succ_edge.
17439 * cfghooks.c (verify_flow_info): Update.
17440 (redirect_edge_succ_nodup): Update.
17441 (split_edge): Update.
17442 (account_profile_record): Update.
17443 * cfgloopanal.c (single_likely_exit): Update.
17444 * cfgloopmanip.c (scale_loop_profile): Update.
17445 (set_zero_probability): Remove.
17446 (duplicate_loop_to_header_edge): Update.
17447 * cfgloopmanip.h (loop_version): Update prototype.
17448 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
17449 (force_nonfallthru_and_redirect): Update.
17450 (update_br_prob_note): Update.
17451 (rtl_verify_edges): Update.
17452 (purge_dead_edges): Update.
17453 (rtl_lv_add_condition_to_bb): Update.
17454 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
17455 * cgraphunit.c (init_lowered_empty_function): Update.
17456 (cgraph_node::expand_thunk): Update.
17457 * cilk-common.c: Include profile-count.h
17458 * dojump.c (inv): Remove.
17459 (jumpifnot): Update.
17460 (jumpifnot_1): Update.
17461 (do_jump_1): Update.
17463 (do_jump_by_parts_greater_rtx): Update.
17464 (do_compare_rtx_and_jump): Update.
17465 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
17466 do_jump_1. do_compare_rtx_and_jump): Update prototype.
17467 * dwarf2cfi.c: Include profile-count.h
17468 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
17469 (sjlj_emit_dispatch_table): Likewise.
17470 * explow.c: Include profile-count.h
17471 * expmed.c (emit_store_flag_force): Update.
17472 (do_cmp_and_jump): Update.
17473 * expr.c (compare_by_pieces_d::generate): Update.
17474 (compare_by_pieces_d::finish_mode): Update.
17475 (emit_block_move_via_loop): Update.
17476 (store_expr_with_bounds): Update.
17477 (store_constructor): Update.
17478 (expand_expr_real_2): Update.
17479 (expand_expr_real_1): Update.
17480 * expr.h (try_casesi, try_tablejump): Update prototypes.
17481 * gimple-pretty-print.c (dump_probability): Update.
17482 (dump_profile): New.
17483 (dump_gimple_label): Update.
17484 (dump_gimple_bb_header): Update.
17485 * graph.c (draw_cfg_node_succ_edges): Update.
17486 * hsa-gen.c (convert_switch_statements): Update.
17487 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
17488 (find_if_case_1): Update.
17489 (find_if_case_2): Update.
17490 * internal-fn.c (expand_arith_overflow_result_store): Update.
17491 (expand_addsub_overflow): Update.
17492 (expand_neg_overflow): Update.
17493 (expand_mul_overflow): Update.
17494 (expand_vector_ubsan_overflow): Update.
17495 * ipa-cp.c (good_cloning_opportunity_p): Update.
17496 * ipa-split.c (split_function): Use make_single_succ_edge.
17497 * ipa-utils.c (ipa_merge_profiles): Update.
17498 * loop-doloop.c (add_test): Update.
17499 (doloop_modify): Update.
17500 * loop-unroll.c (compare_and_jump_seq): Update.
17501 (unroll_loop_runtime_iterations): Update.
17502 * lra-constraints.c (lra_inheritance): Update.
17503 * lto-streamer-in.c (input_cfg): Update.
17504 * lto-streamer-out.c (output_cfg): Update.
17505 * mcf.c (adjust_cfg_counts): Update.
17506 * modulo-sched.c (sms_schedule): Update.
17507 * omp-expand.c (expand_omp_for_init_counts): Update.
17508 (extract_omp_for_update_vars): Update.
17509 (expand_omp_ordered_sink): Update.
17510 (expand_omp_for_ordered_loops): Update.
17511 (expand_omp_for_generic): Update.
17512 (expand_omp_for_static_nochunk): Update.
17513 (expand_omp_for_static_chunk): Update.
17514 (expand_cilk_for): Update.
17515 (expand_omp_simd): Update.
17516 (expand_omp_taskloop_for_outer): Update.
17517 (expand_omp_taskloop_for_inner): Update.
17518 * omp-simd-clone.c (simd_clone_adjust): Update.
17519 * optabs.c (expand_doubleword_shift): Update.
17520 (expand_abs): Update.
17521 (emit_cmp_and_jump_insn_1): Update.
17522 (expand_compare_and_swap_loop): Update.
17523 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
17524 * predict.c (predictable_edge_p): Update.
17525 (edge_probability_reliable_p): Update.
17526 (set_even_probabilities): Update.
17527 (combine_predictions_for_insn): Update.
17528 (combine_predictions_for_bb): Update.
17529 (propagate_freq): Update.
17530 (estimate_bb_frequencies): Update.
17531 (force_edge_cold): Update.
17532 * profile-count.c (profile_count::dump): Add missing space into dump.
17533 (profile_count::debug): Add newline.
17534 (profile_count::differs_from_p): Explicitly convert to unsigned.
17535 (profile_count::stream_in): Update.
17536 (profile_probability::dump): New member function.
17537 (profile_probability::debug): New member function.
17538 (profile_probability::differs_from_p): New member function.
17539 (profile_probability::differs_lot_from_p): New member function.
17540 (profile_probability::stream_in): New member function.
17541 (profile_probability::stream_out): New member function.
17542 * profile-count.h (profile_count_quality): Rename to ...
17543 (profile_quality): ... this one.
17544 (profile_probability): New.
17545 (profile_count): Update.
17546 * profile.c (compute_branch_probabilities): Update.
17547 * recog.c (peep2_attempt): Update.
17548 * sched-ebb.c (schedule_ebbs): Update.
17549 * sched-rgn.c (find_single_block_region): Update.
17550 (compute_dom_prob_ps): Update.
17551 (schedule_region): Update.
17552 * sel-sched-ir.c (compute_succs_info): Update.
17553 * stmt.c (struct case_node): Update.
17554 (do_jump_if_equal): Update.
17555 (get_outgoing_edge_probs): Update.
17556 (conditional_probability): Update.
17557 (emit_case_dispatch_table): Update.
17558 (expand_case): Update.
17559 (expand_sjlj_dispatch_table): Update.
17560 (emit_case_nodes): Update.
17561 * targhooks.c: Update.
17562 * tracer.c (better_p): Update.
17563 (find_best_successor): Update.
17564 * trans-mem.c (expand_transaction): Update.
17565 * tree-call-cdce.c: Update.
17566 * tree-cfg.c (gimple_split_edge): Upate.
17567 (move_sese_region_to_fn): Upate.
17568 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
17569 * tree-eh.c (lower_resx): Upate.
17570 (cleanup_empty_eh_move_lp): Upate.
17571 * tree-if-conv.c (version_loop_for_if_conversion): Update.
17572 * tree-inline.c (copy_edges_for_bb): Update.
17573 (copy_cfg_body): Update.
17574 * tree-parloops.c (gen_parallel_loop): Update.
17575 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
17576 (gimple_gen_time_profiler): Update.
17577 * tree-ssa-dce.c (remove_dead_stmt): Update.
17578 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
17579 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
17580 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
17581 (unloop_loops): Update.
17582 (try_peel_loop): Update.
17583 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
17584 * tree-ssa-loop-split.c (connect_loops): Update.
17585 (split_loop): Update.
17586 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
17587 (hoist_guard): Update.
17588 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
17589 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
17590 (value_replacement): Update.
17591 * tree-ssa-reassoc.c (branch_fixup): Update.
17592 * tree-ssa-tail-merge.c (replace_block_by): Update.
17593 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
17594 (create_edge_and_update_destination_phis): Update.
17595 (compute_path_counts): Update.
17596 (recompute_probabilities): Update.
17597 (update_joiner_offpath_counts): Update.
17598 (freqs_to_counts_path): Update.
17599 (duplicate_thread_path): Update.
17600 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
17601 (struct switch_conv_info): Update.
17602 (gen_inbound_check): Update.
17603 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
17604 (vect_do_peeling): Update.
17605 (vect_loop_versioning): Update.
17606 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
17607 (optimize_mask_stores): Update.
17608 * ubsan.c (ubsan_expand_null_ifn): Update.
17609 * value-prof.c (gimple_divmod_fixed_value): Update.
17610 (gimple_divmod_fixed_value_transform): Update.
17611 (gimple_mod_pow2): Update.
17612 (gimple_mod_pow2_value_transform): Update.
17613 (gimple_mod_subtract): Update.
17614 (gimple_mod_subtract_transform): Update.
17615 (gimple_ic): Update.
17616 (gimple_stringop_fixed_value): Update.
17617 (gimple_stringops_transform): Update.
17618 * value-prof.h: Update.
17620 2017-06-29 Carl Love <cel@us.ibm.com>
17622 * config/rs6000/rs6000-c.c: Add support for built-in functions
17623 vector signed int vec_signed (vector float);
17624 vector signed long long vec_signed (vector double);
17625 vector signed int vec_signed2 (vector double, vector double);
17626 vector signed int vec_signede (vector double);
17627 vector signed int vec_signedo (vector double);
17628 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
17629 instruction generator.
17630 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
17631 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
17632 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
17634 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
17635 vunsignede_v2df): Add define_expands.
17636 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
17637 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
17638 VEC_UNSIGNEDO): Add definitions.
17639 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
17640 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
17641 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
17642 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
17643 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
17644 * config/rs6000/altivec.h (vec_signed, vec_signed2,
17645 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
17646 vec_unsignede, vec_unsignedo): Add builtin defines.
17647 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
17649 * doc/extend.texi: Update the built-in documentation file for the
17650 new built-in functions.
17652 2017-06-29 Richard Biener <rguenther@suse.de>
17654 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
17655 reduction chains to LOOP_VINFO_REDUCTIONS.
17656 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
17657 SLP reductions after processing reduction chains.
17659 2017-06-29 Nathan Sidwell <nathan@acm.org>
17661 * builtins.c (fold_builtin_FUNCTION): Use
17662 lang_hooks.decl_printable_name.
17664 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
17666 PR middle-end/81194
17667 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
17668 with only one label.
17669 * stmt.c (expand_case): Assert NCASES is greater than one.
17671 2017-06-29 Richard Biener <rguenther@suse.de>
17673 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
17675 (group_case_labels): Likewise.
17676 (find_taken_edge): Push sanity checking on val to workers...
17677 (find_taken_edge_cond_expr): ... here
17678 (find_taken_edge_switch_expr): ... and here, handle cases
17679 with just a default label.
17680 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
17681 (group_case_labels): Likewise.
17682 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
17683 group_case_labels does anything cleanup the CFG again.
17685 2017-06-29 Bin Cheng <bin.cheng@arm.com>
17687 PR tree-optimization/81196
17688 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
17689 exit condition comparing two IVs.
17691 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
17693 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
17694 profile to the dummy entry at the end of the list of architectures.
17695 * config/arm/arm-cpu-cdata.h: Regenerated.
17697 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17698 Michael Collison <michael.collison@arm.com>
17701 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
17703 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
17704 (*aarch64_reg_<mode>3_minus_mask): New pattern.
17705 (*aarch64_<optab>_reg_di3_mask2): New pattern.
17706 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
17707 of shift when the shift amount is masked with constant equal to
17708 the size of the mode.
17709 * config/aarch64/predicates.md (subreg_lowpart_operator): New
17712 2017-06-29 Martin Liska <mliska@suse.cz>
17714 * config/i386/i386.opt: Change range from [1,5] to [0,5].
17716 2017-06-29 Yury Gribov <tetra2005@gmail.com>
17719 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
17722 (edge_growth_cache_entry::edge_growth_cache_entry): New
17724 (reset_edge_growth_cache): Update to use constructor.
17726 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17728 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
17729 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
17730 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
17732 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
17734 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
17735 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
17737 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
17739 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
17740 (*-linux-uclibc*): Add t-uclibc tmake_file.
17741 * config/t-musl: New.
17742 * config/t-uclibc: New.
17744 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
17746 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
17748 (gen_comm_data): Emit architectural setting of arch_prof.
17749 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
17751 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
17752 (armv8-m.base, armv8-m.main): Likewise.
17753 * arm-protos.h (arm_build_target): Add profile field.
17754 (arch_option): Likewise.
17755 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
17757 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
17758 arm_active_target.profile.
17760 2017-06-28 Richard Biener <rguenther@suse.de>
17762 PR middle-end/81227
17763 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
17764 TYPE_OVERFLOW_WRAPS.
17765 * match.pd (negate_expr_p): Likewise.
17766 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
17767 fold_build2, not fold_binary.
17769 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
17771 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
17772 Convert memory address to Pmode.
17773 (aarch64_print_operand): Assert MEM operands are always Pmode.
17775 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
17778 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
17779 Remove redundant if.
17780 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
17781 * config/arm/aarch-common-protos.h
17782 (aarch_forward_to_shift_is_not_shifted_re): Remove.
17783 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
17785 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
17788 * multiple_target.c (create_dispatcher_calls): Set the default
17789 clone to be static, not public.
17791 2017-06-28 Richard Biener <rguenther@suse.de>
17793 * tree-vect-loop.c (vectorizable_reduction): Move special
17794 cond reduction IV var creation ...
17795 (vect_create_epilog_for_reduction): ... here. Remove induction_index
17796 parameter. Use STMT_VINFO_VECTYPE.
17797 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
17800 2017-06-28 Martin Liska <mliska@suse.cz>
17803 * ipa-visibility.c (non_local_p): Handle visibility.
17805 2017-06-28 Martin Liska <mliska@suse.cz>
17808 * common.opt: Add IntegerRange to various options.
17809 * opt-functions.awk (integer_range_info): New function.
17810 * optc-gen.awk: Add integer_range_info to cl_options struct.
17811 * opts-common.c (decode_cmdline_option): Handle
17812 CL_ERR_INT_RANGE_ARG.
17813 (cmdline_handle_error): Likewise.
17814 * opts.c (print_filtered_help): Show valid interval in
17815 when --help is provided.
17816 * opts.h (struct cl_option): Add range_min and range_max fields.
17817 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
17819 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
17821 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
17822 (x * C EQ/NE y * C): New transformation.
17824 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
17826 * genmultilib (combination_space): Accept '+' in option names.
17828 2017-06-28 Martin Liska <mliska@suse.cz>
17831 * asan.c (instrument_derefs): Bail out inner references
17832 that are hard register variables.
17834 2017-06-28 Jakub Jelinek <jakub@redhat.com>
17837 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
17838 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
17840 2017-06-28 Richard Biener <rguenther@suse.de>
17842 * tree-vectorizer.h (vect_get_vec_defs): Remove.
17843 (vect_get_slp_defs): Adjust.
17844 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
17846 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
17848 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
17849 get_initial_defs_for_reduction instead of vect_get_vec_defs.
17850 (vectorizable_reduction): Adjust.
17851 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
17853 (vect_get_slp_defs): Likewise.
17854 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
17855 (vectorizable_bswap): Adjust.
17856 (vectorizable_call): Likewise.
17857 (vectorizable_conversion): Likewise.
17858 (vectorizable_assignment): Likewise.
17859 (vectorizable_shift): Likewise.
17860 (vectorizable_operation): Likewise.
17861 (vectorizable_store): Likewise.
17862 (vectorizable_condition): Likewise.
17863 (vectorizable_comparison): Likewise.
17865 2017-06-28 Michael Collison <michael.collison@arm.com>
17868 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
17870 (arm_gen_movmemqi): Removed unused variable 'i'.
17871 Convert 'for' loop into 'while' loop.
17872 (arm_expand_prologue): Remove last unnecessary set of insn.
17873 (thumb_pop): Remove unused variable 'pushed_words'.
17874 (thumb_exit): Remove last unnecessary set of regs_to_pop.
17876 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17878 * config/s390/predicates.md: Use s390_rel_address_ok_p.
17879 * config/s390/s390-protos.h: Add prototype of
17880 s390_rel_address_ok_p.
17881 * config/s390/s390.c (s390_got_symbol): New function.
17882 (s390_rel_address_ok_p): New function.
17883 (legitimize_pic_address): Use s390_rel_address_ok_p.
17884 (s390_load_got): Use s390_got_symbol.
17885 (s390_option_override): Issue error if
17886 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
17887 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
17889 * config/s390/s390.opt: New option mpic-data-is-text-relative.
17891 2017-06-27 Andrew Pinski <apinski@cavium.com>
17893 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
17894 (X * copysign (1.0, X)): New pattern.
17895 (X * copysign (1.0, -X)): New pattern.
17896 (copysign (-1.0, CST)): New pattern.
17898 2017-06-27 Joseph Myers <joseph@codesourcery.com>
17900 * genmultilib (combination_space): Remove variable.
17901 Validate reuse rules against regular expression for any sequence
17902 of multilib options in any order.
17904 2017-06-27 Michael Collison <michael.collison@arm.com>
17906 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
17907 call aarch64_split_simd_combine.
17908 * (aarch64_combine_internal<mode>): Delete pattern.
17909 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
17910 Allow register and subreg operands.
17912 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
17914 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
17915 specific need, just fallback on defaults.
17916 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
17918 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
17919 Olivier Hainque <hainque@adacore.com>
17921 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
17923 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
17924 targets. Pick a default if no particular attempt applied.
17925 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
17928 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
17930 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
17931 (x86_64-wrs-vxworks7): Likewise.
17933 2017-06-27 Marek Polacek <polacek@redhat.com>
17936 * ubsan.c (instrument_null): Check get_base_address's result for null.
17938 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
17940 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
17942 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
17944 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
17945 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
17946 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
17947 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
17948 New function types.
17949 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
17950 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
17951 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
17952 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
17953 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
17954 BUILT_IN_FEUPDATEENV): New builtins.
17955 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
17956 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
17957 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
17958 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
17960 (builtin_structptr_types): Adjust size.
17961 * tree.c (builtin_structptr_types): Add four entries.
17963 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
17964 Olivier Hainque <hainque@adacore.com>
17966 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
17967 (TLS_SYM): New local macro, forcing reference to __tls__ on
17968 link command lines for VxWorks 7 RTPs, triggering initialization
17970 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
17971 OS features TLS support, true for RTPs on VxWorks 7.
17972 * config/vxworks.c (vxworks_override_options): Setup emutls
17975 2017-06-27 Jakub Jelinek <jakub@redhat.com>
17977 * predict.c (test_prediction_value_range): Use -1U instead of -1
17978 to avoid narrowing conversion warning.
17979 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
17980 to avoid narrowing conversion warning.
17981 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
17983 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
17985 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
17987 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
17988 64bit configurations.
17989 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
17990 (SIZE_TYPE): Likewise.
17991 * config/vxworks.c (vxworks_emutls_var_fields): Use
17992 long_unsigned_type_node instead of unsigned_type_node as the offset
17993 field type, which is "pointer" mode in emutls.c.
17995 2017-06-27 Jakub Jelinek <jakub@redhat.com>
17998 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
18000 PR middle-end/81207
18001 * gimple-fold.c (replace_call_with_call_and_fold): Handle
18002 gimple_vuse copying separately from gimple_vdef copying.
18004 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18006 * value-prof.c (free_hist): Remove call to memset and the enclosing if
18009 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18010 Olivier Hainque <hainque@adacore.com>
18012 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
18013 for all vxworks7 targets.
18014 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
18015 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
18016 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
18017 variations for VX6/VX7 and 32/64bits later on in ...
18018 (VXWORKS_LIB_SPEC): Leverage new macros.
18019 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
18020 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
18022 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18024 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
18025 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
18027 2017-06-26 Carl Love <cel@us.ibm.com>
18029 * config/rs6000/rs6000-c.c: Add support for built-in functions
18030 vector bool char vec_reve (vector bool char);
18031 vector signed char vec_reve (vector signed char);
18032 vector unsigned char vec_reve (vector unsigned char);
18033 vector bool int vec_reve (vector bool int);
18034 vector signed int vec_reve (vector signed int);
18035 vector unsigned int vec_reve (vector unsigned int);
18036 vector bool long long vec_reve (vector bool long long);
18037 vector signed long long vec_reve (vector signed long long);
18038 vector unsigned long long vec_reve (vector unsigned long long);
18039 vector bool short vec_reve (vector bool short);
18040 vector signed short vec_reve (vector signed short);
18041 vector double vec_reve (vector double);
18042 vector float vec_reve (vector float);
18043 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
18044 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
18045 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
18046 (altivec_vreve): New pattern.
18047 * config/rs6000/altivec.h (vec_reve): New define.
18048 * doc/extend.texi (vec_rev): Update the built-in documentation file
18049 for the new built-in functions.
18051 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18053 PR tree-optimization/71815
18054 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
18056 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
18058 (slsr_process_phi): Likewise.
18059 (replace_uncond_cands_and_profitable_phis): Don't replace a
18060 multiply candidate with a stride of 1 (copy or cast).
18061 (phi_incr_cost): Call uses_consumed_by_stmt rather than
18063 (lowest_cost_path): Likewise.
18064 (total_savings): Likewise.
18066 2017-06-26 Richard Biener <rguenther@suse.de>
18069 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
18070 Use def_builtin_pure for all gather builtins.
18072 2017-06-26 Richard Biener <rguenther@suse.de>
18074 PR tree-optimization/81203
18075 * tree-tailcall.c (find_tail_calls): Do not move stmts into
18076 non-dominating BBs.
18078 2017-06-26 Marek Polacek <polacek@redhat.com>
18081 * doc/invoke.texi: Document -Wmultistatement-macros.
18083 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
18085 * doc/sourcebuild.texi (ARM-specific attributes): Document new
18086 arm_neon_ok_no_float_abi effective target.
18088 2017-06-26 Richard Biener <rguenther@suse.de>
18090 PR tree-optimization/80928
18091 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
18092 (copy_bbs): Set BB_DUPLICATED flag early.
18093 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
18095 (execute_on_shrinking_pred): Likewise.
18096 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
18097 BB_DUPLICATED blocks.
18098 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
18099 iterate over all PHIs considering removal of *gsi.
18101 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
18103 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
18106 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18108 * config/rs6000/rs6000-string.c: (expand_block_clear,
18109 do_load_for_compare, select_block_compare_mode,
18110 compute_current_alignment, expand_block_compare,
18111 expand_strncmp_align_check, expand_strn_compare,
18112 expand_block_move, rs6000_output_load_multiple)
18113 Move functions related to string/block move/compare
18114 to a separate file.
18115 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
18116 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
18117 for this function which is now used in two files.
18118 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
18119 * config.gcc: Add rs6000-string.o to extra_objs for
18120 targets powerpc*-*-* and rs6000*-*-*.
18122 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
18125 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
18126 32-bit, since indexed is not valid for DImode.
18127 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
18128 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
18129 (define_peephole2 for Altivec d-form load): Add 32-bit support.
18130 (define_peephole2 for Altivec d-form store): Likewise.
18133 * multiple_target.c (create_dispatcher_calls): Only create the
18134 dispatcher call if the function is the default clone of a
18135 versioned function.
18137 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
18139 PR middle-end/80902
18140 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
18141 a call, force the call to not be a tail call.
18143 2017-06-23 Jeff Law <law@redhat.com>
18145 * doc/contrib.texi: Add entry for Steven Pemberton's work on
18148 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
18150 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
18151 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
18152 handling for early expansion of vector shifts (sl,sr,sra,rl).
18153 (builtin_function_type): Add vector shift right instructions
18154 to the unsigned argument list.
18156 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
18158 rtl-optimizatoin/79286
18159 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
18160 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
18161 trap. PIC register plus a const unspec without offset can never trap.
18163 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
18165 * tree.h (builtin_structptr_type): New type.
18166 (builtin_structptr_types): Declare new array.
18167 * tree.c (builtin_structptr_types): New array.
18168 (free_lang_data, build_common_tree_nodes): Use it.
18170 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
18173 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
18176 2017-06-22 Matt Turner <mattst88@gmail.com>
18178 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
18179 Lake models to skylake case. Assume skylake for unknown
18180 models with clflushopt.
18182 2017-06-22 Jeff Law <law@redhat.com>
18184 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
18185 frame sizes that do not satisfy aarch64_uimm12_shift.
18187 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
18189 * profile-count.h (apply_probability,
18190 apply_scale, probability_in): Fix checks for zero.
18192 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18194 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
18195 * doc/cppdiropts.texi (-I @var{dir}): Document it.
18197 2016-06-22 Richard Biener <rguenther@suse.de>
18199 * tree-vect-loop.c (vect_model_reduction_cost): Handle
18200 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
18201 REDUC_MAX_EXPR support.
18202 (vectorizable_reduction): Likewise.
18203 (vect_create_epilog_for_reduction): Likewise.
18205 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
18207 * match.pd (A / (1 << B) -> A >> B): New.
18208 * generic-match-head.c: Include optabs-tree.h.
18209 * gimple-match-head.c: Likewise.
18210 * optabs-tree.h (target_supports_op_p): New.
18211 * optabs-tree.c (target_supports_op_p): New.
18213 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18215 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
18216 $gcc_cv_ld --help output.
18217 (gcc_cv_ld_demangle): Likewise.
18218 (gcc_cv_ld_eh_frame_hdr): Likewise.
18219 (gcc_cv_ld_pie): Likewise.
18220 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
18221 (gcc_cv_ld_buildid): Likewise.
18222 (gcc_cv_ld_sysroot): Likewise.
18223 (ld_bndplt_support): Likewise.
18224 (ld_pushpopstate_support): Likewise.
18225 * configure: Regenerate.
18226 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
18228 2017-06-21 Jakub Jelinek <jakub@redhat.com>
18231 * config/i386/sse.md (round<mode>2): Renumber match_dup and
18232 operands indexes to avoid gap between operands and match_dups.
18234 2017-06-21 Andrew Pinski <apinski@cavium.com>
18236 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
18237 Increment Arith_shift and Arith_shift_reg by 1.
18238 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
18240 * config/aarch64/aarch64.c (thunderx_tunings): Enable
18241 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
18242 (aarch64_strip_extend): Add new argument and test for it.
18243 (aarch64_cheap_mult_shift_p): New function.
18244 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
18245 add a cost if it is true.
18246 Update calls to aarch64_strip_extend.
18247 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
18249 2017-06-21 Andrew Pinski <apinski@cavium.com>
18251 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
18253 (thunderxt88): Likewise.
18254 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
18255 (thunderx_prefetch_tune): New variable.
18256 (thunderx2t99_prefetch_tune): Update for the correct values.
18257 (thunderxt88_tunings): New variable.
18258 (thunderx_tunings): Use thunderx_prefetch_tune instead of
18259 generic_prefetch_tune.
18260 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
18262 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18264 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
18265 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
18266 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
18267 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
18268 (aarch64_atomic_cas<mode>, GPI): Likewise.
18270 2017-06-21 Martin Liska <mliska@suse.cz>
18272 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
18273 statements on cold and hot labels.
18274 * predict.c (tree_estimate_probability_bb): Remove the
18275 prediction from this place.
18277 2017-06-21 Martin Liska <mliska@suse.cz>
18279 PR tree-optimization/79489
18280 * gimplify.c (maybe_add_early_return_predict_stmt): New
18282 (gimplify_return_expr): Call the function.
18283 * predict.c (tree_estimate_probability_bb): Remove handling
18285 * predict.def: Update comment about early return predictor.
18286 * gimple-predict.h (is_gimple_predict): New function.
18287 * predict.def: Change default value of early return to 66.
18288 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
18290 * passes.def: Put pass_strip_predict_hints to the beginning of
18293 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
18295 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
18296 FUNCTION_DECL declarations.
18297 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
18299 (dwaf2out_decl): Likewise.
18300 * godump.c (go_early_global_decl): Skip call to the real debug hook
18301 for FUNCTION_DECL declarations.
18302 * passes.c (rest_of_decl_compilation): Skip call to the
18303 early_global_decl debug hook for FUNCTION_DECL declarations, unless
18304 -fdump-go-spec is passed.
18306 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
18308 * config/i386/i386.c (struct builtin_isa): New field pure_p.
18309 Reorder for compactness.
18310 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
18311 (def_builtin_pure, def_builtin_pure2): New functions.
18312 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
18314 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
18316 * match.pd (nop_convert): New predicate.
18317 ((A +- CST1) +- CST2): Allow some NOP conversions.
18319 2017-06-21 Jakub Jelinek <jakub@redhat.com>
18322 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
18323 with ctors/dtors if GOVD_SHARED is set.
18325 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
18327 * config/aarch64/aarch64.md (movti_aarch64):
18328 Emit mov rather than orr.
18329 (movtf_aarch64): Likewise.
18330 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
18331 Emit mov rather than orr.
18333 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
18335 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
18336 Swap alternatives, make integer dup more expensive.
18338 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
18340 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
18341 Return true for non-tls symbols.
18343 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
18345 * config/aarch64/aarch64-cores.def (cortex-a55): New.
18346 (cortex-a75): Likewise.
18347 (cortex-a75.cortex-a55): Likewise.
18348 * config/aarch64/aarch64-tune.md: Regenerate.
18349 * doc/invoke.texi (-mtune): Document new values for -mtune.
18351 2017-06-21 Tom de Vries <tom@codesourcery.com>
18353 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
18354 stack_size feature.
18355 (Effective-Target Keywords, Other attributes): Suggest using
18356 dg-add-options stack_size feature to get stack limit in stack_size
18357 effective target documentation.
18359 2017-06-21 Julian Brown <julian@codesourcery.com>
18360 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18362 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
18363 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
18364 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
18366 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
18367 attribute type list for neon_multiply.
18368 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
18369 attribute type list for neon_multiply.
18370 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
18371 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
18372 attribute type list for neon_multiply.
18373 * config/arm/types.md (crypto_pmull): Add.
18374 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
18375 attribute type list.
18377 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
18379 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
18382 2017-06-20 Jakub Jelinek <jakub@redhat.com>
18384 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
18385 to make sure not to dereference a NULL cost_classes_ptr pointer.
18387 2017-06-20 Carl Love <cel@us.ibm.com>
18389 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18390 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
18391 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
18392 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
18393 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
18394 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
18395 VMULOSW): New enum "unspec" values.
18396 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
18397 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
18398 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
18399 altivec_vmulosw): New patterns.
18400 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
18401 VMULOSW): Add definitions.
18403 2017-06-20 Julia Koval <julia.koval@intel.com>
18405 * config/i386/i386.c: Fix rounding expand for new pattern.
18406 * config/i386/subst.md: Fix pattern (parallel -> unspec).
18408 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
18410 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
18411 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
18413 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
18415 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
18418 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
18420 * config/aarch64/aarch64-cores.def: Rearrange to sort by
18421 architecture, then by implementer ID.
18422 * config/aarch64/aarch64-tune.md: Regenerate.
18424 2017-06-20 Richard Biener <rguenther@suse.de>
18426 PR middle-end/81097
18427 * fold-const.c (split_tree): Fold to type before negating.
18429 2017-06-20 David Malcolm <dmalcolm@redhat.com>
18431 * diagnostic-show-locus.c
18432 (selftest::test_fixit_deletion_affecting_newline): New function.
18433 (selftest::diagnostic_show_locus_c_tests): Call it.
18435 2017-06-20 Andreas Schwab <schwab@suse.de>
18438 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
18441 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18443 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
18444 __ARM_FEATURE_COPROC according to support.
18446 2017-06-20 Jakub Jelinek <jakub@redhat.com>
18448 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
18449 Rewritten to avoid overflow for > 32-bit pointers.
18452 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
18453 by removing enum keyword.
18454 (ubsan_type_descriptor): Likewise. Formatting fix.
18457 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
18458 splitter): Require TARGET_SSE2 in the condition.
18460 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
18463 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
18464 for doing vector set of SFmode on ISA 3.0.
18465 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
18466 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
18468 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
18469 SFmode value into a V4SF variable that was extracted from another
18470 V4SF variable without converting the element to double precision
18471 and back to single precision vector format.
18472 (vsx_insert_extract_v4sf_p9_2): Likewise.
18474 2017-06-19 Jakub Jelinek <jakub@redhat.com>
18476 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
18477 in UWHI to avoid undefined overflow.
18480 * ubsan.h (enum ubsan_encode_value_phase): New.
18481 (ubsan_encode_value): Change second argument to
18482 enum ubsan_encode_value_phase with default value of
18483 UBSAN_ENCODE_VALUE_GENERIC.
18484 * ubsan.c (ubsan_encode_value): Change second argument to
18485 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
18486 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
18487 create_tmp_var_raw instead of create_tmp_var and use a
18489 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
18490 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
18491 ubsan_encode_value callers.
18494 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
18495 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
18496 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
18498 2017-06-19 Richard Biener <rguenther@suse.de>
18500 PR middle-end/81118
18501 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
18502 estimates if we changed anything.
18504 2017-06-19 Richard Biener <rguenther@suse.de>
18506 PR tree-optimization/80887
18507 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
18508 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
18509 simplified lookups, then reset mprts_hook.
18510 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
18512 (try_to_simplify): Likewise.
18514 2017-06-19 Martin Liska <mliska@suse.cz>
18517 * gimplify.c (gimplify_switch_expr):
18518 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
18520 2017-06-19 Martin Liska <mliska@suse.cz>
18522 * doc/install.texi: Document that PGO runs in 4 stages.
18524 2017-06-19 Martin Liska <mliska@suse.cz>
18527 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
18528 to dispatcher function name.
18529 * multiple_target.c (replace_function_decl): New function.
18530 (create_dispatcher_calls): Redirect both edges and references.
18532 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
18534 * profile-count.c (profile_count::dump): Dump quality.
18535 (profile_count::differs_from_p): Update for unsigned val.
18536 * profile-count.h (profile_count_quality): New enum.
18537 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
18539 2017-06-19 Richard Biener <rguenther@suse.de>
18541 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
18542 struct function as arg.
18543 (estimate_numbers_of_iterations): Export overload with loop arg.
18544 (free_numbers_of_iterations_estimates_loop): Use an overload of
18545 free_numbers_of_iterations_estimates instead.
18546 * tree-cfg.c (remove_bb): Adjust.
18547 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
18548 * tree-parloops.c (gen_parallel_loop): Likewise.
18549 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
18551 (tree_unroll_loops_completely): Likewise.
18552 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
18553 Use an overload instead and export.
18554 (estimated_loop_iterations): Adjust.
18555 (max_loop_iterations): Likewise.
18556 (likely_max_loop_iterations): Likewise.
18557 (estimate_numbers_of_iterations): Take struct function as arg
18559 (loop_exits_before_overflow): Adjust.
18560 (free_numbers_of_iterations_estimates_loop): Use an overload.
18561 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
18562 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
18564 2017-06-19 Richard Biener <rguenther@suse.de>
18567 * ipa-prop.c (find_constructor_constant_at_offset): Handle
18568 RANGE_EXPR conservatively.
18570 2017-06-16 Carl Love <cel@us.ibm.com>
18572 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18573 definitions for vec_float, vec_float2, vec_floato,
18574 vec_floate built-ins.
18575 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
18576 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
18578 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
18579 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
18580 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
18581 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
18582 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
18583 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
18584 vec_floato): Add builtin defines.
18585 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
18586 Update the built-in documentation file for the new built-in
18589 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18591 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
18592 (mthumb): Mark as the negative of -marm.
18594 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18596 * doc/invoke.texi (ARM Options, -mcpu): Document supported
18598 (ARM Options, -mtune): Document that this accepts the same
18599 extension options as -mcpu.
18600 (ARM Options, -mfpu): Document addition of -mfpu=auto.
18602 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18604 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
18605 permitted extensions.
18607 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18609 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
18610 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
18611 (armv8-m.main): Add option +nodsp.
18612 * config/arm/arm-cpu-cdata.h: Regenerated.
18614 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18616 * config/arm/t-fuchsia: New file.
18617 * config.gcc (arm*-*-fuchsia*): Use it.
18619 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18621 * config/arm/t-symbian: Rewrite for new option infrastructure.
18623 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18625 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
18626 (MULTILIB_REQUIRED): Likewise.
18628 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18630 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
18631 (MULTILIB_RESUE): Likewise.
18632 (MULTILIB_MATCHES): Likewise.
18633 (MULTLIB_REQUIRED): Likewise.
18635 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18637 * config/arm/t-rtems: Rewrite for new option framework.
18639 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18641 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
18642 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
18643 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
18644 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
18645 * config/arm/t-multilib: ... here.
18646 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
18647 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
18648 armv7-a and armv8*-a when A-profile libraries have not been built.
18649 * config/arm/t-rmprofile: Rewrite.
18651 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18653 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
18654 with a backslash. Remove the backslash after substituting unescaped
18656 * doc/fragments.texi (MULTILIB_REUSE): Document it.
18658 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18660 * config.gcc: (arm*-*-*): When building a-profile libraries, force
18661 the driver to pass through the default setting of -mfloat-abi.
18662 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
18664 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
18665 (all_feat_combs): New rule.
18666 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
18668 * config/arm/t-aprofile: Rewrite.
18670 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18672 * config/arm/arm.h (FPUTYPE_AUTO): Define.
18673 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
18674 fpu is not specified by the user/command-line.
18675 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
18676 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
18677 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
18678 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
18679 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
18680 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
18682 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18684 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
18685 * config/arm/t-arm-elf: Rewritten.
18687 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18689 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
18690 have some floating-point instructions.
18691 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
18692 (TARGET_MAYBE_HARD_FLOAT): New macro.
18693 * config/arm/arm-builtins.c (arm_init_builtins): Use
18694 TARGET_MAYBE_HARD_FLOAT.
18695 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
18697 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18699 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
18700 (configargs.h): Include it.
18701 (arm_print_hint_for_fpu_option): New function.
18702 (arm_parse_fpu_option): New function.
18703 (candidate_extension): New class.
18704 (arm_canon_for_multilib): New function.
18705 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
18706 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
18707 (ARCH_CANONICAL_SPECS): New macro.
18708 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
18710 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18712 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
18713 are set after handling multilib fragments. Set target_cpu_default2
18716 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18718 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
18720 (arm*-*-*): Set target_cpu_default2 to a quoted string.
18721 * config/arm/parsecpu.awk (check_cpu): Validate any extension
18723 (check_arch): Likewise.
18724 * config/arm/arm.c (arm_configure_build_target): Handle
18725 TARGET_CPU_DEFAULT being a string constant. Scan any feature
18726 options in the default.
18728 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18730 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
18731 when an option is an alias of another.
18732 * config/arm/parsecpu.awk (optalias): New parser token.
18733 (gen_comm_data): Mark non-alias options as such. Emit entries
18734 for extension aliases.
18735 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
18736 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
18737 (armv6kz, armv6zk, armv6t2): Likewise.
18738 (armv7): Make vfpv3-d16 an alias.
18739 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
18741 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
18742 Sort in canonical order.
18743 (armv8-a): Sort in canonical order.
18744 (armv8.1-a, armv8.2-a): Likewise.
18745 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
18747 (cortex-a9): Sort in canonical order.
18748 * config/arm/arm.c (selftests.h): Include it.
18749 (arm_test_cpu_arch_data): New function.
18750 (arm_run_self_tests): New function.
18751 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
18752 (targetm): Move declaration to the end of the file.
18753 * arm-cpu-cdata.h: Regenerated.
18755 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18757 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
18758 call to target_mode_check describing the type of option passed.
18759 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
18760 (arm_target_thumb_only): Use arm_parse_arch_option_name or
18761 arm_parse_cpu_option_name to match parameters against list of
18763 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
18764 arm_arch_core_flags data structure.
18765 * config/arm/arm-cpu_cdata.h: Regenerated.
18767 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18769 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
18771 (arm_print_hint_for_cpu_option): Likewise.
18772 (arm_print_hint_for_arch_option): Likewise.
18773 (arm_parse_cpu_option_name): Likewise.
18774 (arm_parse_arch_option_name): Likewise.
18775 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
18776 of entries in the all_fpus list.
18777 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
18778 (arm_parse_cpu_option_name): Declare.
18779 (arm_parse_arch_option_name): Declare.
18780 (arm_parse_option_features): Declare.
18781 (arm_intialize_isa): Declare.
18782 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
18784 (gen_comm_data): ... here. Make definitions non-static.
18785 * config/arm/arm-cpu-data.h: Regenerated.
18786 * config/arm/arm-cpu-cdata.h: Regenerated.
18788 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18790 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
18791 (cpu_arch_extension): New structure.
18792 (cpu_arch_option, arch_option, cpu_option): New structures.
18793 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
18794 architecture types.
18795 (gen_data): Generate new format data tables.
18796 * config/arm/arm.c (cpu_tune): New structure.
18797 (cpu_option, processors): Delete.
18798 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
18799 (arm_print_hint_for_cpu_option): ... this and ...
18800 (arm_print_hint_for_arch_option): ... this.
18801 (arm_parse_arch_cpu_name): Delete. Replace with ...
18802 (arm_parse_cpu_option_name): ... this and ...
18803 (arm_parse_arch_option_name): ... this.
18804 (arm_unrecognized_feature): Change type of target parameter to
18806 (arm_parse_arch_cpu_features): Delete. Replace with ...
18807 (arm_parse_option_features): ... this.
18808 (arm_configure_build_target): Rework to use new configuration data
18810 (arm_print_tune_info): Rework for new configuration data tables.
18811 * config/arm/arm-cpu-data.h: Regenerated.
18812 * config/arm/arm-cpu.h: Regenerated.
18814 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18816 * Makefile.in (OBJS): Move sbitmap.o from here ...
18817 (OBJS-libcommon): ... to here.
18819 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18821 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
18822 (ISA_ALL_CRYPTO): New macro.
18823 (ISA_ALL_SIMD): New macro
18824 (ISA_ALL_FP): New macro.
18825 * config/arm/arm.c (fpu_bitlist): Update initializer.
18826 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
18828 (arm9e): Add fpu. Add option for nofp
18829 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
18830 (arm926ej-s, arm1026ej-s): Likewise.
18831 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
18832 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
18833 neon-fp16, neon-vfpv4, nofp and nosimd.
18834 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
18835 (cortex-a8): Add fpu. Add option for nofp.
18836 (cortex-a9): Add fpu. Add options for nosimd and nofp.
18837 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
18838 (cortex-r4f): Add fpu.
18839 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
18840 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
18842 (cortex-r8): Likewise.
18843 (cortex-m4): Add fpu. Add option for nofp.
18844 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
18845 (cortex-a17.cortex-a7): Likewise.
18846 (cortex-a32): Add fpu. Add options for crypto and nofp.
18847 (cortex-a35, cortex-a53): Likewise.
18848 (cortex-a57): Add fpu. Add option for crypto.
18849 (cortex-a72, cortex-a73): Likewise.
18850 (exynos-m1): Likewise.
18851 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
18852 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
18853 (cortex-m33): Add fpu. Add option for nofp.
18854 * config/arm/arm-cpu-cdata.h: Regenerated
18855 * config/arm/arm-cpu-data.h: Regenerated.
18857 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18859 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
18860 (armv5te, armv5tej): Likewise.
18861 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
18862 (armv7): Add options fp and vfpv3-d16.
18863 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
18864 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
18866 (armv7ve): Likewise.
18867 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
18868 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
18869 (armv8-a): Add nocrypto option.
18870 (armv8.1-a, armv8.2-a): Likewise.
18871 (armv8-m.main): add options fp, fp.dp and nofp.
18873 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18875 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
18877 (armv8-a+crc): Delete.
18878 (armv8.1-a): Add options simd, crypto and nofp.
18879 (armv8.2-a): Add options fp16, simd, crypto and nofp.
18880 (armv8.2-a+fp16): Delete.
18881 (armv8-m.main): Add option dsp.
18882 (armv8-m.main+dsp): Delete.
18883 (cortex-a8): Add fpu. Add nofp option.
18884 (cortex-a9): Add fpu. Add nofp and nosimd options.
18885 * config/arm/parsecpu.awk (gen_data): Generate option tables and
18886 link to main cpu and architecture data structures.
18887 (gen_comm_data): Only put isa attributes from the main architecture
18889 (option): New statement for architecture and CPU entries.
18890 * arm.c (struct cpu_option): New structure.
18891 (struct processors): Add entry for options.
18892 (arm_unrecognized_feature): New function.
18893 (arm_parse_arch_cpu_name): Ignore any characters after the first
18895 (arm_parse_arch_cpu_feature): New function.
18896 (arm_configure_build_target): Separate out any CPU and architecture
18897 features and parse separately. Don't error out if -mfpu=auto is
18898 used with only an architecture string.
18899 (arm_print_asm_arch_directives): New function.
18900 (arm_file_start): Call it.
18901 * config/arm/arm-cpu-cdata.h: Regenerated.
18902 * config/arm/arm-cpu-data.h: Likewise.
18903 * config/arm/arm-tables.opt: Likewise.
18905 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18907 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
18908 assembler when it is not -mfpu=auto.
18910 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18912 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
18913 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
18914 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
18915 (ASM_CPU_SPEC): Rewrite.
18916 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
18917 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
18918 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
18919 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
18920 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
18921 copied string is NUL-terminated. Also strip any characters prefixed
18923 (arm_rewrite_selected_arch): New function.
18924 (arm_rewrite_march): New function.
18926 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18928 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
18929 (x_arm_cpu_string, x_arm_tune_string): Likewise.
18930 (march, mcpu, mtune): Convert to string-based options.
18931 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
18932 (arm_parse_arch_cpu_name): New function.
18933 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
18934 identify selected architecture or CPU.
18935 (arm_option_save): New function.
18936 (TARGET_OPTION_SAVE): Redefine.
18937 (arm_option_restore): Restore string options.
18938 (arm_option_print): Print string options.
18940 2017-06-16 Martin Sebor <msebor@redhat.com>
18942 PR tree-optimization/80933
18943 PR tree-optimization/80934
18944 * builtins.c (fold_builtin_3): Do not handle bcmp here.
18945 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
18946 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
18947 (gimple_fold_builtin): Call them.
18949 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
18951 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
18952 as unlikely; update profile.
18954 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
18956 * predict.c (force_edge_cold): Handle declaring edges impossible
18959 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
18961 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
18963 (try_unroll_loop_completely): Fix reporting.
18965 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
18967 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
18969 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
18972 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
18973 if given a non-constant argument for an intrinsic which requires a
18976 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
18978 * profile.c (compare_freqs): New function.
18979 (branch_prob): Sort edge list.
18980 (find_spanning_tree): Assume that the list is priority sorted.
18982 2017-06-16 Richard Biener <rguenther@suse.de>
18984 PR tree-optimization/81090
18985 * passes.def (pass_record_bounds): Remove.
18986 * tree-pass.h (make_pass_record_bounds): Likewise.
18987 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
18988 make_pass_record_bounds): Likewise.
18989 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
18990 not free niter estimates at the beginning but at the end.
18991 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
18993 2017-06-16 Richard Biener <rguenther@suse.de>
18995 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
18996 initializer to workaround ICE in host GCC 4.8.
18998 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19000 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
19002 (clone_inlined_nodes): Update.
19004 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19006 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
19007 prefetch settings, and enable prefetching by default at -O3.
19009 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19011 * config/aarch64/aarch64.c (aarch64_override_options_internal):
19012 Set flag_prefetch_loop_arrays according to tuning data.
19014 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19016 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
19017 New tune structure.
19018 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
19019 [Unrelated to main purpose of the patch] Place the pointer field last
19020 to enable type checking errors when tune structure are wrongly merged.
19021 * config/aarch64/aarch64.c (generic_prefetch_tune,)
19022 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
19023 (thunderx2t99_prefetch_tune): New tune constants.
19024 (tune_params *_tunings): Update all tunings (no functional change).
19025 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
19026 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
19027 from tunings structures.
19029 2017-06-16 Jakub Jelinek <jakub@redhat.com>
19032 * ubsan.c (instrument_null): Add T argument, use it instead
19033 of computing it based on IS_LHS.
19034 (instrument_object_size): Likewise.
19035 (pass_ubsan::execute): Adjust instrument_null and
19036 instrument_object_size callers to pass gimple_get_lhs or
19037 gimple_assign_rhs1 result to it. Use instrument_null instead of
19038 calling get_base_address and instrument_mem_ref. Handle
19039 aggregate call arguments for object-size sanitization.
19041 2017-06-16 Yury Gribov <tetra2005@gmail.com>
19043 PR tree-optimization/81089
19044 * tree-vrp.c (is_masked_range_test): Validate operands of
19047 2017-06-15 Martin Sebor <msebor@redhat.com>
19050 * dumpfile.c (dump_register): Avoid calling memset to initialize
19051 a class with a default ctor.
19052 * gcc.c (struct compiler): Remove const qualification.
19053 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
19054 * hash-table.h: Ditto.
19055 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
19057 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
19058 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
19060 * params.h (struct param_info): Make struct members non-const.
19061 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
19062 with default initialization.
19063 * vec.h (vec_copy_construct, vec_default_construct): New helper
19065 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
19066 with vec_copy_construct.
19067 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
19068 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
19069 * doc/invoke.texi (-Wclass-memaccess): Document.
19071 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19073 * emit-rtl.h (is_leaf): Update comment about local
19074 register allocator.
19076 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19079 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
19080 for a variable to have a section before checking if the section has a
19082 Set section to.persistent if persistent attribute is set.
19083 Warn if .persistent attribute is used on an automatic variable.
19085 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
19087 PR rtl-optimization/80474
19088 * reorg.c (update_block): Do not ignore instructions in a delay slot.
19090 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
19092 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
19095 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
19097 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
19098 (casesi): Emit bounds checking as RTL.
19099 (casesi_internal_mips16_<mode>): Remove bounds checking.
19101 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
19103 * config/xtensa/xtensa.c (xtensa_option_override): Append
19104 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
19105 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
19106 xtensa_doloop_hooks): Define unconditionally.
19107 (xtensa_reorg_loops): Only call reorg_loops in the presence of
19109 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
19110 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
19111 for it in xtensa_option_override.
19112 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
19113 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
19115 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
19117 * doc/cppopts.texi: Document '-' special value to -MF.
19119 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
19121 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
19122 (cortex_a53_fconst): Likewise.
19123 (cortex_a53_fpmul): Likewise.
19124 (cortex_a53_f_load_64): Likewise.
19125 (cortex_a53_f_load_many): Likewise.
19126 (cortex_a53_advsimd_alu): Likewise.
19127 (cortex_a53_advsimd_alu_q): Likewise.
19128 (cortex_a53_advsimd_mul): Likewise.
19129 (cortex_a53_advsimd_mul_q): Likewise.
19130 (fpmac bypass): Add new bypass for fpmac-fpmac case.
19131 Add missing fmul, r2f_cvt and fconst cases.
19133 2017-06-14 Richard Biener <rguenther@suse.de>
19135 PR middle-end/81088
19136 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
19138 (fold_binary_loc): When associating do not treat pre-existing
19139 TREE_OVERFLOW on literal constants as a reason to allow
19140 TREE_OVERFLOW on associated literal constants.
19142 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19144 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
19145 (MASK_FEATURES): New macro.
19146 * config/sparc/sparc.c (sparc_option_override): Remove the special
19147 handling of -mfpu and generalize it to all MASK_FEATURES switches.
19149 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19151 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
19152 a division of 0 if non-call exceptions are enabled.
19154 2017-06-14 Andrew Pinski <apinski@cavium.com>
19155 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19158 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
19159 Improve vector initialization code gen for only variable case.
19161 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19163 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
19165 2017-06-14 Richard Biener <rguenther@suse.de>
19167 PR tree-optimization/81083
19168 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
19171 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19173 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
19174 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
19175 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
19176 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
19177 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
19178 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
19180 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19182 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
19183 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
19185 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19187 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
19189 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19191 * config/rs6000/t-rtems: Don't handle SPE.
19193 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19195 * config/rs6000/t-linux: Don't handle SPE.
19197 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19199 * config/rs6000/eabispe.h: Delete file.
19201 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19203 * config/rs6000/t-spe: Delete file.
19205 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19207 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
19208 (rs6000_legitimate_offset_address_p): Return false for anything in
19209 V2SImode or V2SFmode.
19211 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19213 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
19214 except V2SF and V2SI. Rearrange the vector modes, and add comments.
19215 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
19217 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
19218 (rs6000_legitimate_offset_address_p): Ditto.
19219 (rs6000_emit_move): Ditto.
19220 (rs6000_init_builtins): Remove V4HI_type_node.
19222 2017-06-13 Martin Liska <mliska@suse.cz>
19225 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
19226 (gate_asan): Likewise.
19227 * asan.h (asan_no_sanitize_address_p): Remove the function.
19228 (sanitize_flags_p): New function.
19229 * builtins.def: Fix coding style.
19230 * common.opt: Use renamed enum value.
19231 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
19232 * doc/extend.texi: Document no_sanitize attribute.
19233 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
19234 to SANITIZE_UNDEFINED_NONDEFAULT.
19235 * gcc.c (sanitize_spec_function): Use the renamed enum value.
19236 * gimple-fold.c (optimize_atomic_compare_exchange_p):
19237 Use sanitize_flags_p.
19238 * gimplify.c (gimplify_function_tree): Likewise.
19239 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
19240 * opts.c (parse_no_sanitize_attribute): New function.
19241 (common_handle_option): Use renamed enum value.
19242 * opts.h (parse_no_sanitize_attribute): Declare.
19243 * tree.c (sanitize_flags_p): New function.
19244 * tree.h: Declared here.
19245 * tsan.c: Use sanitize_flags_p.
19246 * ubsan.c (ubsan_expand_null_ifn): Likewise.
19247 (instrument_mem_ref): Likewise.
19248 (instrument_bool_enum_load): Likewise.
19249 (do_ubsan_in_current_function): Remove the function.
19250 (pass_ubsan::execute): Use sanitize_flags_p.
19251 * ubsan.h: Remove do_ubsan_in_current_function
19252 * tree-cfg.c (print_no_sanitize_attr_value): New function.
19253 (dump_function_to_file): Use it here.
19255 2017-06-13 Martin Jambor <mjambor@suse.cz>
19257 PR tree-optimization/80803
19258 PR tree-optimization/81063
19259 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
19260 (propagate_subaccesses_across_link): Enqueue subtree whenever
19261 necessary instead of relying on the caller.
19263 2017-06-13 Martin Jambor <mjambor@suse.cz>
19265 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
19266 that have a first_link.
19267 (sort_and_splice_var_accesses): Do not check first_link before
19269 (subtree_mark_written_and_enqueue): Likewise.
19270 (propagate_all_subaccesses): Likewise and do not stop at first
19271 parent with a first_link.
19273 2017-06-13 Martin Jambor <mjambor@suse.cz>
19275 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
19278 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19280 * match.pd: New pattern.
19282 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19284 * tree-vrp.c (is_masked_range_test): New function.
19285 (register_edge_assert_for): Determine ranges for
19288 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19290 PR tree-optimization/67328
19291 * fold-const.c (maskable_range_p): New function.
19292 (build_range_check): Generate bittests if possible.
19294 2017-06-13 Martin Liska <mliska@suse.cz>
19296 * gimple-pretty-print.c (dump_probability): Add new argument.
19297 (dump_edge_probability): Dump both probability and count.
19298 (dump_gimple_label): Likewise.
19299 (dump_gimple_bb_header): Likewise.
19301 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
19304 * config/avr/avr-devices.c: Fix indentation.
19305 * config/avr/gen-avr-mmcu-specs.c: Dito.
19307 2017-06-13 Richard Biener <rguenther@suse.de>
19309 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
19310 instead get vector type from stmt_info.
19311 (vectorizable_reduction): Adjust. Remove dead code.
19313 2017-06-13 Richard Biener <rguenther@suse.de>
19315 PR middle-end/81065
19316 * fold-const.c (extract_muldiv_1): Remove bogus distribution
19317 case of C * (x * C2 + C3).
19318 (fold_addr_of_array_ref_difference): Properly fold index difference.
19320 2017-06-12 David S. Miller <davem@davemloft.net>
19323 * config/sparc/sparc.md (return expander): Emit frame blockage if
19324 function uses alloca.
19326 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
19328 * combine.c (make_field_assignment): Check len rather than the mode
19329 precision when calling force_to_mode.
19331 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
19333 Support multilibs and devices that see flash in RAM address range.
19336 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
19337 (avr_mcu_t) <flash_pm_offset>: New field.
19338 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
19339 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
19340 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
19341 (AVR_TINY_PM_OFFSET): Remove macro.
19342 * config/avr/avr.opt (-mshort-calls): New option.
19343 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
19344 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
19345 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
19346 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
19347 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
19348 instead of avr_arch->have_jmp_call.
19349 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
19350 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
19351 avr_arch->flash_pm_offset to define.
19352 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
19353 new field flash_pm_offset. Add entry for avrxmega3.
19354 (avr_texinfo): Add entry for avrxmega3.
19355 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
19356 attiny212, attiny214,
19357 attiny412, attiny414, attiny416, attiny417,
19358 attiny814, attiny816, attiny817,
19359 attiny1614, attiny1616, attiny1617,
19360 attiny3214, attiny3216, attiny3217.
19361 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
19362 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
19363 (avr_print_operand_address) [AVR_TINY]: Same.
19364 (avr_asm_init_sections) <readonly_data_section>: Only patch
19365 callback if avr_arch->flash_pm_offset = 0.
19366 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
19367 for rodata if avr_arch->flash_pm_offset != 0.
19368 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
19369 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
19370 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
19371 (m_options): Append opt_rcall.
19372 (m_dirnames): Append dir_rcall.
19373 * config/avr/t-multilib: Regenerate.
19375 * configure.ac [target=avr]: Check whether avrxmega3 default
19376 linker description file works as needed.
19377 * configure: Regenerate.
19378 * doc/avr-mmcu.texi: Regenerate.
19379 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
19380 <__AVR_ARCH__>: Document avrxmega3 and 103.
19381 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
19382 <__AVR_SHORT_CALLS__>: Document it.
19383 <__AVR_PM_BASE_ADDRESS__>: Document it.
19384 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
19385 (AVR Variable Attributes) <progmem>: Document this is
19386 not needed for avrxmega3.
19387 (AVR Named Address Spaces) <__flash>: Dito.
19389 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
19391 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
19393 2017-06-12 Doug Rupp <rupp@adacore.com>
19395 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
19396 Append vxworks-stdint.h to the tm_file list.
19397 * config/vxworks-stdint.h: New file.
19399 2017-06-12 Martin Liska <mliska@suse.cz>
19401 PR tree-optimization/81041
19402 * tree-profile.c (gimple_gen_ic_func_profiler):
19403 Create an extra BB in profile-generate
19404 (gimple_gen_time_profiler): Likewise.
19406 2017-06-12 Jakub Jelinek <jakub@redhat.com>
19408 PR tree-optimization/81003
19409 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
19410 (update_range_test): Use it instead of force_gimple_operand_gsi.
19412 2017-06-12 Richard Biener <rguenther@suse.de>
19414 PR tree-optimization/81053
19415 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
19416 with backedge value not defined in loop. Simplify def stmt
19419 2017-06-11 Tom de Vries <tom@codesourcery.com>
19422 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
19424 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
19425 nvptx_cannot_force_const_mem.
19427 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19429 * opts.c (finish_options): Move test for flag_split_stack after
19430 it has been initialized.
19432 2017-06-11 Jason Merrill <jason@redhat.com>
19434 * tree.h (id_equal): New.
19435 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
19436 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
19437 instead of strcmp of IDENTIFIER_POINTER.
19439 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19441 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
19442 (mark_all_inlined_calls_cdtor): Fix formating.
19443 (inline_transform): Rescale profile before inlining.
19445 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19447 * cgraph.h (cgraph_edge::clone): Update prototype.
19448 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
19449 (cgraph_node::create_clone): Update.
19450 (cgraph_node::create_version_clone): Update.
19451 * tree-inline.c (copy_bb): Update.
19452 (expand_call_inline): Update.
19454 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
19456 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
19457 factored out from ...
19458 (rs6000_emit_prologue): ... here.
19460 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
19462 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
19463 factored out from ...
19464 (rs6000_emit_prologue): ... here.
19466 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19468 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
19470 (handle_missing_profiles): Fix computation of tp_first_run.
19471 (counts_to_freqs): Do not touch freqs when count is 0.
19473 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19475 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
19478 2017-06-10 Tom de Vries <tom@codesourcery.com>
19480 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
19481 attributes): Document signal effective target.
19483 2017-06-10 Tom de Vries <tom@codesourcery.com>
19485 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
19486 Document effective target stack_size.
19488 2017-06-09 David Malcolm <dmalcolm@redhat.com>
19490 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
19491 to the edit_context if they can be auto-applied.
19493 2017-06-9 Ian Lance Taylor <iant@golang.org>
19495 * opts.c (finish_options): If -fsplit-stack, disable implicit
19496 -forder-blocks-and-partition.
19497 * doc/invoke.texi (Optimize Options): Document that when using
19498 -fsplit-stack -forder-blocks-and-partition is not implicitly
19501 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
19503 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
19504 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
19505 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
19506 * builtins.def (abort, trap, unreachable): Declare cold.
19507 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
19508 * tree-core.h (ECF_COLD): New.
19509 * tree.c (set_call_expr_flags): Handle ECF_COLD.
19510 (build_common_builtin_nodes): Mark unreachable and abort as cold.
19512 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
19514 * predict.c (unlikely_executed_stmt_p): Cleanup.
19516 2017-06-09 Richard Biener <rguenther@suse.de>
19518 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
19519 model if the ref is always written to.
19521 2017-06-09 Tamar Christina <tamar.christina@arm.com>
19523 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
19525 2017-06-09 Tamar Christina <tamar.christina@arm.com>
19527 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
19530 2017-06-09 Tom de Vries <tom@codesourcery.com>
19533 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
19534 "target cannot support label values" when encountering LABEL_REF.
19536 2017-06-09 Martin Liska <mliska@suse.cz>
19538 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
19539 (gimple_gen_ic_func_profiler): Emit direct comparison
19540 of __gcov_indirect_call_callee with NULL.
19541 (gimple_gen_time_profiler): Change probability from
19542 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
19544 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
19546 * profile.c (edge_gcov_counts): Turn to pointer.
19547 (compute_branch_probabilities, compute_branch_probabilities): Update.
19548 (branch_prob): Do not clear edge_gcov_count.
19549 * profile.h (edge_gcov_counts): Turn to pointer.
19550 (edge_gcov_count): Update.
19552 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
19554 * gimple.h (gimple_check_failed): Mark cold.
19556 2017-06-09 Richard Biener <rguenther@suse.de>
19558 PR tree-optimization/66623
19559 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
19560 refactor check_reduction into two parts, properly computing
19561 whether we have to check reduction validity for outer loop
19564 2017-06-09 Richard Biener <rguenther@suse.de>
19566 PR tree-optimization/79483
19567 * graphite-scop-detection.c (order): New global.
19568 (get_order): Compute bb to order mapping that satisfies code
19569 generation constraints.
19570 (cmp_pbbs): New helper.
19571 (build_scops): Start domwalk at entry block, sort generated
19574 2017-06-09 Richard Biener <rguenther@suse.de>
19576 PR middle-end/81007
19577 * ipa-polymorphic-call.c
19578 (ipa_polymorphic_call_context::restrict_to_inner_class):
19579 Skip FIELD_DECLs with error_mark_node type.
19580 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
19583 2017-06-09 Martin Liska <mliska@suse.cz>
19585 * predict.c (struct branch_predictor): New struct.
19586 (test_prediction_value_range): New test.
19587 (predict_c_tests): New function.
19588 * selftest-run-tests.c (selftest::run_tests): Run the function.
19589 * selftest.h: Declare new tests.
19591 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
19594 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
19595 gen_add3_insn did not fail.
19596 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
19597 r0, construct that number in a temporary reg and add that reg to r0.
19598 If asked to put the result in r0 as well, fail.
19600 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
19602 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
19603 for early expansion of vec_eqv.
19605 2017-06-08 Jakub Jelinek <jakub@redhat.com>
19607 PR middle-end/81005
19608 * ubsan.c (instrument_null): Avoid pointless code temporary.
19609 (pass_ubsan::execute): Instrument aggregate arguments of calls.
19611 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
19615 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
19618 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
19621 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19623 * predict.c (unlikely_executed_edge_p): Move ahead.
19624 (probably_never_executed_edge_p): Use it.
19626 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
19628 PR middle-end/79988
19629 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
19630 gimple_call_builtin_p call.
19632 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19634 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
19635 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
19636 rtl_check_failed_type2, rtl_check_failed_code1,
19637 rtl_check_failed_code2, rtl_check_failed_code_mode,
19638 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
19639 rtvec_check_failed_bounds, rtl_check_failed_flag,
19640 _fatal_insn_not_found, _fatal_insn): Likewise.
19641 * tree.h (tree_contains_struct_check_failed,
19642 tree_check_failed, tree_not_check_failed,
19643 tree_class_check_failed, tree_range_check_failed,
19644 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
19645 tree_vec_elt_check_failed, phi_node_elt_check_failed,
19646 tree_operand_check_failed, omp_clause_check_failed,
19647 omp_clause_operand_check_failed, omp_clause_range_check_failed):
19650 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19652 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
19653 flag_branch_probabilities.
19654 * ipa-inline.c (edge_badness): Likewise.
19655 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
19656 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
19657 * predict.c (maybe_hot_frequency_p): Likewise.
19658 (probably_never_executed): Likewise.
19659 * sched-ebb.c (schedule_ebbs): Likewise.
19660 * sched-rgn.c (find_single_block_region): Likewise.
19661 * tracer.c (tail_duplicate): Likewise.
19663 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19665 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
19666 longer requires x_flag_profile_use.
19668 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19670 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
19671 instead of flag_reorder_blocks_and_partition.
19672 * dbxout.c (dbxout_function_end): Likewise.
19673 * dwarf2out.c (gen_subprogram_die): Likewise.
19674 * haifa-sched.c (sched_create_recovery_edges): Likewise.
19675 * hw-doloop.c (reorg_loops): Likewise.
19676 * varasm.c (assemble_start_function,
19677 assemble_end_function): Likewise.
19678 (decide_function_section): Do not check for
19679 flag_reorder_blocks_and_partition.
19681 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
19683 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
19685 (chkp_get_hard_register_fake_addr_expr): Ditto.
19686 (chkp_build_addr_expr): Add check for hard reg case.
19687 (chkp_parse_array_and_component_ref): Ditto.
19688 (chkp_find_bounds_1): Ditto.
19689 (chkp_process_stmt): Don't generate bounds store for
19692 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19694 * predict.c (maybe_hot_bb_p): Do not check profile status.
19695 (maybe_hot_edge_p): Likewise.
19696 (probably_never_executed): Check for zero counts even if profile
19698 (unlikely_executed_edge_p): New function.
19699 (unlikely_executed_stmt_p): New function.
19700 (unlikely_executed_bb_p): New function.
19701 (set_even_probabilities): Use unlikely predicates.
19702 (combine_predictions_for_bb): Likewise.
19703 (predict_paths_for_bb): Likewise.
19704 (predict_paths_leading_to_edge): Likewise.
19705 (determine_unlikely_bbs): New function.
19706 (estimate_bb_frequencies): Use it.
19707 (compute_function_frequency): Use zero counts even if profile is
19709 * profile-count.h: Fix typo.
19711 2017-08-08 Julia Koval <julia.koval@intel.com>
19713 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
19714 _mm512_mask_cvtsepi16_storeu_epi8,
19715 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
19716 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
19717 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
19718 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
19719 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
19720 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
19721 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
19722 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
19723 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
19724 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
19725 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
19726 __builtin_ia32_pmovuswb256mem_mask,
19727 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
19728 __builtin_ia32_pmovwb512mem_mask): New builtins.
19730 2017-08-08 Julia Koval <julia.koval@intel.com>
19732 PR target/73350,80862
19733 * config/i386/subst.md (round): Fix round pattern.
19734 * config/i386/i386.c (ix86_erase_embedded_rounding):
19735 Fix erasing rounding for the fixed pattern.
19737 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19739 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
19741 2017-06-08 Martin Liska <mliska@suse.cz>
19743 PR gcov-profile/80911
19744 * gcov.c (block_info::block_info): New constructor.
19746 2017-06-07 Carl Love <cel@us.ibm.com>
19748 * config/rs6000/rs6000-c: The return type of the following
19749 built-in functions was implemented as int not long long. Fix sign
19750 of return value for the unsigned version of vec_mulo and vec_mule.
19751 vector unsigned long long vec_bperm (vector unsigned long long,
19752 vector unsigned char)
19753 vector signed long long vec_mule (vector signed int,
19755 vector unsigned long long vec_mule (vector unsigned int,
19756 vector unsigned int)
19757 vector signed long long vec_mulo (vector signed int,
19759 vector unsigned long long vec_mulo (vector unsigned int,
19760 vector unsigned int)
19761 * doc/extend.texi: Fix the documentation for the built-in
19764 2017-06-07 Carl Love <cel@us.ibm.com>
19767 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
19770 2017-06-07 Carl Love <cel@us.ibm.com>
19772 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
19773 support, Generate doublehv for signed int/float for BE case only.
19775 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
19777 * doc/invoke.texi (mcx16): Rewrite.
19779 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19781 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
19782 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
19783 *mov<mode>_softfloat, and an anonymous splitter): Use
19784 nonimmediate_operand instead of rs6000_nonimmediate_operand.
19786 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19788 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
19790 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
19791 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
19792 (rs6000_debug_reg_global): Adjust.
19793 (rs6000_init_hard_regno_mode_ok): Adjust.
19794 (rs6000_dbx_register_number): Adjust.
19795 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
19796 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
19797 Remove SPE_ACC and SPEFSCR.
19798 (REG_ALLOC_ORDER): Ditto.
19799 (FRAME_POINTER_REGNUM): Change to 111.
19800 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
19801 (REG_CLASS_NAMES): Ditto.
19802 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
19803 (REGISTER_NAMES): Ditto.
19804 (ADDITIONAL_REG_NAMES): Ditto.
19805 (rs6000_reg_names): Ditto.
19806 * config/rs6000/rs6000.md: Renumber some register number
19809 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19811 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
19813 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
19814 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
19816 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
19817 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
19818 Delete the SPE high registers.
19819 (REG_ALLOC_ORDER): Ditto.
19820 (enum reg_class): Remove SPE_HIGH_REGS.
19821 (REG_CLASS_NAMES): Ditto.
19822 (REG_CLASS_CONTENTS): Delete the SPE high registers.
19823 (REGISTER_NAMES): Ditto.
19824 (rs6000_reg_names): Ditto.
19825 * doc/tm.texi.in: Remove SPE as example.
19826 * doc/tm.texi: Regenerate.
19828 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19830 * config/rs6000/8540.md (ppc8540_brinc): Delete.
19831 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
19832 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
19833 * config/rs6000/rs6000.md (type): Remove "brinc".
19835 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19837 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
19838 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
19839 * config/rs6000/linuxspe.h: Delete file.
19840 * config/rs6000/rs6000.md: Don't include spe.md.
19841 * config/rs6000/spe.h: Delete file.
19842 * config/rs6000/spe.md: Delete file.
19843 * config/rs6000/t-rs6000: Remove spe.md.
19845 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19847 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
19848 (reg_or_none500mem_operand): Delete.
19849 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
19850 instead of reg_or_none500mem_operand.
19852 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19854 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
19855 handling of SPE flags.
19856 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
19858 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19860 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
19862 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
19863 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
19864 paired_divv2sf3): Similar.
19865 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
19866 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
19867 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
19869 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
19870 Rename the paired_* instruction patterns.
19871 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
19873 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
19874 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
19875 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
19876 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
19877 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
19878 PAIRED_VECTOR_MODE.
19879 (struct machine_function): Delete field spe_insn_chain_scanned_p.
19880 (spe_func_has_64bit_regs_p): Delete.
19881 (spe_expand_predicate_builtin): Delete.
19882 (spe_expand_evsel_builtin): Delete.
19883 (TARGET_DWARF_REGISTER_SPAN): Do not define.
19884 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
19885 (invalid_e500_subreg): Delete.
19886 (rs6000_legitimize_address): Always force_reg op2 as well, for
19887 paired single memory accesses.
19888 (rs6000_member_type_forces_blk): Delete.
19889 (rs6000_spe_function_arg): Delete.
19890 (rs6000_expand_unop_builtin): Delete SPE handling.
19891 (rs6000_expand_binop_builtin): Ditto.
19892 (spe_expand_stv_builtin): Delete.
19893 (bdesc_2arg_spe): Delete.
19894 (spe_expand_builtin): Delete.
19895 (spe_expand_predicate_builtin): Delete.
19896 (spe_expand_evsel_builtin): Delete.
19897 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
19898 (spe_init_builtins): Delete.
19899 (spe_func_has_64bit_regs_p): Delete.
19900 (savres_routine_name): Delete "info" parameter. Adjust callers.
19901 (rs6000_emit_stack_reset): Ditto.
19902 (rs6000_dwarf_register_span): Delete.
19903 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
19904 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
19905 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
19907 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
19909 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
19910 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
19911 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
19912 mulv2sf3, divv2sf3): Delete expanders.
19914 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19916 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
19918 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19920 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
19921 * config/rs6000/rs6000.c: Ditto.
19923 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19925 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
19926 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
19927 comparison_operator.
19929 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19931 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
19932 * config/rs6000/rs6000.opt: Ditto.
19933 * config/rs6000/t-rtems: Ditto.
19935 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19937 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
19938 TARGET_E500_SINGLE by 0, simplify.
19939 * config/rs6000/rs6000.c: Ditto.
19940 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
19941 (spe_build_register_parallel): Delete.
19942 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
19943 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
19944 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
19945 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
19946 (E500_CONVERT): Delete.
19947 * config/rs6000/spe.md: Remove many patterns and all define_constants.
19949 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19951 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
19952 * config/rs6000/dfp.md: Ditto.
19953 (negdd2, *negdd2_fpr): Merge.
19954 (absdd2, *absdd2_fpr): Merge.
19955 (negtd2, *negtd2_fpr): Merge.
19956 (abstd2, *abstd2_fpr): Merge.
19957 * config/rs6000/e500.h: Delete file.
19958 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
19959 TARGET_FPRS by 1 and simplify.
19960 * config/rs6000/rs6000-c.c: Ditto.
19961 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
19962 TARGET_DF_SPE by 0.
19963 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
19965 * config/rs6000/rs6000.md: Ditto.
19966 (floatdidf2, *floatdidf2_fpr): Merge.
19967 (move_from_CR_gt_bit): Delete.
19968 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
19969 (E500_CR_IOR_COMPARE): Delete.
19970 (All patterns that require !TARGET_FPRS): Delete.
19971 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
19973 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19975 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
19977 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19979 * graphds.c (add_edge): Intitialize edge's attached data.
19980 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
19981 pointer parameter. Call pointed function on each edge during
19982 graph traversing. Skip traversing the edge when the function
19984 (graphds_dfs, graphds_scc): Ditto.
19985 (for_each_edge): New parameter. Pass the new parameter to callback
19987 * graphds.h (skip_edge_callback): New function pointer type.
19988 (graphds_dfs, graphds_scc): New function pointer parameter.
19989 (graphds_edge_callback, for_each_edge): New parameter.
19991 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19993 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
19994 out code checking if runtime alias check is possible to below ...
19995 Call the new function.
19996 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
19997 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
19999 2017-06-07 Marek Polacek <polacek@redhat.com>
20002 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
20003 TYPE_OVERFLOW_WRAPS check.
20005 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20007 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
20008 if versioning is required.
20009 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
20010 peeling with the check for versioning.
20012 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20014 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
20015 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
20016 Set true to new parameter if new ssa variable is defined.
20017 (vect_gen_vector_loop_niters): Refactor. Set range information
20018 for the new vector loop bound variable.
20019 (vect_do_peeling): Ditto.
20021 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20023 * tree-affine.c (ssa.h): Include header file.
20024 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
20025 has wrapping overflow behavior.
20027 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20029 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
20031 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20033 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
20034 (tree_to_aff_combination): ... here.
20036 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20038 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
20039 reg_pressure model function.
20040 (ivopts_global_cost_for_size): Delete.
20041 (determine_set_costs, iv_ca_recount_cost): Call new model function
20042 ivopts_estimate_reg_pressure.
20044 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20046 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
20047 expensive than udiv. Remove floating point cases from mod.
20049 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20051 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
20052 Increase idiv cost.
20054 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20056 * config/aarch64/aarch64.md
20057 (copysignsf3): Fix mask generation.
20059 2017-06-07 Jakub Jelinek <jakub@redhat.com>
20061 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
20063 (class dump_manager): Add register_dumps method.
20064 * dumpfile.c: Include langhooks.h.
20065 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
20066 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
20067 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
20068 (dump_manager::dump_register): Start with 512 entries instead of 32.
20069 (dump_manager::register_dumps): New method.
20070 * toplev.c (general_init): Instead of invoking register_dumps
20071 langhook, invoke register_dumps method on the dump manager.
20072 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
20075 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
20077 * doc/md.texi: Clarify the restrictions on a define_insn condition.
20078 Say that # requires an associated define_split to exist, and that
20079 the define_split must be suitable for use after register allocation.
20081 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20083 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
20084 (compute_outgoing_frequencies): Also initialize zero counts.
20085 (find_many_sub_basic_blocks): Do not produce uninitialized profile
20086 around loops; preserve more of profile when nothing changes.
20088 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
20090 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
20092 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
20093 * config/arm/arm-cpu-cdata.h: Regenerate.
20094 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
20095 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
20096 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
20097 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
20098 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
20100 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
20102 * config/arm/t-rmprofile: Likewise.
20103 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
20105 2017-06-06 David S. Miller <davem@davemloft.net>
20108 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
20109 blockage if function uses alloca.
20111 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20113 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
20114 New "uid" fields to hold pretty-print IDs of group and ref.
20115 Memory references are now identified as <group_id>:<ref_id>
20116 instead of using [random] addresses.
20117 (dump_mem_details): Simplify, no functional change.
20118 (dump_mem_ref): Simplify and make output more concise.
20119 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
20120 (find_or_create_group): Initialize group uid.
20121 (record_ref): Initialize ref uid. Improve debug output.
20122 (prune_group_by_reuse, should_issue_prefetch_p,)
20123 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
20124 (mark_nontemporal_store, determine_loop_nest_reuse):
20125 Improve debug output.
20127 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20129 * dbgcnt.def (prefetch): New debug counter.
20130 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
20131 (schedule_prefetches): Stop issueing prefetches if debug counter
20134 2017-06-06 Tom de Vries <tom@codesourcery.com>
20136 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
20137 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
20139 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20141 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
20142 Use aarch64_reg_or_zero predicate for operand 4.
20143 (aarch64_compare_and_swap<mode> define_insn_and_split):
20144 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
20145 (aarch64_store_exclusive<mode>): Likewise for operand 2.
20147 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
20149 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
20150 (arm_compute_save_core_reg_mask): This.
20151 (thumb1_compute_save_reg_mask): Rename into ...
20152 (thumb1_compute_save_core_reg_mask): This.
20153 (arm_compute_save_reg0_reg12_mask): Adapt comment.
20154 (arm_compute_frame_layout): Likewise.
20156 2017-06-06 Richard Biener <rguenther@suse.de>
20158 PR tree-optimization/80974
20159 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
20160 keep or clear leaders SSA info.
20162 2017-06-06 Tom de Vries <tom@codesourcery.com>
20164 * config/nvptx/nvptx.c (split_mode_p): New function.
20165 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
20167 2017-06-06 Tom de Vries <tom@codesourcery.com>
20169 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
20171 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20174 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
20177 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20179 * shrink-wrap.c (handle_simple_exit): Update profile.
20180 (try_shrink_wrapping): Upate profile.
20182 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20184 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
20185 (tree_guess_outgoing_edge_probabilities): New.
20186 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
20187 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
20189 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20191 * ipa-split.c (split_function): Initialize return bb profile.
20193 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20195 * profile.c (compute_branch_probabilities): Also initialize
20196 EXIT_BLOCK profile.
20198 2017-06-06 Richard Biener <rguenther@suse.de>
20200 PR tree-optimization/80928
20201 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
20202 (vect_analyze_loop_operations): Properly guard analysis for
20204 (vect_transform_loop): Likewise.
20205 (vect_analyze_loop_2): Also reset SLP type on PHIs.
20206 (vect_model_induction_cost): Do not cost for pure SLP.
20207 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
20208 of induction in inner loop vectorization.
20209 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
20210 (vect_get_and_check_slp_defs): Handle vect_induction_def.
20211 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
20213 (vect_analyze_slp_cost_1): Cost induction.
20214 (vect_detect_hybrid_slp_stmts): Handle PHIs.
20215 (vect_get_slp_vect_defs): Likewise.
20216 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
20217 (vect_transform_stmt): Handle SLP reductions.
20218 * tree-vectorizer.h (vectorizable_induction): Adjust.
20220 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
20222 * config/rs6000/rs6000.c (make_resolver_func): Update
20223 init_lowered_empty_function call.
20225 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
20227 * doc/invoke.texi: Document the -fprofile-abs-path option.
20228 * common.opt (fprofile-abs-path): New option.
20229 * gcov-io.h (gcov_write_filename): Declare.
20230 * gcov-io.c (gcov_write_filename): New function.
20231 * coverage.c (coverage_begin_function): Use gcov_write_filename.
20232 * profile.c (output_location): Likewise.
20234 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20236 * shring-wrap.c: Revert accidental commit.
20238 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
20240 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
20242 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20244 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
20246 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
20247 profile in callgraph edge.
20248 * profile-count.h (apply_probability): If THIS is 0, then result is 0
20249 (apply_scale): Likewise.
20250 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
20251 Also scale profile when inlining function with zero profile.
20252 (initialize_cfun): Update exit block profile even when it is zero.
20253 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
20254 when profile is read.
20256 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
20258 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
20259 (CLONE_*): New constants to define the processors we can generate
20260 code for with the target_clone attribute.
20261 (rs6000_clone_map): New array to identify which clone processors
20262 the current program is running on.
20263 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
20264 target_clone attribute.
20265 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
20266 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
20267 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
20268 (cpu_expand_builtin): Add support for target_clone attribute.
20269 (rs6000_valid_attribute_p): Allow "default" attribute.
20270 (get_decl_name): New debug function to simplify printing the
20271 current function name in debugging statements.
20272 (rs6000_clone_priority): New functions to support the target_clone
20273 attribute, and be able to generate code to switch between ISA 2.05
20274 through ISA 3.0 (power6 through power9).
20275 (rs6000_compare_version_priority): Likewise.
20276 (rs6000_get_function_versions_dispatcher): Likewise.
20277 (make_resolver_func): Likewise.
20278 (add_condition_to_bb): Likewise.
20279 (dispatch_function_versions): Likewise.
20280 (rs6000_generate_version_dispatcher_body): Likewise.
20281 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
20282 (fusion_gpr_load_p): Fix a spacing issue.
20283 * doc/extend.texi (Common Function Attributes): Document that the
20284 PowerPC supports the target_clone attribute.
20286 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
20288 * config/arm/arm.h: explain F symbol found in description of ARM
20289 register allocation in its legend.
20291 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20293 * config/mips/frame-header-opt.c: Include profile-count.h.
20294 * config/riscv/riscv.c: Include profile-count.h
20296 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20298 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
20300 (sm_set_flag_if_changed): Add bbs field.
20301 (execute_sm_if_changed_flag_set): Pass BBS.
20302 (execute_sm): Update.
20304 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20306 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
20309 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20311 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
20312 (peephole2): New peephole2 to emit the above.
20313 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
20315 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20317 * config/aarch64/aarch64.c (define_peephole2 above
20318 *sub_<shift>_<mode>): New peephole.
20320 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
20322 * config/i386/i386.c (make_resolver_func): Update.
20323 * Makefile.in: Add profile-count.h and profile-count.o
20324 * auto-profile.c (afdo_indirect_call): Update to new API.
20325 (afdo_set_bb_count): Update.
20326 (afdo_propagate_edge): Update.
20327 (afdo_propagate_circuit): Update.
20328 (afdo_calculate_branch_prob): Update.
20329 (afdo_annotate_cfg): Update.
20330 * basic-block.h: Include profile-count.h
20331 (struct edge_def): Turn count to profile_count.
20332 (struct basic_block_def): Likewie.
20333 (REG_BR_PROB_BASE): Move to profile-count.h
20334 (RDIV): Move to profile-count.h
20335 * bb-reorder.c (max_entry_count): Turn to profile_count.
20336 (find_traces): Update.
20337 (rotate_loop):Update.
20338 (connect_traces):Update.
20339 (sanitize_hot_paths):Update.
20340 * bt-load.c (migrate_btr_defs): Update.
20341 * cfg.c (RDIV): Remove.
20342 (init_flow): Use alloc_block.
20343 (alloc_block): Uninitialize count.
20344 (unchecked_make_edge): Uninitialize count.
20345 (check_bb_profile): Update.
20346 (dump_edge_info): Update.
20347 (dump_bb_info): Update.
20348 (update_bb_profile_for_threading): Update.
20349 (scale_bbs_frequencies_int): Update.
20350 (scale_bbs_frequencies_gcov_type): Update.
20351 (scale_bbs_frequencies_profile_count): New.
20352 * cfg.h (update_bb_profile_for_threading): Update.
20353 (scale_bbs_frequencies_profile_count): Declare.
20354 * cfgbuild.c (compute_outgoing_frequencies): Update.
20355 (find_many_sub_basic_blocks): Update.
20356 * cfgcleanup.c (try_forward_edges): Update.
20357 (try_crossjump_to_edge): Update.
20358 * cfgexpand.c (expand_gimple_tailcall): Update.
20359 (construct_exit_block): Update.
20360 * cfghooks.c (verify_flow_info): Update.
20361 (dump_bb_for_graph): Update.
20362 (split_edge): Update.
20363 (make_forwarder_block): Update.
20364 (duplicate_block): Update.
20365 (account_profile_record): Update.
20366 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
20367 (get_estimated_loop_iterations): Update.
20368 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
20369 (single_likely_exit): Update.
20370 * cfgloopmanip.c (scale_loop_profile): Update.
20372 (set_zero_probability): Update.
20373 (lv_adjust_loop_entry_edge): Update.
20374 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
20375 (purge_dead_edges): Update.
20376 (rtl_account_profile_record): Update.
20377 * cgraph.c (cgraph_node::create): Uninitialize count.
20378 (symbol_table::create_edge): Uninitialize count.
20379 (cgraph_update_edges_for_call_stmt_node): Update.
20380 (cgraph_edge::dump_edge_flags): Update.
20381 (cgraph_node::dump): Update.
20382 (cgraph_edge::maybe_hot_p): Update.
20383 * cgraph.h: Include profile-count.h
20384 (create_clone), create_edge, create_indirect_edge): Update.
20385 (cgraph_node): Turn count to profile_count.
20386 (cgraph_edge0: Likewise.
20387 (make_speculative, clone): Update.
20388 (create_edge): Update.
20389 (init_lowered_empty_function): Update.
20390 * cgraphclones.c (cgraph_edge::clone): Update.
20391 (duplicate_thunk_for_node): Update.
20392 (cgraph_node::create_clone): Update.
20393 * cgraphunit.c (cgraph_node::analyze): Update.
20394 (cgraph_node::expand_thunk): Update.
20395 * final.c (dump_basic_block_info): Update.
20396 * gimple-streamer-in.c (input_bb): Update.
20397 * gimple-streamer-out.c (output_bb): Update.
20398 * graphite.c (print_global_statistics): Update.
20399 (print_graphite_scop_statistics): Update.
20400 * hsa-brig.c: Include basic-block.h.
20401 * hsa-dump.c: Include basic-block.h.
20402 * hsa-gen.c (T sum_slice): Update.
20403 (convert_switch_statements):Update.
20404 * hsa-regalloc.c: Include basic-block.h.
20405 * ipa-chkp.c (chkp_produce_thunks): Update.
20406 * ipa-cp.c (struct caller_statistics): Update.
20407 (init_caller_stats): Update.
20408 (gather_caller_stats): Update.
20409 (ipcp_cloning_candidate_p): Update.
20410 (good_cloning_opportunity_p): Update.
20411 (get_info_about_necessary_edges): Update.
20412 (dump_profile_updates): Update.
20413 (update_profiling_info): Update.
20414 (update_specialized_profile): Update.
20415 (perhaps_add_new_callers): Update.
20416 (decide_about_value): Update.
20417 (ipa_cp_c_finalize): Update.
20418 * ipa-devirt.c (struct odr_type_warn_count): Update.
20419 (struct decl_warn_count): Update.
20420 (struct final_warning_record): Update.
20421 (possible_polymorphic_call_targets): Update.
20422 (ipa_devirt): Update.
20423 * ipa-fnsummary.c (redirect_to_unreachable): Update.
20424 * ipa-icf.c (sem_function::merge): Update.
20425 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
20426 * ipa-inline.c (compute_uninlined_call_time): Update.
20427 (compute_inlined_call_time): Update.
20428 (want_inline_small_function_p): Update.
20429 (want_inline_self_recursive_call_p): Update.
20430 (edge_badness): Update.
20431 (lookup_recursive_calls): Update.
20432 (recursive_inlining): Update.
20433 (inline_small_functions): Update.
20434 (dump_overall_stats): Update.
20435 (dump_inline_stats): Update.
20436 * ipa-profile.c (ipa_profile_generate_summary): Update.
20437 (ipa_propagate_frequency): Update.
20438 (ipa_profile): Update.
20439 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
20440 * ipa-utils.c (ipa_merge_profiles): Update.
20441 * loop-doloop.c (doloop_modify): Update.
20442 * loop-unroll.c (report_unroll): Update.
20443 (unroll_loop_runtime_iterations): Update.
20444 * lto-cgraph.c (lto_output_edge): Update.
20445 (lto_output_node): Update.
20446 (input_node): Update.
20447 (input_edge): Update.
20448 (merge_profile_summaries): Update.
20449 * lto-streamer-in.c (input_cfg): Update.
20450 * lto-streamer-out.c (output_cfg): Update.
20451 * mcf.c (create_fixup_graph): Update.
20452 (adjust_cfg_counts): Update.
20453 (sum_edge_counts): Update.
20454 * modulo-sched.c (sms_schedule): Update.
20455 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
20456 * predict.c (maybe_hot_count_p): Update.
20457 (probably_never_executed): Update.
20458 (dump_prediction): Update.
20459 (combine_predictions_for_bb): Update.
20460 (propagate_freq): Update.
20461 (handle_missing_profiles): Update.
20462 (counts_to_freqs): Update.
20463 (rebuild_frequencies): Update.
20464 (force_edge_cold): Update.
20465 * predict.h: Include profile-count.h
20466 (maybe_hot_count_p, counts_to_freqs): UPdate.
20467 * print-rtl-function.c: Do not include cfg.h
20468 * print-rtl.c: Include basic-block.h
20469 * profile-count.c: New file.
20470 * profile-count.h: New file.
20471 * profile.c (is_edge_inconsistent): Update.
20472 (correct_negative_edge_counts): Update.
20473 (is_inconsistent): Update.
20474 (set_bb_counts): Update.
20475 (read_profile_edge_counts): Update.
20476 (compute_frequency_overlap): Update.
20477 (compute_branch_probabilities): Update; Initialize and deinitialize
20479 (branch_prob): Update.
20480 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
20481 (edge_gcov_count): New.
20482 (bb_gcov_count): New.
20483 * shrink-wrap.c (try_shrink_wrapping): Update.
20484 * tracer.c (better_p): Update.
20485 * trans-mem.c (expand_transaction): Update.
20486 (ipa_tm_insert_irr_call): Update.
20487 (ipa_tm_insert_gettmclone_call): Update.
20488 * tree-call-cdce.c: Update.
20489 * tree-cfg.c (gimple_duplicate_sese_region): Update.
20490 (gimple_duplicate_sese_tail): Update.
20491 (gimple_account_profile_record): Update.
20492 (execute_fixup_cfg): Update.
20493 * tree-inline.c (copy_bb): Update.
20494 (copy_edges_for_bb): Update.
20495 (initialize_cfun): Update.
20496 (freqs_to_counts): Update.
20497 (copy_cfg_body): Update.
20498 (expand_call_inline): Update.
20499 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
20500 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
20501 (try_unroll_loop_completely): Update.
20502 (try_peel_loop): Update.
20503 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
20504 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
20505 * tree-ssa-loop-split.c (connect_loops): Update.
20506 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
20507 * tree-ssa-reassoc.c (branch_fixup): Update.
20508 * tree-ssa-tail-merge.c (replace_block_by): Update.
20509 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
20510 (compute_path_counts): Update.
20511 (update_profile): Update.
20512 (recompute_probabilities): Update.
20513 (update_joiner_offpath_counts): Update.
20514 (estimated_freqs_path): Update.
20515 (freqs_to_counts_path): Update.
20516 (clear_counts_path): Update.
20517 (ssa_fix_duplicate_block_edges): Update.
20518 (duplicate_thread_path): Update.
20519 * tree-switch-conversion.c (case_bit_test_cmp): Update.
20520 (struct switch_conv_info): Update.
20521 * tree-tailcall.c (decrease_profile): Update.
20522 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
20523 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
20524 * value-prof.c (check_counter): Update.
20525 (gimple_divmod_fixed_value): Update.
20526 (gimple_mod_pow2): Update.
20527 (gimple_mod_subtract): Update.
20528 (gimple_ic_transform): Update.
20529 (gimple_stringop_fixed_value): Update.
20530 * value-prof.h (gimple_ic): Update.
20532 2017-06-02 Carl Love <cel@us.ibm.com>
20534 * config/rs6000/rs6000-c: Add support for built-in functions
20535 vector double vec_doublee (vector signed int);
20536 vector double vec_doublee (vector unsigned int);
20537 vector double vec_doublee (vector float);
20538 vector double vec_doubleh (vector signed int);
20539 vector double vec_doubleh (vector unsigned int);
20540 vector double vec_doubleh (vector float);
20541 vector double vec_doublel (vector signed int);
20542 vector double vec_doublel (vector unsigned int);
20543 vector double vec_doublel (vector float);
20544 vector double vec_doubleo (vector signed int);
20545 vector double vec_doubleo (vector unsigned int);
20546 vector double vec_doubleo (vector float);.
20547 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
20548 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
20550 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
20551 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
20552 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
20554 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
20555 vec_doublel, vec_doubleh.
20556 * doc/extend.texi: Update the built-in documentation file for the
20557 new built-in functions.
20559 2017-06-02 David Malcolm <dmalcolm@redhat.com>
20562 * ipa-inline-analysis.c (free_growth_caches): Set
20563 edge_removal_hook_holder to NULL after removing it.
20565 2017-06-02 Sudakshina Das <sudi.das@arm.com>
20567 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
20568 comparision with zero.
20570 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
20571 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
20572 for early expansion of vec_min and vec_max builtins.
20573 (builtin_function_type): Add min/max unsigned variants to those
20574 identified as having unsigned arguments.
20576 2017-06-02 Olivier Hainque <hainque@adacore.com>
20578 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
20580 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20582 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
20583 Use VALL_F16 iterator rather than VALL.
20585 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20587 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
20588 Emit CBNZ inside loop when doing a strong exchange and comparing
20589 against zero. Generate the CC flags after the loop.
20591 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
20593 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
20594 (dl_section_ref): New.
20595 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
20596 On AIX, append an expression to subtract the size of the
20597 section length to dl_section_ref.
20599 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
20601 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
20602 for early expansion of vector absolute builtins.
20604 2017-06-02 Richard Biener <rguenther@suse.de>
20606 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
20607 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
20609 2017-06-02 Richard Biener <rguenther@suse.de>
20611 PR tree-optimization/80948
20612 * tree-tailcall.c (find_tail_calls): Track stmts to move in
20613 stmt order as well.
20615 2017-06-02 Richard Biener <rguenther@suse.de>
20617 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
20619 * tree-vect-stmts.c (process_use): Do not mark backedge defs
20620 for inductions as relevant.
20622 2017-06-02 Richard Biener <rguenther@suse.de>
20624 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
20625 (vectorizable_induction): ... this. Remove dead code.
20627 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
20629 * builtins. (expand_builtin_alloca): Remove second parameter and
20630 infer its value from the first parameter instead.
20631 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
20633 2017-06-02 Jakub Jelinek <jakub@redhat.com>
20635 PR rtl-optimization/80903
20636 * loop-doloop.c (add_test): Unshare sequence.
20638 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
20640 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
20642 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
20644 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
20646 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
20647 xlogue_layout::get_instance, logue_layout::xlogue_layout,
20648 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
20649 (xlogue_layout::get_stub_rtx): Make static.
20650 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
20651 (xlogue_layout::compute_stub_managed_regs): Rename to...
20652 (xlogue_layout::count_stub_managed_regs): ...this.
20653 (xlogue_layout::is_stub_managed_reg): New function.
20654 (xlogue_layout::m_stub_names): Rename to...
20655 (xlogue_layout::s_stub_names): ...this, make static.
20656 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
20657 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
20658 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
20659 xlogue_layout::s_stub_names): Instantiate statics.
20660 (stub_managed_regs): Remove.
20661 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
20662 (disable_call_ms2sysv_xlogues): Rename to...
20663 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
20664 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
20666 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
20667 change after reload_completed.
20668 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
20670 (ix86_expand_prologue): Likewise.
20671 (ix86_expand_epilogue): Likewise.
20672 (ix86_expand_split_stack_prologue): Likewise.
20673 (ix86_compute_frame_layout): Remove frame parameter ...
20674 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
20675 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
20677 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
20678 (ix86_frame): Move from here ...
20679 * config/i386/i386.h (ix86_frame): ... to here.
20680 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
20681 complete ix86_frame data structure instead. Remove some_ld_name.
20683 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
20685 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
20686 symbols that hold a DECL_VALUE_EXPR.
20688 2017-06-01 Martin Jambor <mjambor@suse.cz>
20690 PR tree-optimization/80898
20691 * tree-sra.c (process_subtree_disqualification): Removed.
20692 (disqualify_candidate): Do not acll
20693 process_subtree_disqualification.
20694 (subtree_mark_written_and_enqueue): New function.
20695 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
20696 RHS has been disqualified and re-queue LHS if necessary. Apart
20697 from that, ignore disqualified RHS.
20699 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20701 * config/s390/s390.c (s390_emit_epilogue): Disable early return
20702 address fetch for z10 or later.
20704 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20706 * config/arc/arc.md (tst_movb): Add guard when splitting.
20708 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20710 * config/arc/arc.c (arc_can_eliminate): Test against
20711 arc_frame_pointer_needed.
20713 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20715 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
20716 to prevent store reordering.
20717 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
20718 (type): Add block type.
20719 (stack_tie): Define special instruction to be used in
20722 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20724 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
20725 constraint. It is not valid for the pattern.
20726 (noncommutative_binary_comparison): Likewise.
20728 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20730 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
20733 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20735 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
20736 be used by the reg-alloc.
20738 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20740 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
20741 reg-alloc when having mul64 or mul32x16 instructions.
20742 (mulsidi3): Likewise.
20743 (umulsidi3): Likewise.
20744 (mulsi32x16): New pattern.
20745 (mulsi64): Likewise.
20746 (mulsidi64): Likewise.
20747 (umulsidi64): Likewise.
20748 (MUL32x16_REG): Define.
20749 (mul64_600): Use MUL32x16_REG.
20750 (mac64_600): Likewise.
20751 (umul64_600): Likewise.
20752 (umac64_600): Likewise.
20754 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20756 * config/arc/arc.md (mulsi3_700): Make it commutative.
20758 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
20760 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
20762 (*sign_extendsidi2_insn): Likewise for movstosw.
20764 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
20766 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
20767 the type of the input discriminant value. Convert the
20768 discriminant value of signedness vary.
20770 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
20772 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
20773 Add to -Wall section.
20775 2017-06-01 Richard Biener <rguenther@suse.de>
20777 PR middle-end/66313
20778 * fold-const.c (fold_plusminus_mult_expr): If the factored
20779 factor may be zero use a wrapping type for the inner operation.
20780 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
20781 and handle moved defs.
20782 (process_assignment): Properly guard the unary op case. Return a
20783 tri-state indicating that moving the stmt before the call may allow
20784 to continue. Pass through to_move.
20785 (find_tail_calls): Handle moving unrelated defs before
20788 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
20791 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
20792 splitter result in the canonical way.
20794 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
20796 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
20797 also for 32bit target. Update insn attributes.
20798 (zero-extendsidi2 splitter): Allow all registers for operand 1.
20800 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
20802 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
20803 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
20804 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
20805 (_mm_maskz_min_ss): New intrinsics.
20807 2017-05-31 Martin Liska <mliska@suse.cz>
20809 * tree-vect-loop.c (vect_create_epilog_for_reduction):
20810 Change comment style to one we normally use.
20811 (vectorizable_reduction): Likewise.
20812 (vectorizable_induction): Likewise.
20813 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
20814 (vectorizable_call): Likewise.
20815 (vectorizable_simd_clone_call): Likewise.
20816 (vectorizable_conversion): Likewise.
20817 (vectorizable_assignment): Likewise.
20818 (vectorizable_shift): Likewise.
20819 (vectorizable_operation): Likewise.
20820 (vectorizable_store): Likewise.
20821 (vectorizable_load): Likewise.
20822 * tree-vectorizer.h: Likewise.
20824 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
20826 * passes.c (emergency_dump_function): New.
20827 * tree-pass.h (emergency_dump_function): Declare.
20828 * plugin.c (plugins_internal_error_function): Remove.
20829 * plugin.h (plugins_internal_error_function): Remove declaration.
20830 * toplev.c (internal_error_function): New static function. Use it...
20831 (general_init): ...here.
20833 2017-05-31 Graham Markall <graham.markall@embecosm.com>
20835 * config/arc/arc.c (arc_print_operand): Handle constant operands.
20836 (arc_rtx_costs): Add costs for new patterns.
20837 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
20838 * config/arc/predicates.md: Add _1_2_3_operand predicate.
20840 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
20842 * tree-ssa-strlen.c (get_next_strinfo): New function.
20843 (get_stridx_plus_constant): Use it.
20844 (zero_length_string): Likewise.
20845 (adjust_related_strinfos): Likewise.
20846 (adjust_last_stmt): Likewise.
20848 2017-05-31 Richard Biener <rguenther@suse.de>
20851 * config/i386/i386.c (ix86_expand_builtin): Remove assert
20852 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
20854 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
20856 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
20857 loop_vinfo argument and use of dependence distance vectors.
20858 Check instead whether the two references differ only in their
20859 initial value and assume that they have the same alignment if the
20860 difference is a multiple of the vector alignment.
20861 (vect_analyze_data_refs_alignment): Update call accordingly.
20863 2017-05-31 Martin Liska <mliska@suse.cz>
20866 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
20868 2017-05-31 Bin Cheng <bin.cheng@arm.com>
20870 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
20871 (create_intersect_range_checks): Move from ...
20872 * tree-data-ref.c (create_intersect_range_checks_index)
20873 (create_intersect_range_checks): ... to here.
20874 (create_runtime_alias_checks): New function factored from ...
20875 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
20876 here. Call above function.
20877 * tree-data-ref.h (create_runtime_alias_checks): New function.
20879 2017-05-31 Bin Cheng <bin.cheng@arm.com>
20881 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
20882 segment length for dr_b and compute it in wide_int.
20884 2017-05-31 Richard Biener <rguenther@suse.de>
20886 PR tree-optimization/80906
20887 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
20888 and pass through iv_map.
20889 (copy_bb_and_scalar_dependences): Adjust.
20890 (translate_pending_phi_nodes): Likewise.
20891 (copy_loop_close_phi_args): Handle code-generating IVs instead
20894 2017-05-30 David Malcolm <dmalcolm@redhat.com>
20896 * diagnostic-color.c (color_dict): Add "type-diff".
20897 (parse_gcc_colors): Update comment.
20898 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
20899 -fdiagnostics-show-template-tree and -fno-elide-type.
20900 (GCC_COLORS): Add type-diff to example.
20902 (-fdiagnostics-show-template-tree): New.
20903 (-fno-elide-type): New.
20904 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
20905 the pp_format_decoder callback. Call any m_format_postprocessor's
20907 (pretty_printer::pretty_printer): Initialize
20908 m_format_postprocessor.
20909 (pretty_printer::~pretty_printer): Delete any
20910 m_format_postprocessor.
20911 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
20912 (class format_postprocessor): New class.
20913 (struct pretty_printer::format_decoder): Document the new parameters.
20914 (struct pretty_printer::m_format_postprocessor): New field.
20915 * tree-diagnostic.c (default_tree_printer): Update for new
20916 bool and const char ** params.
20917 * tree-diagnostic.h (default_tree_printer): Likewise.
20919 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
20921 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
20922 (lwa_operand): Delete rs6000_gen_cell_microcode test.
20923 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
20924 rs6000_gen_cell_microcode code.
20925 (rs6000_final_prescan_insn): Delete.
20926 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
20927 "warn-cell-microcode" entries.
20928 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
20929 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
20930 throughout. Change cc_reg_not_micro_cr0_operand to
20931 cc_reg_not_cr0_operand throughout.
20932 (*extendhi<mode>2_noload): Delete.
20933 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
20934 (mwarn-cell-microcode): Delete.
20935 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
20936 -mgen-cell-microcode and -mwarn-cell-microcode.
20938 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
20941 * config/i386/constraints.md (Yd): New constraint.
20943 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
20944 and (?Yd, r) alternatives. Update insn attributes.
20945 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
20946 and (?*Yd, r) alternatives. Update insn attributes.
20947 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
20949 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
20951 * gimplify.c (gimplify_modify_expr): Don't create a
20952 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
20955 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
20957 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
20959 2017-05-30 Richard Biener <rguenther@suse.de>
20961 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
20962 and reduc_def fields.
20963 (STMT_VINFO_REDUC_TYPE): New define.
20964 (STMT_VINFO_REDUC_DEF): Likewise.
20965 (vect_force_simple_reduction): Adjust prototype.
20966 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
20967 (vect_is_simple_reduction): Remove check_reduction argument.
20968 (vect_force_simple_reduction): Adjust and set
20969 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
20970 (vectorizable_reduction): Do not re-do reduction analysis
20971 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
20972 * tree-parloops.c (gather_scalar_reductions): Adjust.
20974 2017-05-30 Richard Biener <rguenther@suse.de>
20976 PR middle-end/80901
20977 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
20980 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
20982 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
20983 Introduce unknown_misalignment parameter and remove vf.
20984 (vect_peeling_hash_get_lowest_cost):
20985 Pass unknown_misalignment parameter.
20986 (vect_enhance_data_refs_alignment):
20987 Fix unsupportable data ref treatment.
20989 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
20991 * tree-vect-data-refs.c (vect_get_data_access_cost):
20992 Workaround for SLP handling.
20993 (vect_enhance_data_refs_alignment):
20994 Compute costs for doing no peeling at all, compare to the best
20995 peeling costs so far and avoid peeling if cheaper.
20997 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
20999 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
21000 Return peeling info and set costs to zero for unlimited cost
21002 (vect_enhance_data_refs_alignment): Also inspect all datarefs
21003 with unknown misalignment. Compute and costs for unknown
21004 misalignment, compare them to the costs for known misalignment
21005 and choose the cheapest for peeling.
21007 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21009 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
21010 (vect_get_peeling_costs_all_drs): Create function.
21011 (vect_peeling_hash_get_lowest_cost):
21012 Use vect_get_peeling_costs_all_drs.
21013 (vect_peeling_supportable): Create function.
21014 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
21016 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21018 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
21019 DR_HAS_NEGATIVE_STEP.
21020 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
21021 (vect_enhance_data_refs_alignment): Use.
21022 (vect_duplicate_ssa_name_ptr_info): Use.
21023 * tree-vectorizer.h (dr_misalignment): Use.
21024 (known_alignment_for_access_p): Use.
21026 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21029 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
21031 (has_section_name): New function.
21033 2017-05-30 Martin Liska <mliska@suse.cz>
21036 * auto-profile.c (get_function_decl_from_block): Fix
21039 2017-05-30 Richard Biener <rguenther@suse.de>
21041 PR middle-end/80876
21042 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
21044 2017-05-30 Martin Liska <mliska@suse.cz>
21046 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
21047 * dumpfile.h (struct dump_file_info): Remove ctors.
21049 2017-05-30 Martin Liska <mliska@suse.cz>
21051 * predict.def: Fix GNU coding style.
21053 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21055 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
21056 Mark 'to' argument with ATTRIBUTE_UNUSED.
21058 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21060 * config/xtensa/xtensa.c (xtensa_emit_call): Use
21061 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
21062 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
21065 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
21067 * doc/install.texi (Options specification): Restore entry of
21068 --enable-sjlj-exceptions.
21070 2017-05-27 Michael Eager <eager@eagercon.com>
21073 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
21075 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
21077 * config/microblaze/microblaze.h
21078 (FIXED_REGISTERS): Update in macro.
21079 (CALL_USED_REGISTERS): Update in macro.
21081 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
21083 * doc/install.texi: Add links to macOS binary distributions.
21085 2017-05-27 Jakub Jelinek <jakub@redhat.com>
21089 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21091 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
21093 2017-05-26 Martin Liska <mliska@suse.cz>
21095 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
21097 2017-05-26 Martin Liska <mliska@suse.cz>
21099 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
21100 always leading ';; '.
21101 (dump_bb_info): Likewise.
21102 (brief_dump_cfg): Likewise.
21103 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
21104 * dumpfile.c: Remove usage of TDF_VERBOSE.
21105 * dumpfile.h (enum dump_kind): Likewise.
21106 (dump_gimple_bb_header): Do not use TDF_COMMENT.
21107 * print-tree.c (debug_verbose): Remove.
21108 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
21109 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
21110 * tree-diagnostic.c (default_tree_printer): Replace
21111 TDF_DIAGNOSTIC with TDF_SLIM.
21113 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21115 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
21116 in parameter loop, rather than loop_vinfo.
21117 (create_intersect_range_checks): Ditto.
21118 (vect_create_cond_for_alias_checks): Update call to above functions.
21120 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21122 PR tree-optimization/80815
21123 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
21124 for merging runtime alias checks. Handle negative DR_STEPs.
21126 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21128 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
21130 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
21131 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
21132 out code pruning runtime alias checks.
21133 * tree-data-ref.c (prune_runtime_alias_test_list): New function
21134 factored out from above.
21135 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
21137 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
21139 (prune_runtime_alias_test_list): New decalaration.
21141 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21143 * tree-vect-data-refs.c (compare_tree): Rename and move ...
21144 * tree-data-ref.c (data_ref_compare_tree): ... to here.
21145 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
21146 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
21147 (operator==, comp_dr_with_seg_len_pair): Ditto.
21148 (vect_prune_runtime_alias_test_list): Ditto.
21150 2017-05-26 Martin Liska <mliska@suse.cz>
21153 * params.def: Bound partial-inlining-entry-probability param.
21155 2017-05-26 Marek Polacek <polacek@redhat.com>
21158 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
21161 2017-05-26 Richard Biener <rguenther@suse.de>
21163 PR tree-optimization/80842
21164 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
21167 2017-05-26 Richard Biener <rguenther@suse.de>
21169 PR tree-optimization/80844
21170 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
21172 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
21174 * doc/md.texi (Machine Constraints): Update x86 family
21175 machine constraints section to match 'config/i386/constraints.md'.
21177 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
21179 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
21181 2017-05-25 Nathan Sidwell <nathan@acm.org>
21183 * doc/invoke.texi (--enable-languages): Update documentation.
21185 2017-05-25 Martin Liska <mliska@suse.cz>
21187 * dumpfile.c: Add TDF_FOLDING.
21188 * dumpfile.h (enum dump_kind): Likewise.
21189 * genmatch.c (dt_simplify::gen_1): Use it.
21191 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21193 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
21195 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21197 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
21198 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
21200 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21202 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
21203 * match.pd (X == C): Rewrite it here.
21204 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
21205 with_certain_nonzero_bits2): New predicates.
21206 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
21208 2017-05-24 Nathan Sidwell <nathan@acm.org>
21210 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
21213 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
21216 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
21218 * config/powerpcspe: New port. Files are copied from the rs6000
21219 port, with "rs6000" in filenames replaced by "powerpcspe".
21221 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
21223 PR rtl-optimization/80754
21224 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
21226 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
21228 * config/sparc/sparc.md (length): Return the correct value for -mflat
21229 sibcalls to match output_sibcall.
21231 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
21235 * config/rs6000/rs6000.c (struct machine_function): Add new field
21237 (rs6000_get_separate_components): Init that field, use it.
21238 (rs6000_components_for_bb): Use the field.
21240 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
21242 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
21244 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
21246 PR middle-end/80823
21247 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
21249 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21252 * config/s390/s390.c (s390_check_qrst_address): Check incoming
21253 address against address_operand predicate.
21254 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
21256 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
21258 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
21259 parameters passed indirectly.
21261 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
21263 * config/i386/i386.md (*movdi_internal): Remove SSE4
21264 alternative 18 (?r, *v). Update insn attributes.
21265 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
21266 Update insn attributes.
21267 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
21268 Update insn attributes.
21269 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
21270 alternative 1 (r, v). Remove isa attribute.
21271 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
21272 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
21273 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
21275 2017-05-23 Tom de Vries <tom@codesourcery.com>
21277 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
21280 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21282 * cgraphunit.c (symbol_table::process_new_functions): Update.
21283 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
21284 (inline_generate_summary): Rename to ...
21285 (ipa_fn_summary_generate): ... this one.
21286 (inline_read_summary): Rename to ...
21287 (ipa_fn_summary_read): ... this one.
21288 (inline_write_summary): Rename to ...
21289 (ipa_fn_summary_write): ... this one.
21290 (inline_free_summary): Rename to ...
21291 (ipa_free_fn_summary): ... this one.
21292 (pass_data_local_fn_summary, pass_local_fn_summary,
21293 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
21294 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
21295 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
21296 make_pass_ipa_fn_summary): New.
21297 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
21298 inline_write_summary, inline_free_summary): Remove.
21299 (ipa_free_fn_summary) : New.
21300 * ipa-inline.c (ipa_inline): Update.
21301 (pass_ipa_inline): Do not generate summaries.
21302 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
21304 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
21305 and add pass_ipa_fn_summary.
21306 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
21308 (make_pass_inline_parameters): Remove.
21310 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
21312 * omp-low.c (struct omp_context): Remove "default_kind" member.
21315 * omp-offload.c (execute_oacc_device_lower): Remove the
21316 parallelism dimensions function attributes for unparallelized
21317 OpenACC kernels constructs.
21319 2017-05-23 Martin Liska <mliska@suse.cz>
21321 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
21323 (cgraph_edge::make_speculative): Likewise.
21324 (cgraph_edge::resolve_speculation): Likewise.
21325 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
21326 (cgraph_node::dump): Likewise.
21327 * cgraph.h: Likewise.
21328 * cgraphunit.c (analyze_functions): Likewise.
21329 (symbol_table::compile): Likewise.
21330 * ipa-cp.c (print_all_lattices): Likewise.
21331 (determine_versionability): Likewise.
21332 (initialize_node_lattices): Likewise.
21333 (ipcp_verify_propagated_values): Likewise.
21334 (estimate_local_effects): Likewise.
21335 (update_profiling_info): Likewise.
21336 (create_specialized_node): Likewise.
21337 (perhaps_add_new_callers): Likewise.
21338 (decide_about_value): Likewise.
21339 (decide_whether_version_node): Likewise.
21340 (identify_dead_nodes): Likewise.
21341 (ipcp_store_bits_results): Likewise.
21342 * ipa-devirt.c (dump_targets): Likewise.
21343 (ipa_devirt): Likewise.
21344 * ipa-icf.c (sem_item::dump): Likewise.
21345 (sem_function::equals): Likewise.
21346 (sem_variable::equals): Likewise.
21347 (sem_item_optimizer::read_section): Likewise.
21348 (sem_item_optimizer::execute): Likewise.
21349 (congruence_class::dump): Likewise.
21350 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
21351 (dump_inline_summary): Likewise.
21352 (estimate_node_size_and_time): Likewise.
21353 (inline_analyze_function): Likewise.
21354 * ipa-inline-transform.c (inline_call): Likewise.
21355 * ipa-inline.c (report_inline_failed_reason): Likewise.
21356 (want_early_inline_function_p): Likewise.
21357 (edge_badness): Likewise.
21358 (update_edge_key): Likewise.
21359 (inline_small_functions): Likewise.
21360 * ipa-profile.c (ipa_profile): Likewise.
21361 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
21362 (ipa_make_edge_direct_to_target): Likewise.
21363 (remove_described_reference): Likewise.
21364 (ipa_impossible_devirt_target): Likewise.
21365 (propagate_controlled_uses): Likewise.
21366 (ipa_print_node_params): Likewise.
21367 (ipcp_transform_function): Likewise.
21368 * ipa-pure-const.c (pure_const_read_summary): Likewise.
21369 (propagate_pure_const): Likewise.
21370 * ipa-reference.c (generate_summary): Likewise.
21371 (read_write_all_from_decl): Likewise.
21372 (propagate): Likewise.
21373 (ipa_reference_read_optimization_summary): Likewise.
21374 * ipa-utils.c (ipa_merge_profiles): Likewise.
21375 * ipa.c (walk_polymorphic_call_targets): Likewise.
21376 (symbol_table::remove_unreachable_nodes): Likewise.
21377 (ipa_single_use): Likewise.
21378 * passes.c (execute_todo): Likewise.
21379 * predict.c (drop_profile): Likewise.
21380 * symtab.c (symtab_node::get_dump_name): New function.
21381 (symtab_node::dump_name): Likewise.
21382 (symtab_node::dump_asm_name): Likewise.
21383 (symtab_node::dump_references): Likewise.
21384 (symtab_node::dump_referring): Likewise.
21385 (symtab_node::dump_base): Likewise.
21386 (symtab_node::debug_symtab): Likewise.
21387 * tree-sra.c (convert_callers_for_node): Likewise.
21388 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
21389 * value-prof.c (init_node_map): Likewise.
21391 2017-05-23 Martin Liska <mliska@suse.cz>
21393 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
21394 and symtab_node::debug_symtab to symbol_table::debug.
21395 * cgraphunit.c (analyze_functions): Use the renamed function.
21396 (symbol_table::compile): Likewise.
21397 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
21398 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
21399 * passes.c (execute_todo): Likewise.
21400 * symtab.c (symbol_table::dump): New function.
21401 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
21403 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21405 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
21406 that nonconst implies exec.
21408 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21410 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
21411 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
21412 (inline_edge_summary_vec): Turn into ...
21413 (ipa_call_summaries): ... this one.
21414 (redirect_to_unreachable, edge_set_predicate,
21415 evaluate_properties_for_edge, inline_summary_alloc,
21416 reset_ipa_call_summary, reset_inline_summary,
21417 inline_summary_t::duplicate): Update.
21418 (inline_edge_duplication_hook): Turn to ...
21419 (ipa_call_summary_t::duplicate): ... this one.
21420 (inline_edge_removal_hook): Turn to ...
21421 (ipa_call_summary_t::remove): ... this one.
21422 (dump_inline_edge_summary): Turn to ...
21423 (dump_ipa_call_summary): ... this one.
21424 (estimate_function_body_sizes): Update.
21425 (inline_update_callee_summaries): Update.
21426 (remap_edge_change_prob): Update.
21427 (remap_edge_summaries): Update.
21428 (inline_merge_summary): Update.
21429 (do_estimate_edge_time): Update.
21430 (inline_generate_summary): Update.
21431 (inline_read_section): Update.
21432 (inline_read_summary): Update.
21433 (inline_free_summary): Update.
21434 * ipa-inline.c (can_inline_edge_p): Update.
21435 (compute_inlined_call_time): Update.
21436 (want_inline_small_function_p): Update.
21437 (edge_badness): Update.
21438 (early_inliner): Update.
21439 * ipa-inline.h (inline_edge_summary): Turn to ...
21440 (ipa_call_summary): ... this one.
21441 (ipa_call_summary_t): New class.
21442 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
21443 (ipa_call_summaries): New.
21444 (inline_edge_summary): Remove.
21445 (estimate_edge_growth): Update.
21446 * ipa-profile.c (ipa_propagate_frequency_1): Update.
21447 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21448 * ipa-split.c (execute_split_functions): Update.
21449 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
21451 2017-05-23 Tom de Vries <tom@codesourcery.com>
21453 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
21454 attributes): Document rdrand effective target.
21456 2017-05-23 Tom de Vries <tom@codesourcery.com>
21458 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
21459 attributes): Sort alphabetically.
21461 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
21463 * config/avr/genmultilib.awk: Use gsub instead of gensub.
21465 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
21468 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
21469 V2DF/V2DI splat into two separate patterns, one that handles
21470 registers, and the other that only handles memory. Drop support
21471 for splatting from a GPR on ISA 2.07 and then splitting the
21472 splat into direct move and splat.
21473 (vsx_splat_<mode>_reg): Likewise.
21474 (vsx_splat_<mode>_mem): Likewise.
21476 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
21478 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
21480 2017-05-22 Jakub Jelinek <jakub@redhat.com>
21482 PR middle-end/80809
21483 * omp-low.c (finish_taskreg_remap): New function.
21484 (finish_taskreg_scan): If unit size of ctx->record_type
21485 is non-constant, unshare the size expression and replace
21486 decls in it with possible outer var refs.
21488 PR middle-end/80809
21489 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
21490 GOVD_SHARED rather than GOVD_PRIVATE with it.
21491 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
21492 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
21494 PR middle-end/80853
21495 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
21496 as last argument to build_outer_var_ref for pointer bases of array
21497 section reductions.
21499 2017-05-19 Martin Sebor <msebor@redhat.com>
21501 * print-tree.c (print_node): Print DECL_READ_P flag.
21503 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21505 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
21506 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
21507 * cgraph.c: Likewise.
21508 * cgraphunit.c: Likewise.
21509 * gengtype.c: Likewise.
21510 * ipa-cp.c: Likewise.
21511 * ipa-devirt.c: Likewise.
21512 * ipa-icf.c: Likewise.
21513 * ipa-predicate.c: Likewise.
21514 * ipa-profile.c: Likewise.
21515 * ipa-prop.c: Likewise.
21516 * ipa-split.c: Likewise.
21518 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
21519 edge_predicate_pool, dump_inline_hints,
21520 inline_summary::account_size_time, redirect_to_unreachable,
21521 edge_set_predicate, set_hint_predicate,
21522 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
21523 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
21524 inline_summary_t::remove, remap_hint_predicate_after_duplication,
21525 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
21526 ipa_call_summary_t::remove, initialize_growth_caches,
21527 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
21528 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
21529 mark_modified, unmodified_parm_1, unmodified_parm,
21530 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
21531 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
21532 compute_bb_predicates, will_be_nonconstant_expr_predicate,
21533 will_be_nonconstant_predicate, record_modified_bb_info,
21534 get_minimal_bb, record_modified, param_change_prob,
21535 phi_result_unknown_predicate, predicate_for_phi_result,
21536 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
21537 estimate_function_body_sizes, compute_inline_parameters,
21538 compute_inline_parameters_for_curren, pass_data_inline_parameters,
21539 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
21540 inline_update_callee_summaries, remap_edge_change_prob,
21541 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
21542 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
21543 inline_analyze_function, inline_summary_t::insert,
21544 inline_generate_summary, read_ipa_call_summary, inline_read_section,
21545 inline_read_summary, write_ipa_call_summary, inline_write_summary,
21546 inline_free_summary): Move to ipa-fnsummary.h
21547 (predicate_t): Remove.
21548 * ipa-fnsummary.c: New file.
21549 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
21550 (enum inline_hints_vals, inline_hints, agg_position_info,
21551 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
21552 inline_summaries, ipa_call_summary, ipa_call_summary_t,
21553 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
21554 dump_inline_summary, dump_inline_hints, inline_generate_summary,
21555 inline_read_summary, inline_write_summary, inline_free_summary,
21556 inline_analyze_function, initialize_inline_failed,
21557 inline_merge_summary, inline_update_overall_summary,
21558 compute_inline_parameters): Move to ipa-fnsummary.h
21559 * ipa-fnsummary.h: New file.
21560 * ipa-inline-transform.h: Include ipa-inline.h.
21561 * ipa-inline.c: LIkewise.
21563 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21565 * ipa-inline.c (edge_badness): Use inlined_time instead of
21566 inline_summaries->get.
21568 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21570 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
21572 2017-05-22 Nathan Sidwell <nathan@acm.org>
21574 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
21575 (fdump-lang): Document 'raw' option.
21576 * dumpfile.h (TDI_tu): Delete.
21577 * dumpfile.c (dump_files): Remove translation-unit.
21578 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
21580 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
21582 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
21583 command option from $(AWK) call.
21584 * config/avr/genmultilib.awk: Simplify and rewrite so that it
21585 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
21586 [FORMAT]: Remove handling of variable.
21587 * config/avr/t-multilib: Regenerate.
21589 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21591 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
21593 (dump_inline_summary): Do not print self_time.
21594 (estimate_function_body_sizes): Do not set self_time.
21595 (compute_inline_parameters): Likewise.
21596 (inline_read_section, inline_write_summary): Do not stream self_time.
21597 * ipa-inline.h (inline_summary): Drop self_time.
21599 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21601 * ipa-inline-analysis.c (account_size_time): Rename to ...
21602 (inline_summary::account_size_time): ... this one.
21603 (reset_ipa_call_summary): Turn to ...
21604 (ipa_call_summary::reset): ... this one.
21605 (reset_inline_summary): Turn to ...
21606 (inline_summary::reset): ... this one.
21607 (inline_summary_t::remove): Update.
21608 (inline_summary_t::duplicate): Update.
21609 (ipa_call_summary_t::remove): Update.
21610 (dump_inline_summary): Update.
21611 (estimate_function_body_sizes): Update.
21612 (compute_inline_parameters): Update.
21613 (estimate_node_size_and_time): Update.
21614 (inline_merge_summary): Update.
21615 (inline_update_overall_summary): Update.
21616 (inline_read_section): Update.
21617 (inline_write_summary): Update.
21618 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
21619 add account_size_time and reset member functions.
21620 (ipa_call_summary): Add reset function.
21621 * ipa-predicate.h (predicate::operator &): Constify.
21623 2017-05-22 Richard Biener <rguenther@suse.de>
21625 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
21627 2017-05-19 Jason Merrill <jason@redhat.com>
21629 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
21631 2017-05-19 Marek Polacek <polacek@redhat.com>
21634 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
21635 TYPE_OVERFLOW_WRAPS checks.
21637 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
21639 * tree-core.h (enum omp_clause_default_kind): Add
21640 "OMP_CLAUSE_DEFAULT_PRESENT".
21641 * tree-pretty-print.c (dump_omp_clause): Handle it.
21642 * gimplify.c (enum gimplify_omp_var_data): Add
21643 "GOVD_MAP_FORCE_PRESENT".
21644 (gimplify_adjust_omp_clauses_1): Map it to
21645 "GOMP_MAP_FORCE_PRESENT".
21646 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
21648 * gimplify.c (oacc_default_clause): Clarify.
21650 2017-05-19 Nathan Sidwell <nathan@acm.org>
21652 LANG_HOOK_REGISTER_DUMPS
21653 * toplev.c (general_init): Call register dump lang hook.
21654 * doc/invoke.texi: Document -fdump-lang option family.
21655 * dumpfile.c (dump_files): Remove class dump here.
21656 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
21657 * dumpfile.h (tree_dump_index): Remove TDI_class.
21658 * langhooks-def.h (lhd_register_dumps): Declare.
21659 (LANG_HOOKS_REGISTER_DUMPS): Define.
21660 (LANG_HOOKS_INITIALIZER): Add it.
21661 * langhooks.c (lhd_register_dumps): Define.
21662 * langhooks.h (struct lang_hooks): Add register_dumps.
21664 2017-05-19 Nathan Sidwell <nathan@acm.org>
21666 * context.h (context::set_passes): New.
21667 * context.c (context::context): Do not create pass manager.
21668 * toplev.c (general_init): Create pass manager here.
21670 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
21672 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
21673 use this splitter if two add or or instructions would also work for
21674 the constant we want to generate.
21676 2017-05-19 Richard Biener <rguenther@suse.de>
21679 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
21680 predicate evaluation.
21682 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
21684 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
21686 * ipa-inline.c (want_inline_small_function_p): Do not cast to
21689 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
21691 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
21692 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
21693 (inline_edge_summary_vec): Turn into ...
21694 (ipa_call_summaries): ... this one.
21695 (redirect_to_unreachable, edge_set_predicate,
21696 evaluate_properties_for_edge, inline_summary_alloc,
21697 reset_ipa_call_summary, reset_inline_summary,
21698 inline_summary_t::duplicate): Update.
21699 (inline_edge_duplication_hook): Turn to ...
21700 (ipa_call_summary_t::duplicate): ... this one.
21701 (inline_edge_removal_hook): Turn to ...
21702 (ipa_call_summary_t::remove): ... this one.
21703 (dump_inline_edge_summary): Turn to ...
21704 (dump_ipa_call_summary): ... this one.
21705 (estimate_function_body_sizes): Update.
21706 (inline_update_callee_summaries): Update.
21707 (remap_edge_change_prob): Update.
21708 (remap_edge_summaries): Update.
21709 (inline_merge_summary): Update.
21710 (do_estimate_edge_time): Update.
21711 (inline_generate_summary): Update.
21712 (inline_read_section): Update.
21713 (inline_read_summary): Update.
21714 (inline_free_summary): Update.
21715 * ipa-inline.c (can_inline_edge_p): Update.
21716 (compute_inlined_call_time): Update.
21717 (want_inline_small_function_p): Update.
21718 (edge_badness): Update.
21719 (early_inliner): Update.
21720 * ipa-inline.h (inline_edge_summary): Turn to ...
21721 (ipa_call_summary): ... this one.
21722 (ipa_call_summary_t): New class.
21723 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
21724 (ipa_call_summaries): New.
21725 (inline_edge_summary): Remove.
21726 (estimate_edge_growth): Update.
21727 * ipa-profile.c (ipa_propagate_frequency_1): Update.
21728 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21729 * ipa-split.c (execute_split_functions): Update.
21730 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
21732 2017-05-19 Richard Biener <rguenther@suse.de>
21734 PR middle-end/80764
21735 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
21737 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
21739 * config/rs6000/rs6000.c (struct machine_function): Add field
21740 fpr_is_wrapped_separately.
21741 (rs6000_get_separate_components): Use 64 components. Handle the
21742 new FPR components.
21743 (rs6000_components_for_bb): Handle the FPR components.
21744 (rs6000_emit_prologue_components): Handle the FPR components.
21745 (rs6000_emit_epilogue_components): Handle the FPR components.
21746 (rs6000_set_handled_components): Handle the FPR components.
21747 (rs6000_emit_prologue): Don't output prologue code for those FPRs
21748 that are already separately shrink-wrapped.
21749 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
21750 that are already separately shrink-wrapped.
21752 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
21755 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
21758 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
21759 (define_peephole2 for Altivec d-form load): Add peepholes to catch
21760 cases where the register allocator uses a move and an offsettable
21761 memory operation to/from a FPR register on ISA 2.06/2.07.
21762 (define_peephole2 for Altivec d-form store): Likewise.
21764 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
21767 * config/i386/mmx.md (*mov<mode>_internal): Enable
21768 alternatives 11, 12, 13 and 14 also for 32bit targets.
21769 Remove alternatives 15, 16, 17 and 18.
21770 * config/i386/sse.md (vec_concatv2di): Change
21771 alternative (!x, *y) to (x, ?!*Yn).
21773 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
21775 * dumpfile.h (enum dump_kind): Remove stray comma.
21777 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
21779 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
21780 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
21781 predicate::num_conditions
21782 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
21783 (CHANGED): turn into predicate::changed.
21784 (agg_position_info): Move to ipa-predicate.h
21785 (add_condition, predicate::add_clause, predicate::operator &=,
21786 predicate::or_with, predicate::evaluate, predicate::probability,
21787 dump_condition, dump_clause, predicate::dump,
21788 predicate::remap_after_duplication, predicate::remap_after_inlining,
21789 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
21790 (evaluate_conditions_for_known_args): Update.
21791 (set_cond_stmt_execution_predicate): Update.
21792 * ipa-inline.h: Include ipa-predicate.h
21793 (condition, inline_param_summary, conditions, agg_position_info,
21794 predicate): Move to ipa-predicate.h
21795 * ipa-predicate.c: New file.
21796 * ipa-predicate.h: New file.
21798 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
21800 * final.c (leaf_function_p): Check we are not in a sequence.
21802 2017-05-18 Martin Liska <mliska@suse.cz>
21804 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
21805 * dumpfile.c (dump_register): Use new enum dump_kind.
21806 (get_dump_file_name): Likewise.
21807 (dump_enable_all): Likewise.
21808 (dump_switch_p_1): Likewise.
21809 (enable_rtl_dump_file): Remove usage of TDF_RTL.
21810 * dumpfile.h (enum dump_kind): New enum type.
21811 (struct dump_file_info): Create constructor and
21812 format fields and comments.
21813 * passes.c (pass_manager::register_one_dump_file):
21815 * statistics.c (statistics_early_init): Likewise.
21816 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
21817 TDF_TREE with TDF_SLIM.
21818 (gather_memory_references_ref): Likewise.
21820 2017-05-18 Martin Liska <mliska@suse.cz>
21822 * vec.h (struct vnull): Use it.
21824 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
21826 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
21827 (true_predicate, false_predicate, true_predicate_p,
21828 false_predicate_p): Remove.
21829 (single_cond_predicate, not_inlined_predicate): Turn to member function
21831 (add_condition): Update.
21832 (add_clause): Turn to...
21833 (predicate::add_clause): ... this one; update; allow passing NULL
21835 (and_predicates): Turn to ...
21836 (predicate::operator &=): ... this one.
21837 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
21838 (or_predicates): Turn to ...
21839 (predicate::or_with): ... this one.
21840 (evaluate_predicate): Turn to ...
21841 (predicate::evaluate): ... this one.
21842 (predicate_probability): Turn to ...
21843 (predicate::probability): ... this one.
21844 (dump_condition): Update.
21845 (dump_predicate): Turn to ...
21846 (predicate::dump): ... this one.
21847 (account_size_time): Update.
21848 (edge_set_predicate): Update.
21849 (set_hint_predicate): UPdate.
21850 (evaluate_conditions_for_known_args): Update.
21851 (evaluate_properties_for_edge): Update.
21852 (remap_predicate_after_duplication): Turn to...
21853 (predicate::remap_after_duplication): ... this one.
21854 (remap_hint_predicate_after_duplication): Update.
21855 (inline_summary_t::duplicate): UPdate.
21856 (dump_inline_edge_summary): Update.
21857 (dump_inline_summary): Update.
21858 (set_cond_stmt_execution_predicate): Update.
21859 (set_switch_stmt_execution_predicate): Update.
21860 (compute_bb_predicates): Update.
21861 (will_be_nonconstant_expr_predicate): Update.
21862 (will_be_nonconstant_predicate): Update.
21863 (phi_result_unknown_predicate): Update.
21864 (predicate_for_phi_result): Update.
21865 (array_index_predicate): Update.
21866 (estimate_function_body_sizes): Update.
21867 (estimate_node_size_and_time): Update.
21868 (estimate_ipcp_clone_size_and_time): Update.
21869 (remap_predicate): Rename to ...
21870 (predicate::remap_after_inlining): ... this one.
21871 (remap_hint_predicate): Update.
21872 (inline_merge_summary): Update.
21873 (inline_update_overall_summary): Update.
21874 (estimate_size_after_inlining): Update.
21875 (read_predicate): Rename to ...
21876 (predicate::stream_in): ... this one.
21877 (read_inline_edge_summary): Update.
21878 (write_predicate): Rename to ...
21879 (predicate::stream_out): ... this one.
21880 (write_inline_edge_summary): Update.
21881 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
21882 (clause_t): Turn to uint32_t
21883 (predicate): Turn to class; implement constructor and operators
21885 (size_time_entry): Update.
21886 (inline_summary): Update.
21887 (inline_edge_summary): Update.
21889 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
21891 * fold-const.c (fold_binary_loc): Move transformation...
21892 * match.pd (C - X CMP X): ... here.
21894 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
21896 * config/sparc/sparc.c (sparc_option_override): Set function
21897 alignment for -mcpu=niagara7 to 64 to match the I$ line.
21898 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
21900 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
21902 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
21904 2017-05-18 Marek Polacek <polacek@redhat.com>
21907 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
21908 (pass_ubsan::execute): Call gimple_assign_single_p instead of
21909 gimple_assign_load_p.
21911 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
21913 PR middle-end/80692
21914 * real.c (do_compare): Give decimal_do_compare preference over
21915 comparing just the signs.
21917 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
21919 * doc/md.texi (Canonicalization of Instructions): Describe the
21920 canonical form of instructions that inherently set a condition
21923 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
21925 PR middle-end/80775
21926 * tree-cfg.c: Move deletion of unreachable case statements to after
21927 the merging of consecutive case labels.
21929 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
21931 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
21932 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
21933 restoring of callee-saved registers.
21935 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
21937 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
21938 * config/visium/visium.c (single_set_and_flags): Likewise.
21939 * config/visium/visium.md (Substitutions): Likewise.
21941 2017-05-17 Martin Liska <mliska@suse.cz>
21943 * cfg.c: Introduce dump_flags_t type and
21944 use it instead of int type.
21946 * cfghooks.c: Likewise.
21947 * cfghooks.h (struct cfg_hooks): Likewise.
21948 * cfgrtl.c: Likewise.
21949 * cfgrtl.h: Likewise.
21950 * cgraph.c (cgraph_node::get_body): Likewise.
21951 * coretypes.h: Likewise.
21952 * domwalk.c: Likewise.
21953 * domwalk.h: Likewise.
21954 * dumpfile.c (struct dump_option_value_info): Likewise.
21955 (dump_enable_all): Likewise.
21956 (dump_switch_p_1): Likewise.
21957 (opt_info_switch_p): Likewise.
21958 * dumpfile.h (enum tree_dump_index): Likewise.
21959 (struct dump_file_info): Likewise.
21960 * genemit.c: Likewise.
21961 * generic-match-head.c: Likewise.
21962 * gengtype.c (open_base_files): Likewise.
21963 * gimple-pretty-print.c: Likewise.
21964 * gimple-pretty-print.h: Likewise.
21965 * graph.c (print_graph_cfg): Likewise.
21966 * graphite-scop-detection.c (dot_all_sese): Likewise.
21967 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
21968 * loop-unroll.c (report_unroll): Likewise.
21969 * passes.c (pass_manager::register_one_dump_file): Likewise.
21970 * print-tree.c: Likewise.
21971 * statistics.c: Likewise.
21972 * tree-cfg.c: Likewise.
21973 * tree-cfg.h: Likewise.
21974 * tree-dfa.c: Likewise.
21975 * tree-dfa.h: Likewise.
21976 * tree-dump.c (dump_function): Likewise.
21977 * tree-dump.h (struct dump_info): Likewise.
21978 * tree-pretty-print.c: Likewise.
21979 * tree-pretty-print.h: Likewise.
21980 * tree-ssa-live.c: Likewise.
21981 * tree-ssa-live.h: Likewise.
21982 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
21983 * tree-vect-loop.c: Likewise.
21984 * tree-vect-slp.c: Likewise.
21986 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
21987 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21989 PR tree-optimization/80457
21990 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
21991 of all arguments to a statement as scalar_to_vec operations.
21992 (vectorizable_call): Adjust call to vect_model_simple_cost for
21994 (vectorizable_conversion): Likewise.
21995 (vectorizable_assignment): Likewise.
21996 (vectorizable_shift): Likewise.
21997 (vectorizable_operation): Likewise.
21998 (vectorizable_comparison): Likewise.
21999 (vect_is_simple_cond): Record the def types for operands.
22000 (vectorizable_condition): Likewise, call vect_model_simple_cost.
22001 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
22002 for statement argument count.
22004 2017-05-16 Carl Love <cel@us.ibm.com>
22006 * config/rs6000/rs6000-c: Add support for built-in functions
22007 vector unsigned long long vec_bperm (vector unsigned long long,
22008 vector unsigned char)
22009 vector signed long long vec_mule (vector signed int,
22011 vector unsigned long long vec_mule (vector unsigned int,
22012 vector unsigned int)
22013 vector signed long long vec_mulo (vector signed int,
22015 vector unsigned long long vec_mulo (vector unsigned int,
22016 vector unsigned int)
22017 vector signed char vec_sldw (vector signed char,
22018 vector signed char,
22020 vector unsigned char vec_sldw (vector unsigned char,
22021 vector unsigned char,
22023 vector signed short vec_sldw (vector signed short,
22024 vector signed short,
22026 vector unsigned short vec_sldw (vector unsigned short,
22027 vector unsigned short,
22029 vector signed int vec_sldw (vector signed int,
22032 vector unsigned int vec_sldw (vector unsigned int,
22033 vector unsigned int,
22035 vector signed long long vec_sldw (vector signed long long,
22036 vector signed long long,
22038 vector unsigned long long vec_sldw (vector unsigned long long,
22039 vector unsigned long long,
22041 * config/rs6000/rs6000-c: Add support for built-in functions
22042 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
22043 * config/rs6000/altivec.h: Add defintion for vec_sldw.
22044 * doc/extend.texi: Update the built-in documentation for the
22045 new built-in functions.
22047 2017-05-16 Marek Polacek <polacek@redhat.com>
22051 * tree.c (save_expr): Don't fold the expression.
22053 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
22055 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
22056 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
22057 and (?*y,m). Update insn attributes.
22059 2017-05-16 Martin Liska <mliska@suse.cz>
22061 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
22062 flags argument of print_gimple_stmt, print_gimple_expr,
22063 print_generic_stmt and print_generic_expr.
22064 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
22065 * coretypes.h: Likewise.
22066 * except.c (dump_eh_tree): Likewise.
22067 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
22068 * gimple-pretty-print.h: Likewise.
22069 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
22070 (backprop::push_to_worklist): Likewise.
22071 (backprop::pop_from_worklist): Likewise.
22072 (backprop::process_use): Likewise.
22073 (backprop::intersect_uses): Likewise.
22074 (note_replacement): Likewise.
22075 * gimple-ssa-store-merging.c
22076 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
22077 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
22078 (pass_store_merging::execute): Likewise.
22079 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
22080 (ssa_base_cand_dump_callback): Likewise.
22081 (dump_incr_vec): Likewise.
22082 (replace_refs): Likewise.
22083 (replace_mult_candidate): Likewise.
22084 (create_add_on_incoming_edge): Likewise.
22085 (create_phi_basis): Likewise.
22086 (insert_initializers): Likewise.
22087 (all_phi_incrs_profitable): Likewise.
22088 (introduce_cast_before_cand): Likewise.
22089 (replace_one_candidate): Likewise.
22090 * gimplify.c (gimplify_expr): Likewise.
22091 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
22092 (set_rename): Likewise.
22093 (rename_uses): Likewise.
22094 (copy_loop_phi_nodes): Likewise.
22095 (add_close_phis_to_merge_points): Likewise.
22096 (copy_loop_close_phi_args): Likewise.
22097 (copy_cond_phi_args): Likewise.
22098 (graphite_copy_stmts_from_block): Likewise.
22099 (translate_pending_phi_nodes): Likewise.
22100 * graphite-poly.c (print_pdr): Likewise.
22101 (dump_gbb_cases): Likewise.
22102 (dump_gbb_conditions): Likewise.
22103 (print_scop_params): Likewise.
22104 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
22105 (build_cross_bb_scalars_use): Likewise.
22106 (gather_bbs::before_dom_children): Likewise.
22107 * hsa-dump.c (dump_hsa_immed): Likewise.
22108 * ipa-cp.c (print_ipcp_constant_value): Likewise.
22109 (get_replacement_map): Likewise.
22110 * ipa-inline-analysis.c (dump_condition): Likewise.
22111 (estimate_function_body_sizes): Likewise.
22112 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
22113 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
22114 * ipa-prop.c (ipa_dump_param): Likewise.
22115 (ipa_print_node_jump_functions_for_edge): Likewise.
22116 (ipa_modify_call_arguments): Likewise.
22117 (ipa_modify_expr): Likewise.
22118 (ipa_dump_param_adjustments): Likewise.
22119 (ipa_dump_agg_replacement_values): Likewise.
22120 (ipcp_modif_dom_walker::before_dom_children): Likewise.
22121 * ipa-pure-const.c (check_stmt): Likewise.
22122 (pass_nothrow::execute): Likewise.
22123 * ipa-split.c (execute_split_functions): Likewise.
22124 * omp-offload.c (dump_oacc_loop_part): Likewise.
22125 (dump_oacc_loop): Likewise.
22126 * trans-mem.c (tm_log_emit): Likewise.
22127 (tm_memopt_accumulate_memops): Likewise.
22128 (dump_tm_memopt_set): Likewise.
22129 (dump_tm_memopt_transform): Likewise.
22130 * tree-cfg.c (gimple_verify_flow_info): Likewise.
22131 (print_loop): Likewise.
22132 * tree-chkp-opt.c (chkp_print_addr): Likewise.
22133 (chkp_gather_checks_info): Likewise.
22134 (chkp_get_check_result): Likewise.
22135 (chkp_remove_check_if_pass): Likewise.
22136 (chkp_use_outer_bounds_if_possible): Likewise.
22137 (chkp_reduce_bounds_lifetime): Likewise.
22138 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
22139 (chkp_mark_completed_bounds): Likewise.
22140 (chkp_register_incomplete_bounds): Likewise.
22141 (chkp_mark_invalid_bounds): Likewise.
22142 (chkp_maybe_copy_and_register_bounds): Likewise.
22143 (chkp_build_returned_bound): Likewise.
22144 (chkp_get_bound_for_parm): Likewise.
22145 (chkp_build_bndldx): Likewise.
22146 (chkp_get_bounds_by_definition): Likewise.
22147 (chkp_generate_extern_var_bounds): Likewise.
22148 (chkp_get_bounds_for_decl_addr): Likewise.
22149 * tree-chrec.c (chrec_apply): Likewise.
22150 * tree-data-ref.c (dump_data_reference): Likewise.
22151 (dump_subscript): Likewise.
22152 (dump_data_dependence_relation): Likewise.
22153 (analyze_overlapping_iterations): Likewise.
22154 * tree-inline.c (expand_call_inline): Likewise.
22155 (tree_function_versioning): Likewise.
22156 * tree-into-ssa.c (dump_defs_stack): Likewise.
22157 (dump_currdefs): Likewise.
22158 (dump_names_replaced_by): Likewise.
22159 (dump_update_ssa): Likewise.
22160 (update_ssa): Likewise.
22161 * tree-object-size.c (pass_object_sizes::execute): Likewise.
22162 * tree-parloops.c (build_new_reduction): Likewise.
22163 (try_create_reduction_list): Likewise.
22164 (ref_conflicts_with_region): Likewise.
22165 (oacc_entry_exit_ok_1): Likewise.
22166 (oacc_entry_exit_single_gang): Likewise.
22167 * tree-pretty-print.h: Likewise.
22168 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
22169 (get_scalar_evolution): Likewise.
22170 (add_to_evolution): Likewise.
22171 (get_loop_exit_condition): Likewise.
22172 (analyze_evolution_in_loop): Likewise.
22173 (analyze_initial_condition): Likewise.
22174 (analyze_scalar_evolution): Likewise.
22175 (instantiate_scev): Likewise.
22176 (number_of_latch_executions): Likewise.
22177 (gather_chrec_stats): Likewise.
22178 (final_value_replacement_loop): Likewise.
22179 (scev_const_prop): Likewise.
22180 * tree-sra.c (dump_access): Likewise.
22181 (disqualify_candidate): Likewise.
22182 (create_access): Likewise.
22183 (reject): Likewise.
22184 (maybe_add_sra_candidate): Likewise.
22185 (create_access_replacement): Likewise.
22186 (analyze_access_subtree): Likewise.
22187 (analyze_all_variable_accesses): Likewise.
22188 (sra_modify_assign): Likewise.
22189 (initialize_constant_pool_replacements): Likewise.
22190 (find_param_candidates): Likewise.
22191 (decide_one_param_reduction): Likewise.
22192 (replace_removed_params_ssa_names): Likewise.
22193 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
22194 * tree-ssa-copy.c (dump_copy_of): Likewise.
22195 (copy_prop_visit_cond_stmt): Likewise.
22196 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
22197 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
22198 (record_equivalences_from_stmt): Likewise.
22199 * tree-ssa-dse.c (compute_trims): Likewise.
22200 (delete_dead_call): Likewise.
22201 (delete_dead_assignment): Likewise.
22202 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
22203 (forward_propagate_into_cond): Likewise.
22204 (pass_forwprop::execute): Likewise.
22205 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
22206 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
22208 (move_computations_worker): Likewise.
22209 (execute_sm): Likewise.
22210 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
22211 (remove_exits_and_undefined_stmts): Likewise.
22212 (remove_redundant_iv_tests): Likewise.
22213 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
22214 (adjust_iv_update_pos): Likewise.
22215 * tree-ssa-math-opts.c (bswap_replace): Likewise.
22216 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
22217 (value_replacement): Likewise.
22218 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
22219 * tree-ssa-pre.c (print_pre_expr): Likewise.
22220 (get_representative_for): Likewise.
22221 (create_expression_by_pieces): Likewise.
22222 (insert_into_preds_of_block): Likewise.
22223 (eliminate_insert): Likewise.
22224 (eliminate_dom_walker::before_dom_children): Likewise.
22225 (eliminate): Likewise.
22226 (remove_dead_inserted_code): Likewise.
22227 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
22228 * tree-ssa-reassoc.c (get_rank): Likewise.
22229 (eliminate_duplicate_pair): Likewise.
22230 (eliminate_plus_minus_pair): Likewise.
22231 (eliminate_not_pairs): Likewise.
22232 (undistribute_ops_list): Likewise.
22233 (eliminate_redundant_comparison): Likewise.
22234 (update_range_test): Likewise.
22235 (optimize_range_tests_var_bound): Likewise.
22236 (optimize_vec_cond_expr): Likewise.
22237 (rewrite_expr_tree): Likewise.
22238 (rewrite_expr_tree_parallel): Likewise.
22239 (linearize_expr): Likewise.
22240 (break_up_subtract): Likewise.
22241 (linearize_expr_tree): Likewise.
22242 (attempt_builtin_powi): Likewise.
22243 (attempt_builtin_copysign): Likewise.
22244 (transform_stmt_to_copy): Likewise.
22245 (transform_stmt_to_multiply): Likewise.
22246 (dump_ops_vector): Likewise.
22247 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
22248 (print_scc): Likewise.
22249 (set_ssa_val_to): Likewise.
22250 (visit_reference_op_store): Likewise.
22251 (visit_use): Likewise.
22252 (sccvn_dom_walker::before_dom_children): Likewise.
22253 (run_scc_vn): Likewise.
22254 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
22256 (expr_hash_elt::print): Likewise.
22257 (const_and_copies::pop_to_marker): Likewise.
22258 (const_and_copies::record_const_or_copy_raw): Likewise.
22259 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
22260 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
22261 (dump_predicates): Likewise.
22262 (find_uninit_use): Likewise.
22263 (warn_uninitialized_phi): Likewise.
22264 (pass_late_warn_uninitialized::execute): Likewise.
22265 * tree-ssa.c (verify_vssa): Likewise.
22266 (verify_ssa): Likewise.
22267 (maybe_optimize_var): Likewise.
22268 * tree-vrp.c (dump_value_range): Likewise.
22269 (dump_all_value_ranges): Likewise.
22270 (dump_asserts_for): Likewise.
22271 (register_edge_assert_for_2): Likewise.
22272 (vrp_visit_cond_stmt): Likewise.
22273 (vrp_visit_switch_stmt): Likewise.
22274 (vrp_visit_stmt): Likewise.
22275 (vrp_visit_phi_node): Likewise.
22276 (simplify_cond_using_ranges_1): Likewise.
22277 (fold_predicate_in): Likewise.
22278 (evrp_dom_walker::before_dom_children): Likewise.
22279 (evrp_dom_walker::push_value_range): Likewise.
22280 (evrp_dom_walker::pop_value_range): Likewise.
22281 (execute_early_vrp): Likewise.
22283 2017-05-16 Richard Biener <rguenther@suse.de>
22285 * dwarf2out.c (loc_list_from_tree_1): Do not create
22286 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
22288 2017-05-16 Richard Biener <rguenther@suse.de>
22290 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
22292 (note_variable_value_in_expr): If we resolved the decl ref
22293 do not push to the stack.
22295 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
22297 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
22298 operations in fast-math mode.
22299 (vaddq_f16): Likewise.
22300 (vmul_f16): Likewise.
22301 (vmulq_f16): Likewise.
22302 (vsub_f16): Likewise.
22303 (vsubq_f16): Likewise.
22304 * config/arm/neon.md (add<mode>3): New.
22306 (fma:<VH:mode>3): New. Also remove outdated comment.
22309 2017-05-16 Martin Liska <mliska@suse.cz>
22313 * ipa-devirt.c (warn_types_mismatch): Fix typo.
22314 (odr_types_equivalent_p): Likewise.
22316 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
22318 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
22320 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
22323 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
22324 non-interunit SSE move alternatives with '?'.
22325 (zero-extendsidi peephole2): New peephole to skip intermediate
22326 general register in SSE zero-extend sequence.
22328 2017-05-15 Jeff Law <law@redhat.com>
22330 * reorg.c (relax_delay_slots): Create a new variable to hold
22331 the temporary target rather than clobbering TARGET_LABEL.
22333 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
22334 missing argument to extract_bit_field call.
22335 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
22337 2017-05-15 Martin Liska <mliska@suse.cz>
22340 * gcc.c (process_command): Do not allow empty argument of -o option.
22342 2017-05-15 Renlin Li <renlin.li@arm.com>
22344 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
22345 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
22346 * config/aarch64/constraints.md (Usf): Add long call check.
22347 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
22348 (call_value): Likewise.
22349 (sibcall): Likewise.
22350 (sibcall_value): Likewise.
22352 (call_value_insn): New.
22353 (sibcall_insn): Update rtx pattern.
22354 (sibcall_value_insn): Likewise.
22355 (call_internal): Remove.
22356 (call_value_internal): Likewise.
22357 (sibcall_internal): Likewise.
22358 (sibcall_value_internal): Likewise.
22359 (call_reg): Likewise.
22360 (call_symbol): Likewise.
22361 (call_value_reg): Likewise.
22362 (call_value_symbol): Likewise.
22364 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
22367 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
22369 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
22371 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
22372 compatible with CCGOCmode and with CCZmode.
22374 2017-05-14 Martin Sebor <msebor@redhat.com>
22376 PR middle-end/77671
22377 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
22378 (gimple_fold_builtin_snprintf): Same.
22379 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
22380 (gimple_fold_builtin_snprintf): Same.
22381 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
22382 of character types.
22383 (is_call_safe): New function.
22384 (try_substitute_return_value): Call it.
22385 (try_simplify_call): New function.
22386 (pass_sprintf_length::handle_gimple_call): Call it.
22388 2017-05-14 Martin Sebor <msebor@redhat.com>
22390 PR middle-end/80669
22391 * builtins.c (expand_builtin_stpncpy): Simplify.
22393 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
22395 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
22396 * config/i386/i386.h
22397 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
22398 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
22399 (struct machine_function): Add new members call_ms2sysv,
22400 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
22401 (struct machine_frame_state): New fields sp_realigned and
22402 sp_realigned_offset.
22403 * config/i386/i386.c
22404 (enum xlogue_stub): New enum.
22405 (enum xlogue_stub_sets): New enum.
22406 (class xlogue_layout): New class.
22407 (struct ix86_frame): New fields stack_realign_allocate_offset,
22408 stack_realign_offset and outlined_save_offset. Modify comments to
22409 detail stack layout when using out-of-line stubs.
22410 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
22411 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
22412 -mcall-ms2sysv-xlogues.
22413 (stub_managed_regs): New static variable.
22414 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
22415 registers managed by out-of-line stub.
22416 (disable_call_ms2sysv_xlogues): New function.
22417 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
22418 m->call_ms2sysv when appropriate and compute frame layout for
22420 (sp_valid_at, fp_valid_at): New inline functions.
22421 (choose_basereg): New function.
22422 (choose_baseaddr): Add align parameter, use choose_basereg and modify
22424 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
22425 Use align parameter of choose_baseaddr to generated aligned SSE movs
22427 (pro_epilogue_adjust_stack): Modify to track
22428 machine_frame_state::sp_realigned.
22429 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
22430 (ix86_nsaved_sseregs): Likewise.
22431 (ix86_emit_save_regs): Likewise.
22432 (ix86_emit_save_regs_using_mov): Likewise.
22433 (ix86_emit_save_sse_regs_using_mov): Likewise.
22434 (get_scratch_register_on_entry): Likewise.
22435 (gen_frame_set): New function.
22436 (gen_frame_load): Likewise.
22437 (gen_frame_store): Likewise.
22438 (emit_outlined_ms2sysv_save): Likewise.
22439 (emit_outlined_ms2sysv_restore): Likewise.
22440 (ix86_expand_prologue): Modify stack re-alignment code and call
22441 emit_outlined_ms2sysv_save when appropriate.
22442 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
22443 parameter rtx_insn *insn, which allows the function to be used to only
22444 generate the notes.
22445 (ix86_expand_epilogue): Modify validity checks of frame and stack
22446 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
22447 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
22448 * config/i386/predicates.md
22449 (save_multiple): New predicate.
22450 (restore_multiple): Likewise.
22451 * config/i386/sse.md
22452 (save_multiple<mode>): New pattern.
22453 (save_multiple_realign<mode>): Likewise.
22454 (restore_multiple<mode>): Likewise.
22455 (restore_multiple_and_return<mode>): Likewise.
22456 (restore_multiple_leave_return<mode>): Likewise.
22457 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
22459 2017-05-14 Julia Koval <julia.koval@intel.com>
22461 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
22462 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
22463 (__builtin_ia32_xsetbv): New builtins.
22464 * config/i386/i386.c (ix86_expand_special_args_builtin):
22466 (ix86_expand_builtin): Special expand for new intrinsics.
22467 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
22468 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
22469 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
22471 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22473 * cfganal.c (inverted_post_order_compute): Change argument type
22475 * cfganal.h (inverted_post_order_compute): Adjust prototype.
22476 * df-core.c (rest_of_handle_df_initialize): Adjust.
22477 (rest_of_handle_df_finish): Likewise.
22478 (df_analyze_1): Likewise.
22479 (df_analyze): Likewise.
22480 (loop_inverted_post_order_compute): Change argument to be a vec *.
22481 (df_analyze_loop): Adjust.
22482 (df_get_n_blocks): Likewise.
22483 (df_get_postorder): Likewise.
22484 * df.h (struct df_d): Change field to be a vec.
22485 * lcm.c (compute_laterin): Adjust.
22486 (compute_available): Likewise.
22487 * lra-lives.c (lra_create_live_ranges_1): Likewise.
22488 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
22489 * tree-ssa-pre.c (compute_antic): Likewise.
22491 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22493 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
22494 (depth_first_search::depth_first_search): Change structure init
22495 function to this constructor.
22496 (depth_first_search::add_bb): Rename function to this member.
22497 (depth_first_search::execute): Likewise.
22498 (flow_dfs_compute_reverse_finish): Adjust.
22500 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22502 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
22503 (longest_simple_path): Likewise.
22504 * shrink-wrap.c (spread_components): Likewise.
22505 (disqualify_problematic_components): Likewise.
22506 (emit_common_heads_for_components): Likewise.
22507 (emit_common_tails_for_components): Likewise.
22508 (insert_prologue_epilogue_for_components): Likewise.
22510 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22512 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
22515 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22517 * df-core.c (df_set_blocks): Start using auto_bitmap.
22518 (df_compact_blocks): Likewise.
22519 * df-problems.c (df_rd_confluence_n): Likewise.
22520 * df-scan.c (df_insn_rescan_all): Likewise.
22521 (df_process_deferred_rescans): Likewise.
22522 (df_update_entry_block_defs): Likewise.
22523 (df_update_exit_block_uses): Likewise.
22524 (df_entry_block_bitmap_verify): Likewise.
22525 (df_exit_block_bitmap_verify): Likewise.
22526 (df_scan_verify): Likewise.
22527 * lra-constraints.c (lra_constraints): Likewise.
22528 (undo_optional_reloads): Likewise.
22529 (lra_undo_inheritance): Likewise.
22530 * lra-remat.c (calculate_gen_cands): Likewise.
22531 (do_remat): Likewise.
22532 * lra-spills.c (assign_spill_hard_regs): Likewise.
22533 (spill_pseudos): Likewise.
22534 * tree-ssa-pre.c (bitmap_set_and): Likewise.
22535 (bitmap_set_subtract_values): Likewise.
22537 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22539 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
22540 management with auto_bitmap.
22541 (fix_inter_tick): Likewise.
22542 (fix_recovery_deps): Likewise.
22543 * ira.c (add_store_equivs): Likewise.
22544 (find_moveable_pseudos): Likewise.
22545 (split_live_ranges_for_shrink_wrap): Likewise.
22546 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
22547 (rtx_reuse_manager::seen_def_p): Likewise.
22548 (rtx_reuse_manager::set_seen_def): Likewise.
22549 * print-rtl.h (class rtx_reuse_manager): Likewise.
22551 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22553 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
22555 (migrate_btr_def): Likewise.
22556 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
22557 * df-core.c (loop_post_order_compute): Likewise.
22558 (loop_inverted_post_order_compute): Likewise.
22559 * hsa-common.h: Likewise.
22560 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
22561 * init-regs.c (initialize_uninitialized_regs): Likewise.
22562 * ipa-inline.c (resolve_noninline_speculation): Likewise.
22563 (inline_small_functions): Likewise.
22564 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
22565 * ira.c (combine_and_move_insns): Likewise.
22566 (build_insn_chain): Likewise.
22567 * loop-invariant.c (find_invariants): Likewise.
22568 * lower-subreg.c (propagate_pseudo_copies): Likewise.
22569 * predict.c (tree_predict_by_opcode): Likewise.
22570 (predict_paths_leading_to): Likewise.
22571 (predict_paths_leading_to_edge): Likewise.
22572 (estimate_loops_at_level): Likewise.
22573 (estimate_loops): Likewise.
22574 * shrink-wrap.c (try_shrink_wrapping): Likewise.
22575 (spread_components): Likewise.
22576 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
22577 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
22578 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
22579 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
22580 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
22581 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
22582 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
22583 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
22584 (mark_threaded_blocks): Likewise.
22585 (thread_through_all_blocks): Likewise.
22586 * tree-ssa.c (verify_ssa): Likewise.
22587 (execute_update_addresses_taken): Likewise.
22588 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
22590 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22592 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
22594 (post_order_compute): Likewise.
22595 (inverted_post_order_compute): Likewise.
22596 (pre_and_rev_post_order_compute_fn): Likewise.
22598 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22600 * genrecog.c (int_set::int_set): Explicitly construct our
22601 auto_vec base class.
22602 * vec.h (auto_vec::auto_vec): New constructor.
22604 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22606 * bitmap.h (class auto_bitmap): New constructor taking
22607 bitmap_obstack * argument.
22609 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22611 * bitmap.h (class auto_bitmap): Change type of m_bits to
22612 bitmap_head, and adjust ctor / dtor and member operators.
22614 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
22616 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
22617 when returned register mode doesn't match original mode.
22619 2017-05-12 Jeff Law <law@redhat.com>
22620 Jakub Jelinek <jakub@redhat.com>
22622 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
22623 we look for cc setter after the compare-elim changes.
22624 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
22625 within the vector to match what compare-elim now expects.
22626 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
22627 (xorsi3_flags, one_cmplsi2_flags): Likewise.
22629 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
22630 after the compare-elim changes.
22631 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
22632 the vector to match what compare-elim now expects.
22633 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
22634 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
22635 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
22636 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
22637 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
22639 * config/visium/visium.c (single_set_and_flags): Fix where
22640 we look for cc setter after the compare-elim changes.
22641 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
22642 with the vector to match what compare-elim now expects.
22643 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
22644 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
22645 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
22646 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
22647 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
22648 (neg<mode>2_insn_set_overflow): Likewise.
22650 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
22652 PR middle-end/79794
22653 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
22654 maybe_expand_insn call, set ops[0].target. If still set after call,
22655 set alt_rtl. Add extra arg to recursive calls.
22656 (extract_bit_field): Add alt_rtl argument. Pass to
22658 * expmed.h (extract_bit_field): Fix prototype.
22659 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
22660 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
22661 to extract_bit_field_calls.
22662 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
22663 Pass alt_rtl to extract_bit_field calls.
22664 * calls.c (store_unaligned_arguments_into_psuedos)
22665 load_register_parameters): Pass extra NULL to extract_bit_field calls.
22666 * optabs.c (maybe_legitimize_operand): Clear op->target when call
22668 * optabs.h (struct expand_operand): Add target bitfield.
22670 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
22672 * compare-elim.c (try_eliminate_compare): Canonicalize
22673 operation with embedded compare to
22674 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
22675 (set (reg) (operation)].
22677 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
22679 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
22682 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
22683 cost of adding a carry flag for ADC instruction.
22684 [case MINUS]: Ignore the cost of subtracting a carry flag
22685 for SBB instruction.
22687 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
22689 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
22691 * config/rs6000/bmiintrin.h: New file.
22692 * config/rs6000/bmi2intrin.h: New file.
22693 * config/rs6000/x86intrin.h: New file.
22695 2017-05-12 Jeff Law <law@redhat.com>
22697 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
22700 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
22702 PR middle-end/80707
22703 * tree-cfg.c: Remove cfg edges of unreachable case statements.
22705 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
22707 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
22708 early expansion of vector divide builtins.
22709 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
22710 builtins identified as having unsigned arguments.
22712 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
22714 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
22715 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
22716 expansion of vector logical operations (and, andc, or, xor,
22719 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
22721 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
22722 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
22724 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
22726 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
22727 early GIMPLE expansion of vector multiplies.
22729 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22731 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
22732 TARGET_HAVE_MOVT conditional.
22733 (movt splitter): Likewise.
22735 2017-05-12 Richard Biener <rguenther@suse.de>
22737 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
22738 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
22739 Fold all stmts not inplace.
22741 2017-05-12 Richard Biener <rguenther@suse.de>
22743 PR tree-optimization/80713
22744 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
22745 inserted_exprs bit for not removed stmts.
22747 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
22749 PR middle-end/69921
22750 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
22751 parallelized" attribute for parallelized OpenACC kernels.
22752 * omp-offload.c (execute_oacc_device_lower): Use it.
22754 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
22755 Set "oacc kernels" attribute.
22756 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
22757 parameter. Adjust all users.
22758 (oacc_fn_attrib_kernels_p): Remove function.
22759 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
22760 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
22761 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
22762 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
22763 assert "oacc kernels" attribute is set.
22765 2017-05-11 Carl Love <cel@us.ibm.com>
22767 * config/rs6000/rs6000-c: Add support for built-in functions
22768 vector unsigned char vec_popcnt (vector signed char)
22769 vector unsigned char vec_popcnt (vector unsigned char)
22770 vector unsigned short vec_popcnt (vector signed short)
22771 vector unsigned short vec_popcnt (vector unsigned short)
22772 vector unsigned int vec_popcnt (vector signed int)
22773 vector unsigned int vec_popcnt (vector unsigned int)
22774 vector unsigned long long vec_popcnt (vector signed long long)
22775 vector unsigned long long vec_popcnt (vector unsigned long long)
22776 vector signed long long vec_slo (vector signed long long,
22777 vector signed char)
22778 vector signed long long vec_slo (vector signed long long,
22779 vector unsigned char)
22780 vector unsigned long long vec_slo (vector unsigned long long,
22781 vector signed char)
22782 vector unsigned long long vec_slo (vector unsigned long long,
22783 vector unsigned char)
22784 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
22785 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
22786 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
22787 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
22788 * doc/extend.texi: Update the built-in documentation file for the
22789 new built-in functions.
22791 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
22793 * attribs.h (sorted_attr_string): Move machine independent
22794 functions for target clone support from the i386 port to common
22795 code. Rename ix86_function_versions to common_function_versions.
22796 Rename make_name to make_unique_name.
22797 (common_function_versions): Likewise.
22798 (make_unique_name): Likewise.
22799 (make_dispatcher_decl): Likewise.
22800 (is_function_default_version): Likewise.
22801 * attribs.c (attr_strcmp): Likewise.
22802 (sorted_attr_string): Likewise.
22803 (common_function_versions): Likewise.
22804 (make_unique_name): Likewise.
22805 (make_dispatcher_decl): Likewise.
22806 (is_function_default_version): Likewise.
22807 * config/i386/i386.c (attr_strcmp): Likewise.
22808 (sorted_attr_string): Likewise.
22809 (ix86_function_versions): Likewise.
22810 (make_name): Likewise.
22811 (make_dispatcher_decl): Likewise.
22812 (is_function_default_version): Likewise.
22813 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
22815 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22818 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
22819 Account for direct move costs for vec_construct of integer
22822 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
22825 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
22826 (UNSPEC_STX_ATOMIC): Ditto.
22827 (loaddi_via_sse): New insn.
22828 (storedi_via_sse): Ditto.
22829 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
22830 Update corresponding peephole2 patterns.
22831 (atomic_storedi_fpu): Ditto.
22833 2017-05-11 Julia Koval <julia.koval@intel.com>
22835 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
22836 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
22838 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
22839 (__builtin_ia32_rsqrt14ss_mask): New builtins.
22840 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
22842 2017-05-11 Nathan Sidwell <nathan@acm.org>
22844 * graphite-poly.c: Include dumpfile.h.
22846 * dumpfle.h (dump_function): Declare here ...
22847 * tree-dump.h (dump_function): ... not here.
22848 * dumpfile.c: #include tree-cfg.h.
22849 (dump_function): Move here from ...
22850 * tree-dump.c (dump_function): ... here.
22851 * gimplify.c: #include splay-tree.h, not tree-dump.h.
22852 * graphite-poly.c: Don't include tree-dump.h.
22853 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
22854 * print-tree.c: Likewise.
22855 * stor-layout.c: Likewise.
22856 * tree-nested.c: Likewise.
22858 * dumpfile.c (dump_start): Use TDF_FLAGS.
22859 (dump_enable_all): Fix TDF_KIND check thinko.
22861 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
22863 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22864 array entries to represent two legal parameterizations of the
22865 overloaded __builtin_cmpb function, as represented by the
22866 P6_OV_BUILTIN_CMPB constant.
22867 (altivec_resolve_overloaded_builtin): Add special case handling
22868 for the __builtin_cmpb function, as represented by the
22869 P6_OV_BUILTIN_CMPB constant.
22870 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
22871 (BU_P6_64BIT_2): New macro.
22872 (BU_P6_OVERLOAD_2): New macro
22873 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
22874 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
22875 (CMPB): Add overload support to represent both 32-bit and 64-bit
22876 compare-bytes function.
22877 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
22878 support for TARGET_CMPB.
22879 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
22880 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
22881 documentation of the __builtin_cmpb overloaded built-in function.
22883 2017-05-11 Richard Biener <rguenther@suse.de>
22885 PR tree-optimization/80705
22886 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
22887 bases are not vectorizable.
22889 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22891 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
22892 when counting register pressure.
22894 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22896 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
22897 (iv_ca_more_deps): Renamed to ...
22898 (iv_ca_compare_deps): ... this.
22899 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
22901 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22903 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
22905 (determine_group_iv_costs): ... here.
22906 (find_inv_vars_cb): Record inv var if it's not recorded before.
22908 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22910 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
22911 (get_shiftadd_cost): Ditto.
22913 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22915 * tree-ssa-address.c: Include header file.
22916 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
22918 (add_to_parts): Refactor.
22919 (addr_to_parts): New parameter. Update use of move_hint_to_base.
22920 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
22923 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22925 PR tree-optimization/53090
22926 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
22928 (extract_cond_operands): Detect condition with IV on both sides
22929 and return COMP_IV_EXPR_2.
22930 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
22931 (rewrite_use_compare): Simplify by removing call to function
22932 extract_cond_operands.
22934 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22936 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
22937 (extract_cond_operands): Detect condition comparing against non-
22938 invariant bound and return appropriate enum value.
22939 (find_interesting_uses_cond): Update use of extract_cond_operands.
22940 Handle its return value accordingly.
22941 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
22943 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22945 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
22946 nonlinear iv_use computation in loop invariant sensitive way.
22948 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22950 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
22951 (find_iv_candidates): Call relate_compare_use_with_all_cands.
22953 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22955 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
22956 (dump_cand): Support iv_cand.inv_exprs.
22957 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
22959 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
22962 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22964 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
22966 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
22967 as local function. Include necessary header files.
22968 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
22970 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22972 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
22974 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22976 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
22977 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
22978 RSHIFT_EXPR and BIT_NOT_EXPR.
22980 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22982 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
22983 (adjust_setup_cost): New parameter supporting round up adjustment.
22984 (struct address_cost_data): Delete.
22985 (force_expr_to_var_cost): Don't bound cost with spill_cost.
22986 (split_address_cost, ptr_difference_cost): Delete.
22987 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
22988 (struct ainc_cost_data): New struct.
22989 (get_address_cost_ainc): New function.
22990 (get_address_cost, get_computation_cost): Reimplement.
22991 (determine_group_iv_cost_address): Record inv_expr for all uses of
22993 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
22994 (iv_ca_has_deps): Reimplemented to ...
22995 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
22997 (iv_ca_extend): Call iv_ca_more_deps.
22999 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23001 * tree-ssa-address.c (struct mem_address): Move to header file.
23002 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
23003 * tree-ssa-address.h (struct mem_address): Move from C file.
23004 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
23006 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23008 * tree-affine.h (aff_combination_type): New interface.
23009 (aff_combination_zero_p): Remove static.
23010 (aff_combination_const_p): New interface.
23011 (aff_combination_singleton_var_p): New interfaces.
23013 2017-05-11 Richard Biener <rguenther@suse.de>
23015 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23016 Skip unreachable blocks and destinations.
23017 (eliminate): Move stmt removal and fixup ...
23018 (fini_eliminate): ... here. Skip inserted exprs.
23019 (pass_pre::execute): Move fini_pre after fini_eliminate.
23020 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
23021 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
23022 PRE to get rid of dead code that has invalid SSA form and
23023 split critical edges again.
23025 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23027 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
23029 2017-05-11 Richard Biener <rguenther@suse.de>
23031 * passes.c (execute_function_todo): Verify loops if they are
23032 said to be up-to-date.
23033 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
23034 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
23036 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
23039 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
23040 handle calling assemble_external ourself.
23043 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
23044 modes with zero size. Enhance comment.
23046 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23048 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
23049 built-ins for vec_xl and vec_xst with short and char pointer
23052 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
23054 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
23055 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
23056 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
23057 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
23058 (_mm_maskz_min_round_ss): New intrinsics.
23059 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
23060 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
23061 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
23062 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
23063 (__builtin_ia32_minss_mask_round): New builtins.
23064 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23065 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23066 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
23068 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
23069 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
23071 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
23074 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
23076 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
23077 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
23078 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
23079 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
23080 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
23081 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
23082 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
23083 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23084 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
23085 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
23086 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
23087 (__builtin_ia32_mulss_mask_round): New builtins.
23088 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23089 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23090 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
23092 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
23093 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
23095 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
23098 2017-05-10 Julia Koval <julia.koval@intel.com>
23100 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
23101 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
23102 (_mm256_setr_m128i): New intrinsics.
23104 2017-05-10 Julia Koval <julia.koval@intel.com>
23106 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
23107 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
23108 (_mm_maskz_rcp14_ss): New intrinsics.
23109 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
23110 (__builtin_ia32_rcp14ss_mask): New builtins.
23111 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
23113 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
23115 PR tree-optimization/51513
23116 * tree-cfg.c (gimple_seq_unreachable_p): New function.
23117 (assert_unreachable_fallthru_edge_p): Use it.
23118 (group_case_labels_stmt): Likewise.
23119 * tree-cfg.h: Prototype it.
23120 * stmt.c: Include cfghooks.h and tree-cfg.h.
23121 (emit_case_dispatch_table) <gap_label>: New local variable.
23122 Use it to fill dispatch table gaps.
23123 Test for default_label before updating probabilities.
23124 (expand_case) <default_label>: Remove unneeded initialization.
23125 Test for unreachable default case statement and remove its edge.
23126 Set default_label accordingly.
23127 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
23129 2017-05-10 Carl Love <cel@us.ibm.com>
23131 * config/rs6000/rs6000-c: Add support for built-in functions
23132 vector signed char vec_neg (vector signed char)
23133 vector signed short int vec_neg (vector short int)
23134 vector signed int vec_neg (vector signed int)
23135 vector signed long long vec_neg (vector signed long long)
23136 vector float vec_neg (vector float)
23137 vector double vec_neg (vector double)
23138 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
23140 * config/rs6000/altivec.h: Add define for vec_neg
23141 * doc/extend.texi: Update the built-in documentation for the
23142 new built-in functions.
23144 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23146 PR tree-optimization/77644
23147 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
23149 2017-05-10 Nathan Sidwell <nathan@acm.org>
23151 * dumpfile.h (TDI_lang_all): New.
23152 (TDF_KIND): New. Renumber others
23153 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
23155 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
23157 (get_dump_file_name): Adjust suffix generation.
23158 (dump_enable_all): Use TDF_KIND.
23159 * doc/invoke.texi (-fdump-lang-all): Document.
23161 * dumpfile.h: Tabify.
23163 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
23166 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
23167 Move member access before delete.
23169 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
23171 * tree-inline.c (expand_call_inline): Split block at stmt
23174 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
23177 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
23178 are now unused after splitting mov{sf,sd}_hardfloat.
23179 (f32_lr2): Likewise.
23180 (f32_lm): Likewise.
23181 (f32_lm2): Likewise.
23182 (f32_li): Likewise.
23183 (f32_li2): Likewise.
23184 (f32_lv): Likewise.
23185 (f32_sr): Likewise.
23186 (f32_sr2): Likewise.
23187 (f32_sm): Likewise.
23188 (f32_sm2): Likewise.
23189 (f32_si): Likewise.
23190 (f32_si2): Likewise.
23191 (f32_sv): Likewise.
23192 (f32_dm): Likewise.
23193 (f32_vsx): Likewise.
23194 (f32_av): Likewise.
23195 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
23196 For movsf, order stores so the VSX stores occur before the GPR
23197 store which encourages the register allocator to use a traditional
23198 FPR instead of a GPR. For movsd, order the stores so that the GPR
23199 store comes before the VSX stores to allow the power6 to work.
23200 This is due to the power6 not having a 32-bit integer store
23201 instruction from a FPR.
23202 (movsf_hardfloat): Likewise.
23203 (movsd_hardfloat): Likewise.
23205 2017-05-09 Martin Sebor <msebor@redhat.com>
23207 PR translation/80280
23208 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
23211 PR translation/80280
23212 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
23213 data member added in r247778.
23214 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
23216 2017-05-09 Nathan Sidwell <nathan@acm.org>
23218 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
23220 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
23223 2017-05-09 Marek Polacek <polacek@redhat.com>
23225 * doc/invoke.texi: Fix typo.
23227 2017-05-09 Richard Biener <rguenther@suse.de>
23229 * tree-vrp.c (vrp_val_is_max): Adjust comment.
23230 (vrp_val_is_min): Likewise.
23231 (set_value_range_to_value): Likewise.
23232 (set_value_range_to_nonnegative): Likewise.
23233 (gimple_assign_nonzero_p): Likewise.
23234 (gimple_stmt_nonzero_p): Likewise.
23235 (vrp_int_const_binop): Likewise. Remove unreachable case.
23236 (adjust_range_with_scev): Adjust comments.
23237 (compare_range_with_value): Likewise.
23238 (extract_range_from_phi_node): Likewise.
23239 (test_for_singularity): Likewise.
23241 2017-05-09 Richard Biener <rguenther@suse.de>
23243 * tree-vrp.c (get_single_symbol): Add assert that we don't
23244 get overflowed constants as invariant part.
23245 (compare_values_warnv): Add comment before the TREE_NO_WARNING
23246 checks. Use wi::cmp instead of recursing for integer constants.
23247 (compare_values): Just ignore whether we assumed undefined
23248 overflow instead of failing the compare.
23249 (extract_range_for_var_from_comparison_expr): Add comment before the
23250 TREE_NO_WARNING sets.
23251 (test_for_singularity): Likewise.
23252 (extract_range_from_comparison): Do not disable optimization
23253 when we assumed undefined overflow.
23254 (extract_range_basic): Remove init of unused var.
23256 2017-05-09 Richard Biener <rguenther@suse.de>
23258 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
23259 (extract_range_from_multiplicative_op_1): Adjust.
23260 (extract_range_from_binary_expr_1): Use int_const_binop.
23262 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
23265 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
23266 rs6000_store_data_bypass_p in seven define_bypass directives and
23267 in several comments.
23268 * config/rs6000/rs6000-protos.h: Add prototype for
23269 rs6000_store_data_bypass_p function.
23270 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
23271 function implements slightly different (rs6000-specific) semantics
23272 than store_data_bypass_p, returning false rather than aborting
23273 with assertion error when arguments do not satisfy the
23274 requirements of store data bypass.
23275 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
23276 rs6000_store_data_bypass_p.
23278 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
23280 * config/xtensa/xtensa-protos.h
23281 (xtensa_initial_elimination_offset): New declaration.
23282 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
23283 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
23284 macro definition, add case for FRAME_POINTER_REGNUM when
23285 FRAME_GROWS_DOWNWARD.
23286 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
23287 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
23288 xtensa_initial_elimination_offset.
23290 2017-05-08 Nathan Sidwell <nathan@acm.org>
23292 * doc/invoke.texi: Alphabetize -fdump options.
23294 2017-05-08 Martin Sebor <msebor@redhat.com>
23296 PR translation/80280
23297 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
23299 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
23301 * target.def (compute_frame_layout): New optional target hook.
23302 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
23303 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
23304 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
23306 * reload1.c (verify_initial_elim_offsets): Likewise.
23307 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
23308 (use_simple_return_p): Call arm_compute_frame_layout if needed.
23309 (arm_get_frame_offsets): Split up into this ...
23310 (arm_compute_frame_layout): ... and this function.
23312 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
23314 * config/aarch64/constraints.md (Usa): New constraint.
23315 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
23317 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
23319 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
23320 with_multilib_list after it has been checked.
23322 2017-05-08 Richard Biener <rguenther@suse.de>
23324 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
23325 (bitmap_set_subtract_values): Likewise.
23327 2017-05-08 Richard Biener <rguenther@suse.de>
23329 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
23330 (gimple_assign_nonzero): ... this and remove strict_overflow_p
23332 (gimple_stmt_nonzero_warnv_p): Rename to ...
23333 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
23335 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
23336 (extract_range_basic): Adjust, do not disable propagation on
23337 strict overflow sensitive simplification.
23338 (vrp_visit_cond_stmt): Likewise.
23340 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
23342 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
23343 body size unconditionally.
23345 2017-05-07 Jeff Law <law@redhat.com>
23348 2017-05-06 Jeff Law <law@redhat.com>
23349 PR tree-optimization/78496
23350 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
23353 PR tree-optimization/78496
23354 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
23355 (simplify_stmt_using_ranges): Call it.
23356 (vrp_dom_walker::before_dom_children): Extract equivalences
23357 from an ASSERT_EXPR with an equality comparison against a
23360 2017-05-06 Jeff Law <law@redhat.com>
23362 PR tree-optimization/78496
23363 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
23366 PR tree-optimization/78496
23367 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
23368 (simplify_stmt_using_ranges): Call it.
23369 (vrp_dom_walker::before_dom_children): Extract equivalences
23370 from an ASSERT_EXPR with an equality comparison against a
23373 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
23375 * lra-constraints.c (lra_copy_reg_equiv): New function.
23376 (split_reg): Use it to copy equivalence information from the
23377 original register to the spill register.
23379 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
23381 PR rtl-optimization/75964
23382 * simplify-rtx.c (simplify_const_relational_operation): Remove
23383 invalid handling of comparisons of integer ABS.
23385 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
23387 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
23388 initialize to zero.
23389 (init_regs): Remove declaration.
23390 (function_arg_advance_32): Initialize error_p as boolean variable.
23392 2017-05-05 Nathan Sidwell <nathan@acm.org>
23394 * store-motion.c (remove_reachable_equiv_notes): Reformat long
23395 lines. Use for (;;).
23397 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23399 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
23400 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
23401 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
23402 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
23403 VF=2 that require versioning.
23405 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23407 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
23410 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23412 * diagnostic.h (diagnostic_override_option_index): Convert from
23413 macro to inline function.
23415 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23417 * diagnostic.c (last_module_changed_p): New function.
23418 (set_last_module): New function.
23419 (diagnostic_report_current_module): Convert macro usage to
23420 the above functions.
23421 * diagnostic.h (diagnostic_context::last_module): Strengthen
23422 from const line_map * to const line_map_ordinary *.
23423 (diagnostic_last_module_changed): Delete macro.
23424 (diagnostic_set_last_module): Delete macro.
23426 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23428 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
23429 with diagnostic_report_diagnostic.
23430 (diagnostic_n_impl_richloc): Likewise.
23431 * diagnostic.h (report_diagnostic): Delete macro.
23432 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
23433 with diagnostic_report_diagnostic.
23434 * substring-locations.c (format_warning_va): Likewise.
23436 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23438 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
23439 save/restor of format_spec. Move option-printing code to...
23440 (print_option_information): ...this new function, and
23441 reimplement by simply printing to the pretty_printer,
23442 rather than appending to the format string.
23444 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23446 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
23447 handling logic into...
23448 (update_effective_level_from_pragmas): ...this new function.
23450 2017-05-04 Andrew Waterman <andrew@sifive.com>
23452 * config/riscv/riscv.opt (mstrict-align): New option.
23453 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
23454 (SLOW_UNALIGNED_ACCESS): Define.
23455 (riscv_slow_unaligned_access): Declare.
23456 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
23458 (riscv_slow_unaligned_access): New variable.
23459 (rocket_tune_info): Set slow_unaligned_access to true.
23460 (optimize_size_tune_info): Set slow_unaligned_access to false.
23461 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
23462 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
23463 (riscv_option_override): Set riscv_slow_unaligned_access.
23464 * doc/invoke.texi: Add -mstrict-align to RISC-V.
23466 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
23468 * config/riscv/riscv.md: Unify indentation.
23470 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
23475 * config/rs6000/rs6000.md (u code attribute): Add FIX and
23477 (extendsi<mode>2): Add support for doing sign extension via
23478 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
23479 don't have ISA 3.0 instructions.
23480 (extendsi<mode>2 splitter): Likewise.
23481 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
23482 generate the normal insns since SImode can now go in vector
23483 registers. Disallow the special UNSPECs needed for previous
23484 machines to hide SImode being used. Add new insns
23485 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
23486 (fix_trunc<mode>si2_stfiwx): Likewise.
23487 (fix_trunc<mode>si2_internal): Likewise.
23488 (fixuns_trunc<mode>si2): Likewise.
23489 (fixuns_trunc<mode>si2_stfiwx): Likewise.
23490 (fctiw<u>z_<mode>_smallint): Likewise.
23491 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
23492 of floating point to 32-bit integer from doing a direct move to
23493 the GPR registers to do a store.
23494 (fctiwz_<mode>): Break long line.
23496 2017-05-05 Bin Cheng <bin.cheng@arm.com>
23498 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
23499 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
23500 (addr_list, addr_offset_valid_p): New.
23501 (split_address_groups): Check offset validity with above function.
23502 (gt-tree-ssa-loop-ivopts.h): Include header file.
23504 2017-05-05 Nathan Sidwell <nathan@acm.org>
23506 * config.gcc (arm*-*-*): Add missing 'fi'.
23508 2017-05-05 Steve Ellcey <sellcey@cavium.com>
23510 * doc/invoke.texi (-fopt-info): Explicitly say order of options
23511 included in -fopt-info does not matter.
23512 * doc/optinfo.texi (-fopt-info): Fix description of default
23513 behavour. Explicitly say order of options included in -fopt-info
23516 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
23518 * config.gcc: Allow combinations of aprofile and rmprofile values for
23519 --with-multilib-list.
23520 * config/arm/t-multilib: New file.
23521 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
23522 variables. Remove setting of ISA and floating-point ABI in
23523 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
23524 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
23525 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
23526 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
23528 * config/arm/t-rmprofile: Likewise except for the matches changes.
23529 * doc/install.texi (--with-multilib-list): Document the combination of
23530 aprofile and rmprofile values and warn about pitfalls in doing that.
23532 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
23534 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
23535 (movdi_aarch64): Likewise.
23537 2017-05-05 Jakub Jelinek <jakub@redhat.com>
23539 PR tree-optimization/80632
23540 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
23542 (build_arrays): Initialize it for virtual phis.
23543 (fix_phi_nodes): Use it for virtual phis.
23545 PR tree-optimization/80558
23546 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
23547 [x, y] op z into [x op, y op z] for op & or | if conditions
23550 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
23551 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23554 * config/arm/arm.md (use_literal_pool): Remove.
23555 (64-bit immediate split): No longer takes cost into consideration
23556 if arm_disable_literal_pool is enabled.
23557 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
23558 used when arm_disable_literal_pool is enabled.
23559 (arm_max_const_double_inline_cost): Remove use of
23560 arm_disable_literal_pool.
23561 (push_minipool_fix): Add assert.
23562 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
23563 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
23564 (no_literal_pool_sf_immediate): New.
23566 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23568 PR tree-optimization/80613
23569 * tree-ssa-dce.c (propagate_necessity): Remove cases for
23570 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
23572 2017-05-05 Richard Biener <rguenther@suse.de>
23574 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
23576 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
23578 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
23579 of this flag from insn conditions due to removal from r247495.
23581 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
23583 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
23585 (arm_early_store_addr_dep_ptr): Likewise.
23586 * config/arm/aarch-common-protos.h
23587 (arm_early_load_addr_dep_ptr): Add prototype.
23588 (arm_early_store_addr_dep_ptr): Likewise.
23589 * config/arm/cortex-a53.md: Add new bypasses.
23591 2017-05-05 Jakub Jelinek <jakub@redhat.com>
23593 * tree.c (next_type_uid): Change type to unsigned.
23594 (type_hash_canon): Decrement back next_type_uid if
23595 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
23596 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
23599 2017-05-04 Martin Sebor <msebor@redhat.com>
23601 * builtins.c: Fix a trivial typo in a comment.
23603 PR middle-end/79234
23604 * builtins.c (check_sizes): Adjust to handle reading past the end.
23605 Avoid printing excessive upper bound of ranges. Use %E to print
23606 tree nodes instead of converting them to %wu.
23607 (expand_builtin_memchr): New function.
23608 (compute_dest_size): Rename...
23609 (compute_objsize): ...to this.
23610 (expand_builtin_memcpy): Adjust.
23611 (expand_builtin_mempcpy): Adjust.
23612 (expand_builtin_strcat): Adjust.
23613 (expand_builtin_strcpy): Adjust.
23614 (check_strncat_sizes): Adjust.
23615 (expand_builtin_strncat): Adjust.
23616 (expand_builtin_strncpy): Adjust and simplify.
23617 (expand_builtin_memset): Adjust.
23618 (expand_builtin_bzero): Adjust.
23619 (expand_builtin_memcmp): Adjust.
23620 (expand_builtin): Handle memcmp.
23621 (maybe_emit_chk_warning): Check strncat just once.
23623 2017-05-04 Martin Sebor <msebor@redhat.com>
23625 PR preprocessor/79214
23626 PR middle-end/79222
23627 PR middle-end/79223
23628 * builtins.c (check_sizes): Add inlining context and issue
23629 warnings even when -Wno-system-headers is set.
23630 (check_strncat_sizes): Same.
23631 (expand_builtin_strncat): Same.
23632 (expand_builtin_memmove): New function.
23633 (expand_builtin_stpncpy): Same.
23634 (expand_builtin): Handle memmove and stpncpy.
23636 2017-05-04 Bin Cheng <bin.cheng@arm.com>
23638 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
23639 which is not used any more.
23641 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
23643 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
23645 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
23647 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
23648 (cortexa53_tunings): Likewise.
23649 (cortexa57_tunings): Likewise.
23650 (cortexa72_tunings): Likewise.
23651 (cortexa73_tunings): Likewise.
23653 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
23655 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
23656 Set loop alignment to 8.
23658 2017-05-04 Martin Sebor <msebor@redhat.com>
23660 PR translation/80280
23661 * builtins.c (expand_builtin_object_size): Add missing quoting to
23662 %D and like directives.
23663 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
23664 (hsa_type_for_tree_type): Same.
23665 (verify_function_arguments): Same.
23666 * symtab.c (symbol_table::change_decl_assembler_name): Same.
23667 * varasm.c (get_section): Same.
23670 2017-05-04 Martin Sebor <msebor@redhat.com>
23672 PR translation/80280
23673 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
23675 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
23677 * config/aarch64/aarch64.c (generic_addrcost_table):
23678 Change HI/TI mode setting.
23680 2017-05-04 Martin Jambor <mjambor@suse.cz>
23682 PR tree-optimization/80622
23683 * tree-sra.c (comes_initialized_p): New function.
23684 (build_accesses_from_assign): Only set write lazily when
23685 comes_initialized_p is false.
23686 (analyze_access_subtree): Use comes_initialized_p.
23687 (propagate_subaccesses_across_link): Assert !comes_initialized_p
23688 instead of testing for PARM_DECL.
23690 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23692 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
23693 constraint on operand 0 to allow more general addressing modes.
23694 Adjust output template.
23695 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
23697 * config/aarch64/aarch64-protos.h
23698 (aarch64_address_valid_for_prefetch_p): Declare prototype.
23699 * config/aarch64/constraints.md (Dp): New address constraint.
23700 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
23703 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
23705 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
23706 update use of estimate_ipcp_clone_size_and_time.
23707 (estimate_local_effects): Update use of
23708 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
23709 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
23710 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
23711 Return nonspecialized time.
23713 2017-05-04 Richard Biener <rguenther@suse.de>
23715 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
23716 for the last VUSE which def dominates the PHI. Directly call
23718 (get_continuation_for_phi_1): Remove.
23720 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
23722 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
23723 to explain the use of truncating division. Cap the number of
23724 iterations to the maximum given by nb_iterations_upper_bound,
23727 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
23729 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
23730 * configure: Regenerate.
23731 * config.in: Regenerate.
23732 * config/i386/driver-mingw32.c: new file.
23733 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
23734 * config.host: Link driver-mingw32.o on MinGW host.
23735 * doc/install.texi: Document new --enable-mingw-wildcard configure
23738 2017-05-04 Marek Polacek <polacek@redhat.com>
23740 PR tree-optimization/80612
23741 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
23743 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23744 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
23746 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
23747 (movt splitter): Likewise.
23748 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
23749 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
23750 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
23751 block for Thumb-1 with MOVT.
23752 (thumb2_legitimate_address_p): Move code block ...
23753 (can_avoid_literal_pool_for_label_p): ... into this new function.
23754 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
23756 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
23757 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
23758 "M-profile targets with the MOVT instruction".
23760 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23762 * config/arm/arm-builtins.c (arm_init_builtins): Rename
23763 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
23764 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
23766 2017-05-04 Martin Liska <mliska@suse.cz>
23768 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
23769 variable cond_code.
23771 2017-05-04 Richard Biener <rguenther@suse.de>
23773 * tree.c (array_at_struct_end_p): Handle arrays at struct
23774 end with flexarrays more conservatively. Refactor and treat
23775 arrays of arrays or aggregates more strict. Fix
23776 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
23777 * tree.c (array_at_struct_end_p): Adjust prototype.
23778 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
23779 * gimple-fold.c (get_range_strlen): Likewise.
23780 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
23782 2017-05-04 Richard Biener <rguenther@suse.de>
23784 PR tree-optimization/31130
23785 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
23787 (supports_overflow_infinity): Likewise.
23788 (is_negative_overflow_infinity): Likewise.
23789 (is_positive_overflow_infinity): Likewise.
23790 (is_overflow_infinity): Likewise.
23791 (stmt_overflow_infinity): Likewise.
23792 (overflow_infinity_range_p): Likewise.
23793 (usable_range_p): Remove as always returning true.
23794 (make_overflow_infinity): Remove.
23795 (negative_overflow_infinity): Likewise.
23796 (positive_overflow_infinity): Likewise.
23797 (avoid_overflow_infinity): Likewise.
23798 (set_value_range): Adjust accordingly.
23799 (set_value_range_to_nonnegative): Likewise, remove now unused
23800 overflow_infinity arg.
23801 (vrp_operand_equal_p): Adjust.
23802 (update_value_range): Likewise.
23803 (range_int_cst_singleton_p): Likewise.
23804 (operand_less_p): Likewise.
23805 (compare_values_warnv): Likewise.
23806 (extract_range_for_var_from_comparison_expr): Likewise.
23807 (vrp_int_const_binop): Likewise.
23808 (zero_nonzero_bits_from_vr): Likewise.
23809 (extract_range_from_multiplicative_op_1): Likewise.
23810 (extract_range_from_binary_expr_1): Likewise.
23811 (extract_range_from_unary_expr): Likewise.
23812 (extract_range_from_comparison): Likewise.
23813 (extract_range_basic): Likewise.
23814 (adjust_range_with_scev): Likewise.
23815 (compare_ranges): Likewise.
23816 (compare_range_with_value): Likewise.
23817 (dump_value_range): Likewise.
23818 (test_for_singularity): Likewise, remove strict_overflow_p parameter
23820 (simplify_cond_using_ranges): Adjust.
23822 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
23824 * brig-builtins.def: Added a builtin for class_f64.
23825 * builtin-types.def: Added a builtin type needed by class_f64.
23827 2017-05-03 Jason Merrill <jason@redhat.com>
23829 * timevar.def: Add TV_CONSTEXPR.
23831 2017-05-03 David Malcolm <dmalcolm@redhat.com>
23833 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
23835 2017-05-03 Martin Jambor <mjambor@suse.cz>
23837 * ipa-prop.c (ipa_update_after_lto_read): Removed.
23838 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
23839 * ipa-cp.c (ipcp_propagate_stage): Do not call
23840 ipa_update_after_lto_read.
23841 * ipa-inline.c (ipa_inline): Likewise.
23843 2017-05-03 Martin Jambor <mjambor@suse.cz>
23845 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
23846 tag. Added a default constructor and a destructor.
23847 (ipa_edge_args_sum_t): New class;
23848 (ipa_edge_args_sum): Declare.
23849 (ipa_edge_args_vector): Remove declaration.
23850 (IPA_EDGE_REF): Use ipa_edge_args_sum.
23851 (ipa_free_edge_args_substructures): Remove declaration.
23852 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
23853 (ipa_edge_args_info_available_for_edge_p): Likewise.
23854 * ipa-prop.c (ipa_edge_args_vector): Removed.
23855 (edge_removal_hook_holder): Likewise.
23856 (edge_duplication_hook_holder): Likewise.
23857 (ipa_edge_args_sum): New variable.
23858 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
23859 ipa_edge_args_vector.
23860 (ipa_free_edge_args_substructures): Likewise.
23861 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
23862 ipa_edge_args_vector.
23863 (ipa_edge_removal_hook): Turned into method
23864 ipa_edge_args_sum_t::remove.
23865 (ipa_edge_duplication_hook): Turned into method
23866 ipa_edge_args_sum_t::duplicate.
23867 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
23868 registering edge hooks.
23869 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
23870 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
23871 ipa_edge_args_sum instead of ipa_edge_args_vector.
23872 * ipa-profile.c (ipa_profile): Likewise.
23874 2017-05-03 Martin Jambor <mjambor@suse.cz>
23876 * symbol-summary.h (function_summary): New method exists.
23877 (function_summary::symtab_removal): Deallocate through release.
23878 (call_summary): New class.
23879 (gt_ggc_mx): New overload.
23880 (gt_pch_nx): Likewise.
23881 (gt_pch_nx): Likewise.
23883 2017-05-03 Jeff Law <law@redhat.com>
23885 PR tree-optimization/78496
23886 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
23887 from simplify_cond_using_ranges. Split off code to walk
23888 backwards through casts into ...
23889 (simplify_cond_using_ranges_2): New function.
23890 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
23891 (execute_vrp): After identifying jump threads, call
23892 simplify_cond_using_ranges_2.
23894 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
23897 * ipa-inline.h (inline_summary): Add ctor.
23898 (create_ggc): Do not use ggc_cleared_alloc.
23900 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
23901 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23903 * gcc.c (handle_braces): Support escaping in switch matching
23905 * doc/invoke.texi (Spec Files): Document it.
23906 Remove superfluous @code markup in items.
23908 2017-05-03 David Malcolm <dmalcolm@redhat.com>
23910 * diagnostic-show-locus.c (struct column_range): New struct.
23911 (get_affected_columns): New function.
23912 (get_printed_columns): New function.
23913 (struct correction): New struct.
23914 (correction::ensure_capacity): New function.
23915 (correction::ensure_terminated): New function.
23916 (struct line_corrections): New struct.
23917 (line_corrections::~line_corrections): New dtor.
23918 (line_corrections::add_hint): New function.
23919 (layout::print_trailing_fixits): Reimplement in terms of the new
23921 (selftest::test_overlapped_fixit_printing): New function.
23922 (selftest::diagnostic_show_locus_c_tests): Call it.
23924 2017-05-03 Nathan Sidwell <nathan@acm.org>
23926 Canonicalize canonical type hashing
23927 * tree.h (type_hash_canon_hash): Declare.
23928 * tree.c (type_hash_list, attribute_hash_list): Move into
23929 type_hash_canon_hash.
23930 (build_type_attribute_qual_variant): Break out hash code calc into
23931 type_hash_canon_hash.
23932 (type_hash_canon_hash): New. Generic type hash computation.
23933 (build_range_type_1, build_array_type_1, build_function_type,
23934 build_method_type_directly, build_offset_type, build_complex_type,
23935 make_vector_type): Call it.
23937 2017-05-03 Richard Biener <rguenther@suse.de>
23939 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
23940 When all DRs have unknown misaligned do not always peel
23941 when there is a store but apply the same costing model as if
23942 there were only loads.
23944 2017-05-03 Richard Biener <rguenther@suse.de>
23947 PR tree-optimization/80492
23948 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
23949 compare_base_decls returning dont-know properly.
23951 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
23953 * config/arm/iterators.md (CCSI): New mode iterator.
23954 (arch): New mode attribute.
23955 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
23956 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
23957 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
23958 code iterator for success result mode.
23959 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
23960 the corresponding new insn generators.
23962 2017-05-03 Bin Cheng <bin.cheng@arm.com>
23965 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23966 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
23968 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
23970 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
23971 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
23972 (DDR_A): Wrap DDR argument in brackets.
23973 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
23974 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
23975 (DDR_REVERSED_P): Likewise.
23977 2017-05-03 Jakub Jelinek <jakub@redhat.com>
23979 PR tree-optimization/79472
23980 * tree-switch-conversion.c (struct switch_conv_info): Add
23981 contiguous_range and default_case_nonstandard fields.
23982 (collect_switch_conv_info): Compute contiguous_range and
23983 default_case_nonstandard fields, don't clear final_bb if
23984 contiguous_range and only the default case doesn't have the required
23986 (check_all_empty_except_final): Set default_case_nonstandard instead
23987 of failing if contiguous_range and the default case doesn't have empty
23989 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
23990 and only the default case doesn't have the required constants. Skip
23992 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
23993 if default_case_nonstandard.
23994 (build_constructors): Build constant 1 just once. Assert that default
23995 values aren't inserted in between cases if contiguous_range. Skip
23997 (build_arrays): Skip virtual phis.
23998 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
23999 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
24000 Handle virtual phis.
24001 (gen_inbound_check): Handle default_case_nonstandard case.
24002 (process_switch): Adjust check_final_bb caller. Call
24003 gather_default_values with the first non-default case instead of
24004 default case if default_case_nonstandard.
24006 2017-05-02 Nathan Sidwell <nathan@acm.org>
24008 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
24009 check. Fix formatting.
24011 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
24013 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
24014 errors when comparing specialized and unspecialized times.
24016 2017-05-02 David Malcolm <dmalcolm@redhat.com>
24018 * diagnostic-show-locus.c
24019 (layout::should_print_annotation_line_p): Make private.
24020 (layout::print_annotation_line): Make private.
24021 (layout::annotation_line_showed_range_p): Make private.
24022 (layout::show_ruler): Make private.
24023 (layout::print_source_line): Make private. Pass in line and
24024 line_width, rather than calling location_get_source_line. Drop
24026 (layout::print_leading_fixits): New method.
24027 (layout::print_any_fixits): Rename to...
24028 (layout::print_trailing_fixits): ...this, and make private.
24029 Don't print newline fixits.
24030 (diagnostic_show_locus): Move logic for printing one row into...
24031 (layout::print_line): ...this new function. Move the
24032 location_get_source_line call and error-handling from
24033 print_source_line to here. Call print_leading_fixits, and rename
24034 print_any_fixits to print_trailing_fixits.
24035 (selftest::test_fixit_insert_containing_newline): Update now that
24036 newlines are partially supported.
24037 (selftest::test_fixit_insert_containing_newline_2): New test.
24038 (selftest::test_fixit_replace_containing_newline): Update comments.
24039 (selftest::diagnostic_show_locus_c_tests): Call the new test.
24040 * edit-context.c (class added_line): New class.
24041 (class edited_line): Describe newline handling in comment.
24042 (edited_line::actually_edited_p): New method.
24043 (edited_line::print_content): Delete redundant decl.
24044 (edited_line::m_predecessors): New field.
24045 (edited_file::print_content): Call edited_line::print_content.
24046 (edited_file::print_diff): Update to support newlines.
24047 (edited_file::print_diff_hunk): Likewise.
24048 (edited_file::print_run_of_changed_lines): New function.
24049 (edited_file::print_diff_line): Convert to...
24050 (print_diff_line): ...this.
24051 (edited_file::get_effective_line_count): New function.
24052 (edited_line::edited_line): Initialize new field m_predecessors.
24053 (edited_line::~edited_line): Clean up m_predecessors.
24054 (edited_line::apply_fixit): Handle newlines.
24055 (edited_line::get_effective_line_count): New function.
24056 (edited_line::print_content): New function.
24057 (edited_line::print_diff_lines): New function.
24058 (selftest::test_applying_fixits_insert_containing_newline): New
24060 (selftest::test_applying_fixits_replace_containing_newline): New
24062 (selftest::insert_line): New function.
24063 (selftest::test_applying_fixits_multiple_lines): Add example of
24065 (selftest::edit_context_c_tests): Call the new tests.
24067 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24069 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
24070 parameter cand. Update dump information.
24071 (get_computation_cost): Update uses.
24073 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24075 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
24076 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
24077 (get_computation_at, rewrite_use_address): Update use of
24078 get_computation_aff.
24080 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24082 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
24083 (get_computation): Delete.
24084 (get_computation_cost): Implement like get_computation_cost_at.
24085 Use get_computation_at.
24086 (get_computation_cost_at): Delete.
24087 (rewrite_use_nonlinear_expr): Use get_computation_at.
24088 (rewrite_use_compare, remove_unused_ivs): Ditto.
24090 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24092 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
24094 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24096 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
24097 (ivopts_global_cost_for_size): Rename parameter and update uses.
24098 (iv_ca_recount_cost): Update uses.
24099 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
24100 candidates seperately in n_invs and n_cands.
24101 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
24103 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24105 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
24106 (find_inv_vars_cb): New.
24107 (find_depends): Renamed to ...
24108 (find_inv_vars): ... this.
24109 (add_candidate_1, force_var_cost): Call find_inv_vars.
24110 (split_address_cost, determine_group_iv_cost_cond): Ditto.
24112 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24114 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
24115 inv_vars. Add inv_exprs.
24116 (struct iv_cand): Rename depends_on to inv_vars.
24117 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
24118 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
24119 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
24120 (dump_cand): Dump inv_vars.
24121 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
24122 (record_invariant, find_depends, add_candidate_1): Ditto.
24123 (set_group_iv_cost, force_var_cost): Ditto.
24124 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
24125 (get_computation_cost_at, get_computation_cost): Ditto.
24126 (determine_group_iv_cost_generic): Ditto.
24127 (determine_group_iv_cost_address): Ditto.
24128 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
24129 (determine_group_iv_costs): Ditto.
24130 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
24131 (iv_ca_set_remove_invariants): Renamed to ...
24132 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
24133 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
24134 (iv_ca_set_add_invariants): Renamed to ...
24135 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
24136 (iv_ca_set_cp): Use iv_ca_set_add_invs.
24137 (iv_ca_has_deps): Support inv_vars and inv_exprs.
24138 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
24139 (create_new_ivs): Remove useless dump.
24141 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24143 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
24145 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
24146 (iv_ca_set_no_cp, create_new_iv): Ditto.
24148 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24150 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24152 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
24154 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
24155 function tree_check2.
24157 2017-05-02 Martin Liska <mliska@suse.cz>
24159 * doc/gcov.texi: Add missing preposition.
24160 * gcov.c (function_info::function_info): Properly fill up
24161 all member variables.
24163 2017-05-02 Tamar Christina <tamar.christina@arm.com>
24165 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
24167 2017-05-02 Tamar Christina <tamar.christina@arm.com>
24169 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
24171 2017-05-02 Martin Liska <mliska@suse.cz>
24174 * lto-streamer-in.c (lto_read_tree_1): Remove
24175 LTO_STREAMER_DEBUG.
24176 * lto-streamer.c (struct tree_hash_entry): Likewise.
24177 (struct tree_entry_hasher): Likewise.
24178 (tree_entry_hasher::hash): Likewise.
24179 (tree_entry_hasher::equal): Likewise.
24180 (lto_streamer_init): Likewise.
24181 (lto_orig_address_map): Likewise.
24182 (lto_orig_address_get): Likewise.
24183 (lto_orig_address_remove): Likewise.
24184 * lto-streamer.h: Likewise.
24185 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
24186 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
24188 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
24190 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
24191 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
24192 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
24193 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
24194 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
24195 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
24196 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
24197 (mm_maskz_sub_ss): New intrinsics.
24198 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24199 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
24200 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
24201 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
24202 (__builtin_ia32_subss_mask_round): New builtins.
24203 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24204 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24205 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
24207 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
24208 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
24210 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
24213 2017-05-02 Martin Jambor <mjambor@suse.cz>
24215 PR tree-optimization/78687
24216 * tree-sra.c (access): New field parent.
24217 (process_subtree_disqualification): New function.
24218 (disqualify_candidate): Call it.
24219 (build_accesses_from_assign): Reset write flag if creating an
24221 (build_access_subtree): Fill in parent field and also prpagate
24222 down grp_write flag.
24223 (create_artificial_child_access): New parameter set_grp_write, set
24224 grp_write to its value.
24225 (propagate_subaccesses_across_link): Also propagate grp_write flag
24227 (propagate_all_subaccesses): Push the closest parent back to work
24228 queue if add_access_to_work_queue returned true.
24230 2017-05-02 Richard Biener <rguenther@suse.de>
24232 * common.opt (fstrict-overflow): Alias negative to fwrapv.
24233 * doc/invoke.texi (fstrict-overflow): Remove all traces of
24234 -fstrict-overflow documentation.
24235 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
24236 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
24237 flag_strict_overflow.
24238 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
24239 * lto-opts.c (lto_write_options): Do not stream it.
24240 * lto-wrapper.c (merge_and_complain): Do not handle it.
24241 * opts.c (default_options_table): Do not set -fstrict-overflow.
24242 (finish_options): Likewise do not clear it when sanitizing.
24243 * simplify-rtx.c (simplify_const_relational_operation): Do not
24244 test flag_strict_overflow.
24246 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
24248 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
24249 using enabled attribute.
24250 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
24251 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
24252 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
24253 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
24254 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
24255 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
24256 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
24257 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
24258 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
24259 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
24261 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
24263 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
24265 2017-05-02 Richard Biener <rguenther@suse.de>
24267 PR tree-optimization/80591
24269 2017-04-10 Richard Biener <rguenther@suse.de>
24271 * tree-ssa-structalias.c (find_func_aliases): Properly handle
24274 2017-05-02 Richard Biener <rguenther@suse.de>
24276 PR tree-optimization/80549
24277 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
24278 (cleanup_tree_cfg_noloop): Create forwarders to known loop
24279 headers if they do not have a preheader.
24281 2017-05-02 Martin Liska <mliska@suse.cz>
24284 * common.opt: Fix typo.
24285 * doc/invoke.texi: Likewise.
24287 2017-05-01 Jan Beulich <jbeulich@suse.com>
24289 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
24290 swapping, add (x,x,m,x,n) alternative.
24292 2017-05-01 Nathan Sidwell <nathan@acm.org>
24294 * calls.c (combine_pending_stack_adjustment_and_call): Remove
24295 unnecessary unadjusted_alignment check.
24297 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
24300 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
24302 * gimplify.c (gimplify_cilk_detach): New function.
24303 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
24304 * tree-core.h: Document EXPR_CILK_SPAWN.
24305 * tree.h (EXPR_CILK_SPAWN): Define.
24307 2017-05-01 David Malcolm <dmalcolm@redhat.com>
24309 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
24310 to use new fixit_hint representation, using the "replace" logic.
24311 (get_line_span_for_fixit_hint): Likewise.
24312 (layout::print_any_fixits): Likewise.
24313 (selftest::test_one_liner_many_fixits): Rename to...
24314 (selftest::test_one_liner_many_fixits_1): ...this, and update
24315 comment and expected output to reflect that the multiple fix-it
24316 hints are now consolidated into one insertion.
24317 (selftest::test_one_liner_many_fixits_2): New test.
24318 (selftest::test_diagnostic_show_locus_one_liner): Update for
24320 (selftest::test_fixit_consolidation): Update for fix-it API
24322 * diagnostic.c (print_parseable_fixits): Likewise.
24323 * edit-context.c (edited_line::m_line_events): Convert from
24324 auto_vec <line_event *> to auto_vec <line_event>.
24325 (class line_event): Convert from abstract base class to a concrete
24326 class, taking over the role of replace_event.
24327 (class insert_event): Delete.
24328 (class replace_event): Rename to class line_event. Convert to
24330 (edit_context::add_fixits): Reimplement.
24331 (edit_context::apply_insert): Delete.
24332 (edit_context::apply_replace): Rename to...
24333 (edit_context::apply_fixit): ...this. Convert to half-open range.
24334 (edited_file::apply_insert): Delete.
24335 (edited_file::apply_replace): Rename to...
24336 (edited_file::apply_fixit): ...this.
24337 (edited_line::~edited_line): Drop deletion of events.
24338 (edited_line::apply_insert): Delete.
24339 (edited_line::apply_replace): Rename to...
24340 (edited_line::apply_fixit): ...this. Convert to half-open range.
24341 Update for change to type of m_line_events.
24342 * edit-context.h (edit_context::apply_insert): Delete.
24343 (edit_context::apply_replace): Rename to...
24344 (edit_context::apply_fixit): ...this.
24346 2017-05-01 Martin Sebor <msebor@redhat.com>
24348 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
24351 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
24354 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
24355 __get_cpuid_max returns 0.
24356 (__get_cpuid_count): Ditto.
24358 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
24360 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
24361 replacement expression is another instance of one of its arguments.
24363 2017-05-01 Jakub Jelinek <jakub@redhat.com>
24366 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
24367 check for stack push/pop autoinc.
24368 * config/i386/i386.c (ix86_agi_dependent): Return false
24369 if the only reason why modified_in_p returned true is that
24370 addr is SP based and set_insn is a push or pop.
24372 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
24374 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
24377 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
24380 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
24381 (account_size_time): Use two predicates - exec_pred and
24383 (evaluate_conditions_for_known_args): Compute both clause and
24385 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
24386 (inline_summary_t::duplicate): Update.
24387 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
24389 (compute_inline_parameters): Likewise.
24390 (estimate_edge_size_and_time): Update caluclation of time.
24391 (estimate_node_size_and_time): Compute both time and nonspecialized
24393 (estimate_ipcp_clone_size_and_time): Update.
24394 (inline_merge_summary): Update.
24395 (do_estimate_edge_time): Update.
24396 (do_estimate_edge_size): Update.
24397 (do_estimate_edge_hints): Update.
24398 (inline_read_section, inline_write_summary): Stream both new predicates.
24399 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
24401 (compute_inlined_call_time): Cleanup.
24402 (big_speedup_p): Update.
24403 (edge_badness): Update.
24404 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
24405 (size_time_entry): Replace predicate by exec_predicate and
24406 nonconst_predicate.
24407 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
24408 (estimate_edge_time): Return also nonspec_time.
24409 (reset_edge_growth_cache): Update.
24411 2017-04-29 Jakub Jelinek <jakub@redhat.com>
24413 PR rtl-optimization/80491
24414 * ifcvt.c (noce_process_if_block): When looking for x setter
24415 with missing else_bb, don't check only the insn right before
24416 cond_earliest, but look for the last insn that x is modified in
24417 within the same bb.
24419 PR rtl-optimization/80491
24420 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
24422 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
24424 PR tree-optimization/80487
24425 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
24427 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24429 PR tree-optimization/79697
24430 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
24431 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
24432 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
24434 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
24435 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
24437 2017-04-28 Martin Sebor <msebor@redhat.com>
24439 PR tree-optimization/80523
24440 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
24441 (init_target_to_host_charmap, target_to_host, target_strtol10): New
24443 (maybe_warn, format_directive, parse_directive): Use new functions.
24444 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
24446 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
24448 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
24450 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
24452 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
24453 target_header_dir): Set correctly.
24454 * configure: Regenerated.
24455 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
24456 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
24457 instead of SYSTEM_HEADER_DIR.
24459 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
24461 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
24462 (estimate_local_effects): Likewise.
24463 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
24464 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
24465 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
24466 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
24467 do_estimate_edge_time, estimate_edge_time): Likewise.
24468 * ipa-inline-analysis.c (estimate_node_size_and_time,
24469 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
24470 (estimate_time_after_inlining): Remove.
24472 2017-04-28 Martin Liska <mliska@suse.cz>
24474 * doc/gcov.texi: Enhance documentation of gcov.
24476 2017-04-28 Martin Liska <mliska@suse.cz>
24478 * doc/gcov.texi: Sort options in alphabetic order.
24479 * doc/gcov-dump.texi: Likewise.
24480 * doc/gcov-tool.texi: Likewise.
24481 * gcov.c (print_usage): Likewise.
24482 * gcov-dump.c (print_usage): Likewise.
24483 * gcov-tool.c (print_merge_usage_message): Likewise.
24484 (print_rewrite_usage_message): Likewise.
24485 (print_overlap_usage_message): Likewise.
24487 2017-04-28 Martin Liska <mliska@suse.cz>
24489 PR gcov-profile/53915
24490 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
24492 2017-04-28 Martin Liska <mliska@suse.cz>
24494 PR gcov-profile/79891
24495 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
24496 is marked by compiler as living on a line.
24497 (get_cycles_count): Remove usage of the union.
24498 (output_intermediate_file): Likewise.
24499 (find_source): Fix GNU coding style.
24500 (accumulate_line_counts): Remove old non-all block mode.
24501 (output_lines): Remove usage of the union.
24502 * profile.c (output_location): Include all BBs, even if
24503 belonging to a same line (and file) as a previous BB.
24505 2017-04-28 Martin Liska <mliska@suse.cz>
24507 * gcov.c (process_args): Handle new argument 'w'.
24508 (read_graph_file): Assign ID to BBs.
24509 (output_branch_count): Display BB # if verbose flag is set.
24510 (output_lines): Likewise for arcs.
24511 (print_usage): Add '--verbose' option help.
24512 * doc/gcov.texi: Document --verbose (-w) option.
24514 2017-04-28 Martin Liska <mliska@suse.cz>
24516 * gcov.c (struct block_location_info): New struct.
24517 (process_file): Fill up the new structure.
24518 (read_graph_file): Replace usage of encoding by the newly added
24520 (add_line_counts): Likewise.
24521 (accumulate_line_counts): Remove usage of the union.
24522 (function_info::function_info): New function.
24523 (function_info::~function_info): Likewise.
24524 (process_file): Call delete instead of release_function.
24525 (release_function): Release the function.
24526 (release_structures): Call delete instead of release_function.
24527 (solve_flow_graph): Replace usage of num_blocks.
24528 (find_exception_blocks): Likewise.
24529 (output_lines): Fix GNU coding style.
24531 2017-04-28 Martin Liska <mliska@suse.cz>
24534 * coverage.c (coverage_remove_note_file): New function.
24535 * coverage.h: Declare the function.
24536 * toplev.c (finalize): Clean if an error has been seen.
24538 2017-04-28 Martin Liska <mliska@suse.cz>
24540 PR gcov-profile/80031
24541 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
24542 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
24543 * gcov.c (read_graph_file): Read just number of blocks.
24544 * profile.c (branch_prob): Do not stream 0 flags per a basic
24547 2017-04-28 Martin Liska <mliska@suse.cz>
24549 * gcov-dump.c (tag_*): Add new argument to declarations.
24550 (dump_gcov_file): Likewise.
24551 (tag_blocks): Add and use new argument depth.
24552 (tag_arcs): Likewise.
24553 (tag_lines): Likewise.
24554 (tag_counters): Likewise.
24555 (tag_summary): Likewise.
24556 (dump_working_sets): Use depth to do a proper indentation.
24558 2017-04-28 Jakub Jelinek <jakub@redhat.com>
24561 * cgraph.h (symtab_node::debug_symtab): No longer inline.
24562 * symtab.c (symtab_node::debug_symtab): Move definition here.
24564 2017-04-28 Richard Biener <rguenther@suse.de>
24566 * lto-streamer.h (LTO_major_version): Bump to 7.
24568 2017-04-28 Richard Biener <rguenther@suse.de>
24570 * tree-vrp.c (assert_info): New struct.
24571 (add_assert_info): New helper.
24572 (register_edge_assert_for_2): Refactor to add asserts to a vector
24574 (register_edge_assert_for_1): Likewise.
24575 (register_edge_assert_for): Likewise.
24576 (finish_register_edge_assert_for): New helper actually registering
24577 asserts where live on edge.
24578 (find_conditional_asserts): Adjust.
24579 (find_switch_asserts): Likewise.
24580 (evrp_dom_walker::try_find_new_range): Generalize.
24581 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
24583 2017-04-27 Marek Polacek <polacek@redhat.com>
24586 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
24587 arg10 and arg11 to itype.
24589 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
24591 * doc/extend.texi (Object Size Checking): Improve grammar.
24593 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
24596 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
24597 that the logic for permitting reciprocal estimates matches that
24600 2017-04-27 Jakub Jelinek <jakub@redhat.com>
24603 * tree.c (type_cache_hasher::equal): Only compare
24604 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
24605 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
24606 non-aggregate element types.
24607 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
24608 about the flag on ARRAY_TYPEs in the comment, formatting fix.
24610 2017-04-27 Richard Biener <rguenther@suse.de>
24612 PR middle-end/80533
24613 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
24614 stripping ARRAY_REFs from MEM_EXPR make sure we're not
24615 keeping a reference to a trailing array.
24617 2017-04-27 Richard Biener <rguenther@suse.de>
24619 PR middle-end/80539
24620 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
24621 being in loop-closed SSA form conservatively.
24622 (chrec_fold_multiply_poly_poly): Likewise.
24624 2017-04-27 Tamar Christina <tamar.christina@arm.com>
24626 PR middle-end/79665
24627 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
24628 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
24630 2017-04-27 Jakub Jelinek <jakub@redhat.com>
24633 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
24634 (aarch64_function_arg_alignment): Return unsigned int again, but still
24635 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
24636 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
24637 Don't emit -Wpsabi note.
24638 (aarch64_function_arg_boundary): Likewise.
24639 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
24642 2017-04-26 Nathan Sidwell <nathan@acm.org>
24644 * tree.h (crc32_unsigned_n): Declare.
24645 (crc32_unsigned, crc32_unsigned): Make inline.
24646 * tree.c (crc32_unsigned_bits): Replace with ...
24647 (crc32_unsigned_n): ... this.
24648 (crc32_unsigned, crc32_byte): Remove.
24649 (crc32_string): Remove unnecessary braces.
24651 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
24653 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
24654 * ipa-inline-analysis.c (MAX_TIME): Remove.
24655 (account_size_time): Use sreal for time.
24656 (dump_inline_summary): Update.
24657 (estimate_function_body_sizes): Update.
24658 (estimate_edge_size_and_time): Update.
24659 (estimate_calls_size_and_time): Update.
24660 (estimate_node_size_and_time): Update.
24661 (inline_merge_summary): Update.
24662 (inline_update_overall_summary): Update.
24663 (estimate_time_after_inlining): Update.
24664 (inline_read_section): Update.
24665 (inline_write_summary): Update.
24666 * ipa-inline.c (compute_uninlined_call_time): Update.
24667 (compute_inlined_call_time): Update.
24668 (recursive_inlining): Update.
24669 (inline_small_functions): Update.
24670 (dump_overall_stats): Update.
24671 * ipa-inline.h: Include sreal.h.
24672 (size_time_entry): Turn time to sreal.
24673 (inline_summary): Turn self_time nad time to sreal.
24675 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
24677 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
24679 (sreal::stream_out, sreal::stream_in): New.
24680 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
24682 2017-04-25 Jakub Jelinek <jakub@redhat.com>
24684 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
24687 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
24690 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
24692 (dimode_scalar_chain::compute_convert_gain): Ditto.
24693 (dimode_scalar_chain::make_vector_copies): Ditto.
24694 (dimode_scalar_chain::convert_reg): Ditto.
24695 (dimode_scalar_chain::convert_insn): Ditto.
24696 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
24697 (VI248_AVX512BW_1): New mode iterator.
24698 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
24699 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
24702 2017-04-25 Martin Sebor <msebor@redhat.com>
24704 PR tree-optimization/80497
24705 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
24706 constants are representable in HOST_WIDE_INT.
24707 (parse_directive): Ditto.
24709 2017-04-25 Martin Sebor <msebor@redhat.com>
24712 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
24713 (new_zero_array): Adjust signature.
24714 (dom_info::dom_init): Used unsigned rather that size_t.
24715 (dom_info::dom_info): Same.
24717 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24718 Jakub Jelinek <jakub@redhat.com>
24721 * config/arm/arm.c: Include gimple.h.
24722 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
24723 returns negative, increment ncrn only if it returned positive.
24724 (arm_needs_doubleword_align): Return int instead of bool,
24725 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
24726 members, but if there is any such non-FIELD_DECL
24727 > PARM_BOUNDARY aligned decl, return -1 instead of false.
24728 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
24729 returns negative, increment nregs only if it returned positive.
24730 (arm_setup_incoming_varargs): Likewise.
24731 (arm_function_arg_boundary): Emit -Wpsabi note if
24732 arm_needs_doubleword_align returns negative, return
24733 DOUBLEWORD_ALIGNMENT only if it returned positive.
24735 2017-04-25 Marek Polacek <polacek@redhat.com>
24738 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
24739 first argument to type.
24741 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
24744 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
24745 type checks to test for compatibility instead of equality.
24747 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24748 Jakub Jelinek <jakub@redhat.com>
24751 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
24753 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
24754 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
24755 the alignment computation, but return their maximum in warn_alignment.
24756 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
24757 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
24759 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
24760 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
24763 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24765 * config/arc/simdext.md (dmpyh): Fix typo.
24767 2017-04-25 Richard Biener <rguenther@suse.de>
24769 PR tree-optimization/80492
24770 * alias.c (compare_base_decls): Handle registers with asm
24771 specification conservatively.
24772 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
24773 compare_base_decls returning dont-know properly.
24775 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24777 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
24778 (legitimate_offset_address_p): New function.
24779 (arc_legitimate_address_p): Use above function.
24781 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24783 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
24785 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24787 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
24788 ACCH registers whenever they are available.
24790 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24792 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
24793 double regs fix when not used.
24795 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24797 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
24799 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
24800 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
24802 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24804 * config/arc/arc.c (arc_output_addsi): Check for h-register class
24805 when emitting short ADD instructions.
24807 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24809 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
24811 (cmpsi_cc_c_insn): Likewise.
24812 (cbranchsi4_scratch): Compute proper instruction length using
24813 compact_hreg_operand.
24814 * config/arc/predicates.md (compact_hreg_operand): New predicate.
24816 2017-04-25 Richard Biener <rguenther@suse.de>
24818 PR middle-end/80509
24819 * passes.c (pass_manager::pass_manager): Initialize
24820 m_name_to_pass_map.
24822 2017-04-25 Richard Biener <rguenther@suse.de>
24824 PR tree-optimization/79201
24825 * tree-ssa-sink.c (statement_sink_location): Handle calls.
24827 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24830 * config/s390/vector.md: Split MEM->GPR vector moves for
24831 non-s_operand addresses.
24833 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24836 * config/s390/predicates.md (reload_const_wide_int_operand): New
24838 * config/s390/s390.md ("movti"): Remove d/P alternative.
24839 ("movti_bigconst"): New pattern definition.
24841 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
24844 * s390-protos.h (s390_expand_cs_hqi): Removed.
24845 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
24846 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
24847 modes as well as CCZ1mode and CCZmode.
24848 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
24849 signature of s390_emit_compare_and_swap.
24850 (s390_expand_cs_hqi): Likewise, make static.
24851 (s390_expand_cs_tdsi): Generate an explicit compare before trying
24852 compare-and-swap, in some cases.
24853 (s390_expand_cs): Wrapper function.
24854 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
24856 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
24857 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
24858 patterns for small and large integers. Forbid symref memory operands.
24859 Move expander to s390.c. Require cc register.
24860 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
24861 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
24862 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
24863 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
24864 symref memory operands. Remove CC mode and call s390_match_ccmode
24866 ("atomic_exchange<mode>"): Allow and implement all integer modes.
24868 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
24870 * config/s390/s390.md (define_peephole2): New peephole to help
24871 combining the load-and-test pattern with volatile memory.
24873 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
24875 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
24876 with CCZmode for TARGET_Z196.
24878 2017-04-25 Jakub Jelinek <jakub@redhat.com>
24880 PR rtl-optimization/80501
24881 * combine.c (make_compound_operation_int): Set subreg_code to SET
24882 even for AND with mask of the sign bit of mode.
24884 PR rtl-optimization/80500
24885 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
24886 sum's initial value.
24888 2017-04-25 Julian Brown <julian@codesourcery.com>
24889 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24891 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
24893 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
24895 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
24897 2017-04-25 Julian Brown <julian@codesourcery.com>
24898 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24900 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
24901 (thunderx2t99_sha): New Reservation.
24903 2017-04-25 Julian Brown <julian@codesourcery.com>
24904 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24906 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
24907 type for 1-element load.
24909 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
24911 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
24913 2017-04-24 Martin Jambor <mjambor@suse.cz>
24915 PR tree-optimization/80293
24916 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
24917 char arrays not totally scalarizable if it is false.
24918 (analyze_all_variable_accesses): Pass correct value in the new
24919 parameter. Add a statistics counter.
24921 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
24923 PR middle-end/79931
24924 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
24926 2017-04-24 Richard Biener <rguenther@suse.de>
24928 PR tree-optimization/80494
24929 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
24930 out for complex types.
24932 2017-04-24 Richard Biener <rguenther@suse.de>
24934 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
24935 * tree-ssa-sccvn.c (print_scc): Print SCC size.
24936 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
24937 (DFS): Adjust and never fail.
24938 (sccvn_dom_walker::fail): Remove.
24939 (sccvn_dom_walker::before_dom_children): Adjust.
24940 (run_scc_vn): Likewise and never fail.
24941 * tree-ssa-pre.c (pass_pre::execute): Adjust.
24942 (pass_fre::execute): Likewise.
24944 2017-04-24 Richard Biener <rguenther@suse.de>
24946 PR tree-optimization/79725
24947 * tree-ssa-sink.c (statement_sink_location): Return whether
24948 failure reason was zero uses. Move that check later.
24949 (sink_code_in_bb): Deal with zero uses by removing the stmt
24952 2017-04-24 Richard Biener <rguenther@suse.de>
24955 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
24956 pointer-based references.
24958 2017-04-24 Richard Biener <rguenther@suse.de>
24961 * pass_manager.h (pass_manager::operator new): Remove.
24962 (pass_manager::operator delete): Likewise.
24963 * passes.c (pass_manager::operator new): Remove.
24964 (pass_manager::operator delete): Likewise.
24965 (pass_manager::pass_manager): Zero individual pass members.
24967 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
24970 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
24971 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
24972 Check "XEXP (src, 1)" operand here.
24973 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
24974 Check "XEXP (src, 1)" operand here.
24975 (dimode_scalar_chain::make_vector_copies): Detect count register
24976 of a shift instruction. Zero extend count register from QImode
24977 to DImode to satisfy vector shift pattern count operand predicate.
24978 Substitute vector shift count operand with a DImode copy.
24979 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
24982 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
24984 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
24985 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
24986 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
24987 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
24988 (UNSPEC_NOREX_MEM): Remove definition.
24990 2017-04-21 Richard Biener <rguenther@suse.de>
24992 PR tree-optimization/79547
24993 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
24994 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
24995 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
24996 without any constraints.
24998 2017-04-21 Richard Biener <rguenther@suse.de>
25000 PR tree-optimization/78847
25001 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
25003 2017-04-21 Richard Biener <rguenther@suse.de>
25005 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
25006 (build_distinct_type_copy): Likewise.
25007 (build_variant_type_copy): Likewise.
25008 * tree.c (build_qualified_type): Pass down mem-stat info.
25009 (build_distinct_type_copy): Likewise.
25010 (build_variant_type_copy): Likewise.
25012 2017-04-21 Richard Biener <rguenther@suse.de>
25014 PR tree-optimization/80237
25015 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
25017 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
25018 for a simplified result.
25020 2016-04-21 Richard Biener <rguenther@suse.de>
25022 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
25023 sth as strict as a simple_iv but a chrec without symbols and an
25024 operand defined in the loop we are peeling (and not some subloop).
25025 (propagate_constants_for_unrolling): Propagate all constants.
25027 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25030 * config/i386/i386.c (print_reg): Remove assert for disalowed
25031 regno values, call output_operand_lossage instead.
25033 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25036 * config/i386/constraints.md (Yc): New register constraint.
25037 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
25038 Use Yc constraint for alternative 2 of operand 0. Remove
25039 preferred_for_speed attribute.
25041 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
25043 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
25044 lastprivate clauses in SIMT case.
25046 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
25048 * doc/invoke.texi (-Wextra-semi): Document new warning option.
25050 2017-04-20 Richard Biener <rguenther@suse.de>
25052 PR tree-optimization/57796
25053 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
25054 as N scalar stores.
25055 (vect_model_load_cost): Cost gathers as N scalar loads.
25057 2017-04-20 Richard Biener <rguenther@suse.de>
25059 * ggc-page.c (ggc_allocated_p): Rename to ...
25060 (safe_lookup_page_table_entry): ... this and return the lookup
25062 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
25064 2017-04-20 Richard Biener <rguenther@suse.de>
25066 PR tree-optimization/80453
25067 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
25068 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
25069 from the conditions.
25070 (vn_phi_eq): Pass them down.
25071 (vn_phi_lookup): Record them.
25072 (vn_phi_insert): Likewise.
25074 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
25076 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
25077 uninitialized variable warning to avoid buffer overrun.
25079 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
25082 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
25083 is suppressed for '{ 0 }' in C.
25085 2017-04-20 Jakub Jelinek <jakub@redhat.com>
25087 * BASE-VER: Set to 8.0.0.
25089 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
25091 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
25092 priority .init_array and .fini_array section with SECTION_NOTYPE
25095 2017-04-20 Jakub Jelinek <jakub@redhat.com>
25097 PR middle-end/80423
25098 * tree.h (build_array_type): Add typeless_storage default argument.
25099 * tree.c (type_cache_hasher::equal): Also compare
25100 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
25101 (build_array_type): Add typeless_storage argument, set
25102 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
25104 (build_nonshared_array_type): Adjust build_array_type_1 caller.
25105 (build_array_type): Likewise. Add typeless_storage argument.
25107 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
25108 Jakub Jelinek <jakub@redhat.com>
25110 PR tree-optimization/80426
25111 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
25112 operation on symbolic operands, also compute the overflow for the
25113 invariant part when the operation degenerates into a negation.
25115 2017-04-19 Jakub Jelinek <jakub@redhat.com>
25118 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
25119 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
25122 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
25124 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
25127 * config/avr/avr.c (tree.h): Include it.
25128 (cgraph.h): Include it.
25129 (avr_encode_section_info): Don't warn for uninitialized progmem
25130 variable if it's just an alias.
25132 2017-04-19 Richard Biener <rguenther@suse.de>
25135 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
25136 when needed by AutoPGO.
25138 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
25141 * doc/lto.texi: Remove an extra 'that'.
25143 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
25145 PR rtl-optimization/80429
25146 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
25147 are only used in debug insns.
25149 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
25150 Vladimir Makarov <vmakarov@redhat.com>
25152 * config/sparc/predicates.md (input_operand): Add comment. Return
25153 true for any memory operand when LRA is in progress.
25154 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
25156 2017-04-18 Jeff Law <law@redhat.com>
25159 * mips.md ({return,simple_return}_internal): Do not overwrite
25162 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25164 PR tree-optimization/80443
25165 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
25166 instead of adding 1, subtract -1 and similarly instead of subtracting
25169 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
25171 PR rtl-optimization/80357
25172 * haifa-sched.c (tmp_bitmap): New variable.
25173 (model_recompute): Handle duplicate use records.
25174 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
25175 (free_global_sched_pressure_data): Free it.
25177 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
25180 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
25181 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25182 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25183 instead of SYSTEM_HEADER_DIR.
25185 2017-04-18 Jeff Law <law@redhat.com>
25187 PR middle-end/80422
25188 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
25189 predecessors after walking up the insn chain.
25191 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25194 * dwarf2out.c (modified_type_die): Try harder not to emit internal
25195 sizetype type into debug info.
25197 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
25200 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
25201 unneeded test for TARGET_UPPER_REGS_SF.
25202 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
25204 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25207 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
25208 instead of gsi_after_labels.
25210 2017-04-18 Jeff Law <law@redhat.com>
25212 * regcprop.c (maybe_mode_change): Avoid creating copies of the
25216 2017-04-13 Jeff Law <law@redhat.com>
25217 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
25218 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
25220 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
25223 * config/avr/avr.c (intl.h): Include it.
25224 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
25226 2017-04-18 Martin Liska <mliska@suse.cz>
25228 PR gcov-profile/78783
25229 * gcov-tool.c (gcov_output_files): Validate that destination
25230 file is either removed by the tool or by a user.
25232 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
25233 Guy Benyei <guybe@mellanox.com>
25235 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
25236 block, and do not negate it, the stored id is already negative.
25238 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
25240 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
25242 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
25245 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
25246 masks of options that should be turned off if the VSX vector
25247 options are turned off.
25248 (OTHER_P8_VECTOR_MASKS): Likewise.
25249 (OTHER_VSX_VECTOR_MASKS): Likewise.
25250 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
25251 rs6000_disable_incompatible_switches to validate no type switches
25253 (rs6000_incompatible_switch): New function to disallow turning on
25254 other vector options if -mno-vsx, -mno-power8-vector, or
25255 -mno-power9-vector are specified.
25257 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25259 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
25261 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25263 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
25264 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
25265 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
25266 (ARG_POINTER_CFA_OFFSET): Likewise.
25268 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25270 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
25271 conditions to take advantage of various optimizations.
25273 2017-04-13 Jeff Law <law@redhat.com>
25275 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
25276 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
25277 (zero_extendsidi2_dext): Likewise.
25279 2017-04-13 Jakub Jelinek <jakub@redhat.com>
25282 * fold-const.c (fold_ternary_loc): Revert
25283 use op0 instead of fold_convert_loc (loc, type, arg0) part of
25286 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
25288 PR rtl-optimization/80343
25289 * lra-remat.c (update_scratch_ops): Assign original hard reg to
25290 new scratch pseudo.
25292 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
25295 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
25296 to ubsan_encode_value.
25298 2017-04-13 Jeff Law <law@redhat.com>
25300 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
25301 appearing in DEBUG_INSNs.
25303 2017-04-13 Martin Liska <mliska@suse.cz>
25305 PR gcov-profile/80413
25306 * gcov-io.c (gcov_write_string): Copy to buffer just when
25307 allocated size is greater than zero.
25309 2017-04-13 Jakub Jelinek <jakub@redhat.com>
25312 * dwarf2out.c (decls_for_scope): Ignore declarations of
25313 current_function_decl in BLOCK_NONLOCALIZED_VARS.
25315 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
25318 * ipa-visibility.c (non_local_p): Fix typos.
25319 (localize_node): When localizing symbol in same comdat group,
25320 dissolve the group only when we know external symbols are going
25322 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
25324 2017-04-12 Jakub Jelinek <jakub@redhat.com>
25326 PR tree-optimization/79390
25327 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
25328 order does not result in usable sequence, retry with reversed operand
25334 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
25335 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
25336 op0 instead of fold_convert_loc (loc, type, arg0).
25338 2017-04-12 Jeff Law <law@redhat.com>
25340 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
25341 has a delay slot in the generated code.
25343 * config/cris/cris.md (cris_preferred_reload_class): Return
25344 GENNONACR_REGS rather than GENERAL_REGS.
25346 2017-04-12 Jakub Jelinek <jakub@redhat.com>
25349 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
25350 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
25351 signedness of the result type.
25353 2017-04-12 Richard Biener <rguenther@suse.de>
25354 Jeff Law <law@redhat.com>
25356 PR tree-optimization/80359
25357 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
25358 trim stores to TARGET_MEM_REFs.
25360 2017-04-12 Richard Biener <rguenther@suse.de>
25362 PR tree-optimization/79390
25363 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
25364 threading case even more.
25366 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
25369 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
25370 for quad_address_p for TImode, instead of just not indexed_address.
25372 2017-04-12 Richard Biener <rguenther@suse.de>
25373 Bernd Edlinger <bernd.edlinger@hotmail.de>
25375 PR middle-end/79671
25376 * alias.c (component_uses_parent_alias_set_from): Handle
25377 TYPE_TYPELESS_STORAGE.
25378 (get_alias_set): Likewise.
25379 * tree-core.h (tree_type_common): Add typeless_storage flag.
25380 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
25381 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
25382 for types containing members with TYPE_TYPELESS_STORAGE.
25383 (place_field): Likewise.
25384 (layout_type): Likewise for ARRAY_TYPE.
25385 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
25386 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
25387 TYPE_TYPELESS_STORAGE.
25388 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
25390 2017-04-12 Jakub Jelinek <jakub@redhat.com>
25393 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
25394 first argument to type.
25396 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25400 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
25401 CONST0_RTX (mode) rather than const0_rtx where appropriate.
25402 (rs6000_expand_binop_builtin): Likewise.
25403 (rs6000_expand_ternop_builtin): Likewise; also add missing
25404 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
25405 vshasigma built-ins.
25406 * doc/extend.texi: Document that vec_xxpermdi's third argument
25407 must be a constant.
25409 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
25411 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
25412 Use shift_const cost parameter when calculating gain of STV shifts.
25414 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
25416 PR rtl-optimization/70478
25417 * lra-constraints.c (process_alt_operands): Check memory for
25418 disfavoring memory insn operand.
25420 2017-04-11 Jakub Jelinek <jakub@redhat.com>
25422 PR middle-end/80100
25423 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
25424 left shift in unsigned HOST_WIDE_INT type.
25426 PR rtl-optimization/80385
25427 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
25428 (not (neg X)) into (plus X -1) for complex or non-integral modes.
25431 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
25432 if they have any depend clauses.
25434 2017-04-11 Martin Liska <mliska@suse.cz>
25437 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
25438 * ipa-split.c (split_function): Create a local comdat symbol
25439 if caller is in a comdat group.
25441 2017-04-11 Martin Liska <mliska@suse.cz>
25444 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
25447 2017-04-11 Martin Sebor <msebor@redhat.com>
25449 PR middle-end/80364
25450 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
25451 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
25453 (directive::set_width, directive::set_precision, format_character):
25455 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
25458 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
25461 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
25462 conflict, set target->arch_name instead of target->cpu_name.
25464 2017-04-11 Richard Biener <rguenther@suse.de>
25466 PR tree-optimization/80374
25467 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
25468 build_zero_cst, remove fold_convertible_p check again.
25470 2017-04-11 Martin Liska <mliska@suse.cz>
25473 * ubsan.c (instrument_object_size): Do not instrument register
25476 2017-04-11 Jakub Jelinek <jakub@redhat.com>
25479 * config/i386/i386-builtin-types.def
25480 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
25481 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
25482 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
25483 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
25484 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
25485 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
25486 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
25487 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
25488 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
25489 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
25490 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
25491 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
25492 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
25493 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
25494 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
25495 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
25496 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
25497 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
25498 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
25499 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
25500 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
25501 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
25502 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
25503 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
25504 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
25505 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
25506 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
25507 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
25508 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
25509 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
25510 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
25511 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
25512 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
25513 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
25514 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
25515 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
25516 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
25517 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
25518 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
25519 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
25520 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
25521 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
25522 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
25523 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
25524 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
25525 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
25527 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
25528 flag to second_arg_count, handle 4 argument function type _COUNT
25529 aliases, handle second_arg_count on second argument rather than last.
25531 2017-04-10 Jeff Law <law@redhat.com>
25533 PR tree-optimization/80374
25534 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
25535 record anything if we can not convert integer_zero_node to the
25538 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
25541 * config/rs6000/rs6000.c (rs6000_option_override_internal):
25542 Enhance special handling given to the TARGET_P9_MINMAX option in
25543 relation to certain other options.
25545 2017-04-10 Bin Cheng <bin.cheng@arm.com>
25547 PR tree-optimization/80153
25548 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
25549 remove POINTER_PLUS_EXPR's base part directly, rather than through
25552 2017-04-10 Richard Biener <rguenther@suse.de>
25553 Bin Cheng <bin.cheng@arm.com>
25555 PR tree-optimization/80153
25556 * tree-affine.c (aff_combination_to_tree): Get base pointer from
25557 the first element of pointer type aff_tree. Build result expr in
25559 (add_elt_to_tree): Convert to type unconditionally. Remove other
25560 fold_convert calls.
25561 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
25562 (rewrite_use_nonlinear_expr): Check invariant using iv information.
25564 2017-04-10 Richard Biener <rguenther@suse.de>
25566 * tree-ssa-structalias.c (find_func_aliases): Properly handle
25569 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
25571 PR rtl-optimization/70478
25572 * lra-constraints.c (curr_small_class_check): New.
25573 (update_and_check_small_class_inputs): New.
25574 (process_alt_operands): Update curr_small_class_check. Disfavor
25575 alternative insn memory operands. Check available regs for small
25578 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
25581 * config/mips/mips.opt (-mvirt): Update description.
25582 * doc/invoke.texi (-mvirt): Likewise.
25584 2017-04-10 Richard Biener <rguenther@suse.de>
25586 PR middle-end/80362
25587 * fold-const.c (fold_binary_loc): Look at unstripped ops when
25588 looking for NEGATE_EXPR in -A / -B to A / B folding.
25590 2017-04-10 Martin Liska <mliska@suse.cz>
25592 PR gcov-profile/80224
25593 * gcov.c (print_usage): Fix usage string.
25594 (get_gcov_intermediate_filename): Remove.
25595 (output_gcov_file): Use both for normal and intermediate format.
25596 (generate_results): Do not initialize special file for
25597 intermediate format.
25599 2017-04-10 Richard Biener <rguenther@suse.de>
25601 PR tree-optimization/80304
25602 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
25605 2017-04-10 Nathan Sidwell <nathan@acm.org>
25608 * config/rs6000/rs6000.c (rs6000_vector_type): New.
25609 (rs6000_init_builtins): Use it.
25611 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25613 * config/arm/arm.md (<mrc>): Add mode to SET source.
25614 (<mrrc>): Likewise.
25616 2017-04-10 Richard Biener <rguenther@suse.de>
25618 PR middle-end/80344
25619 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
25621 2017-04-10 Jakub Jelinek <jakub@redhat.com>
25624 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
25625 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
25626 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
25627 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
25628 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
25629 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
25630 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
25631 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
25632 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
25633 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
25634 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
25635 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
25636 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
25637 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
25638 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
25639 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
25640 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
25641 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
25642 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
25643 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
25644 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
25645 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
25646 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
25648 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
25650 PR rtl-optimization/70478
25651 * lra-constraints.c: Reverse the last patch.
25653 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
25655 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
25656 Add comment for WCHAR_T.
25658 2017-04-08 Martin Liska <mliska@suse.cz>
25661 2017-04-07 Martin Liska <mliska@suse.cz>
25664 * ipa-split.c (split_function): Add function part to a same comdat
25667 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25670 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
25672 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
25674 * rs6000/rs6000.c (vec_load_pendulum): Rename...
25675 (vec_pairing): ...to this.
25676 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
25677 (rs6000_sched_init): Adjust for name change.
25678 (struct rs6000_sched_context): Likewise.
25679 (rs6000_init_sched_context): Likewise.
25680 (rs6000_set_sched_context): Likewise.
25682 2017-04-07 Jakub Jelinek <jakub@redhat.com>
25688 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
25690 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
25691 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
25692 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
25694 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
25696 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
25698 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
25700 PR rtl-optimization/70703
25701 * ira-color.c (update_conflict_hard_regno_costs): Use
25702 int64_t instead of HOST_WIDE_INT.
25704 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
25706 PR rtl-optimization/70478
25707 * lra-constraints.c (process_alt_operands): Disfavor alternative
25708 insn memory operands.
25710 2017-04-07 Jeff Law <law@redhat.com>
25712 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
25713 CALL and NOTE_INSN_CALL_ARG_LOCATION.
25715 2017-04-07 Martin Liska <mliska@suse.cz>
25718 * config/aarch64/aarch64.c (aarch64_process_target_attr):
25719 Show error message instead of an ICE.
25721 2017-04-07 Martin Liska <mliska@suse.cz>
25724 * ipa-split.c (split_function): Add function part to a same comdat
25727 2017-04-07 Richard Biener <rguenther@suse.de>
25729 PR middle-end/80341
25730 * tree.c (get_unwidened): Also handle ! for_type case for
25732 * convert.c (do_narrow): Split out from ...
25733 (convert_to_integer_1): ... here. Do not pass final truncation
25734 type to get_unwidened for TRUNC_DIV_EXPR.
25736 2017-04-07 Richard Biener <rguenther@suse.de>
25738 * tree-affine.c (wide_int_ext_for_comb): Take type rather
25740 (aff_combination_const): Adjust.
25741 (aff_combination_scale): Likewise.
25742 (aff_combination_add_elt): Likewise.
25743 (aff_combination_add_cst): Likewise.
25744 (aff_combination_convert): Likewise.
25745 (add_elt_to_tree): Likewise. Remove unused argument.
25746 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
25748 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
25750 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
25752 * config/arm/arm.c (arm_default_short_enums): Use
25753 ARM_DEFAULT_SHORT_ENUMS.
25754 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
25756 2017-04-06 Jakub Jelinek <jakub@redhat.com>
25759 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
25760 members with redundant out-of-class redeclaration.
25762 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
25765 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
25766 * config/i386/i386.md (*zero_extendsidi2):
25767 Add (?*x,*x) and (?*v,*v) alternatives.
25769 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
25772 * config/i386/i386.c (ix86_expand_builtin)
25773 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
25774 mode from insn data. Convert operands to insn operand mode.
25775 Copy operands that don't satisfy insn predicate to a register.
25777 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
25779 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
25782 2017-04-06 Richard Biener <rguenther@suse.de>
25784 PR tree-optimization/80334
25785 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
25786 preserve alignment of accesses.
25788 2017-04-06 Richard Biener <rguenther@suse.de>
25790 PR tree-optimization/80262
25791 * tree-sra.c (build_ref_for_offset): Preserve address-space
25793 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
25794 Drop useless address-space information on MEM_REF offsets.
25796 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
25798 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
25800 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
25802 PR rtl-optimization/70703
25803 * ira-color.c (update_conflict_hard_regno_costs): Use
25804 HOST_WIDE_INT instead of long.
25806 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
25809 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
25810 not defined for x86_64 target. Add -mmmx target option when __SSE2__
25812 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
25813 for x86_64 target. Handle -m3dnowa option.
25815 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
25817 PR rtl-optimization/70703
25818 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
25819 (update_conflict_hard_regno_costs): Use long instead of unsigned
25820 arithmetic for cost calculation.
25822 2017-04-05 Jakub Jelinek <jakub@redhat.com>
25823 Bernd Edlinger <bernd.edlinger@hotmail.de>
25826 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
25829 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
25832 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
25833 ptr_mode with Pmode throughout.
25834 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
25835 into probe_stack_range and use DImode.
25837 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
25840 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
25841 call_eh_return is true.
25843 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25845 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
25846 Initialize last_match_fntype_index.
25848 2017-04-05 Jakub Jelinek <jakub@redhat.com>
25851 * tree-nvr.c: Include internal-fn.h.
25852 (pass_return_slot::execute): Ignore internal calls without
25855 2017-04-04 Jakub Jelinek <jakub@redhat.com>
25856 Richard Biener <rguenther@suse.de>
25859 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
25860 captures used multiple times, except for the last use.
25861 * generic-match-head.c: Include gimplify.h.
25863 2017-04-04 Jakub Jelinek <jakub@redhat.com>
25865 PR tree-optimization/79390
25866 * target.h (struct noce_if_info): Declare.
25867 * targhooks.h (default_noce_conversion_profitable_p): Declare.
25868 * target.def (noce_conversion_profitable_p): New target hook.
25869 * ifcvt.h (struct noce_if_info): New type, moved from ...
25870 * ifcvt.c (struct noce_if_info): ... here.
25871 (noce_conversion_profitable_p): Renamed to ...
25872 (default_noce_conversion_profitable_p): ... this. No longer
25874 (noce_try_store_flag_constants, noce_try_addcc,
25875 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
25876 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
25877 instead of noce_conversion_profitable_p.
25878 * config/i386/i386.c: Include ifcvt.h.
25879 (ix86_option_override_internal): Don't override
25880 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
25881 (ix86_noce_conversion_profitable_p): New function.
25882 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
25883 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
25884 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
25885 * doc/tm.texi: Regenerated.
25887 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25889 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
25892 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25895 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
25896 instructions for small multiply cores.
25898 2017-04-04 Jeff Law <law@redhat.com>
25900 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
25902 (mips_expand_vec_perm_const): Initialize elements in orig_perm
25903 that are not set by the loop over the elements.
25905 2017-04-04 Jakub Jelinek <jakub@redhat.com>
25908 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
25909 int mode, convert_modes it to mode as unsigned, otherwise use
25910 lowpart_subreg to mode rather than SImode.
25911 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
25912 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
25913 Use DImode instead of SImode for the shift count operand.
25914 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
25917 2017-04-04 Richard Biener <rguenther@suse.de>
25919 PR middle-end/80281
25920 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
25921 arithmetic done for the negate or the plus. Simplify.
25922 (A - (-B) -> A + B): Likewise.
25923 * fold-const.c (split_tree): Make sure to not negate pointers.
25925 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
25927 PR rtl-optimization/60818
25928 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
25929 a compare of comparisons with the thing compared if this results
25930 in a different machine mode.
25932 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
25934 * alias.c (base_alias_check): Fix typo in comment.
25935 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
25936 * cgraphunit.c (symbol_table::compile): Likewise.
25937 * collect2.c (maybe_run_lto_and_relink): Likewise.
25938 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
25939 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
25940 * config/avr/avr.c (avr_map_op_t): Likewise.
25941 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
25942 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
25943 * config/epiphany/epiphany.md (movcc): Likewise.
25944 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
25945 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
25947 * config/mips/mips.c (mips_save_restore_reg): Likewise.
25948 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
25949 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
25950 * config/sh/sh.c (sh_rtx_costs): Likewise.
25951 * fold-const.c (fold_truth_andor): Likewise.
25952 * genautomata.c (collapse_flag): Likewise.
25953 * gengtype.h (struct type::u::s): Likewise.
25954 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
25955 * input.c (FORMAT_AMOUNT): Likewise.
25956 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
25957 (known_aggs_to_agg_replacement_list): Likewise.
25958 * ipa-inline-analysis.c: Likewise.
25959 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
25960 * ipa-polymorphic-call.c
25961 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
25962 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
25963 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
25965 * modulo-sched.c (apply_reg_moves): Likewise.
25966 * omp-expand.c (build_omp_regions_1): Likewise.
25967 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
25968 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
25969 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
25970 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
25971 * value-prof.c: Likewise.
25972 * var-tracking.c (val_reset): Likewise.
25974 2017-04-03 Richard Biener <rguenther@suse.de>
25976 PR tree-optimization/80275
25977 * fold-const.c (split_address_to_core_and_offset): Handle
25980 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
25982 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
25983 descriptors is at least equal to that of functions.
25985 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
25987 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
25989 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
25992 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
25993 (mov<IMOD4:mode>): New expander.
25994 (*mov<IMOD4:mode>_internal): New insn and split pattern.
25996 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
25998 PR rtl-optimization/79405
25999 * fwprop.c (propagations_left): New variable.
26000 (forward_propagate_into): Decrement it.
26001 (fwprop_init): Initialize it.
26002 (fw_prop): If the variable has reached zero, stop propagating.
26003 (fwprop_addr): Ditto.
26005 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26008 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
26009 a FUNCTION_DECL, pass it as decl instead of origin to
26012 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
26014 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
26017 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
26020 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
26021 TARGET_VSX_SMALL_INTEGER.
26023 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26025 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
26026 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
26028 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26030 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
26031 extraction from odd-numbered MSA register.
26033 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26035 PR middle-end/80173
26036 * expmed.c (store_bit_field_1): Don't attempt to create
26037 a word subreg out of hard registers wider than word if they
26038 have HARD_REGNO_NREGS of 1 for their mode.
26040 PR middle-end/80163
26041 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
26042 conversions to integer types wider than word and pointer.
26045 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
26046 (rtx_equal_for_cselib_p): Pass 0 to it.
26047 * cselib.c (cselib_hasher::equal): Likewise.
26048 (rtx_equal_for_cselib_1): Add depth argument. If depth
26049 is 128, don't look up VALUE locs and punt. Increment
26050 depth in recursive calls when walking VALUE locs.
26052 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
26054 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
26055 (make_gcov_file_name): Use the canonical path name for generating
26057 (read_line): Fix handling of files with ascii null bytes.
26059 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
26061 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
26062 to initialise a vector register instead
26063 of using a const_int.
26065 2017-03-30 Jakub Jelinek <jakub@redhat.com>
26067 PR translation/80189
26068 * gimplify.c (omp_default_clause): Use %qs instead of %s in
26069 diagnostic messages.
26071 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
26074 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
26075 (dfp_diex_<mode>): Update mode of operand 1.
26076 * doc/extend.texi (dxex, dxexq): Document change to return type.
26077 (diex, diexq): Document change to argument type.
26079 2017-03-30 Martin Jambor <mjambor@suse.cz>
26082 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
26083 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
26084 it reflects the signature changes performed at the callee side.
26085 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
26086 to cgraph_build_function_type_skip_args.
26087 (build_function_decl_skip_args): Adjust call to the above function.
26089 2017-03-30 Jakub Jelinek <jakub@redhat.com>
26092 * config/i386/sse.md
26093 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
26094 register as dest whenever it is a MEM not rtx_equal_p to the
26095 corresponding dup operand, and when forcing into reg move the
26096 reg into the memory afterwards.
26097 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
26098 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
26099 for the force_reg mode.
26100 (avx512vl_vextractf128<mode>): Use register as dest either
26101 always when a MEM, or when it is a MEM not rtx_equal_p to the
26102 corresponding dup operand, or even not when it is a CONST_VECTOR
26103 depending on the mode and lo vs. hi.
26104 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
26106 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
26107 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
26108 Likewise. Require that operands[2] is even.
26109 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
26110 Remove extraneous parens. Require that operands[2] is a multiple
26112 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
26113 operands[0] is a MEM if <mask_applied>, the predicates/constraints
26114 disallow memory then.
26116 2017-03-30 Richard Biener <rguenther@suse.de>
26118 PR tree-optimization/77498
26119 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
26120 to non-constants over backedges.
26122 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
26124 PR rtl-optimization/80233
26125 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
26126 as last_combined_insn. Do not test for BARRIER_P separately.
26128 2017-03-29 Andreas Schwab <schwab@suse.de>
26131 * calls.c (prepare_call_address): Convert funexp to Pmode before
26132 copying to temp reg.
26134 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26136 PR tree-optimization/80158
26137 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
26138 Handle possible future case of more than one alternate
26140 (replace_rhs_if_not_dup): Likewise.
26141 (replace_one_candidate): Likewise.
26143 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
26145 PR rtl-optimization/80193
26146 * ira.c (ira): Do not check allocation for LRA.
26148 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
26150 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
26151 (nvptx_output_simt_exit): Declare.
26152 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
26153 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
26154 (init_softstack_frame): Move initialization of crtl->is_leaf to...
26155 (nvptx_declare_function_name): ...here. Emit declaration of local
26156 memory space buffer for omp_simt_enter insn.
26157 (nvptx_output_unisimt_switch): New.
26158 (nvptx_output_softstack_switch): New.
26159 (nvptx_output_simt_enter): New.
26160 (nvptx_output_simt_exit): New.
26161 * config/nvptx/nvptx.h (struct machine_function): New fields
26162 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
26163 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
26164 (UNSPECV_SIMT_EXIT): Ditto.
26165 (omp_simt_enter_insn): New insn.
26166 (omp_simt_enter): New expansion.
26167 (omp_simt_exit): New insn.
26168 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
26170 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
26171 (expand_GOMP_SIMT_ENTER_ALLOC): New.
26172 (expand_GOMP_SIMT_EXIT): New.
26173 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
26174 (GOMP_SIMT_ENTER_ALLOC): Ditto.
26175 (GOMP_SIMT_EXIT): Ditto.
26176 * target-insns.def (omp_simt_enter): New insn.
26177 (omp_simt_exit): Ditto.
26178 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
26180 (lower_rec_simd_input_clauses): Implement SIMT privatization.
26181 (lower_rec_input_clauses): Likewise.
26182 (lower_lastprivate_clauses): Handle SIMT privatization.
26184 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
26185 (ompdevlow_adjust_simt_enter): New.
26186 (find_simtpriv_var_op): New.
26187 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
26188 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
26190 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
26191 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
26192 (copy_decl_for_dup_finish): Ditto.
26194 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
26196 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
26199 * config/i386/i386.c (ix86_option_override_internal): Always
26200 allow -mpreferred-stack-boundary=3 for 64-bit targets.
26202 2017-03-28 Bin Cheng <bin.cheng@arm.com>
26204 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
26206 2017-03-28 Bin Cheng <bin.cheng@arm.com>
26208 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
26209 mark new edge's irreducible flag accordign to it.
26210 (vect_do_peeling): Check loop preheader edge's irreducible flag
26211 and pass it to function slpeel_add_loop_guard.
26213 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
26215 PR tree-optimization/80218
26216 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
26217 Update block frequencies and counts.
26219 2017-03-28 Richard Biener <rguenther@suse.de>
26221 PR tree-optimization/78644
26222 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
26223 of a simplification result we may not use it at all.
26225 2017-03-28 Richard Biener <rguenther@suse.de>
26228 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
26229 without arguments, generate default definition of a SSA name.
26231 2017-03-28 Richard Biener <rguenther@suse.de>
26233 PR middle-end/80222
26234 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
26235 TYPE_REF_CAN_ALIAS_ALL references.
26236 * fold-const.c (fold_indirect_ref_1): Likewise.
26238 2017-03-28 Martin Liska <mliska@suse.cz>
26241 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
26242 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
26244 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
26245 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
26247 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
26248 (EXTRA_SPECS): Define.
26249 (SUBTARGET_EXTRA_SPECS): Likewise.
26250 (SUBTARGET_CPP_SPEC): Likewise.
26251 * config/arc/elf.h (EXTRA_SPECS): Renamed to
26252 SUBTARGET_EXTRA_SPECS.
26253 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
26255 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
26257 * config/arc/simdext.md (vst64_insn): Update pattern.
26258 (vld32wh_insn): Likewise.
26259 (vld32wl_insn): Likewise.
26260 (vld64_insn): Likewise.
26261 (vld32_insn): Likewise.
26263 2017-03-28 Marek Polacek <polacek@redhat.com>
26266 * fold-const.c (fold_comparison): Use protected_set_expr_location
26267 instead of SET_EXPR_LOCATION.
26269 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
26271 * tree.c (add_expr): Avoid name lookup warning.
26273 2017-03-27 Jeff Law <law@redhat.com>
26275 PR tree-optimization/80216
26276 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
26277 function name. Limit recursion depth.
26278 (record_temporary_equivalences): Corresponding changes.
26280 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
26282 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
26285 2017-03-27 Jakub Jelinek <jakub@redhat.com>
26288 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
26290 * cfgcleanup.c (reg_note_cfa_p): New array.
26291 (insns_have_identical_cfa_notes): New function.
26292 (old_insns_match_p): Don't cross-jump in between /f
26293 and non-/f instructions. If both i1 and i2 are frame related,
26294 verify all CFA notes, their order and content.
26296 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
26299 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
26300 HImode and SImode with zero extend to DImode to one insn.
26301 (bswap<mode>2_extenddi): Likewise.
26302 (bswapsi2_extenddi): Likewise.
26303 (bswaphi2_extendsi): Likewise.
26304 (bswaphi2): Combine bswap HImode and SImode into one insn.
26305 Separate memory insns from swapping register.
26306 (bswapsi2): Likewise.
26307 (bswap<mode>2): Likewise.
26308 (bswaphi2_internal): Delete, no longer used.
26309 (bswapsi2_internal): Likewise.
26310 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
26311 store, and gpr<-gpr swap insns.
26312 (bswap<mode>2_store): Likewise.
26313 (bswaphi2_reg): Register only splitter, combine with the splitter.
26314 (bswaphi2 splitter): Likewise.
26315 (bswapsi2_reg): Likewise.
26316 (bswapsi2 splitter): Likewise.
26317 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
26318 the insns into load, store, and register/register insns.
26319 (bswapdi2_ldbrx): Likewise.
26320 (bswapdi2_load): Likewise.
26321 (bswapdi2_store): Likewise.
26322 (bswapdi2_reg): Likewise.
26324 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
26326 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
26327 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
26329 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
26332 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
26334 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
26335 special handling for target option conflicts between dform
26336 options (-mpower9-dform, -mpower9-dform-vector,
26337 -mpower9-dform-scalar) and -mno-direct-move.
26339 2017-03-27 Richard Biener <rguenther@suse.de>
26341 PR tree-optimization/80181
26342 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
26344 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26346 * config/arc/predicates.md (move_double_src_operand): Replace the
26347 call to move_double_src_operand with a call to address_operand.
26349 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26351 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
26352 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
26353 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
26355 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26357 * config/arc/predicates.md (long_immediate_loadstore_operand):
26358 Consider scaled addresses cases.
26360 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26362 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
26363 restored when in interrupt.
26364 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
26365 doesn't have delay slot.
26367 2017-03-27 Richard Biener <rguenther@suse.de>
26370 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
26371 inlined thunk clones.
26373 2017-03-27 Jakub Jelinek <jakub@redhat.com>
26376 * asan.c (instrument_derefs): Copy over last operand from
26377 original COMPONENT_REF to the new COMPONENT_REF with
26378 DECL_BIT_FIELD_REPRESENTATIVE.
26379 * ubsan.c (instrument_object_size): Likewise.
26381 2017-03-27 Richard Biener <rguenther@suse.de>
26383 PR tree-optimization/80170
26384 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
26385 sure DR/SCEV didnt fold in constants we do not see when looking
26386 at the reference base alignment.
26388 2017-03-27 Richard Biener <rguenther@suse.de>
26390 PR middle-end/80171
26391 * gimple-fold.c (fold_ctor_reference): Properly guard against
26392 NULL return value from canonicalize_constructor_val.
26394 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
26397 * config/i386/i386.c (ix86_expand_builtin)
26398 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
26399 flags reg setting and flags reg using instructions.
26400 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
26401 clobbering instructions to zero extend op2.
26403 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
26405 * doc/install.texi (Configuration) <--with-aix-soname>:
26406 Update link to AIX ld.
26408 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
26410 PR rtl-optimization/80160
26411 PR rtl-optimization/80159
26412 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
26413 reg_alternate_class into account.
26415 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
26418 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
26419 to consider in curr_insn_transform.
26421 2017-03-24 Jakub Jelinek <jakub@redhat.com>
26423 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
26424 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
26425 and emit_mode_inner.
26427 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26429 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
26430 argument to the overloaded builtin variants. Use the new flag to
26431 deprecate certain builtin variants.
26432 * config/s390/s390-builtin-types.def: Add new builtin types.
26433 * config/s390/s390-builtins.h: Support new flags field for
26434 overloaded builtins.
26435 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
26436 (s390_macro_to_expand): Enable vector float data type.
26437 (s390_cpu_cpp_builtins_internal): Indicate support of the new
26438 builtins by incrementing the __VEC__ version number.
26439 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
26441 (s390_resolve_overloaded_builtin): Emit error messages depending
26442 on the builtin flags.
26443 * config/s390/s390.c (s390_expand_builtin): Support additional
26444 flags argument. Change error message to match the messages
26445 emitted in s390-c.c.
26446 * config/s390/s390.md: New UNSPEC_* constants.
26447 (op_type): Add new instruction types.
26448 * config/s390/vecintrin.h: Add new builtins and test data class
26450 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
26451 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
26452 (VEC_INEXACT, VEC_NOINEXACT): New constants.
26453 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
26454 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
26455 ("vec_mergel<mode>"): V_HW -> VEC_HW.
26457 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
26458 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
26459 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
26460 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
26462 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
26463 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
26464 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
26465 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
26467 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
26468 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
26469 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
26470 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
26471 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
26472 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
26473 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
26475 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
26476 ("vec_scatter_element<V_HW_4:mode>_DI")
26477 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
26478 ("vec_fpint<mode>", "vflls")
26479 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
26480 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
26481 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
26482 ("*vec_cmphe<mode>_cc"): ... these.
26484 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
26485 mode constant instead of magic value.
26487 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26489 * config/s390/s390.c (s390_expand_vec_compare): Support other
26490 vector floating point modes than just V2DF.
26491 (s390_expand_vcond): Likewise.
26492 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
26493 (s390_cannot_change_mode_class): Prevent mode changes between TF
26494 and V1TF in vector registers.
26495 * config/s390/s390.md (DF, SF): New mode attributes.
26496 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
26497 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
26498 SFmode support for VRs.
26499 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
26501 (VFT, VF_HW): New mode iterators.
26502 (vw, sdx): New mode attributes.
26503 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
26504 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
26505 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
26506 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
26507 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
26508 also the new vector floating point modes. Renaming to ...
26510 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
26511 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
26512 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
26513 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
26514 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
26515 ("vec_unordered<mode>"): ... these.
26517 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
26518 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
26519 ("*vec_extendv2df"): New insn definitions.
26521 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26523 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
26524 ("mulditi3_2", "*muldi3_sign"): New patterns.
26525 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
26526 rename the pattern definition.
26528 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26530 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
26532 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
26534 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26536 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
26537 instruction if possible.
26538 * config/s390/vector.md (vec_halfnumelts): New mode
26540 ("*vec_vllezlf<mode>"): New pattern.
26542 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26544 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
26545 ("popcountv4si2", "popcountv2di2"): Rename to ...
26546 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
26547 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
26549 ("popcount<mode>2_vxe"): New pattern.
26551 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26553 * common/config/s390/s390-common.c (processor_flags_table): Add
26555 * config.gcc: Add arch12.
26556 * config/s390/driver-native.c (s390_host_detect_local_cpu):
26557 Default to arch12 for unknown CPU model numbers.
26558 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
26559 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
26560 PROCESSOR_max sanity check.
26561 * config/s390/s390-opts.h (enum processor_type): Add
26563 * config/s390/s390.c (processor_table): Add arch12.
26564 (s390_expand_builtin): Add check for B_VXE flag.
26565 (s390_issue_rate): Add PROCESSOR_ARCH12.
26566 (s390_get_sched_attrmask): Likewise.
26567 (s390_get_unit_mask): Likewise.
26568 (s390_sched_score): Enable z13 scheduling for arch12.
26569 (s390_sched_reorder): Likewise.
26570 (s390_sched_variable_issue): Likewise.
26571 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
26573 (s390_tune_attr): Use z13 scheduling also for arch12.
26574 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
26575 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
26576 (TARGET_VXE_P): New macros.
26577 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
26578 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
26579 * config/s390/s390.opt: Add arch12 as processor_type.
26581 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26583 * config/s390/s390.md
26584 ("fixuns_truncdddi2", "fixuns_trunctddi2")
26585 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
26586 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
26588 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
26589 Rename expanders to ...
26591 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
26592 ("fixuns_truncdddi2_emu"): ... these.
26594 ("fixuns_trunc<mode>si2_emu"): New expander.
26596 ("*fixuns_truncdfdi2_z13"): Rename to ...
26597 ("*fixuns_truncdfdi2_vx"): ... this.
26599 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26601 * config/s390/2964.md: Remove the single element vector compare
26602 instructions which are no longer used.
26603 * config/s390/s390.c (s390_select_ccmode): Remove handling of
26605 (s390_canonicalize_comparison): Remove handling of DFmode
26607 (s390_expand_vec_compare_scalar): Remove function.
26608 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
26609 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
26611 ("*cmp<mode>_ccs"): Add wfcdb instruction.
26613 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26615 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
26617 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
26618 will anyway by matched by mov<mode>_64dfp.
26620 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26622 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
26623 vlef/vstef. Add missing operand to vleif.
26625 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26627 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
26628 pair for all vector types with 64 bit elements.
26629 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
26630 * config/s390/vector.md (V_HW_64): ... here.
26631 (V_128_NOSINGLE): New mode iterator.
26632 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
26633 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
26634 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
26635 ("*vec_load_pairv2di"): Change to ...
26636 ("*vec_load_pair<mode>"): ... this one.
26638 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26640 * config/s390/constraints.md: Add comments.
26641 (jKK): Reject element sizes > 8 bytes.
26642 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
26644 * config/s390/s390.md: Add the s_operand checks formerly in
26645 s390_split_ok_p to various splitters where they are still
26647 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
26648 for 128 bit vectors. Plus two splitters.
26650 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26652 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
26655 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26658 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
26659 error if the boundary argument is not constant.
26661 2017-03-24 Jakub Jelinek <jakub@redhat.com>
26663 PR rtl-optimization/80112
26664 * loop-doloop.c (doloop_condition_get): Don't check condition
26665 if cmp isn't SET with IF_THEN_ELSE src.
26667 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26669 PR tree-optimization/80158
26670 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
26671 replacing a candidate statement, also replace it for the
26672 candidate's alternate interpretation.
26673 (replace_rhs_if_not_dup): Likewise.
26674 (replace_one_candidate): Likewise.
26676 2017-03-24 Richard Biener <rguenther@suse.de>
26678 PR tree-optimization/80167
26679 * graphite-isl-ast-to-gimple.c
26680 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
26682 (translate_isl_ast_to_gimple::get_rename): Likewise.
26684 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
26686 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
26687 handling of certain combinations of target options, including the
26688 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
26689 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
26691 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26694 * config/arm/arm.md (*load_multiple): Add reload_completed to
26695 matching condition.
26697 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26698 Richard Biener <rguenth@suse.de>
26700 PR tree-optimization/79908
26701 PR tree-optimization/80136
26702 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
26703 been cast away, gimplify_and_add suffices.
26705 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
26707 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
26709 2017-03-23 Richard Biener <rguenther@suse.de>
26711 PR tree-optimization/80032
26712 * gimplify.c (gimple_push_cleanup): Forced unconditional
26713 cleanups still have to go to the conditional_cleanups
26716 2017-03-22 Jakub Jelinek <jakub@redhat.com>
26718 PR tree-optimization/80072
26719 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
26721 (next_operand_entry_id): Change type to unsigned int.
26722 (sort_by_operand_rank): Make sure to return the right return value
26723 even if unsigned fields are bigger than INT_MAX.
26724 (struct oecount): Change cnt and id type to unsigned int.
26725 (oecount_hasher::equal): Formatting fix.
26726 (oecount_cmp): Make sure to return the right return value
26727 even if unsigned fields are bigger than INT_MAX.
26728 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
26731 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
26732 TREE_READONLY on result if writing it more than once.
26735 * doc/invoke.texi (-fsanitize=thread): Document that with
26736 -fnon-call-exceptions atomics are not able to throw
26740 * tsan.c: Include tree-eh.h.
26741 (instrument_builtin_call): Call maybe_clean_eh_stmt or
26742 maybe_clean_or_replace_eh_stmt where needed.
26743 (instrument_memory_accesses): Add cfg_changed argument.
26744 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
26745 if it returned true.
26746 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
26748 PR rtl-optimization/63191
26749 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
26750 wrapper function, moved the whole old content into ...
26751 (ix86_delegitimize_address_1): ... this. New inline function.
26752 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
26753 true as last argument instead of ix86_delegitimize_address.
26755 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
26757 * config/aarch64/aarch64.c (generic_branch_cost): Copy
26758 cortexa57_branch_cost.
26760 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
26762 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
26764 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26767 * doc/md.texi (Constraints): Document wA constraint.
26768 * config/rs6000/constraints.md (wA): New.
26769 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
26770 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
26771 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
26772 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
26774 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
26777 * gimplify.c (is_oacc_declared): New function.
26778 (oacc_default_clause): Use it to set default flags for acc declared
26779 variables inside parallel regions.
26780 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
26781 declared variables.
26782 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
26783 declare attribute to any decl as necessary.
26785 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
26788 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
26789 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
26790 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
26791 (arm_arch_lpae): This.
26792 * config/arm/arm.c (arm_arch7ve): Rename into ...
26793 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
26794 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
26797 2017-03-22 Martin Liska <mliska@suse.cz>
26800 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
26801 error message instead of an ICE.
26803 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26805 * doc/extend.texi (6.11 Additional Floating Types): Revise.
26807 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
26809 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
26811 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
26814 2017-03-21 Martin Sebor <msebor@redhat.com>
26816 * doc/extend.texi: Use "cannot" instead of "can't."
26817 * doc/hostconfig.texi: Same.
26818 * doc/install.texi: Same.
26819 * doc/invoke.texi: Same.
26820 * doc/loop.texi: Same.
26821 * doc/md.texi: Same.
26822 * doc/objc.texi: Same.
26823 * doc/rtl.texi: Same.
26824 * doc/tm.texi: Same.
26825 * doc/tm.texi.in: Same.
26826 * doc/trouble.texi: Same.
26828 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
26831 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
26832 (collect_checksum_attributes): Set it.
26833 (die_checksum_ordered): Use it.
26835 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26837 PR tree-optimization/79908
26838 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
26839 change: For a VA_ARG whose LHS has been cast away, use
26840 force_gimple_operand to construct the side effects.
26842 2017-03-21 David Malcolm <dmalcolm@redhat.com>
26844 PR translation/80001
26845 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
26846 more amenable to translation.
26847 (oacc_loop_auto_partitions): Likewise.
26849 2017-03-21 Marek Polacek <polacek@redhat.com>
26850 Martin Sebor <msebor@redhat.com>
26852 PR tree-optimization/80109
26853 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
26854 on INTEGRAL_TYPE_P.
26856 2017-03-21 Jakub Jelinek <jakub@redhat.com>
26857 Segher Boessenkool <segher@kernel.crashing.org>
26860 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
26861 check reg_used_between_p between insn and one of succ or succ2
26862 depending on if succ is artificial insn not inserted into insn
26865 2017-03-21 Martin Liska <mliska@suse.cz>
26867 PR gcov-profile/80081
26868 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
26869 * doc/gcc.texi: Include gcov-dump stuff.
26870 * doc/gcov-dump.texi: New file.
26872 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
26874 PR rtl-optimization/79150
26875 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
26876 conditional jump, if the jump is the last insn of the loop.
26878 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26879 Richard Biener <rguenth@suse.de>
26881 PR tree-optimization/79908
26882 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
26883 been cast away, use force_gimple_operand to construct the side
26886 2017-03-21 Martin Liska <mliska@suse.cz>
26888 PR libfortran/79956
26889 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
26892 2017-03-21 Brad Spengler <spender@grsecurity.net>
26895 * plugin.c (htab_hash_plugin): New function.
26896 (add_new_plugin): Use it and adjust.
26897 (parse_plugin_arg_opt): Adjust.
26898 (init_one_plugin): Likewise.
26900 2017-03-21 Richard Biener <rguenther@suse.de>
26902 PR tree-optimization/80032
26903 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
26904 if set force the cleanup to happen unconditionally.
26905 (gimplify_target_expr): Push inserted clobbers with force_uncond
26906 to avoid them being removed by control-dependent DCE.
26908 2017-03-21 Richard Biener <rguenther@suse.de>
26910 PR tree-optimization/80122
26911 * tree-inline.c (copy_bb): Do not expans va-arg packs or
26912 va_arg_pack_len when the inlined call stmt requires pack
26914 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
26916 2017-03-21 Jakub Jelinek <jakub@redhat.com>
26919 * tsan.c (instrument_builtin_call): If the memory model argument
26920 is not a constant, assume it is valid.
26923 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
26926 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
26928 PR rtl-optimization/79910
26929 * combine.c (can_combine_p): Do not allow combining an I0 or I1
26930 if its dest is used by an insn before I2 (other than the combined
26931 insns themselves, which are properly handled already).
26933 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
26936 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
26938 * combine.c (record_used_regs): New static function.
26939 (try_combine): Handle situations where there is an additional
26940 instruction between I2 and I3 which needs to have a LOG_LINK
26944 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
26946 * combine.c (try_combine): Delete redundant i1 test. Call
26947 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
26949 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26952 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
26953 alternatives 13/14.
26955 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26957 PR tree-optimization/80054
26958 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
26959 the optimization if a PHI or any of its arguments is not dominated
26960 by the candidate's basis. Use gphi* rather than gimple* as
26962 (replace_profitable_candidates): Clean up a gimple* variable that
26963 should be a gphi* variable.
26965 2017-03-20 Martin Sebor <msebor@redhat.com>
26968 * doc/extend.texi (attribute constructor): Document present limitation.
26970 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
26973 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
26974 __POWER9_VECTOR__ #ifdef control, change template definition to
26975 use Power9-specific built-in function.
26976 (vec_any_eq): Likewise.
26977 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
26978 to control outcomes from this test.
26979 (vector_ae_<mode>p): For VEC_F modes, likewise.
26981 2017-03-20 Ian Lance Taylor <iant@google.com>
26983 * config/i386/i386.c (ix86_function_regparm): Save an extra
26984 register for -fsplit-stack with DECL_STATIC_CHAIN.
26986 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
26989 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
26990 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
26992 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
26994 * config/riscv/riscv.c (riscv_print_operand): Use "fence
26996 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
26999 2017-03-20 Marek Polacek <polacek@redhat.com>
27002 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
27004 2017-03-20 Richard Biener <rguenther@suse.de>
27006 PR tree-optimization/80113
27007 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
27008 allocate extra SSA name for PHI def.
27009 (add_close_phis_to_outer_loops): Likewise.
27010 (add_close_phis_to_merge_points): Likewise.
27011 (copy_loop_close_phi_args): Likewise.
27012 (copy_cond_phi_nodes): Likewise.
27014 2017-03-20 Martin Liska <mliska@suse.cz>
27016 PR middle-end/79753
27017 * tree-chkp.c (chkp_build_returned_bound): Do not build
27018 returned bounds for a LHS that's not a BOUNDED_P type.
27020 2017-03-20 Martin Liska <mliska@suse.cz>
27024 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
27025 COMPLEX_CST and VECTOR_CST.
27027 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27030 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
27031 target operand. A new splitter adds the clobber statement in case
27032 the target operand is dead anyway.
27034 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
27036 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
27037 to age-old versions of binutils and glibc.
27039 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
27041 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
27043 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27045 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
27047 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27049 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
27050 requirement for binutils 2.13.
27052 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27054 * combine.c (try_combine): Delete redundant i1 test. Call
27055 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27057 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
27059 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
27060 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
27062 <riscv64-*-elf>: Re-arrange section
27063 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
27064 <riscv32-*-linux>: Likewise.
27065 <riscv64-*-elf>: Likewise
27066 <riscv64-*-linux>: Likewise.
27068 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
27071 * aarch64.opt(verbose-cost-dump): Fix typo.
27073 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
27076 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
27077 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
27079 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27081 * reload.c (find_reloads): When reloading a nonoffsettable address,
27082 use RELOAD_OTHER for it and its address reloads.
27084 PR rtl-optimization/79910
27085 * combine.c (record_used_regs): New static function.
27086 (try_combine): Handle situations where there is an additional
27087 instruction between I2 and I3 which needs to have a LOG_LINK
27090 2017-03-17 Jeff Law <law@redhat.com>
27092 PR tree-optimization/71437
27093 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
27094 conditional in the hash table first.
27095 (vrp_dom_walker::before_dom_children): Extract condition from
27096 ASSERT_EXPR. Record condition, its inverion and any implied
27097 conditions as well.
27099 2017-03-17 Marek Polacek <polacek@redhat.com>
27100 Markus Trippelsdorf <markus@trippelsdorf.de>
27102 PR tree-optimization/80079
27103 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
27106 2017-03-17 Richard Biener <rguenther@suse.de>
27108 PR middle-end/80075
27109 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
27110 Properly verify the LHS before the RHS possibly claims to be
27112 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
27115 2017-03-17 Martin Jambor <mjambor@suse.cz>
27117 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
27118 (List of -O2 options): Likewise.
27119 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
27122 2017-03-17 Tom de Vries <tom@codesourcery.com>
27124 * gcov-dump.c (print_usage): Print bug_report_url.
27126 2017-03-17 Richard Biener <rguenther@suse.de>
27128 PR middle-end/80050
27129 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
27130 (parser::peek): Likewise.
27132 2017-03-17 Richard Biener <rguenther@suse.de>
27134 PR tree-optimization/80048
27135 * sese.c (free_sese_info): Properly release rename_map and
27136 copied_bb_map elements.
27138 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
27140 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
27141 Add linked-list forward and backlinks. Insert on
27142 construction, remove on destruction.
27143 (class pass_store_merging): Add m_stores_head field.
27144 (pass_store_merging::terminate_and_process_all_chains):
27145 Iterate over m_stores_head list.
27146 (pass_store_merging::terminate_all_aliasing_chains):
27148 (pass_store_merging::execute): Check for debug stmts first.
27149 Push new chains onto the m_stores_head stack.
27151 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
27154 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
27155 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
27156 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
27158 2017-03-16 Jeff Law <law@redhat.com>
27160 PR tree-optimization/71437
27161 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
27162 member function. Implementation moved into after_dom_children
27163 member function and into the threader's thread_outgoing_edges
27165 (dom_opt_dom_walker::after_dom_children): Simplify by moving
27166 some code into new thread_outgoing_edges.
27167 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
27168 definition. Simplify marker handling (do it here). Assume we always
27169 have the available expression and the const/copies tables.
27170 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
27172 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
27173 * tree-vrp.c (equiv_stack): No longer file scoped.
27174 (vrp_dom_walker): New class.
27175 (vrp_dom_walker::before_dom_children): New member function.
27176 (vrp_dom_walker::after_dom_children): Likewise.
27177 (identify_jump_threads): Setup domwalker. Use it rather than
27178 walking edges in a random order by hand. Simplify setup/finalization.
27179 (finalize_jump_threads): Remove.
27180 (vrp_finalize): Do not call identify_jump_threads here.
27181 (execute_vrp): Do it here instead and call thread_through_all_blocks
27184 PR tree-optimization/71437
27185 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
27187 (simplify_stmt_for_jump_threading): Add basic_block argument. All
27189 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
27190 (dom_opt_dom_walker::thread_across_edge): Remove
27191 handle_dominating_asserts argument. All callers changed.
27192 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
27193 changes. Remove calls to lhs_of_dominating_assert. Other
27194 uses of handle_dominating_asserts turn into unconditional code
27195 (simplify_control_stmt_condition_1): Likewise.
27196 (simplify_control_stmt_condition): Likewise.
27197 (thread_through_normal_block, thread_across_edge): Likewise.
27198 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
27199 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
27200 object if it is not an SSA_NAME.
27201 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
27202 before calling into the VRP specific simplifiers.
27203 (identify_jump_threads): Remove handle_dominating_asserts
27206 2017-03-16 Jakub Jelinek <jakub@redhat.com>
27209 * tree-diagnostic.c (default_tree_printer): No longer static.
27210 * tree-diagnostic.h (default_tree_printer): New prototype.
27212 2017-03-16 Tamar Christina <tamar.christina@arm.com>
27214 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
27215 Change ins into fmov.
27217 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27219 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
27220 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
27221 Use h_con constraint for operand 1.
27222 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
27223 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
27225 2017-03-15 Jeff Law <law@redhat.com>
27227 PR tree-optimization/71437
27228 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
27229 (record_temporary_equivalences): Use it.
27231 PR tree-optimization/71437
27232 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
27233 tree-ssa-scopedtables.
27234 (lookup_avail_expr, build_and_record_new_cond): Likewise.
27235 (record_conditions, record_cond, vuse_eq): Likewise.
27236 (record_edge_info): Adjust to API tweak of record_conditions.
27237 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
27238 (record_temporary_equivalences, optimize_stmt): Likewise.
27239 (eliminate_redundant_computations): Likewise.
27240 (record_equivalences_from_stmt): Likewise.
27241 * tree-ssa-scopedtables.c: Include options.h and params.h.
27242 (vuse_eq): New function, moved from tree-ssa-dom.c
27243 (build_and_record_new_cond): Likewise.
27244 (record_conditions): Likewise. Accept vector of conditions rather
27245 than edge_equivalence structure for first argument.
27246 for the first argument.
27247 (avail_exprs_stack::lookup_avail_expr): New member function, moved
27248 from tree-ssa-dom.c.
27249 (avail_exprs_stack::record_cond): Likewise.
27250 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
27251 from tree-ssa-dom.c.
27252 (avail_exprs_stack): Add new member functions lookup_avail_expr
27254 (record_conditions): Declare.
27256 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
27259 * lra-constraints.c (process_alt_operands): Increase reject for
27260 reloading an input/output operand.
27262 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
27265 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
27266 insns to convert from signed/unsigned char/short to IEEE 128-bit
27268 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
27270 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
27273 * config/i386/i386.c (ix86_vector_duplicate_value): Create
27274 subreg of inner mode for values already in registers.
27276 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
27278 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
27279 iteration reg is used after the loop.
27281 2017-03-14 Martin Sebor <msebor@redhat.com>
27283 PR tree-optimization/79800
27284 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
27285 precision in negative-positive range.
27286 (format_floating): Call non-const overload with adjusted precision.
27288 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
27291 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
27294 2017-03-14 Martin Sebor <msebor@redhat.com>
27296 PR middle-end/80020
27297 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
27298 * builtins.def (aligned_alloc): Use it.
27301 * Makefile.in (GTFILES): Add calls.c.
27302 * calls.c: Include "gt-calls.h".
27304 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
27306 PR rtl-optimization/79728
27307 * regs.h (struct target_regs): New field
27308 x_contains_allocatable_regs_of_mode.
27309 (contains_allocatable_regs_of_mode): New macro.
27310 * reginfo.c (init_reg_sets_1): Initialize it, and change
27311 contains_reg_of_mode so it includes global regs as well.
27312 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
27313 rather than contains_regs_of_mode.
27315 2017-03-14 Martin Liska <mliska@suse.cz>
27317 * doc/invoke.texi: Document options that can't be combined with
27318 -fcheck-pointer-bounds.
27320 2017-03-14 Martin Liska <mliska@suse.cz>
27322 PR middle-end/79831
27323 * doc/invoke.texi (-Wchkp): Document the option.
27325 2017-03-14 Martin Liska <mliska@suse.cz>
27327 * Makefile.in: Install gcov-dump.
27329 2017-03-14 Martin Liska <mliska@suse.cz>
27331 * multiple_target.c (expand_target_clones): Bail out for
27332 an invalid attribute.
27334 2017-03-14 Richard Biener <rguenther@suse.de>
27336 * alias.c (struct alias_set_entry): Pack properly.
27337 * cfgloop.h (struct loop): Likewise.
27338 * cse.c (struct set): Likewise.
27339 * ipa-utils.c (struct searchc_env): Likewise.
27340 * loop-invariant.c (struct invariant): Likewise.
27341 * lra-remat.c (struct cand): Likewise.
27342 * recog.c (struct change_t): Likewise.
27343 * rtl.h (struct address_info): Likewise.
27344 * symbol-summary.h (function_summary): Likewise.
27345 * tree-loop-distribution.c (struct partition): Likewise.
27346 * tree-object-size.c (struct object_size_info): Likewise.
27347 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
27348 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
27349 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
27350 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
27351 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
27352 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
27353 (struct _stmt_vec_info): Likewise.
27355 2017-03-14 Martin Liska <mliska@suse.cz>
27358 * multiple_target.c (create_dispatcher_calls): Check that
27359 a target can create a function dispatcher.
27361 2017-03-14 Martin Liska <mliska@suse.cz>
27364 * multiple_target.c (expand_target_clones): Drop local.local
27365 flag for default implementation.
27367 2017-03-14 Richard Biener <rguenther@suse.de>
27369 PR tree-optimization/80030
27370 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
27372 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
27374 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
27375 gcc_fallthrough() instead of __attribute__((fallthrough));
27377 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
27379 * doc/gcc.texi: Remove "up" link to (DIR).
27380 * doc/gccint.texi: Ditto.
27382 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
27384 * doc/install.texi (Specific) <avr>: Remove reference to
27387 2017-03-13 Jeff Law <law@redhat.com>
27389 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
27390 attribute rather than comments.
27392 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
27393 match_scratch operand is highest.
27395 2017-03-13 Martin Liska <mliska@suse.cz>
27397 PR middle-end/78339
27398 * ipa-pure-const.c (warn_function_noreturn): If the declarations
27399 is a CHKP clone, use original declaration.
27401 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27403 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
27404 (arc_conditional_register_usage): Use a different allocation order
27405 when optimizing for size.
27406 * common/config/arc/arc-common.c (arc_option_optimization_table):
27407 Section anchors default on when optimizing for size.
27409 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27411 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
27413 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27415 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
27416 * config/arc/arc.md (cpu_facility): Add cd variant.
27417 (*movqi_insn): Add code density variant.
27418 (*movhi_insn): Likewise.
27419 (*movqi_insn): Likewise.
27420 (*addsi3_mixed): Likewise.
27421 (subsi3_insn): Likewise.
27423 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27425 * config/arc/arc.md (movsi_cond_exec): Update constraint.
27427 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27429 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
27430 expressions with MINUS and UNARY ops.
27432 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27435 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
27437 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
27438 between vec_select and vector argument.
27439 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
27440 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
27441 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
27442 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
27443 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
27444 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
27446 2017-03-13 Richard Biener <rguenther@suse.de>
27449 * params.def (vect-max-peeling-for-alignment): Fix typo.
27451 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
27453 * doc/install.texi (Specific) <mips-*-*>: Remove description of
27454 issue that only occurred with binutils below 2.18.
27456 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
27458 * doc/install.texi (Specific) <cris-axis-elf>: No longer
27459 refer to binutils 2.11/2.12 minimum.
27461 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
27463 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
27464 ftp.kernel.org and simplify binutils requirement.
27466 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
27468 * doc/invoke.texi (Warning Options): Fix spelling of link-time
27470 (Optimize Options): Ditto. Also remove redundancy.
27472 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27474 PR translation/79848
27475 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
27477 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
27478 to G_ to avoid double translation.
27480 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27482 PR translation/79923
27483 * auto-profile.c (get_combined_location): Convert leading
27484 character of diagnostics to lower case and remove trailing period.
27485 (read_profile): Likewise for various diagnostics.
27486 * config/arm/arm.c (arm_option_override): Remove trailing period
27487 from various diagnostics.
27488 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
27489 (msp430_expand_delay_cycles): Likewise.
27491 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27494 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
27495 full command-line argument, rather than just "str".
27496 (aarch64_validate_march): Likewise.
27497 (aarch64_validate_mtune): Likewise.
27499 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
27501 PR rtl-optimization/78911
27502 * lra-assigns.c (must_not_spill_p): New function.
27503 (spill_for): Use it.
27505 2017-03-10 Jakub Jelinek <jakub@redhat.com>
27507 PR tree-optimization/79981
27508 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
27509 ATOMIC_COMPARE_EXCHANGE ifn result.
27510 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
27511 IFN_ATOMIC_COMPARE_EXCHANGE.
27513 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27516 * opts.c (parse_sanitizer_options): Add missing question mark to
27517 "did you mean" message.
27519 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27521 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
27523 (VMULEUH_UNS): Likewise.
27524 (VMULOUB_UNS): Likewise.
27525 (VMULOUH_UNS): Likewise.
27526 * config/rs6000/rs6000.c (builtin_function_type): Remove
27527 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
27529 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27532 * read-rtl-function.c (function_reader::read_rtx_operand): Update
27533 x with result of extra_parsing_for_operand_code_0.
27534 (function_reader::extra_parsing_for_operand_code_0): Convert
27535 return type from void to rtx, returning x. When reading
27536 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
27537 larger size containing struct block_symbol.
27539 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
27541 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
27542 -mfloat128-hardware without -m64.
27544 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
27547 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
27548 entries to the case statement that marks unsigned arguments to
27549 overloaded functions.
27551 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
27553 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
27554 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
27556 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
27559 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
27560 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
27562 2017-03-10 Martin Liska <mliska@suse.cz>
27566 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
27567 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
27570 2017-03-10 Olivier Hainque <hainque@adacore.com>
27572 * tree-switch-conversion (array_value_type): Start by resetting
27573 candidate type to it's main variant.
27575 2017-03-10 Jakub Jelinek <jakub@redhat.com>
27577 PR rtl-optimization/79909
27578 * combine.c (try_combine): Use simplify_replace_rtx on individual
27579 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
27580 of the whole CALL_INSN_FUNCTION_USAGE.
27582 PR tree-optimization/79972
27583 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
27584 get_range_info on SSA_NAMEs. Formatting fixes.
27586 2017-03-10 Richard Biener <rguenther@suse.de>
27587 Jakub Jelinek <jakub@redhat.com>
27589 PR tree-optimization/77975
27590 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
27591 edge to be constant.
27592 (get_val_for): For constant x return it. Formatting fix.
27593 (loop_niter_by_eval): Avoid pointless looping if the next iteration
27594 would use the same bases as the current one.
27596 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27598 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
27599 instead of vec_select for V1TImode.
27600 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
27602 (VSX_LE_128): Add V1TI to this mode iterator.
27603 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
27604 (*vsx_le_perm_store_<mode>): Likewise.
27605 (pre-reload splitter for VSX stores): Likewise.
27606 (post-reload splitter for VSX stores): Likewise.
27607 (*vsx_xxpermdi2_le_<mode>): Likewise.
27608 (*vsx_lxvd2x2_le_<mode>): Likewise.
27609 (*vsx_stxvd2x2_le_<mode>): Likewise.
27611 2017-03-09 Michael Eager <eager@eagercon.com>
27613 Correct failures with --enable-checking=yes,rtl.
27615 * config/microblaze/microblaze.c (microblaze_expand_shift):
27616 Replace GET_CODE test with CONST_INT_P and INTVAL test with
27617 test for const0_rtx.
27618 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
27619 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
27621 2017-03-09 Richard Biener <rguenther@suse.de>
27623 PR tree-optimization/79977
27624 * graphite-scop-detection.c (scop_detection::merge_sese):
27625 Handle the case of extra exits to blocks dominating the entry.
27627 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
27629 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
27632 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
27634 PR rtl-optimization/79949
27635 * lra-constraints.c (process_alt_operands): Check memory when
27636 trying to predict a cycle. Print about the overall increase.
27638 2017-03-09 Richard Biener <rguenther@suse.de>
27640 PR middle-end/79971
27641 * gimple-expr.c (useless_type_conversion_p): Preserve
27642 TYPE_SATURATING for fixed-point types.
27644 2017-03-09 Richard Biener <rguenther@suse.de>
27647 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
27648 alignment of BLKmode params.
27650 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27653 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
27654 (VALL_NO_V2Q): Likewise.
27656 * config/aarch64/aarch64-simd.md
27657 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
27659 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
27660 VALL_NO_V2Q mode iterator.
27661 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
27663 2017-03-09 Martin Liska <mliska@suse.cz>
27665 PR tree-optimization/79631
27666 * tree-chkp-opt.c (chkp_is_constant_addr): Call
27667 tree_int_cst_sign_bit just for INTEGER constants.
27669 2017-03-09 Martin Liska <mliska@suse.cz>
27673 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
27676 2017-03-09 Marek Polacek <polacek@redhat.com>
27679 * tree.c (inchash::add_expr): Handle TREE_VEC.
27681 2017-03-09 Martin Liska <mliska@suse.cz>
27684 (chkp_narrow_size_and_offset): New function.
27685 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
27686 (void chkp_parse_bit_field_ref): New function.
27687 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
27688 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
27690 2017-03-09 Martin Liska <mliska@suse.cz>
27693 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
27694 (chkp_find_bounds_1): Remove gcc_unreachable.
27696 2017-03-09 Jakub Jelinek <jakub@redhat.com>
27699 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
27700 BUILT_IN_SYNC*, determine the access type from the size suffix and
27701 always build a MEM_REF with that type. Handle forgotten
27702 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
27705 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
27706 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
27707 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
27708 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
27709 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
27710 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
27711 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
27712 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
27713 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
27714 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
27715 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
27716 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
27717 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
27718 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
27719 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
27720 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
27721 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
27722 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
27723 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
27724 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
27725 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
27726 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
27727 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
27728 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
27729 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
27730 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
27731 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
27732 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
27733 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
27734 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
27735 definitions outside of __OPTIMIZE__ guarded section.
27738 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
27739 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
27740 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
27741 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
27744 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27746 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
27747 ("vfenez<mode>"): Add missing constraints.
27749 2017-03-08 Martin Sebor <msebor@redhat.com>
27752 * config/nds32/nds32.c (nds32_option_override):
27753 Fix misspelled diagnostic.
27755 2017-03-08 Jakub Jelinek <jakub@redhat.com>
27758 * gimplify.c (gimplify_omp_for): Replace index var in outer
27759 taskloop statement with an artificial variable and add
27760 OMP_CLAUSE_PRIVATE clause for it.
27762 2017-03-08 Richard Biener <rguenther@suse.de>
27764 PR tree-optimization/79955
27765 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
27766 for accesses that are completely outside of the variable.
27768 2017-03-08 Andrew Haley <aph@redhat.com>
27770 PR tree-optimization/79943
27771 * tree-ssa-loop-split.c (compute_new_first_bound): When
27772 calculating the new upper bound, (END-BEG) should be added, not
27775 2017-03-08 Jakub Jelinek <jakub@redhat.com>
27777 * config/avr/avr.md (setmemhi): Make sure match_dup
27778 operand number comes before match_scratch.
27780 2017-03-08 Richard Biener <rguenther@suse.de>
27782 PR tree-optimization/79920
27783 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
27784 with ncopies == 1 to ...
27785 (vect_transform_slp_perm_load): ... here. Properly compute
27786 all element loads by iterating VF times over the group. Do
27787 not handle ncopies (computed in a broken way) in
27788 vect_create_mask_and_perm.
27790 2017-03-08 Jakub Jelinek <jakub@redhat.com>
27793 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
27794 is a uniform vector, use uniform_vector_p return value instead of
27795 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
27797 2017-03-07 Marek Polacek <polacek@redhat.com>
27799 PR middle-end/79809
27800 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
27801 (alloca_call_type): Likewise.
27803 2017-03-07 Martin Liska <mliska@suse.cz>
27805 * gcov.c (process_args): Put comment to correct location.
27807 2017-03-07 Martin Liska <mliska@suse.cz>
27809 PR middle-end/68270
27810 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
27811 Use array_at_struct_end_p instead of DECL_CHAIN (field).
27812 (chkp_narrow_bounds_for_field): Likewise.
27813 (chkp_parse_array_and_component_ref): Pass one more argument to
27816 2017-03-07 Richard Biener <rguenther@suse.de>
27818 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
27821 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
27823 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
27824 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
27826 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27829 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
27830 to end of description.
27831 (PARAM_MAX_STORES_TO_MERGE): Likewise.
27833 2017-03-07 Jakub Jelinek <jakub@redhat.com>
27835 PR rtl-optimization/79901
27836 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
27838 (*avx512f_<code><mode>3<mask_name>): ... this.
27839 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
27840 iterator instead of VI8_AVX2_AVX512BW.
27842 PR rtl-optimization/79901
27843 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
27844 min/max expander, expand it using expand_vec_cond_expr.
27847 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
27850 2017-03-06 Jakub Jelinek <jakub@redhat.com>
27853 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
27854 to void * for PCH reasons.
27855 * dwarf2out.c (output_loc_operands, output_die): Cast
27856 v.val_vec.array to unsigned char *.
27858 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
27861 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
27864 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
27866 PR rtl-optimization/79571
27867 * lra-constraints.c (process_alt_operands): Calculate static
27868 reject and subtract it from overall when only addresses will be
27871 2017-03-06 Julia Koval <julia.koval@intel.com>
27874 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
27875 incoming stack boundary to 128 for 64-bit targets.
27877 2017-03-06 Richard Biener <rguenther@suse.de>
27879 PR tree-optimization/79894
27880 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
27881 to NULL after folding it.
27883 2017-03-06 Richard Biener <rguenther@suse.de>
27885 PR tree-optimization/79824
27886 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
27887 check disabling peeling for gaps.
27889 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
27891 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
27892 attributes): Document gettimeofday.
27894 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
27896 * config/s390/s390.c (s390_option_override_internal): Set
27897 PARAM_MIN_VECT_LOOP_BOUND
27899 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
27901 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
27902 * config/s390/s390.md: Likewise.
27904 2017-03-06 Jakub Jelinek <jakub@redhat.com>
27907 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
27908 (<avx2_avx512>_perm<mode>): Rename to ...
27909 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
27911 (<avx512>_perm<mode>_mask): Rename to ...
27912 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
27914 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
27915 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
27916 instead of VI8F_256_512.
27917 (avx512f_perm<mode>): New define_expand.
27918 (avx512f_perm<mode>_mask): Likewise.
27919 (avx512f_perm<mode>_1<mask_name>): New define_insn.
27920 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
27922 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
27924 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
27925 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
27927 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
27929 2017-03-06 Martin Liska <mliska@suse.cz>
27932 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
27933 when having a SSA NAME w/o VAR_DECL assigned to it.
27935 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
27937 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
27938 msa_dpsub_<su>_d): Fix MODE for vec_select.
27940 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
27942 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
27944 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
27946 2017-03-06 Richard Biener <rguenther@suse.de>
27948 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
27949 * plugin.c (register_plugin_info): Likewise.
27950 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
27952 2017-03-05 Jakub Jelinek <jakub@redhat.com>
27954 * config/i386/sse.md (sse_storehps, sse_storelps,
27955 avx_<castmode><avxsizesuffix>_<castmode>,
27956 avx512f_<castmode><avxsizesuffix>_<castmode>,
27957 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
27958 in condition that at least one operand is not a MEM.
27960 2017-03-03 Jakub Jelinek <jakub@redhat.com>
27962 PR middle-end/79805
27963 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
27964 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
27966 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
27967 gimple_call_nothrow_p flag based on whether original builtin can throw.
27968 If it can, emit following stmts on the fallthrough edge.
27969 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
27970 don't create new bb if inserting just debug stmts on the edge, try to
27971 insert them on the fallthru bb or just reset debug stmts.
27973 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
27976 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
27977 restore recog_data (including the operand rtxes inside it) around
27978 the call to get_insn_template.
27980 2017-03-03 Martin Sebor <msebor@redhat.com>
27982 PR tree-optimization/79699
27983 * context.c (context::~context): Free MPFR caches to avoid
27984 a memory leak on program exit.
27986 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27988 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
27989 Use wide_int::ulow () instead of .elt (0).
27991 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
27993 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
27994 (*pushxf): Limit oF constraint to 32bit targets and add oC
27995 constraint for 64bit targets.
27996 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
27997 (*pushdf): Change rmF constraint to rmC.
27999 2017-03-03 Martin Liska <mliska@suse.cz>
28001 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
28002 Remove unused variable.
28004 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28007 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
28008 is a memory operand, increase num_memory.
28009 (ix86_expand_args_builtin): Likewise.
28011 2017-03-03 Jan Hubicka <jh@suse.cz>
28014 * ipa-devirt.c (maybe_record_node): Properly handle
28015 __cxa_pure_virtual visibility.
28017 2017-03-03 Martin Liska <mliska@suse.cz>
28019 PR tree-optimization/79803
28020 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
28022 (pass_loop_prefetch::execute): Disabled optimization if an
28023 assumption about L1 cache size is not met.
28025 2017-03-03 Martin Liska <mliska@suse.cz>
28027 PR rtl-optimization/79574
28028 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
28029 (hash_scan_set): Likewise.
28030 (dump_hash_table): Likewise.
28031 (hoist_code): Likewise.
28033 2017-03-03 Richard Biener <rguenther@suse.de>
28035 * fixed-value.c (fixed_from_string): Restore use of elt (1)
28036 in place of uhigh ().
28037 (fixed_convert_from_real): Likewise.
28039 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28042 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
28044 2017-03-03 Richard Biener <rguenther@suse.de>
28046 PR middle-end/79818
28047 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
28048 TYPE_OVERFLOW_UNDEFINED check.
28050 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28052 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
28054 (vector_ae_<mode>_p): Likewise.
28055 (vector_nez_<mode>_p): Likewise.
28056 (vector_ne_v2di_p): Likewise.
28057 (vector_ae_v2di_p): Likewise.
28058 (vector_ne_<mode>_p): Likewise.
28059 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
28061 (vsx_tsqrt<mode>2_fe): Likewise.
28063 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
28066 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
28068 2017-03-02 Jakub Jelinek <jakub@redhat.com>
28070 PR rtl-optimization/79780
28071 * cprop.c (one_cprop_pass): When second and further conditional trap
28072 in a single basic block is turned into an unconditional trap, turn it
28073 into a deleted note to avoid RTL verification failures.
28075 2017-03-02 Richard Biener <rguenther@suse.de>
28077 * fold-const.c (const_binop): Use ulow () instead of elt (0).
28079 2017-03-02 Richard Biener <rguenther@suse.de>
28081 PR tree-optimization/79345
28083 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
28084 param and abort the walk, returning -1 if it is hit.
28085 (walk_aliased_vdefs): Take a limit param and pass it on.
28086 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
28087 defaulting to 0 and return a signed int.
28088 * tree-ssa-uninit.c (struct check_defs_data): New struct.
28089 (check_defs): New helper.
28090 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
28091 about uninitialized memory.
28092 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
28093 bogus uninitialized warning.
28094 (fixed_convert_from_real): Likewise.
28096 2017-03-02 Bin Cheng <bin.cheng@arm.com>
28098 PR tree-optimization/66768
28099 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
28100 iv_use if base object can't be determined.
28102 2017-03-02 Jakub Jelinek <jakub@redhat.com>
28104 PR tree-optimization/79345
28105 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
28106 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
28107 (get_pattern_stats): Initialize it.
28108 * genemit.c (gen_expand): Verify match_scratch numbers come after
28109 match_operand/match_dup numbers.
28110 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
28111 match_scratch numbers.
28112 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
28114 * config/s390/s390.md (trunctdsd2): Likewise.
28116 2017-03-02 Richard Biener <rguenther@suse.de>
28118 * wide-int.h (wide_int_storage::operator=): Implement in terms
28121 2017-03-02 Richard Biener <rguenther@suse.de>
28123 PR tree-optimization/79777
28124 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
28125 the to insert expression to sth existing.
28127 2017-03-01 Martin Sebor <msebor@redhat.com>
28129 PR middle-end/79692
28130 * gimple-ssa-sprintf.c
28131 (directive::known_width_and_precision): New function.
28132 (format_integer): Use it.
28133 (get_mpfr_format_length): Consider the full range of precision
28134 when computing %g output with the # flag. Set the likely byte
28135 count to 3 rather than 1 when precision is indeterminate.
28136 (format_floating): Correct the lower bound of precision.
28138 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28140 * doc/invoke.texi: Document default code model for 64-bit Linux.
28142 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28145 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
28146 udiv rather than div since input pattern is unsigned.
28148 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
28150 * config/i386/i386.c (print_reg): Warn for values of
28151 unsupported size in integer register.
28153 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
28156 * config/rs6000/predicates.md (current_file_function_operand): Do
28157 not allow self calls to be local if the function is replaceable.
28159 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
28162 * config/rs6000/altivec.h (vec_ctz and others): Change the
28163 preprocessor macro that controls conditional compilation from
28164 _ARCH_PWR9 to __POWER9_VECTOR__.
28165 (vec_all_ne): Change parameterization of __altivec_scalar_pred
28166 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
28167 control (instead of _ARCH_PWR9 control) so that template
28168 definition uses power9-specific function.
28169 (vec_any_eq): Likewise.
28170 (vec_all_ne): Change macro definition to use a power9-specific
28171 expansion under #ifdef __POWER9_VECTOR__ control (instead of
28172 _ARCH_PWR9 control).
28173 (vec_any_eq) Likewise.
28174 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
28175 expansion for CMPNEF to remove support for xvcmpnesp instruction.
28176 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
28177 support for xvcmpnedp instruction.
28178 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
28179 macro expansion so that Power9 implementation of vec_all_ne does
28180 not use the AltiVec predicate framework.
28181 (VCMPNEH_P): Likewise.
28182 (VCMPNEW_P): Likewise.
28183 (VCMPNED_P): Likewise.
28184 (VCMPNEFP_P): Likewise.
28185 (VCMPNEDP_P): Likewise.
28186 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
28187 implementation of vec_any_eq to not use AltiVec predicate
28189 (VCMPAEH_P): Likewise.
28190 (VCMPAEW_P): Likewise.
28191 (VCMPAED_P): Likewise.
28192 (VCMPAEFP_P): Likewise.
28193 (VCMPAEDP_P): Likewise.
28194 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
28195 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
28196 not use the AltiVec predicate framework.
28197 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
28198 of vec_any_eq to not use AltiVec predicate framework.
28199 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
28200 support for predefined __POWER9_VECTOR__ macro to indicate that
28201 Power9 instruction selection is enabled.
28202 (altivec_overloaded_builtins): Remove extraneous
28203 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
28204 function argument types RS6000_BTI_bool_V16QI and
28205 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
28206 entry for overloaded function argument types RS6000_BTI_bool_V4SI
28207 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
28208 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
28209 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
28210 Power9 for implementations of vec_cmpne. Change the signature for
28211 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
28212 (representing vec_all_ne) to remove the previously described first
28213 argument of type RS6000_BTI_INTSI, as this was an artifact of
28214 reliance on the AltiVec predicate framework, which is no longer
28215 used in the implementation of these functions. Add
28216 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
28217 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
28218 since, unlike the AltiVec predicate framework implementation, we
28219 do not share function descriptors between vec_alle and vec_anyeq.
28220 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
28221 set of modes that receive special treatment even when
28222 TARGET_P9_VECTOR is true. The special treatment emits code that
28223 does not depend on Power9 instructions.
28224 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
28225 define_expand to not rely on AltiVec predicate framework.
28226 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
28228 (vector_ne_v2di_p): Change this define_expand to not rely on
28229 AltiVec predicate framework.
28230 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
28232 (vector_ne_<mode>_p): Change this define_expand to not rely on
28233 AltiVec predicate framework.
28234 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
28236 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
28237 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
28238 define_insn pattern.
28239 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
28240 define_insn pattern because the xvcmpne<VSs>. instruction is not
28242 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
28243 instruction is not supported.
28245 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28247 * config/nvptx/nvptx.c: Include intl.h.
28249 2017-03-01 Martin Jambor <mjambor@suse.cz>
28252 * ipa-prop.h (ipa_bits): Removed field known.
28253 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
28254 to pointers. Adjusted their comments to warn about their sharing.
28255 (ipcp_transformation_summary): Change bits to a vector of pointers.
28256 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
28257 (ipa_get_ipa_bits_for_value): Declare.
28258 * tree-vrp.h (value_range): Mark as GTY((for_user)).
28259 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
28260 (ipa_bits_hash_table): Likewise.
28261 (ipa_vr_ggc_hash_traits): Likewise.
28262 (ipa_vr_hash_table): Likewise.
28263 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
28264 being pointers and vr_known being removed.
28265 (ipa_set_jf_unknown): Likewise.
28266 (ipa_get_ipa_bits_for_value): New function.
28267 (ipa_set_jfunc_bits): Likewise.
28268 (ipa_get_value_range): New overloaded functions.
28269 (ipa_set_jfunc_vr): Likewise.
28270 (ipa_compute_jump_functions_for_edge): Use the above functions to
28271 construct bits and vr parts of jump functions.
28272 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
28273 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
28275 (ipcp_grow_transformations_if_necessary): Also allocate
28276 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
28278 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
28279 them. Fix too long lines.
28280 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
28281 vr_known being removed.
28282 (ipa_read_jump_function): Use new setter functions to construct bits
28283 and vr parts of jump functions or set them to NULL.
28284 (write_ipcp_transformation_info): Adjust for bits being pointers.
28285 (read_ipcp_transformation_info): Likewise.
28286 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
28288 Include gt-ipa-prop.h.
28289 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
28291 (ipcp_store_bits_results): Likewise.
28292 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
28293 Do not write to existing jump functions but use a temporary instead.
28295 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28298 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
28299 attempt to use its first operand as BIT_FIELD_REF base.
28301 2017-03-01 Richard Biener <rguenther@suse.de>
28303 PR middle-end/79721
28304 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
28305 interpolating formula in wrapping arithmetic.
28306 (chrec_apply): Convert chrec_evaluate return value to wanted type.
28308 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28310 PR tree-optimization/79734
28311 * tree-vect-generic.c (expand_vector_condition): Optimize
28312 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
28313 Handle VEC_COND_EXPR where comparison has different inner width from
28314 type's inner width.
28316 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
28318 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
28319 markup, and similar issues. Remove @opindex entries for things
28320 that aren't options. Add missing -mmpy-option entries.
28322 2017-02-28 Jakub Jelinek <jakub@redhat.com>
28324 PR tree-optimization/79737
28325 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
28326 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
28327 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
28328 instead of byte_size. Formatting fix.
28329 (shift_bytes_in_array_right): Formatting fix.
28331 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
28334 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
28335 condition on optimize for the leaf function test.
28337 2017-02-28 Martin Liska <mliska@suse.cz>
28340 * read-rtl-function.c (function_reader::handle_unknown_directive):
28341 Bail out when one uses -flto.
28343 2017-02-28 Martin Liska <mliska@suse.cz>
28345 * common.opt: Replace space with tabular for options of <number>
28347 * config/i386/i386.opt: Show <number> value for
28348 -mlarge-data-threshold.
28349 * opts.c (print_filtered_help): Do not display number in hexadecimal
28352 2017-02-28 Martin Liska <mliska@suse.cz>
28354 * common.opt: Fix --help=option -Q for options which are of
28357 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
28359 * config/i386/i386.c (print_reg): Error out for values
28360 of 8-bit size in invalid integer register.
28362 2017-02-28 Martin Sebor <msebor@redhat.com>
28364 PR tree-optimization/79691
28365 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
28367 2017-02-28 Jakub Jelinek <jakub@redhat.com>
28370 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
28371 gcc_unreachable with output_operand_lossage.
28373 2017-02-28 Richard Biener <rguenther@suse.de>
28375 PR tree-optimization/79740
28376 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
28378 (visit_nary_op): Insert the nary into the hashtable if we
28379 pattern-matched sth.
28380 * tree-ssa-pre.c (eliminate_insert): Robustify.
28382 2017-02-28 Richard Biener <rguenther@suse.de>
28384 PR middle-end/79731
28385 * fold-const.c (decode_field_reference): Reject out-of-bound
28388 2017-02-28 Jakub Jelinek <jakub@redhat.com>
28390 * config/i386/i386.c: Include intl.h.
28391 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
28392 instead of just cond ? "..." : "...".
28393 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
28394 * coverage.c (read_counts_file): Likewise.
28395 * omp-offload.c: Include intl.h.
28396 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
28397 of just cond ? "..." : "...".
28398 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
28399 of just cond ? "..." : "...".
28401 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
28404 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
28406 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
28407 'tune for' CPU name.
28408 * config/arm/arm-cpu-data.h: Regenerated.
28410 2017-02-28 Richard Biener <rguenther@suse.de>
28412 PR tree-optimization/79732
28413 * tree-inline.c (expand_call_inline): Do not shadow var.
28415 2017-02-28 Richard Biener <rguenther@suse.de>
28417 PR tree-optimization/79723
28418 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
28419 address-space properly.
28421 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
28423 * doc/optinfo.texi (Optimization groups): Fix option used for
28425 * doc/invoke.texi (-fopt-info): Document "omp".
28426 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
28427 (OPTGROUP_ALL): Add OPTGROUP_OMP.
28428 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
28429 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
28430 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
28432 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
28434 * dumpfile.c (optgroup_options): Instead of "openmp", associate
28435 OPTGROUP_OMP with "omp".
28437 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
28440 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
28441 for arithmetic shift of unsigned V2DI.
28443 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
28445 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
28446 arc/linux.h headers.
28447 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
28448 (LINK_SPEC): Likewise.
28449 (ARC_TLS_EXTRA_START_SPEC): Likewise.
28450 (EXTRA_SPECS): Likewise.
28451 (STARTFILE_SPEC): Likewise.
28452 (ENDFILE_SPEC): Likewise.
28453 (LIB_SPEC): Likewise.
28454 (TARGET_SDATA_DEFAULT): Likewise.
28455 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
28456 (MULTILIB_DEFAULTS): Likewise.
28457 (DWARF2_UNWIND_INFO): Likewise.
28458 * config/arc/big.h: New file.
28459 * config/arc/elf.h: Likewise.
28460 * config/arc/linux.h: Likewise.
28461 * config/arc/t-uClibc: Remove.
28463 2017-02-27 Bin Cheng <bin.cheng@arm.com>
28465 PR tree-optimization/77536
28466 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
28467 (tree_transform_and_unroll_loop): Use above function to compute the
28468 estimated niter of unrolled loop and use it when scaling profile.
28469 Also use count info rather than frequency if it's non-zero.
28470 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
28471 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
28472 (vect_transform_loop): Call above function.
28474 2017-02-27 Richard Biener <rguenther@suse.de>
28476 PR tree-optimization/45397
28477 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
28478 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
28479 (visit_nary_op): Add pattern matching for CSEing sign-changed
28480 or truncated operations with wider ones.
28482 2017-02-27 Richard Biener <rguenther@suse.de>
28484 PR tree-optimization/79690
28485 * tree-vect-stmts.c (vectorizable_store): Use vector type
28486 built from the DR with address-space.
28488 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
28490 * doc/invoke.texi (Optimize Options): Refine the description
28491 of asan-use-after-return.
28493 2017-02-25 Alan Modra <amodra@gmail.com>
28495 PR rtl-optimization/79584
28496 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
28497 base, not ad->base_term, the reg within base. Remove assertion
28498 that ad->base == ad->base_term. Replace gen_int_mode using
28499 bogus mode with const0_rtx.
28501 2017-02-25 Jakub Jelinek <jakub@redhat.com>
28503 PR middle-end/79396
28504 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
28505 FMA_EXPR like tcc_binary or tcc_unary.
28507 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
28510 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
28512 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
28513 (output_loc_operands): Handle DW_OP_call_ref and
28514 DW_OP_GNU_variable_value.
28515 (struct variable_value_struct): New type.
28516 (struct variable_value_hasher): Likewise.
28517 (variable_value_hash): New variable.
28518 (string_types): Remove.
28519 (copy_loc_descr): New function.
28520 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
28521 (prepend_loc_descr_to_each): New function.
28522 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
28523 instead of add_loc_descr_to_each if the first argument is single
28524 location list and the second has multiple.
28525 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
28526 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
28527 when looking for variable value which doesn't have other location info.
28528 (loc_list_from_tree): Formatting fix.
28529 (gen_array_type_die): Simplify DW_AT_string_length handling.
28530 (adjust_string_types): Remove.
28531 (gen_subprogram_die): Don't call adjust_string_types nor test/set
28532 string_types. Call resolve_variable_values.
28533 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
28534 (resolve_addr_in_expr): Likewise. Add A argument.
28535 (copy_deref_exprloc): Remove deref argument. Adjust for the
28536 original expression being DW_OP_GNU_variable_value with optionally
28537 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
28538 optionally after it.
28539 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
28540 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
28541 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
28542 (variable_value_hasher::hash, variable_value_hasher::equal): New
28544 (resolve_variable_value_in_expr, resolve_variable_value,
28545 resolve_variable_values, note_variable_value_in_expr,
28546 note_variable_value): New functions.
28547 (dwarf2out_early_finish): Call note_variable_value on all toplevel
28550 2017-02-24 Jakub Jelinek <jakub@redhat.com>
28553 * opts.h (handle_generated_option): Add GENERATED_P argument.
28554 * opts-common.c (handle_option): Adjust function comment.
28555 (handle_generated_option): Add GENERATED_P argument, pass it to
28557 (control_warning_option): Pass false to handle_generated_option
28559 * opts.c (maybe_default_option): Pass true to handle_generated_option
28561 * optc-gen.awk: Likewise.
28563 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28565 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
28566 a REG, look at the REG it is a SUBREG of.
28567 (splitter for cmpeqsi_t): Ditto.
28569 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28571 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
28572 the special USEs with the pattern of the insn, not the insn itself.
28574 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
28577 * doc/invoke.texi: Document -mload-store-pairs.
28579 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28580 Sandra Loosemore <sandra@codesourcery.com>
28582 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
28583 argument isn't a CONST_INT.
28584 (nios2_alternate_compare_const): Assert op is a CONST_INT.
28585 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
28586 (nios2_validate_compare): Bypass alternate compare logic if *op2
28587 is not a CONST_INT.
28588 (ldstwm_operation_p): Return false if first_base is not a REG or
28589 if first_offset is not a CONST_INT.
28591 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28593 * config/cris/cris.md: Use correct operand in a define_peephole2.
28595 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28597 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
28599 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28601 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
28602 this_insn if it is an INSN or JUMP_INSN.
28603 (force_offsettable): Look at base, not at addr.
28604 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
28605 on things that aren't necessarily CONST_INTs.
28607 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
28609 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
28610 -mfpmath=sse is the default also for x86-32 targets with SSE2
28611 instruction set when @option{-ffast-math} is enabled
28613 2017-02-24 Jeff Law <law@redhat.com>
28615 PR rtl-optimizatoin/79286
28616 * ira.c (update_equiv_regs): Drop may_trap_p exception to
28619 2017-02-24 Richard Biener <rguenther@suse.de>
28621 PR tree-optimization/79389
28622 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
28625 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
28627 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
28628 function comment to reflect reality.
28629 (loop_exits_before_overflow): Fix typo in function description.
28631 2017-02-24 Richard Biener <rguenther@suse.de>
28633 PR tree-optimization/79389
28634 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
28635 properly that a threading opportunity exists. Detect conditional
28636 copy/constant propagation opportunities.
28638 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
28640 * config/visium/visium.md (type): Add trap.
28641 (b): New mode attribute.
28642 (*btst): Rename into...
28643 (*btst<mode>): ...this and adjust.
28644 (*cbranchsi4_btst_insn): Rename into...
28645 (*cbranch<mode>4_btst_insn): ...this and adjust.
28646 (trap): New define_insn.
28648 2017-02-23 Jakub Jelinek <jakub@redhat.com>
28650 PR tree-optimization/79389
28651 * ifcvt.c (struct noce_if_info): Add rev_cond field.
28652 (noce_reversed_cond_code): New function.
28653 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
28654 reversed_comparison_code. Formatting fix.
28655 (noce_try_store_flag): Test rev_cond != NULL in addition to
28656 reversed_comparison_code.
28657 (noce_try_store_flag_constants): Likewise.
28658 (noce_try_store_flag_mask): Likewise.
28659 (noce_try_addcc): Use rev_cond if non-NULL instead of
28660 reversed_comparison_code.
28661 (noce_try_cmove_arith): Likewise. Formatting fixes.
28662 (noce_try_minmax, noce_try_abs): Clear rev_cond.
28663 (noce_find_if_block): Initialize rev_cond.
28664 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
28665 instead of false as last argument never attempt to reverse it
28668 2017-02-23 Bin Cheng <bin.cheng@arm.com>
28670 PR tree-optimization/79663
28671 * tree-predcom.c (combine_chains): Process refs in reverse order
28672 only for ZERO length chains, and add explaining comment.
28674 2017-02-23 Jeff Law <law@redhat.com>
28676 PR tree-optimization/79578
28677 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
28678 in call to operand_equal_p.
28680 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
28683 * config/i386/cpuid.h: Fix another undefined behavior.
28685 2017-02-23 Richard Biener <rguenther@suse.de>
28687 PR tree-optimization/79683
28688 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
28689 vector types for data-refs.
28691 2017-02-23 Martin Liska <mliska@suse.cz>
28693 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
28695 2017-02-23 Jakub Jelinek <jakub@redhat.com>
28697 PR middle-end/79665
28698 * internal-fn.c (get_range_pos_neg): Moved to ...
28699 * tree.c (get_range_pos_neg): ... here. No longer static.
28700 * tree.h (get_range_pos_neg): New prototype.
28701 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
28702 are known to be in between 0 and signed maximum inclusive, try to
28703 expand both unsigned and signed divmod and use the cheaper one from
28706 2017-02-22 Jeff Law <law@redhat.com>
28708 PR tree-optimization/79578
28709 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
28710 to compare base operands.
28712 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
28715 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
28716 gpc_reg_operand instead of fpr_reg_operand.
28718 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
28720 * config/mips/mips.c (mips_return_in_memory): Force FP
28721 vector types to be returned in memory for o32 ABI.
28723 2017-02-22 Jakub Jelinek <jakub@redhat.com>
28725 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
28726 instead of DW_TAG_member for static data member declarations and don't
28727 set no_linkage_name for static inline data members.
28728 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
28731 2017-02-22 Martin Liska <mliska@suse.cz>
28733 * doc/invoke.texi: Replace inequality signs with square brackets
28736 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
28739 * lra-constraints.c (simplify_operand_subreg): Handle
28740 WORD_REGISTER_OPERATIONS targets.
28742 2017-02-22 Jakub Jelinek <jakub@redhat.com>
28745 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
28746 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
28747 elimination by swapping fld*.
28749 2017-02-22 Richard Biener <rguenther@suse.de>
28751 PR tree-optimization/79673
28752 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
28753 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
28754 irrelevant address-space qualifiers and avoiding a
28755 ADDR_SPACE_CONVERT_EXPR from fold_convert.
28757 2017-02-22 Richard Biener <rguenther@suse.de>
28759 PR tree-optimization/79666
28760 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
28761 to not symbolically negate if that may introduce undefined
28764 2017-02-22 Martin Liska <mliska@suse.cz>
28767 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
28768 * data-streamer-out.c (streamer_write_gcov_count_stream):
28770 * value-prof.c (stream_out_histogram_value): Make assert more
28771 precise based on type of counter.
28773 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
28776 * config/i386/i386.md (standard_x87sse_constant_load splitter):
28777 Use nonimmediate_operand instead of memory_operand for operand 1.
28778 (float-extend standard_x87sse_constant_load splitter): Ditto.
28780 2017-02-21 Jeff Law <law@redhat.com>
28782 PR tree-optimization/79621
28783 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
28784 blocks with edges to themselves.
28786 2017-02-21 Jakub Jelinek <jakub@redhat.com>
28789 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
28790 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
28791 Use gimple_call_builtin_p.
28794 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
28795 on temporarily removed DEBUG_INSNs.
28797 PR tree-optimization/79649
28798 * tree-loop-distribution.c (classify_partition): Give up on
28799 non-generic address space loads/stores.
28801 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
28803 * doc/loop.texi (Loop manipulation): Remove nonexistent
28804 tree_ssa_loop_version from the documentation.
28805 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
28807 2017-02-21 Jakub Jelinek <jakub@redhat.com>
28810 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
28811 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
28812 * config/rs6000/rs6000.c: Include except.h.
28813 (rs6000_expand_split_stack_prologue): Call
28814 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
28816 2017-02-21 Martin Jambor <mjambor@suse.cz>
28819 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
28820 have been analyzed.
28822 2017-02-21 Martin Jambor <mjambor@suse.cz>
28824 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
28825 for backward compatibility only.
28826 * doc/invoke.texi (Option Summary): Remove all references to
28827 -fipa-cp-alignment.
28829 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
28833 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
28835 * lra-constraints.c (curr_insn_transform): Handle
28836 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
28838 2017-02-21 Martin Liska <mliska@suse.cz>
28840 * config/i386/i386.opt: Replace -masm-dialect with -masm.
28842 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
28844 PR translation/79638
28845 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
28847 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
28850 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
28851 (arm_function_ok_for_sibcall): Return false for an indirect call by
28852 descriptor if all the argument registers are used.
28853 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
28854 alignment of the function.
28856 2017-02-21 Jakub Jelinek <jakub@redhat.com>
28858 PR tree-optimization/61441
28859 * simplify-rtx.c (simplify_const_unary_operation): For
28860 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
28861 the sNaN unmodified.
28863 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
28865 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
28866 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
28867 instead of SYSTEM_HEADER_DIR.
28869 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
28870 Martin Liška <mliska@suse.cz>
28872 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
28873 Fix typos and grammar, use active voice, and clarify.
28875 2017-02-20 Marek Polacek <polacek@redhat.com>
28877 PR middle-end/79537
28878 * gimplify.c (gimplify_expr): Handle unused *&&L;.
28881 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
28883 2017-02-20 Jakub Jelinek <jakub@redhat.com>
28886 * config/i386/i386.c (ix86_expand_builtin): Handle
28887 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
28888 ix86_builtins_isa[fcode].isa as a requirement of those
28889 flags and any other flag in the bitmask.
28890 (ix86_init_mmx_sse_builtins): Use 0 instead of
28891 ~OPTION_MASK_ISA_64BIT as mask.
28892 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
28893 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
28894 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
28895 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
28897 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
28900 * lra-constraints.c (split_reg): Check requested split mode
28901 is supported by the register.
28903 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
28905 * lra-constraints.c (simplify_operand_subreg): Remove early
28908 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
28911 * lra-constraints.c (curr_insn_transform): Tighten condition
28912 for converting SUBREG reloads from OP_OUT to OP_INOUT.
28914 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
28917 * lra-constraints.c (curr_insn_transform): Handle
28918 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
28920 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
28923 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
28925 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
28927 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
28930 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
28933 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28935 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
28938 2017-02-18 Jakub Jelinek <jakub@redhat.com>
28940 * final.c (last_columnnum, override_columnnum): New variables.
28941 (final_start_function): Set last_columnnum, pass it to begin_prologue
28942 hook and pass 0 to dwarf2out_begin_prologue.
28943 (final_scan_insn): Update override_columnnum. Pass last_columnnum
28944 to source_line debug hook.
28945 (notice_source_line): Compute last_columnnum and for debug_column_info
28946 return true on column changes.
28947 * debug.h (struct gcc_debug_hooks): Add column argument to
28948 source_line and begin_prologue hooks.
28949 (debug_nothing_int_charstar_int_bool): Remove prototype.
28950 (debug_nothing_int_int_charstar,
28951 debug_nothing_int_int_charstar_int_bool): New prototypes.
28952 (dwarf2out_begin_prologue): Add column argument.
28953 * debug.c (do_nothing_debug_hooks): Adjust source_line and
28954 begin_prologue hooks.
28955 (debug_nothing_int_charstar_int_bool): Remove.
28956 (debug_nothing_int_int_charstar,
28957 debug_nothing_int_int_charstar_int_bool): New functions.
28958 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
28959 through to dwarf2out_source_line.
28960 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
28961 (dwarf2out_source_line): Add column argument, emit it if requested.
28962 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
28964 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
28965 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
28966 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
28967 through to dwarf2out_begin_prologue.
28968 (vmsdbgout_source_line): Add column argument, pass it through to
28969 dwarf2out_source_line.
28970 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
28971 dbxout_source_line caller.
28972 (dbxout_source_line): Add column argument.
28974 * common.opt (gno-column-info, gcolumn-info): New options.
28975 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
28976 (check_die): Also test for multiple DW_AT_decl_column attributes.
28977 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
28978 DW_AT_decl_column if requested.
28979 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
28981 (gen_variable_die): Likewise.
28982 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
28983 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
28986 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
28987 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
28988 (ix86_handle_option): Handle OPT_m3dnowa.
28989 * doc/invoke.texi (-m3dnowa): Document.
28990 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
28991 -m3dnowa instead of -m3dnow -march=athlon.
28994 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
28995 instead of gcc_assert for K, r and R code checks. Formatting fixes.
28997 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29000 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
29001 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
29002 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
29003 generator for vsx_xxpermdi_<mode>_be.
29004 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
29005 force big-endian semantics.
29006 (vsx_xxpermdi_<mode>_be): New define_expand with same
29007 implementation as previous version of vsx_xxpermdi_<mode>.
29009 2017-02-17 Jakub Jelinek <jakub@redhat.com>
29011 PR tree-optimization/79327
29012 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
29013 variable, its initialization and use.
29015 2017-02-17 Julia Koval <julia.koval@intel.com>
29017 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
29018 (OPTION_MASK_ISA_PKU_UNSET): New.
29019 (ix86_handle_option): Handle -mrdpid.
29020 * config/i386/cpuid.h (bit_RDPID): New.
29021 * config/i386/driver-i386.c (host_detect_local_cpu):
29022 Detect RDPID feature.
29023 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
29024 * config/i386/i386-c.c (ix86_target_macros_internal):
29026 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
29027 (ix86_valid_target_attribute_inner_p): Add "rdpid".
29028 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
29029 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
29030 * config/i386/i386.md (define_insn "rdpid"): New.
29031 * config/i386/i386.opt Add -mrdpid.
29032 * config/i386/immintrin.h (_rdpid_u32): New.
29034 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
29036 PR rtl-optimization/79541
29037 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
29038 instead of transforming it into USE.
29040 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
29042 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
29043 If HONOR_SNANS (SFmode) force the input to a register.
29044 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
29045 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
29046 an frsp or similar insn.
29048 2017-02-17 Martin Liska <mliska@suse.cz>
29050 PR rtl-optimization/79577
29051 * params.def (selsched-max-sched-times): Increase minimum to 1.
29053 2017-02-17 Martin Liska <mliska@suse.cz>
29055 PR rtl-optimization/79574
29056 * gcse.c (want_to_gcse_p): Prevent integer overflow.
29058 2017-02-17 Martin Liska <mliska@suse.cz>
29060 PR tree-optimization/79529
29061 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
29062 ssa_defined_default_def_p to handle cases which are implicitly
29064 * tree-ssa.c (ssa_defined_default_def_p): New function.
29065 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
29066 which are implicitly defined.
29067 * tree-ssa.h (ssa_defined_default_def_p): Declare.
29069 2017-02-17 Richard Biener <rguenther@suse.de>
29071 PR middle-end/79576
29072 * params.def (max-ssa-name-query-depth): Limit to 10.
29074 2017-02-17 Richard Biener <rguenther@suse.de>
29076 PR tree-optimization/79552
29077 * tree-ssa-structalias.c (visit_loadstore): Properly verify
29080 2017-02-17 Richard Biener <rguenther@suse.de>
29083 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
29085 2017-02-17 Marek Polacek <polacek@redhat.com>
29087 PR middle-end/79536
29088 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
29089 (fold_negate_expr): New wrapper.
29091 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
29093 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
29094 Correct terminology and de-emphasize pre-standard behavior.
29096 2017-02-16 Alan Modra <amodra@gmail.com>
29098 PR rtl-optimization/79286
29099 * ira.c (def_dominates_uses): New function.
29100 (update_equiv_regs): Don't create an equivalence for insns that
29101 may trap where the register def does not dominate the use.
29103 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
29105 PR rtl-optimization/78127
29106 * lra.c (lra): Call lra_eliminate before finish the loop after
29109 2017-02-16 Richard Biener <rguenther@suse.de>
29111 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
29113 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
29114 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
29115 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
29116 (isl_val_int_from_wi): New function.
29117 (extract_affine_gmp): Rename to ...
29118 (extract_affine_wi): ... this, take a widest_int.
29119 (extract_affine_int): Just wrap extract_affine_wi.
29120 (add_param_constraints): Use isl_val_int_from_wi.
29121 (add_loop_constraints): Likewise, and extract_affine_wi.
29123 2017-02-15 Jeff Law <law@redhat.com>
29125 PR middle-end/79521
29126 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
29127 ira_init_register_move_cost_if_necessary.
29129 2017-02-15 Martin Sebor <msebor@redhat.com>
29131 PR middle-end/32003
29132 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
29133 removed in a prior commit.
29135 2017-02-15 Bin Cheng <bin.cheng@arm.com>
29137 PR tree-optimization/79347
29138 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
29139 counters during peeling.
29141 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
29143 * Makefile.in (site.exp): Remove "set ISLVER".
29145 2017-02-15 Jakub Jelinek <jakub@redhat.com>
29148 * real.c (real_from_integer): Call real_convert even for decimal.
29150 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
29153 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
29155 2017-02-14 Andrew Pinski <apinski@cavium.com>
29157 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
29158 cores and change the partno/implementer to be correct.
29159 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
29160 the 'B" as the implementer.
29161 * config/aarch64/aarch64-tune.md: Regenerate.
29163 2017-02-14 Carl Love <cel@us.ibm.com>
29165 * config/rs6000/rs6000.c: Add case statement entry to make the
29166 xvcvuxdsp built-in argument unsigned.
29167 * config/rs6000/vsx.md: Fix the source and return operand types so they
29168 match the instruction definitions from the ISA document. Fix typo
29169 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
29172 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
29175 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
29176 member early_clobber_alts.
29177 * lra-lives.c (reg_early_clobber_p): New.
29178 (process_bb_lives): Use it.
29179 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
29180 (debug_operand_data): Initialize early_clobber_alts.
29181 (setup_operand_alternative): Set up early_clobber_alts.
29182 (collect_non_operand_hard_regs): Ditto. Pass early clobber
29183 alternatives to new_insn_reg.
29184 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
29186 (lra_update_insn_regno_info): Pass the new arg.
29188 2017-02-14 Jakub Jelinek <jakub@redhat.com>
29190 PR middle-end/79505
29191 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
29192 (new_oacc_loop_raw): Don't clear already cleared fields.
29195 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
29196 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
29197 _mm512_prefetch_i64gather_ps): New inline functions and macros.
29199 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
29202 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
29204 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
29207 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
29208 the extra instruction to the right place to store 128-bit constant
29211 2017-02-14 Martin Sebor <msebor@redhat.com>
29213 PR middle-end/79448
29214 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
29215 warning for strings of unknown length.
29217 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
29219 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
29221 2017-02-14 Jeff Law <law@redhat.com>
29224 * ira-costs.c (scan_one_insn): Initialize register move costs
29225 for pseudos seen in USE/CLOBBER insns.
29227 PR tree-optimization/79095
29228 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
29229 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
29230 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
29231 if the operands are known to be not equal, then the resulting range
29233 (intersect_ranges): If the new range is ~[0,0] and the old range is
29234 wide, then prefer ~[0,0].
29235 * tree-vrp.c (overflow_comparison_p_1): New function.
29236 (overflow_comparison_p): New function.
29237 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
29238 if NAME is used in an overflow test.
29239 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
29240 overflow check that can be expressed as an equality test, then adjust
29241 ops to be that equality test.
29243 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29245 * config/s390/s390-builtin-types.def: Remove flags argument.
29246 * config/s390/s390.c (s390_init_builtins): Likewise.
29248 2017-02-14 Martin Liska <mliska@suse.cz>
29250 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
29251 vector. Fix trailing white spaces.
29253 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
29255 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
29258 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29260 PR rtl-optimization/68664
29261 * config/arm/arm.c (arm_sched_can_speculate_insn):
29262 New function. Declare prototype.
29263 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
29265 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29267 PR rtl-optimization/68664
29268 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
29270 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
29272 2017-02-14 Amit Pawar <amit.pawar@amd.com>
29274 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
29275 max skip bytes for function, loop and jump.
29277 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29279 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
29280 ABS_EXPR for gimple dump.
29282 2017-02-14 Jakub Jelinek <jakub@redhat.com>
29285 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
29287 PR tree-optimization/79408
29288 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
29289 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
29290 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
29291 also if rhs1 is INTEGER_CST.
29293 2017-02-14 Richard Biener <rguenther@suse.de>
29295 PR middle-end/79432
29296 * tree-into-ssa.c (insert_phi_nodes): When the function can
29297 have abnormal edges rewrite SSA names with broken use-def
29298 dominance out of SSA and register them for PHI insertion.
29300 2017-02-13 Martin Sebor <msebor@redhat.com>
29302 PR middle-end/79496
29303 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
29304 clearing info.nowrite flag when snprintf size argument is a range.
29306 2017-02-13 Jakub Jelinek <jakub@redhat.com>
29308 * cprop.c (cprop_jump): Add missing space in string literal.
29309 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
29310 (get_constraint_for_component_ref): Likewise.
29311 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
29312 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
29313 * lra-constraints.c (process_alt_operands): Likewise.
29314 * ipa-inline.c (inline_small_functions): Likewise.
29315 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
29316 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
29317 * trans-mem.c (diagnose_tm_1_op): Likewise.
29318 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
29319 (grid_parallel_clauses_gridifiable): Likewise.
29321 * config/nvptx/mkoffload.c (process): Add space in between
29324 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
29325 "MOD4_SSE_REGS" and "ALL_REGS".
29327 * spellcheck.c (test_data): Add , in between "foo" and "food".
29329 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29332 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
29333 boundary crossing check and subsequent code generation agree.
29335 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29337 * config/aarch64/aarch64.c (has_memory_op): Delete.
29338 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
29341 2017-02-13 Jakub Jelinek <jakub@redhat.com>
29343 PR rtl-optimization/79388
29344 PR rtl-optimization/79450
29345 * combine.c (distribute_notes): When removing TEM_INSN for which
29346 corresponding dest has last value recorded, invalidate that last
29349 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29351 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
29352 of explicit '@'. Add missing assembly comment marker on branch costs
29355 2017-02-13 Nathan Sidwell <nathan@acm.org>
29357 * gengtype-lex.l (<in_struct>): Add '/'.
29359 2017-02-13 Martin Liska <mliska@suse.cz>
29362 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
29364 2017-02-13 Richard Biener <rguenther@suse.de>
29366 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
29368 * configure: Re-generate.
29369 * config.in: Likewise.
29370 * graphite-dependences.c: Simplify as if
29371 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
29372 * graphite-isl-ast-to-gimple.c: Likewise.
29373 * graphite-optimize-isl.c: Likewise.
29374 * graphite-poly.c: Likewise.
29375 * graphite-sese-to-poly.c: Likewise.
29376 * graphite.h: Likewise.
29377 * toplev.c: Include isl/version.h and use isl_version () for
29378 printing the ISL version.
29379 * doc/install.texi: Update ISL requirement.
29381 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
29383 * doc/standards.texi (Standards): Update reference to
29386 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
29388 * doc/extend.texi (Named Address Spaces): sourceware.org now
29390 * doc/install.texi (Binaries): Ditto.
29393 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
29395 * doc/cpp.texi: Replace "stringify"/"stringification" with C
29396 standard terminology "stringize"/"stringizing" throughout.
29397 * doc/cppinternals.texi: Likewise.
29399 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
29401 * doc/extend.texi: Fix some spelling mistakes and typos.
29402 * doc/invoke.texi: Likewise.
29404 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
29407 * params.def (inline-min-speedup) Change from 10 to 8.
29409 2017-02-11 Jakub Jelinek <jakub@redhat.com>
29411 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
29414 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
29417 * ipa-inline-analysis.c (get_minimal_bb): New function.
29418 (record_modified): Use it.
29419 (remap_edge_change_prob): Handle also ancestor functions.
29421 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
29423 * doc/contrib.texi (Contributors): Remove broken link into
29424 the Mauve CVS repository.
29426 2017-02-11 Jakub Jelinek <jakub@redhat.com>
29428 PR middle-end/79454
29429 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
29430 result computation whenever lhs doesn't have vector mode, not
29431 just when it has BLKmode.
29433 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
29435 * doc/makefile.texi (profiledbootstrap): Refer to the
29436 installation instructions only in textual form.
29438 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29441 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
29443 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
29445 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
29446 (Specific): Update mingw-w64 reference.
29448 (Specific): Remove broken link to Renesas RX processor.
29450 2017-02-10 Richard Biener <rguenther@suse.de>
29452 * toplev.c (process_options): Do not mention obsolete graphite
29453 options when printing sorry message about missing graphite support.
29454 Mention -floop-nest-optimize.
29456 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
29458 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
29459 (vtst_p16): Likewise.
29460 (vtstq_p8): Likewise.
29461 (vtstq_p16): Likewise.
29463 (vtstq_p64): Likewise.
29464 * config/arm/arm_neon.h (vgetq_lane_p64): New.
29465 (vset_lane_p64): New.
29466 (vsetq_lane_p64): New.
29468 2017-02-10 Jakub Jelinek <jakub@redhat.com>
29470 PR tree-optimization/79411
29471 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
29472 stmt operands are SSA_NAMEs used in abnormal phis.
29473 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
29476 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
29479 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
29482 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
29484 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
29486 2017-02-09 Jakub Jelinek <jakub@redhat.com>
29488 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
29492 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
29493 not arbitrary TREE_CONSTANT.
29496 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
29497 "omp declare target link" attribute unless is_global_var.
29498 * omp-offload.c (find_link_var_op): Likewise.
29500 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
29501 Chung-Lin Tang <cltang@codesourcery.com>
29503 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
29505 (gimplify_adjust_omp_clauses): Don't delete TILE.
29506 (gimplify_omp_for): Deal with TILE.
29507 * internal-fn.c (expand_GOACC_TILE): New function.
29508 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
29510 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
29511 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
29513 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
29514 avoid DIV for outermost collapse var.
29515 (expand_oacc_for): Insert tile element loop as needed. Adjust.
29516 Remove out of date comments, fix whitespace.
29517 * omp-general.c (omp_extract_for_data): Deal with tiling.
29518 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
29519 adjust OLF_DIM_BASE value.
29520 (struct omp_for_data): Add tiling field.
29521 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
29522 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
29523 for auto loops. Remove default auto determining, moved to
29524 oacc_loop_fixed_partitions.
29525 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
29526 stmts, add e_mask field.
29527 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
29528 (oacc_thread_numbers): Use oacc_dim_call.
29529 (oacc_xform_tile): New.
29530 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
29531 (finish_oacc_loop): Adjust for ifns vector.
29532 (oacc_loop_discover_walk): Append loop abstraction sites to list,
29533 add case for GOACC_TILE fns.
29534 (oacc_loop_xform_loop): Delete.
29535 (oacc_loop_process): Iterate over call list directly, and add
29536 handling for GOACC_TILE fns.
29537 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
29539 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
29540 vector partitioning to outer loops. Assign 2 partitions to loops
29541 when available. Add TILE handling.
29542 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
29543 (execite_oacc_device_lower): Process GOACC_TILE fns,
29544 ignore unknown specs.
29545 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
29546 * tree.c (omp_clause_num_ops): Adjust TILE ops.
29547 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
29549 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
29551 * configure.ac (ACX_BUGURL): Update.
29552 * configure: Regenerate.
29554 2017-02-09 Richard Biener <rguenther@suse.de>
29556 PR tree-optimization/69823
29557 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
29558 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
29560 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
29562 * config/arc/arc-c.def: Add __NPS400__ definition.
29563 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
29564 (TARGET_NPS400): Define.
29566 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
29568 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
29570 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
29571 pointer, arch_info.
29572 (arc_cpu_types): Fill the arch_info field with a pointer into the
29573 arc_arch_types table.
29574 (arc_selected_cpu): Declare.
29575 * config/arc/arc.c (arc_selected_cpu): Make global.
29576 (arc_selected_arch): Delete.
29577 (arc_base_cpu): Delete.
29578 (arc_override_options): Remove references to deleted variables,
29579 update access to arch information.
29580 (ARC_OPT): Update access to arch information.
29581 (ARC_OPTX): Likewise.
29582 * config/arc/arc.h (arc_base_cpu): Remove declaration.
29583 (TARGET_ARC600): Update access to arch information.
29584 (TARGET_ARC601): Likewise.
29585 (TARGET_ARC700): Likewise.
29586 (TARGET_EM): Likewise.
29587 (TARGET_HS): Likewise.
29588 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
29591 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
29594 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
29595 condition/operands for integer GE/LE/GEU/LEU operations.
29597 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
29599 PR translation/79397
29600 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
29603 2017-02-08 Martin Jambor <mjambor@suse.cz>
29606 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
29607 whether allocation happened.
29608 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
29609 nothing was allocated.
29611 2017-02-08 Jakub Jelinek <jakub@redhat.com>
29613 PR tree-optimization/79408
29614 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
29615 constant, but SSA_NAME with a known integer range, use the minimum
29616 of that range instead of op1 to determine if modulo can be replaced
29617 with its first operand.
29619 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29621 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
29623 2017-02-08 Richard Biener <rguenther@suse.de>
29625 PR tree-optimization/71824
29626 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
29627 Check all loops contained in the merged region.
29629 2017-02-07 Andrew Pinski <apinski@cavium.com>
29631 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
29633 2017-02-07 Andrew Pinski <apinski@cavium.com>
29635 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
29636 (thunderxt88): Likewise.
29637 (thunderxt81): Disable LSE and change v8.1 to v8.
29638 (thunderxt83): Likewise.
29640 2017-02-07 Jakub Jelinek <jakub@redhat.com>
29641 Richard Biener <rguenther@suse.de>
29643 PR middle-end/79399
29644 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
29645 type from int to size_t.
29646 * ira-costs.c (struct_costs_size): Change type from int to size_t.
29648 2017-02-07 Jakub Jelinek <jakub@redhat.com>
29650 PR rtl-optimization/79386
29651 * cprop.c (bypass_conditional_jumps): Initialize
29652 bypass_last_basic_block already before splitting bbs after
29653 unconditional traps...
29654 (bypass_conditional_jumps): ... rather than here.
29657 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
29658 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
29659 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
29660 fix -masm=intel patterns.
29662 2017-02-07 Richard Biener <rguenther@suse.de>
29664 PR tree-optimization/79256
29665 PR middle-end/79278
29666 * builtins.c (get_object_alignment_2): Use min_align_of_type
29667 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
29668 and ADJUST_FIELD_ALIGN.
29670 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
29672 * doc/tm.texi: Regenerate.
29673 * stor-layout.c (layout_decl): Adjust.
29674 (update_alignment_for_field): Likewise.
29675 (place_field): Likewise.
29676 (min_align_of_type): Likewise.
29677 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
29678 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
29679 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
29680 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
29681 * config/frv/frv.c (frv_adjust_field_align): Likewise.
29682 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
29683 * config/i386/i386.c (x86_field_alignment): Likewise.
29684 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
29685 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
29686 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
29687 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
29688 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
29689 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
29693 2017-01-30 Richard Biener <rguenther@suse.de>
29695 PR tree-optimization/79256
29696 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
29697 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
29700 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
29702 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
29703 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
29704 builtins to SImode and emit a zero-extend, if necessary.
29706 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
29708 * docs/invoke.texi (RISC-V Options): Alphabetize.
29710 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
29712 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
29715 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
29717 * config/riscv/riscv.c: New file.
29718 * common/config/riscv/riscv-common.c: Likewise.
29719 * config.gcc: Likewise.
29720 * config/riscv/constraints.md: Likewise.
29721 * config/riscv/elf.h: Likewise.
29722 * config/riscv/generic.md: Likewise.
29723 * config/riscv/linux.h: Likewise.
29724 * config/riscv/multilib-generator: Likewise.
29725 * config/riscv/peephole.md: Likewise.
29726 * config/riscv/pic.md: Likewise.
29727 * config/riscv/predicates.md: Likewise.
29728 * config/riscv/riscv-builtins.c: Likewise.
29729 * config/riscv/riscv-c.c: Likewise.
29730 * config/riscv/riscv-ftypes.def: Likewise.
29731 * config/riscv/riscv-modes.def: Likewise.
29732 * config/riscv/riscv-opts.h: Likewise.
29733 * config/riscv/riscv-protos.h: Likewise.
29734 * config/riscv/riscv.h: Likewise.
29735 * config/riscv/riscv.md: Likewise.
29736 * config/riscv/riscv.opt: Likewise.
29737 * config/riscv/sync.md: Likewise.
29738 * config/riscv/t-elf-multilib: Likewise.
29739 * config/riscv/t-linux: Likewise.
29740 * config/riscv/t-linux-multilib: Likewise.
29741 * config/riscv/t-riscv: Likewise.
29742 * configure.ac: Likewise.
29743 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
29744 Waterman as RISC-V maintainers.
29745 * doc/install.texi: Add RISC-V entries.
29746 * doc/invoke.texi: Add RISC-V options section.
29747 * doc/md.texi: Add RISC-V constraints section.
29748 * configure: Regenerated.
29750 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
29753 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
29754 false values to be constant vectors with all 0 or all 1 bits set.
29755 (vcondu<mode><mode>): Likewise.
29756 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
29758 (fpmask_comparison_operator): Update comment.
29759 (vecint_comparison_operator): New predicate.
29760 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
29761 vector conditionals when the true and false values are constant
29762 vectors with all 0 bits or all 1 bits set.
29764 2017-02-06 Martin Sebor <msebor@redhat.com>
29766 PR tree-optimization/79376
29767 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
29769 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
29771 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
29772 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
29773 to simplify split condition.
29775 2017-02-06 Jakub Jelinek <jakub@redhat.com>
29777 * omp-expand.c (oxpand_omp_atomic_fetch_op,
29778 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
29781 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
29783 PR rtl-optimization/68664
29784 * target.def (can_speculate_insn): New hook.
29785 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
29786 * doc/tm.texi: Regenerate.
29787 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
29788 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
29789 (rs6000_sched_can_speculate_insn): New function.
29791 2017-02-06 Jakub Jelinek <jakub@redhat.com>
29793 PR tree-optimization/79284
29794 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
29795 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
29796 vectorizable_mask_load_store, vectorizable_operation,
29797 vect_is_simple_cond, get_same_sized_vectype): Use it instead
29798 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
29799 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
29800 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
29801 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
29802 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
29803 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
29804 is_gimple_assign (stmt). Replace another such test with
29805 is_gimple_assign (stmt).
29807 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
29810 * config/avr/avr.c (rtl-iter.h): Include it.
29811 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
29812 (avr_legitimate_combined_insn): ...and implementation.
29814 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
29816 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
29817 * config/s390/s390.c (s390_const_operand_ok)
29818 (s390_canonicalize_comparison, s390_extract_part)
29819 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
29820 (s390_contiguous_bitmask_p, s390_rtx_costs)
29821 (legitimize_pic_address): Likewise.
29822 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
29823 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
29824 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
29825 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
29826 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
29828 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
29830 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
29831 REGNO($0) == REGNO($1).
29833 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29835 * config/s390/linux.h(SIZE_TYPE): Add comment.
29837 2017-02-06 Julian Brown <julian@codesourcery.com>
29838 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29839 Virendra Pathak <virendra.pathak@broadcom.com>
29841 * config/aarch64/aarch64-cores.def: Change the scheduler
29843 * config/aarch64/aarch64.md: Include thunderx2t99.md.
29844 * config/aarch64/thunderx2t99.md: New file.
29846 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
29848 * doc/standards.texi (Go Language): Update link to language
29851 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
29853 * tree-eh.c (lower_resx): Sanitize profile.
29854 (cleanup_empty_eh_move_lp): Likewise.
29856 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
29859 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
29861 * cfgloopmanip.h (loop_version): Update prototype.
29862 * modulo-sched.c (sms_schedule): Update call of loop_version.
29863 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
29864 * tree-parloops.c (gen_parallel_loop): Likewise.
29865 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
29866 * tree-ssa-loop-split.c (split_loop): Likewise.
29867 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
29868 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
29870 2017-02-05 Martin Liska <mliska@suse.cz>
29873 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
29875 (print_operand_address): Initialize a struct to zero.
29877 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
29879 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
29880 garbage collector only in textual form.
29882 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
29884 * doc/extend.texi (x86 specific memory model extensions for
29885 transactional memory): Simplify a phrase.
29887 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
29890 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
29891 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
29892 (atomic_storedi_1): Likewise.
29894 2017-02-04 Jakub Jelinek <jakub@redhat.com>
29896 PR tree-optimization/79338
29897 * tree-parloops.c (gather_scalar_reductions): Don't call
29898 vect_analyze_loop_form for loop->inner before destroying loop's
29901 2017-02-03 Martin Sebor <msebor@redhat.com>
29903 PR tree-optimization/79327
29904 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
29905 when precision has resulted in leading zeros.
29906 (format_integer): Adjust the likely counter to assume an unknown
29907 argument that may be zero is non-zero.
29909 2017-02-03 Jason Merrill <jason@redhat.com>
29912 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
29913 avoid copying non-taken branch.
29915 2017-02-03 Jakub Jelinek <jakub@redhat.com>
29917 PR tree-optimization/79340
29918 * tree-vect-loop.c (vectorizable_reduction): Release
29919 vec_defs elements after safe_splicing them into other vectors.
29922 PR tree-optimization/79327
29923 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
29924 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
29926 (format_integer): Use wide_int_to_tree instead of build_int_cst
29927 + to_?hwi. If argmin is NULL, just set argmin and argmax to
29928 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
29929 of shortest and longest sequence.
29931 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
29933 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
29934 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
29936 2017-02-03 Walter Lee <walt@tilera.com>
29939 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
29940 after initial stackframe link reg save.
29941 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
29943 2017-02-03 Jakub Jelinek <jakub@redhat.com>
29946 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
29947 wu for stxssp alternative.
29949 2017-02-03 Martin Sebor <msebor@redhat.com>
29951 PR tree-optimization/79352
29952 * gimple-fold.c (get_range_strlen): Add argument.
29953 (get_range_strlen): Change return type to bool.
29954 (get_maxval_strlen): Pass in a dummy argument.
29955 * gimple-fold.h (get_range_strlen): Change return type to bool.
29956 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
29957 * tree.h (array_at_struct_end_p): Add argument.
29958 * tree.c (array_at_struct_end_p): Handle it.
29960 2017-02-03 Martin Liska <mliska@suse.cz>
29963 * multiple_target.c (create_dispatcher_calls): Redirect edge
29964 from a caller of a dispatcher.
29965 (expand_target_clones): Make the clones local.
29966 (ipa_target_clone): Do both target clones and resolvers.
29967 (ipa_dispatcher_calls): Remove the pass.
29968 (pass_dispatcher_calls::gate): Likewise.
29969 (make_pass_dispatcher_calls): Likewise.
29970 * passes.def (pass_target_clone): Put as very first IPA early
29973 2017-02-03 Martin Liska <mliska@suse.cz>
29975 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
29976 in case of a function with ifunc attribute.
29978 2017-02-03 Martin Liska <mliska@suse.cz>
29980 * cgraph.c (cgraph_node::dump): Dump function version info.
29981 * symtab.c (symtab_node::dump_base): Add missing new line.
29983 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
29985 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
29986 (ifcombine_ifandif): Use it.
29988 2017-02-03 Martin Liska <mliska@suse.cz>
29990 * doc/invoke.texi: Document default value for
29991 use-after-scope-direct-emission-threshold.
29993 2017-02-03 Martin Liska <mliska@suse.cz>
29995 PR tree-optimization/79339
29996 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
29997 (format_floating): Likewise.
29999 2017-02-03 Martin Liska <mliska@suse.cz>
30002 * ipa-prop.c (ipa_node_params_t::insert): Remove current
30004 (ipa_node_params_t::remove): Likewise.
30005 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
30006 initialization from removed ipa_node_params_t::insert.
30007 (ipa_node_params::~ipa_node_params): Move from removed
30008 ipa_node_params_t::release.
30009 * symbol-summary.h (symbol_summary::m_released): New member.
30010 Do not release a summary twice. Do not allow to call finalizer
30011 for types of a summary that live in GGC memory.
30013 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30015 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
30018 2017-02-02 Martin Sebor <msebor@redhat.com>
30020 PR middle-end/79275
30021 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
30022 (format_string): Tighten up the range of output for non-constant
30023 strings and correct the expected range for wide non-constant strings.
30025 2017-02-02 Martin Sebor <msebor@redhat.com>
30027 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
30029 PR middle-end/32003
30030 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
30032 (-fdump-tree-@var): Add to index and document how to come up
30033 with pass-specific option and dump file names.
30034 (-fdump-passes): Clarify where to look for output.
30036 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30038 PR middle-end/77445
30039 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
30040 statistics of the analyzed path; allow threading for speed when
30041 any of BBs along the path are optimized for speed.
30043 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
30045 PR middle-end/78468
30046 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
30047 settings of the virtual registers.
30050 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
30052 * explow.c (get_dynamic_stack_size): Take known alignment of stack
30053 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
30056 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30058 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
30059 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
30061 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30063 * config/s390/s390.md: Add missing comments with the expanded
30065 * config/s390/vector.md: Likewise.
30066 * config/s390/vx-builtins.md: Likewise.
30068 2017-02-02 Jakub Jelinek <jakub@redhat.com>
30071 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
30072 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
30073 conditions on a single line.
30075 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30077 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
30078 __S390_VX__ to __VX__.
30080 2017-02-01 Andrew Pinski <apinski@cavium.com>
30082 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
30083 stmt_info to record_stmt_cost.
30084 (vect_get_known_peeling_cost): Pass stmt_info if known to
30086 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
30087 cpu_vector_cost field into
30088 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
30089 field into vec_int_stmt_cost and vec_fp_stmt_cost.
30090 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
30091 splitting of scalar_stmt_cost and vec_stmt_cost.
30092 (thunderx_vector_cost): Likewise.
30093 (cortexa57_vector_cost): LIkewise.
30094 (exynosm1_vector_cost): Likewise.
30095 (xgene1_vector_cost): Likewise.
30096 (thunderx2t99_vector_cost): Improve after the splitting of the two
30098 (aarch64_builtin_vectorization_cost): Update for the splitting of
30099 scalar_stmt_cost and vec_stmt_cost.
30101 2017-02-01 Torvald Riegel <triegel@redhat.com>
30102 Richard Henderson <rth@redhat.com>
30104 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
30105 conditional on existance of a fast atomic load.
30106 * optabs-query.c (can_atomic_load_p): New function.
30107 * optabs-query.h (can_atomic_load_p): Declare it.
30108 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
30109 no fast atomic load is available for the particular size of access.
30110 (expand_atomic_compare_and_swap): Likewise.
30111 (expand_atomic_load): Likewise.
30112 (expand_atomic_store): Likewise.
30113 (expand_atomic_fetch_op): Likewise.
30114 * testsuite/lib/target-supports.exp
30115 (check_effective_target_sync_int_128): Remove x86 because it provides
30116 no fast atomic load.
30117 (check_effective_target_sync_int_128_runtime): Likewise.
30119 2017-02-01 Richard Biener <rguenther@suse.de>
30121 * graphite.c: Include tree-vectorizer.h for find_loop_location.
30122 (graphite_transform_loops): Provide opt-info for optimized nests.
30123 * tree-parloop.c (parallelize_loops): Provide opt-info for
30124 parallelized loops.
30126 2017-02-01 Richard Biener <rguenther@suse.de>
30128 PR middle-end/79315
30129 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
30130 was not set before.
30132 2017-02-01 Richard Biener <rguenther@suse.de>
30134 PR tree-optimization/71824
30135 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
30136 Verify the loops are valid in the merged SESE region.
30137 (scop_detection::can_represent_loop_1): Check analyzing the
30138 evolution of the number of iterations in the region succeeds.
30140 2017-01-31 Ian Lance Taylor <iant@golang.org>
30142 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
30143 REG_ARGS_SIZE note to 32-bit push insns and call insn.
30145 2017-01-31 David Malcolm <dmalcolm@redhat.com>
30147 PR preprocessor/79210
30148 * input.c (get_substring_ranges_for_loc): Replace line_width
30149 assertion with error-handling.
30151 2017-01-31 Richard Biener <rguenther@suse.de>
30153 PR tree-optimization/77318
30154 * graphite-sese-to-poly.c (extract_affine): Fix assert.
30155 (create_pw_aff_from_tree): Take loop parameter.
30156 (add_condition_to_pbb): Pass loop of the condition to
30157 create_pw_aff_from_tree.
30159 2017-01-31 Jakub Jelinek <jakub@redhat.com>
30161 * config/s390/s390.c (s390_asan_shadow_offset): New function.
30162 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
30164 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
30168 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
30170 (convert_int_to_float128): Likewise.
30171 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
30172 (convert_int_to_float128): Likewise.
30173 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
30174 (UNSPEC_IEEE128_CONVERT): Likewise.
30175 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
30176 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
30177 Use local variables for IBM extended format.
30178 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
30179 (fix_trunc<mode>si2_fprs): Likewise.
30180 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
30181 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
30182 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
30183 to know that we can now have integers of all sizes in vector
30185 (fix<uns>_<mode>di2_hw): Likewise.
30186 (float<uns>_<mode>si2_hw): Likewise.
30187 (fix_<mode>si2_hw): Likewise.
30188 (fixuns_<mode>si2_hw): Likewise.
30189 (float<uns>_<mode>di2_hw): Likewise.
30190 (float_<mode>di2_hw): Likewise.
30191 (float_<mode>si2_hw): Likewise.
30192 (floatuns_<mode>di2_hw): Likewise.
30193 (floatuns_<mode>si2_hw): Likewise.
30194 (xscvqp<su>wz_<mode>): Delete, no longer used.
30195 (xscvqp<su>dz_<mode>): Likewise.
30196 (xscv<su>dqp_<mode>): Likewise.
30197 (ieee128_mfvsrd_64bit): Likewise.
30198 (ieee128_mfvsrd_32bit): Likewise.
30199 (ieee128_mfvsrwz): Likewise.
30200 (ieee128_mtvsrw): Likewise.
30201 (ieee128_mtvsrd_64bit): Likewise.
30202 (ieee128_mtvsrd_32bit): Likewise.
30204 2017-01-31 Martin Liska <mliska@suse.cz>
30207 * ipa-prop.c (ipa_free_all_node_params): Call release method
30208 instead of ~sumbol_summary to not to trigger double times
30211 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
30213 PR tree-optimization/71691
30214 * bitmap.h (class auto_bitmap): New.
30215 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
30216 is_maybe_undefined instead of ssa_undefined_value_p.
30218 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30220 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
30221 __S390_ARCH_LEVEL__ to __ARCH__.
30223 2017-01-31 Jakub Jelinek <jakub@redhat.com>
30225 PR tree-optimization/79267
30226 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
30227 if should_remove_lhs_p is true.
30229 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
30232 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
30233 (add_alignment_attribute): New.
30234 (base_type_die): Add alignment attribute.
30235 (subrange_type_die): Likewise.
30236 (modified_type_die): Likewise.
30237 (gen_array_type_die): Likewise.
30238 (gen_descr_array_type_die: Likewise.
30239 (gen_enumeration_type_die): Likewise.
30240 (gen_subprogram_die): Likewise.
30241 (gen_variable_die): Likewise.
30242 (gen_field_die): Likewise.
30243 (gen_ptr_to_mbr_type_die): Likewise.
30244 (gen_struct_or_union_type_die): Likewise.
30245 (gen_subroutine_type_die): Likewise.
30246 (gen_typedef_die): Likewise.
30247 (base_type_cmp): Compare alignment attribute.
30249 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30252 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
30253 (setb_unsigned) New pattern for setb with CCUNS.
30254 * config/rs6000/rs6000.c (expand_block_compare): Use a different
30255 subfc./subfe sequence to avoid overflow problems. Generate a
30256 shorter sequence with cmpld/setb for power9.
30257 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
30258 for generating subfc. instruction.
30259 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
30260 now uses this instruction.
30262 2017-01-30 Ian Lance Taylor <iant@google.com>
30265 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
30266 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
30268 2017-01-30 Martin Sebor <msebor@redhat.com>
30270 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
30271 Move constant to the right of a relational operator.
30272 (get_mpfr_format_length, format_character, format_string): Ditto.
30273 (should_warn_p, maybe_warn): Same.
30275 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
30277 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
30280 * asan.c (get_translation_unit_decl): Remove function.
30281 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
30283 2017-01-30 Martin Liska <mliska@suse.cz>
30285 PR gcov-profile/79259
30286 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
30287 -fprofile-generate.
30289 2017-01-30 Martin Liska <mliska@suse.cz>
30292 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
30293 Initialize variables with NULL value.
30295 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
30298 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
30300 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
30302 2017-01-30 Richard Biener <rguenther@suse.de>
30304 PR tree-optimization/79276
30305 * tree-vrp.c (process_assert_insertions): Properly adjust common
30306 when removing a duplicate.
30308 2017-01-30 Richard Biener <rguenther@suse.de>
30310 PR tree-optimization/79256
30311 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
30312 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
30314 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
30316 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
30319 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
30320 ("*r<noxa>sbg_<mode>_sll_bitmask")
30321 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
30322 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
30323 Use contiguous_bitmask_nowrap_operand.
30325 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30328 * config/rs6000/altivec.h (vec_xl): Revise #define.
30329 (vec_xst): Likewise.
30331 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
30333 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
30335 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
30337 PR rtl-optimization/79194
30338 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
30339 traps before call to bypass_conditional_jumps.
30341 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
30343 PR tree-optimization/71374
30344 * lra-constraints.c (check_conflict_input_operands): New.
30345 (match_reload): Use it.
30347 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
30350 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
30351 account to calculate conflict_set.
30353 2017-01-27 Bin Cheng <bin.cheng@arm.com>
30355 PR rtl-optimization/78559
30356 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
30357 other_insn in combine.
30359 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
30361 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
30362 uint16_type_node for BT_UINT16.
30364 2017-01-27 David Malcolm <dmalcolm@redhat.com>
30366 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
30367 "RTL Tests" to menu.
30368 (GIMPLE Tests): New node.
30369 (RTL Tests): New node.
30371 2017-01-27 Richard Biener <rguenther@suse.de>
30373 PR tree-optimization/79245
30374 * tree-loop-distribution.c (distribute_loop): Apply cost
30375 modeling also to detected patterns.
30377 2017-01-27 Richard Biener <rguenther@suse.de>
30379 PR tree-optimization/71433
30380 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
30381 (compare_assert_loc): New function.
30382 (process_assert_insertions): Sort and optimize assert locations
30383 to remove duplicates and push down identical assertions on
30384 edges to their destination block.
30386 2017-01-27 Richard Biener <rguenther@suse.de>
30388 PR tree-optimization/79244
30389 * tree-vrp.c (remove_range_assertions): Forcefully propagate
30390 out SSA names even if abnormal.
30392 2017-01-27 Jakub Jelinek <jakub@redhat.com>
30394 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
30395 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
30396 instead of MPFR_RNDN.
30398 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
30401 * arm.c (arm_option_override): Don't call build_target_option_node
30402 until after doing all option overrides.
30403 (arm_valid_target_attribute_tree): Likewise.
30405 2017-01-27 Martin Liska <mliska@suse.cz>
30407 * doc/invoke.texi (-fprofile-arcs): Document profiling support
30408 for {cd}tors and C++ {cd}tors.
30410 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
30412 * config/s390/s390.md ("*setmem_long_and")
30413 ("*setmem_long_and_31z"): Use zero_extend instead of and.
30415 2017-01-26 Martin Sebor <msebor@redhat.com>
30417 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
30420 2017-01-26 Martin Sebor <msebor@redhat.com>
30422 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
30423 HAVE_DFmode before using XFmode or DFmode.
30424 (parse_directive): Avoid using the z length modifier to avoid
30425 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
30427 PR middle-end/78703
30428 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
30429 to accept adjustment as an array.
30430 (get_int_range): New function.
30431 (struct directive): Make width and prec arrays.
30432 (directive::set_width, directive::set_precision): Call get_int_range.
30433 (format_integer, format_floating): Handle width and precision ranges.
30434 (format_string, parse_directive): Same.
30436 2017-01-26 Jakub Jelinek <jakub@redhat.com>
30439 * dwarf2out.c (generate_skeleton_bottom_up): For children with
30440 comdat_type_p set, just clone them, but keep the children in the
30444 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
30445 which have direct callers with -fvar-tracking-assignments enabled
30447 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
30448 inside of type units.
30450 2017-01-26 Martin Sebor <msebor@redhat.com>
30452 PR middle-end/78703
30453 * gimple-ssa-sprintf.c (struct result_range): Add likely and
30455 (struct format_result): Replace number_chars, number_chars_min,
30456 and number_chars_max with a single member of struct result_range.
30458 (format_result::operator+=): Adjust.
30459 (struct fmtresult): Remove bounded. Handle likely and unlikely
30461 (fmtresult::adjust_for_width_or_precision): New function.
30462 (fmtresult:type_max_digits): New function.
30463 (bytes_remaining): Handle likely and unlikely counters.
30464 (min_bytes_remaining): Remove.
30465 (format_percent): Simplify.
30466 (format_integer, format_floating): Set likely and unlikely counters.
30467 (get_string_length, format_character, format_string): Same.
30468 (format_plain, should_warn_p): New function.
30469 (maybe_warn): Call should_warn_p. Update diagnostic messages
30470 and handle those for all directives, including plain strings.
30471 (format_directive): Handle likely and unlikely counters.
30472 Remove unnecessary quoting from diagnostics. Add an informational
30474 (add_bytes): Remove.
30475 (pass_sprintf_length::compute_format_length): Simplify.
30476 (try_substitute_return_value): Handle likely and unlikely counters.
30478 2017-01-26 Carl Love <cel@us.ibm.com>
30480 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
30481 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
30483 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
30486 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
30487 endianess for subregs into account.
30488 * lra-constraints.c (lra_constraints): Do risky transformations
30489 always on the first iteration.
30490 * lra-lives.c (check_pseudos_live_through_calls): Add arg
30491 last_call_used_reg_set.
30492 (process_bb_lives): Define and use last_call_used_reg_set.
30493 * lra.c (lra): Always continue after lra_constraints on the first
30496 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
30498 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
30500 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
30502 2017-01-26 Jakub Jelinek <jakub@redhat.com>
30504 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
30505 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
30506 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
30507 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
30508 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
30509 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
30510 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
30511 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
30512 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
30514 2017-01-26 Marek Polacek <polacek@redhat.com>
30517 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
30518 for the third operand.
30520 2017-01-26 Jakub Jelinek <jakub@redhat.com>
30522 PR middle-end/79236
30523 * omp-low.c (struct omp_context): Add simt_stmt field.
30524 (scan_omp_for): Return omp_context *.
30525 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
30526 context to the _simt_ SIMD stmt.
30527 (lower_omp_for): For combined SIMD with sibling _simt_
30528 SIMD, make sure to use the same decls in _looptemp_
30529 clauses as in the sibling.
30531 2017-01-26 David Sherwood <david.sherwood@arm.com>
30533 PR middle-end/79212
30534 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
30537 2017-01-26 Jakub Jelinek <jakub@redhat.com>
30540 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
30541 emit fld b; fld a; if possible.
30543 * brig-builtins.def: Update copyright years.
30544 * config/arm/arm_acle_builtins.def: Update copyright years.
30546 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
30549 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
30550 constraint instead of o for the stxsd instruction.
30552 2017-01-25 Carl Love <cel@us.ibm.com>
30554 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
30555 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
30557 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
30559 * doc/invoke.texi (C++ Dialect Options): Fix typo.
30561 2017-01-25 Richard Biener <rguenther@suse.de>
30563 PR tree-optimization/69264
30564 * target.def (vector_alignment_reachable): Improve documentation.
30565 * doc/tm.texi: Regenerate.
30566 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
30568 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
30569 earlier changes with respect to TYPE_USER_ALIGN.
30570 (vector_alignment_reachable_p): Likewise. Improve dumping.
30572 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30575 * config/arm/arm.md (xordi3): Force constant operand into a register
30578 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30580 * doc/invoke.texi (-fstore-merging): Correct default optimization
30581 levels at which it is enabled.
30582 (-O): Move -fstore-merging from list to...
30585 2017-01-25 Richard Biener <rguenther@suse.de>
30588 * omp-expand.c: Include debug.h.
30589 (expand_omp_taskreg): Make sure to generate early debug before
30590 outlining anything from a function.
30591 (expand_omp_target): Likewise.
30592 (grid_expand_target_grid_body): Likewise.
30594 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
30597 * asan.c (get_translation_unit_decl): New function.
30598 (asan_add_global): Extract modules file name from globals
30599 TRANSLATION_UNIT_DECL name.
30601 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
30604 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
30605 for long calls with APCS frame and VFP.
30607 2017-01-24 David Malcolm <dmalcolm@redhat.com>
30609 * cfg.c (original_copy_tables_initialized_p): New function.
30610 * cfg.h (original_copy_tables_initialized_p): New decl.
30611 * cfgrtl.c (relink_block_chain): Guard the call to
30612 free_original_copy_tables with a call to
30613 original_copy_tables_initialized_p.
30614 * cgraph.h (symtab_node::native_rtl_p): New decl.
30615 * cgraphunit.c (symtab_node::native_rtl_p): New function.
30616 (symtab_node::needed_p): Don't assert for early assembly output
30617 for __RTL functions.
30618 (cgraph_node::finalize_function): Set "force_output" for __RTL
30620 (cgraph_node::analyze): Bail out early for __RTL functions.
30621 (analyze_functions): Update assertion to support __RTL functions.
30622 (cgraph_node::expand): Bail out early for __RTL functions.
30623 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
30625 * function.h (struct function): Update comment for field
30627 * gimple-expr.c: Include "tree-pass.h".
30628 (gimple_has_body_p): Return false for __RTL functions.
30629 * Makefile.in (OBJS): Add run-rtl-passes.o.
30630 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
30632 (gcc::pass_manager::get_clean_slate): New accessor.
30633 * passes.c: Include "insn-addr.h".
30634 (should_skip_pass_p): Add logging. Update logic for running
30635 "expand" to be compatible with both __GIMPLE and __RTL. Guard
30636 property-provider override so it is only done for gimple passes.
30638 (skip_pass): New function.
30639 (execute_one_pass): Call skip_pass when skipping passes.
30640 * read-md.c (md_reader::read_char): Support filtering
30641 the input to a subset of line numbers.
30642 (md_reader::md_reader): Initialize fields
30643 m_first_line and m_last_line.
30644 (md_reader::read_file_fragment): New function.
30645 * read-md.h (md_reader::read_file_fragment): New decl.
30646 (md_reader::m_first_line): New field.
30647 (md_reader::m_last_line): New field.
30648 * read-rtl-function.c (function_reader::create_function): Only
30649 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
30650 curr_properties. Set DECL_INITIAL to a dummy block.
30651 (read_rtl_function_body_from_file_range): New function.
30652 * read-rtl-function.h (read_rtl_function_body_from_file_range):
30654 * run-rtl-passes.c: New file.
30655 * run-rtl-passes.h: New file.
30657 2017-01-24 Jeff Law <law@redhat.com>
30659 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
30662 2017-01-24 Bin Cheng <bin.cheng@arm.com>
30664 PR tree-optimization/79159
30665 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
30666 (record_nonwrapping_iv): Improve boundary using above function if no
30667 value range information.
30669 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
30670 Martin Jambor <mjambor@suse.cz>
30672 * brig-builtins.def: New file.
30673 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
30674 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
30675 (DEF_HSAIL_SAT_BUILTIN): Likewise.
30676 (DEF_HSAIL_INTR_BUILTIN): Likewise.
30677 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
30678 * builtin-types.def (BT_INT8): New.
30679 (BT_INT16): Likewise.
30680 (BT_UINT8): Likewise.
30681 (BT_UINT16): Likewise.
30682 (BT_FN_ULONG): Likewise.
30683 (BT_FN_UINT_INT): Likewise.
30684 (BT_FN_UINT_ULONG): Likewise.
30685 (BT_FN_UINT_LONG): Likewise.
30686 (BT_FN_UINT_PTR): Likewise.
30687 (BT_FN_ULONG_PTR): Likewise.
30688 (BT_FN_INT8_FLOAT): Likewise.
30689 (BT_FN_INT16_FLOAT): Likewise.
30690 (BT_FN_UINT32_FLOAT): Likewise.
30691 (BT_FN_UINT16_FLOAT): Likewise.
30692 (BT_FN_UINT8_FLOAT): Likewise.
30693 (BT_FN_UINT64_FLOAT): Likewise.
30694 (BT_FN_UINT16_UINT32): Likewise.
30695 (BT_FN_UINT32_UINT16): Likewise.
30696 (BT_FN_UINT16_UINT16_UINT16): Likewise.
30697 (BT_FN_INT_PTR_INT): Likewise.
30698 (BT_FN_UINT_PTR_UINT): Likewise.
30699 (BT_FN_LONG_PTR_LONG): Likewise.
30700 (BT_FN_ULONG_PTR_ULONG): Likewise.
30701 (BT_FN_VOID_UINT64_UINT64): Likewise.
30702 (BT_FN_UINT8_UINT8_UINT8): Likewise.
30703 (BT_FN_INT8_INT8_INT8): Likewise.
30704 (BT_FN_INT16_INT16_INT16): Likewise.
30705 (BT_FN_INT_INT_INT): Likewise.
30706 (BT_FN_UINT_FLOAT_UINT): Likewise.
30707 (BT_FN_FLOAT_UINT_UINT): Likewise.
30708 (BT_FN_ULONG_UINT_UINT): Likewise.
30709 (BT_FN_ULONG_UINT_PTR): Likewise.
30710 (BT_FN_ULONG_ULONG_ULONG): Likewise.
30711 (BT_FN_UINT_UINT_UINT): Likewise.
30712 (BT_FN_VOID_UINT_PTR): Likewise.
30713 (BT_FN_UINT_UINT_PTR: Likewise.
30714 (BT_FN_UINT32_UINT64_PTR): Likewise.
30715 (BT_FN_INT_INT_UINT_UINT): Likewise.
30716 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
30717 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
30718 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
30719 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
30720 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
30721 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
30722 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
30723 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
30724 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
30725 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
30726 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
30727 * doc/frontends.texi: List BRIG FE.
30728 * doc/install.texi (Testing): Add BRIG tesring requirements.
30729 * doc/invoke.texi (Overall Options): Mention BRIG.
30730 * doc/standards.texi (Standards): Doucment BRIG HSA version.
30732 2017-01-24 Richard Biener <rguenther@suse.de>
30734 PR translation/79208
30735 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
30737 2017-01-24 Martin Jambor <mjambor@suse.cz>
30740 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
30741 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
30742 and known_contexts.
30744 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
30746 PR middle-end/79123
30747 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
30748 casts from signed to unsigned really don't have a range.
30750 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
30752 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
30753 GMP_RNDx for compatiblity.
30755 2017-01-24 Martin Liska <mliska@suse.cz>
30758 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
30759 that would prevent us to call alloca with -1 as argument.
30761 2017-01-24 Jakub Jelinek <jakub@redhat.com>
30763 * dwarf2out.c (output_compilation_unit_header, output_file_names):
30764 Avoid -Wformat-security warning.
30766 2017-01-23 Andrew Pinski <apinski@cavium.com>
30768 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
30771 2017-01-23 Martin Sebor <msebor@redhat.com>
30773 PR middle-end/78703
30774 * gimple-ssa-sprintf.c (warn_level): New global.
30775 (format_integer): Use it here and throughout the rest of the file.
30776 Use the same switch to compute sign as base.
30777 (maybe_warn): New function.
30778 (format_directive): Factor out warnings into maybe_warn.
30779 Add debugging output. Use warn_level.
30780 (add_bytes): Use warn_level.
30781 (pass_sprintf_length::compute_format_length): Add debugging output.
30782 (try_substitute_return_value): Same.
30783 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
30785 PR middle-end/78703
30786 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
30787 (struct fmtresult, format_integer, format_floating): Adjust.
30788 (fmtresult::fmtresult): Set max correctly in two argument ctor.
30789 (get_string_length, format_string,format_directive): Same.
30790 (pass_sprintf_length::compute_format_length): Same.
30791 (try_substitute_return_value): Simplify slightly.
30793 PR middle-end/78703
30794 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
30795 (fmtresult::operator+=): Outlined.
30796 (struct fmtresult): Add ctors.
30797 (struct conversion_spec): Rename...
30798 (struct directive): ...to this. Add and remove data members.
30799 (directive::set_width, directive::set_precision): New functions.
30800 (format_percent): Use fmtresult ctor.
30801 (get_width_and_precision): Remove.
30802 (format_integer): Make naming changes. Avoid computing width and
30804 (format_floating): Same. Adjust indentation.
30805 (format_character, format_none): New functions.
30806 (format_string): Moved character handling to format_character.
30807 (format_directive): Remove arguments, change return type.
30808 (parse_directive): New function.
30809 (pass_sprintf_length::compute_format_length): Move directive
30810 parsing to parse_directive.
30812 2017-01-23 Jakub Jelinek <jakub@redhat.com>
30814 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
30815 (assign_assembler_name_if_needed): ... this.
30816 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
30817 (assign_assembler_name_if_needed): ... this.
30818 (free_lang_data_in_cgraph): Adjust callers.
30819 * cgraphunit.c (cgraph_node::analyze): Likewise.
30820 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
30823 2017-01-23 Richard Biener <rguenther@suse.de>
30825 PR tree-optimization/79088
30826 PR tree-optimization/79188
30827 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
30828 resetting loop bounds after last path deletion. Reset loop
30829 bounds of the target loop, make code match the comments.
30830 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
30831 Make sure loops need no fixups.
30833 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
30835 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
30836 exponent support with double type for first argument.
30837 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
30838 type returned by __builtin_vec_extract_sig,
30839 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
30840 functions from "vector int" to "vector unsigned int" or from
30841 "vector long long int" to "vector unsigned long long int".
30842 Changed type returned by __builtin_vec_extract_exp,
30843 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
30844 functions from "vector int" to "vector unsigned int" or from
30845 "vector long long int" to "vector unsigned long long int".
30846 Changed return type of __builtin_vec_test_data_class,
30847 __builtin_vec_test_data_class_sp, and
30848 __builtin_vec_test_data_class_dp from "vector int" to
30849 "vector bool int" or from "vector long long int" to "vector bool
30850 long long int" and changed second argument type from "unsigned
30851 int" to "int". Added new overloaded function forms "vector float
30852 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
30853 "vector float __builtin_vec_insert_exp_sp (vector float, vector
30854 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
30855 double, vector unsigned long long int)" and "vector double
30856 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
30857 long int)". Changed return type of
30858 __builtin_scalar_test_data_class and
30859 __builtin_scalar_test_data_class_sp and
30860 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
30861 int" and changed second argument from "unsigned int" to "int".
30862 Changed type returned by __builtin_scalar_test_neg,
30863 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
30864 from "int" to "bool int". Added new overloaded function form
30865 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
30866 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
30867 exponent double-precision with floating point first argument.
30868 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
30869 documentation of scalar_test_data_class, scalar_test_neg,
30870 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
30871 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
30872 vec_test_data_class built-in functions to reflect refinements in
30873 their type signatures.
30875 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
30877 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
30879 (aarch64_elf_asm_destructor): Likewise.
30881 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
30883 PR rtl-optimization/78634
30884 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
30885 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
30886 * ifcvt.c (noce_try_cmove): Add missing cost check.
30888 PR rtl-optimization/71724
30889 * combine.c (if_then_else_cond): Look for situations where it is
30890 beneficial to undo the work of one of the recursive calls.
30892 2017-01-23 Bin Cheng <bin.cheng@arm.com>
30894 PR tree-optimization/70754
30895 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
30896 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
30897 combined stmt before it if not NULL.
30898 (combine_chains): Process refs reversely and compute dominance point
30901 2017-01-23 Martin Liska <mliska@suse.cz>
30903 PR tree-optimization/79196
30904 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
30905 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
30907 (strlen_optimize_stmt): Call the renamed function.
30909 2017-01-23 Michael Matz <matz@suse.de>
30911 PR tree-optimization/78384
30912 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
30914 2017-01-23 Richard Biener <rguenther@suse.de>
30916 PR tree-optimization/79186
30917 * tree-vrp.c (register_new_assert_for): Make sure we've seen
30918 both incoming edges before moving an assert.
30920 2017-01-23 Martin Jambor <mjambor@suse.cz>
30922 * ipa-prop.c (load_from_param_1): Removed.
30923 (load_from_unmodified_param): Bits from load_from_param_1 put back
30925 (load_from_param): Removed.
30926 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
30927 with stmt. Reverted back to use of load_from_unmodified_param.
30929 2017-01-23 Martin Jambor <mjambor@suse.cz>
30932 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
30933 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
30934 field a pointer to garbage collected vector, mark lattices and
30935 ipcp_orig_node with GTY((skip)).
30936 (ipa_get_param_count): Adjust to descriptors being a pointer.
30937 (ipa_get_param): Likewise.
30938 (ipa_get_type): Likewise.
30939 (ipa_get_param_move_cost): Likewise.
30940 (ipa_set_param_used): Likewise.
30941 (ipa_get_controlled_uses): Likewise.
30942 (ipa_set_controlled_uses): Likewise.
30943 (ipa_is_param_used): Likewise.
30944 (ipa_node_params_t): Move into garbage collector. New methods insert
30946 (ipa_node_params_sum): Annotate wth GTY(()).
30947 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
30949 (ipa_load_from_parm_agg): Adjust declaration.
30950 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
30951 * ipa-profile.c (ipa_profile): Likewise.
30952 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
30953 (ipa_populate_param_decls): Make descriptors parameter garbage
30955 (ipa_dump_param): Adjust to descriptors being a pointer.
30956 (ipa_alloc_node_params): Likewise.
30957 (ipa_initialize_node_params): Likewise.
30958 (load_from_param_1): Make descriptors parameter garbage collected.
30959 (load_from_unmodified_param): Likewise.
30960 (load_from_param): Likewise.
30961 (ipa_load_from_parm_agg): Likewise.
30962 (ipa_node_params::~ipa_node_params): Removed.
30963 (ipa_free_all_node_params): Remove call to delete operator.
30964 (ipa_node_params_t::insert): New.
30965 (ipa_node_params_t::remove): Likewise.
30966 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
30967 copy known_csts and known_contexts vectors.
30968 (ipa_read_node_info): Adjust to descriptors being a pointer.
30969 (ipcp_modif_dom_walker): Make m_descriptors field garbage
30971 (ipcp_transform_function): Make descriptors variable garbage
30974 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
30976 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
30977 * config/i386/avx512dqintrin.h: Ditto.
30978 * config/i386/avx512fintrin.h: Ditto.
30979 * config/i386/i386.c: Handle new builtins.
30980 * config/i386/i386-builtin.def: Add new builtins.
30981 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
30982 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
30984 2017-01-23 Jakub Jelinek <jakub@redhat.com>
30985 Martin Liska <mliska@suse.cz>
30987 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
30988 * asan.c (asan_expand_poison_ifn): Support stores and use
30989 appropriate ASAN report function.
30990 * internal-fn.c (expand_ASAN_POISON_USE): New function.
30991 * internal-fn.def (ASAN_POISON_USE): Declare.
30992 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
30993 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
30994 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
30995 ASAN_POISON calls w/o LHS.
30996 * tree-ssa.c (execute_update_addresses_taken): Create clobber
30997 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
30998 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
30999 * gimplify.c (asan_poison_variables): Add attribute
31000 use_after_scope_memory to variables that really needs to live
31002 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
31003 having the attribute.
31005 2017-01-23 Martin Liska <mliska@suse.cz>
31007 * asan.c (create_asan_shadow_var): New function.
31008 (asan_expand_poison_ifn): Likewise.
31009 * asan.h (asan_expand_poison_ifn): New declaration.
31010 * internal-fn.c (expand_ASAN_POISON): Likewise.
31011 * internal-fn.def (ASAN_POISON): New builtin.
31012 * sanopt.c (pass_sanopt::execute): Expand
31013 asan_expand_poison_ifn.
31014 * tree-inline.c (copy_decl_for_dup_finish): Make function
31016 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
31017 * tree-ssa.c (is_asan_mark_p): New function.
31018 (execute_update_addresses_taken): Rewrite local variables
31019 (identified just by use-after-scope as addressable) into SSA.
31021 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31023 * doc/install.texi (Specific): opensource.apple.com uses https
31024 now. Remove trailing slash.
31026 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31028 * README.Portability: Remove note on an Irix compatibility issue.
31030 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
31032 * gcov.c (INCLUDE_ALGORITHM): Define.
31033 (INCLUDE_VECTOR): Define.
31034 No longer include <vector> and <algorithm> directly.
31036 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
31038 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
31040 * doc/invoke.texi (Code Gen Options): Ditto.
31042 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
31045 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
31047 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
31049 rtl-optimization/79125
31050 * cprop.c (local_cprop_pass): Handle cases where we make an
31051 unconditional trap.
31053 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
31057 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
31058 read from, for big endian.
31060 2017-01-20 Jiong Wang <jiong.wang@arm.com>
31062 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
31063 register pauth builtins for LP64 only.
31065 2017-01-20 Marek Polacek <polacek@redhat.com>
31068 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
31071 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
31073 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
31075 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
31076 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
31077 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
31079 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31082 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
31083 in big-endian mode when they are not a single duplicated value.
31085 2017-01-20 Richard Biener <rguenther@suse.de>
31087 * BASE-VER: Bump to 7.0.1.
31089 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
31091 * omp-low.c (omplow_simd_context): New struct. Use it...
31092 (lower_rec_simd_input_clauses): ...here and...
31093 (lower_rec_input_clauses): ...here to hold common data. Adjust all
31094 references to idx, lane, max_vf, is_simt.
31096 2017-01-20 Graham Markall <graham.markall@embecosm.com>
31098 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
31101 2017-01-20 Martin Jambor <mjambor@suse.cz>
31103 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
31104 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
31106 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
31107 (GTFILES): Rename hsa.c to hsa-common.c.
31108 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
31109 * hsa-dump.c: Likewise.
31110 * hsa-gen.c: Likewise.
31111 * hsa-regalloc.c: Likewise.
31112 * ipa-hsa.c: Likewise.
31113 * omp-expand.c: Likewise.
31114 * omp-low.c: Likewise.
31115 * toplev.c: Likewise.
31117 2017-01-20 Marek Polacek <polacek@redhat.com>
31120 * doc/invoke.texi: Document -Wduplicated-branches.
31121 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
31122 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
31123 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
31124 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
31125 return 0 only when not OEP_LEXICOGRAPHIC.
31126 (fold_build_cleanup_point_expr): Use the expression
31127 location when building CLEANUP_POINT_EXPR.
31128 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
31129 * tree.c (add_expr): Handle error_mark_node.
31131 2017-01-20 Martin Liska <mliska@suse.cz>
31134 * tree-profile.c (init_ic_make_global_vars): Do not call
31136 (gimple_init_gcov_profiler): Likewise.
31138 2017-01-20 Martin Liska <mliska@suse.cz>
31141 * cgraph.h (maybe_create_reference): Remove argument and
31143 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
31145 * ipa-cp.c (create_specialized_node): Likewise.
31146 * symtab.c (symtab_node::maybe_create_reference): Handle
31147 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
31149 2017-01-20 Martin Liska <mliska@suse.cz>
31151 * read-rtl-function.c (function_reader::create_function): Use
31152 build_decl instread of build_decl_stat.
31154 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
31156 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
31157 * config/i386/avx512dqintrin.h: Ditto.
31158 * config/i386/avx512fintrin.h: Ditto.
31159 * config/i386/i386-builtin-types.def: Add new types.
31160 * config/i386/i386.c: Handle new types.
31161 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
31162 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
31163 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
31164 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
31165 (__builtin_ia32_kshiftridi): New.
31166 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
31168 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
31172 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
31173 define to rs6000_init_stack_protect_guard.
31174 (rs6000_init_stack_protect_guard): New function.
31176 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
31177 Yunqiang Su <yunqiang.su@imgtec.com>
31179 * config.gcc (supported_defaults): Add madd4.
31180 (with_madd4): Add validation.
31181 (all_defaults): Add madd4.
31182 * config/mips/mips.opt (mmadd4): New option.
31183 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
31185 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
31187 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
31188 (ISA_HAS_FUSED_MADD4): Likewise.
31189 * doc/invoke.texi (-mmadd4): Document the new option.
31190 * doc/install.texi (--with-madd4): Document the new option.
31192 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31194 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
31195 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
31196 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
31197 (aarch64_init_pauth_hint_builtins): New.
31198 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
31199 (aarch64_expand_builtin): Expand new builtins.
31201 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31203 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
31204 * combine-stack-adj.c (no_unhandled_cfa): Handle
31205 REG_CFA_TOGGLE_RA_MANGLE.
31206 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
31207 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
31208 info for return address signing.
31209 (aarch64_expand_epilogue): Likewise.
31211 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31213 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
31214 * config/aarch64/aarch64-protos.h
31215 (aarch64_return_address_signing_enabled): New declaration.
31216 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
31218 (aarch64_expand_prologue): Sign return address before it's pushed onto
31220 (aarch64_expand_epilogue): Authenticate return address fetched from
31222 (aarch64_override_options): Sanity check for ILP32 and ISA level.
31223 (aarch64_attributes): New function attributes for "sign-return-address".
31224 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
31225 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
31226 ("*do_return"): Generate combined instructions according to key index.
31227 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
31228 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
31230 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
31231 * config/aarch64/aarch64.opt (msign-return-address=): New.
31232 * doc/extend.texi (AArch64 Function Attributes): Documents
31233 "sign-return-address=".
31234 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
31236 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
31238 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
31239 overall option summary.
31241 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31243 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
31244 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
31245 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
31246 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
31248 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
31250 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
31251 -mpower9-minmax by default for -mcpu=power9.
31252 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
31253 128-bit floating point.
31255 2017-01-20 Alan Modra <amodra@gmail.com>
31257 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
31258 optimizing for size.
31260 2017-01-20 Alan Modra <amodra@gmail.com>
31263 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
31264 for strcmp and strncmp from corresponding builtin decl.
31266 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
31268 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
31269 instead of i386/rtems-64.h.
31270 * config/i386/rtems-64.h: Remove.
31272 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
31276 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
31278 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
31280 2017-01-19 Tamar Christina <tamar.christina@arm.com>
31282 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
31283 Change int to HOST_WIDE_INT.
31284 * config/aarch64/aarch64-protos.h
31285 (aarch64_simd_gen_const_vector_dup): Likewise.
31286 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
31288 2017-01-19 David Malcolm <dmalcolm@redhat.com>
31290 * langhooks-def.h (lhd_type_for_size): New decl.
31291 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
31292 * langhooks.c (lhd_type_for_size): New function, taken from
31295 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
31297 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
31298 define_bypass for CR latency.
31299 (power9-cracked-alu): Update bypass latency and remove power9-branch.
31300 (power9-alu2): Add define_bypass for CR latency.
31302 (power9-mul): Update insn latency.
31303 (power9-mul-compare): Update insn latency, bypass latency and remove
31306 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31308 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
31310 * config/aarch64/aarch64.md
31311 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
31312 aarch64_nopcrelative_literal_loads.
31313 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
31315 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
31317 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
31318 TARGET_LOONGSON_3A.
31319 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
31321 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
31324 * config.gcc (supported_defaults): Add lxc1-sxc1.
31325 (with_lxc1_sxc1): Add validation.
31326 (all_defaults): Add lxc1-sxc1.
31327 * config/mips/mips.opt (mlxc1-sxc1): New option.
31328 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
31330 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
31331 __mips_no_lxc1_sxc1.
31332 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
31333 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
31334 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
31336 2017-01-19 Richard Biener <rguenther@suse.de>
31338 PR tree-optimization/72488
31339 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
31340 sure to restore SSA info.
31341 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
31343 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
31345 PR rtl-optimization/79121
31346 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
31347 of the inner type when shifting an extended value.
31349 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
31352 * symtab.c (symtab_node::equal_address_to): Fix comparing of
31353 interposable aliases.
31355 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
31358 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
31359 Use the evmergelohi instruction.
31360 (mov_si<mode>_e500_subreg4_2_le): Likewise.
31361 (mov_sitf_e500_subreg8_2_be): Likewise.
31362 (mov_sitf_e500_subreg12_2_le): Likewise.
31363 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
31364 (mov_si<mode>_e500_subreg4_2_be): Likewise.
31365 (mov_sitf_e500_subreg8_2_le): Likewise.
31366 (mov_sitf_e500_subreg12_2_be): Likewise.
31368 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31370 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
31371 attribute from vecsimple to vecperm.
31372 (altivec_vbpermq2): Likewise.
31374 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31377 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
31379 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31380 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
31381 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
31382 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
31383 case where N arg is SIZE_MAX.
31384 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
31385 (cmpstrsi): Add pattern.
31387 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
31389 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
31390 __builtin_vec_revb builtins.
31391 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
31392 built-in functions to support generation of the ISA 3.0 XXBR<x>
31393 vector byte reverse instructions.
31394 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
31395 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
31396 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
31397 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
31398 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
31399 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
31400 (P9V_BUILTIN_VEC_REVB): Likewise.
31401 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
31402 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
31403 (p9_xxbrq_v16qi): Likewise.
31404 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
31405 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
31406 (p9_xxbrh_v8hi): Likewise.
31407 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
31408 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
31409 vec_revb built-in functions.
31411 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
31413 PR rtl-optimization/78952
31414 * config/i386/i386.md (any_extract): New code iterator.
31415 (*insvqi_2): Use any_extract for source operand.
31416 (*insvqi_3): Use any_shiftrt for source operand.
31418 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
31420 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
31422 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
31424 2017-01-18 Matthias Klose <doko@ubuntu.com>
31426 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
31428 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31430 * config/rs6000/altivec.h (vec_bperm): Change #define.
31431 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
31432 (altivec_vbpermq2): New define_insn.
31433 (altivec_vbpermd): Likewise.
31434 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
31435 function interface.
31436 (VBPERMD): Likewise.
31437 (VBPERM): New polymorphic function interface.
31438 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
31439 Add entries for P9V_BUILTIN_VEC_VBPERM.
31440 * doc/extend.texi: Add interfaces for vec_bperm.
31442 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31444 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
31445 first letter of error messages.
31446 (s390_resolve_overloaded_builtin): Likewise.
31447 * config/s390/s390.c (s390_expand_builtin): Likewise.
31448 (s390_invalid_arg_for_unprototyped_fn): Likewise.
31449 (s390_valid_target_attribute_inner_p): Likewise.
31450 * config/s390/s390.md ("tabort"): Likewise.
31452 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
31454 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
31455 (ISA_AVOID_DIV_HILO): New macro.
31456 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
31457 (ISA_HAS_DDIV): Likewise.
31459 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
31461 * doc/invoke.texi (fabi-version): Correct number of occurrences.
31463 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
31465 * doc/invoke.texi (fabi-version): Spelling fix.
31467 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
31470 * doc/invoke.texi (fabi-version): Mention mangling fix for
31473 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
31476 * doc/invoke.texi (fabi-version): Document discriminator mangling.
31478 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
31481 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
31482 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
31484 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
31486 (stack_protect_test): Ditto.
31487 * config/rs6000/rs6000.opt (mstack-protector-guard=,
31488 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
31490 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
31491 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
31492 -mstack-protector-guard-offset=.
31493 (RS/6000 and PowerPC Options): Ditto.
31495 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
31497 * config/i386/i386.h (MASK_CLASS_P): New define.
31498 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
31499 there are no registers from different register sets also when
31500 mask registers are used. Update function comment.
31501 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
31502 to (*k/*r) and (*k/*km) alternatives.
31504 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
31506 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
31507 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
31508 (EH_RETURN_HANDLER_RTX): New define.
31509 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
31510 Force frame pointer in EH return functions.
31511 (aarch64_expand_epilogue): Add barrier for eh_return.
31512 (aarch64_final_eh_return_addr): Remove.
31513 (aarch64_eh_return_handler_rtx): New function.
31514 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
31516 (aarch64_eh_return_handler_rtx): New prototype.
31518 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31520 * config/rs6000/altivec.h (vec_rlmi): New #define.
31521 (vec_vrlnm): Likewise.
31522 (vec_rlnm): Likewise.
31523 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
31524 (UNSPEC_VRLNM): Likewise.
31525 (VIlong): New mode iterator.
31526 (altivec_vrl<VI_char>mi): New define_insn.
31527 (altivec_vrl<VI_char>nm): Likewise.
31528 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
31530 (VRLDNM): Likewise.
31531 (RLNM): New polymorphic function entry.
31532 (VRLWMI): New monomorphic function entry.
31533 (VRLDMI): Likewise.
31534 (RLMI): New polymorphic function entry.
31535 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
31536 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
31537 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
31540 2017-01-17 Jakub Jelinek <jakub@redhat.com>
31543 * dwarf2out.c (field_byte_offset): Restore the
31544 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
31545 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
31546 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
31549 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
31552 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
31554 2017-01-17 Jakub Jelinek <jakub@redhat.com>
31557 * dwarf2out.c (add_data_member_location_attribute): For constant
31558 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
31559 instead of DW_AT_data_member_location, DW_AT_bit_offset and
31560 DW_AT_byte_size attributes.
31562 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
31564 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
31565 after forcing to constant memory when the code model is medium.
31567 2017-01-17 Julia Koval <julia.koval@intel.com>
31570 * config/i386/avx512fintrin.h
31571 (_mm512_i32gather_ps): Change __addr type to void const*.
31572 (_mm512_mask_i32gather_ps): Ditto.
31573 (_mm512_i32gather_pd): Ditto.
31574 (_mm512_mask_i32gather_pd): Ditto.
31575 (_mm512_i64gather_ps): Ditto.
31576 (_mm512_mask_i64gather_ps): Ditto.
31577 (_mm512_i64gather_pd): Ditto.
31578 (_mm512_mask_i64gather_pd): Ditto.
31579 (_mm512_i32gather_epi32): Ditto.
31580 (_mm512_mask_i32gather_epi32): Ditto.
31581 (_mm512_i32gather_epi64): Ditto.
31582 (_mm512_mask_i32gather_epi64): Ditto.
31583 (_mm512_i64gather_epi32): Ditto.
31584 (_mm512_mask_i64gather_epi32): Ditto.
31585 (_mm512_i64gather_epi64): Ditto.
31586 (_mm512_mask_i64gather_epi64): Ditto.
31587 (_mm512_i32scatter_ps): Change __addr type to void*.
31588 (_mm512_mask_i32scatter_ps): Ditto.
31589 (_mm512_i32scatter_pd): Ditto.
31590 (_mm512_mask_i32scatter_pd): Ditto.
31591 (_mm512_i64scatter_ps): Ditto.
31592 (_mm512_mask_i64scatter_ps): Ditto.
31593 (_mm512_i64scatter_pd): Ditto.
31594 (_mm512_mask_i64scatter_pd): Ditto.
31595 (_mm512_i32scatter_epi32): Ditto.
31596 (_mm512_mask_i32scatter_epi32): Ditto.
31597 (_mm512_i32scatter_epi64): Ditto.
31598 (_mm512_mask_i32scatter_epi64): Ditto.
31599 (_mm512_i64scatter_epi32): Ditto.
31600 (_mm512_mask_i64scatter_epi32): Ditto.
31601 (_mm512_i64scatter_epi64): Ditto.
31602 (_mm512_mask_i64scatter_epi64): Ditto.
31603 * config/i386/avx512pfintrin.h
31604 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
31605 (_mm512_mask_prefetch_i32gather_ps): Ditto.
31606 (_mm512_mask_prefetch_i64gather_pd): Ditto.
31607 (_mm512_mask_prefetch_i64gather_ps): Ditto.
31608 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
31609 (_mm512_prefetch_i32scatter_ps): Ditto.
31610 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
31611 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
31612 (_mm512_prefetch_i64scatter_pd): Ditto.
31613 (_mm512_prefetch_i64scatter_ps): Ditto.
31614 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
31615 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
31616 * config/i386/avx512vlintrin.h
31617 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
31618 (_mm_mmask_i32gather_ps): Ditto.
31619 (_mm256_mmask_i32gather_pd): Ditto.
31620 (_mm_mmask_i32gather_pd): Ditto.
31621 (_mm256_mmask_i64gather_ps): Ditto.
31622 (_mm_mmask_i64gather_ps): Ditto.
31623 (_mm256_mmask_i64gather_pd): Ditto.
31624 (_mm_mmask_i64gather_pd): Ditto.
31625 (_mm256_mmask_i32gather_epi32): Ditto.
31626 (_mm_mmask_i32gather_epi32): Ditto.
31627 (_mm256_mmask_i32gather_epi64): Ditto.
31628 (_mm_mmask_i32gather_epi64): Ditto.
31629 (_mm256_mmask_i64gather_epi32): Ditto.
31630 (_mm_mmask_i64gather_epi32): Ditto.
31631 (_mm256_mmask_i64gather_epi64): Ditto.
31632 (_mm_mmask_i64gather_epi64): Ditto.
31633 (_mm256_i32scatter_ps): Change __addr type to void*.
31634 (_mm256_mask_i32scatter_ps): Ditto.
31635 (_mm_i32scatter_ps): Ditto.
31636 (_mm_mask_i32scatter_ps): Ditto.
31637 (_mm256_i32scatter_pd): Ditto.
31638 (_mm256_mask_i32scatter_pd): Ditto.
31639 (_mm_i32scatter_pd): Ditto.
31640 (_mm_mask_i32scatter_pd): Ditto.
31641 (_mm256_i64scatter_ps): Ditto.
31642 (_mm256_mask_i64scatter_ps): Ditto.
31643 (_mm_i64scatter_ps): Ditto.
31644 (_mm_mask_i64scatter_ps): Ditto.
31645 (_mm256_i64scatter_pd): Ditto.
31646 (_mm256_mask_i64scatter_pd): Ditto.
31647 (_mm_i64scatter_pd): Ditto.
31648 (_mm_mask_i64scatter_pd): Ditto.
31649 (_mm256_i32scatter_epi32): Ditto.
31650 (_mm256_mask_i32scatter_epi32): Ditto.
31651 (_mm_i32scatter_epi32): Ditto.
31652 (_mm_mask_i32scatter_epi32): Ditto.
31653 (_mm256_i32scatter_epi64): Ditto.
31654 (_mm256_mask_i32scatter_epi64): Ditto.
31655 (_mm_i32scatter_epi64): Ditto.
31656 (_mm_mask_i32scatter_epi64): Ditto.
31657 (_mm256_i64scatter_epi32): Ditto.
31658 (_mm256_mask_i64scatter_epi32): Ditto.
31659 (_mm_i64scatter_epi32): Ditto.
31660 (_mm_mask_i64scatter_epi32): Ditto.
31661 (_mm256_i64scatter_epi64): Ditto.
31662 (_mm256_mask_i64scatter_epi64): Ditto.
31663 (_mm_i64scatter_epi64): Ditto.
31664 (_mm_mask_i64scatter_epi64): Ditto.
31665 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
31666 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
31667 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
31668 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
31669 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
31670 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
31671 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
31672 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
31673 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
31674 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
31675 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
31676 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
31677 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
31678 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
31679 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
31680 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
31681 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
31682 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
31683 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
31684 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
31685 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
31686 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
31687 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
31688 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
31689 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
31690 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
31691 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
31692 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
31693 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
31694 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
31695 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
31696 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
31697 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
31698 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
31699 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
31700 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
31701 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
31702 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
31703 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
31704 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
31705 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
31706 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
31707 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
31708 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
31709 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
31710 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
31711 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
31712 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
31713 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
31714 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
31715 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
31716 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
31717 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
31718 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
31719 definitions accordingly.
31721 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
31722 Kuan-Lin Chen <kuanlinchentw@gmail.com>
31725 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
31728 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
31731 * ira-conflicts.c (ira_build_conflicts): Update total conflict
31732 hard regs for inner regno.
31734 2017-01-17 Martin Liska <mliska@suse.cz>
31737 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
31738 assumption and add comment.
31740 2017-01-17 Nathan Sidwell <nathan@acm.org>
31742 * ipa-visibility.c (localize_node): New function, broken out of ...
31743 (function_and_variable_visibility): ... here. Call it.
31745 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
31747 PR middle-end/77445
31748 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
31749 correctly set frequency of oudgoing edge.
31750 (duplicate_thread_path): Fix profile updating.
31752 2017-01-17 Jakub Jelinek <jakub@redhat.com>
31755 * configure.ac: Add GCC_BASE_VER.
31756 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
31757 version from BASE-VER file.
31758 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
31759 (gcc.o): Depend on $(BASEVER).
31760 * common.opt (dumpfullversion): New option.
31761 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
31762 * doc/invoke.texi: Document -dumpfullversion.
31763 * doc/install.texi: Document --with-gcc-major-version-only.
31764 * configure: Regenerated.
31766 2017-01-17 Richard Biener <rguenther@suse.de>
31768 PR tree-optimization/71433
31769 * tree-vrp.c (register_new_assert_for): Merge same asserts
31770 on all incoming edges.
31771 (process_assert_insertions_for): Handle insertions at the
31774 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
31776 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
31777 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
31779 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
31782 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
31785 2017-01-17 Alan Modra <amodra@gmail.com>
31788 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
31789 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
31790 symbolic stack limit when pic.
31792 2017-01-16 Martin Sebor <msebor@redhat.com>
31794 PR tree-optimization/78608
31795 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
31797 2017-01-16 Jeff Law <law@redhat.com>
31800 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
31801 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
31802 for several include directories that may be relative to sysroot.
31803 * config/i386/x-mingw32 (gplus_includedir): Define.
31804 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
31805 (native_system_includedir): Likewise.
31806 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
31807 override if TARGET_SYSTEM_ROOT is defined.
31808 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
31810 PR tree-optimization/79090
31811 PR tree-optimization/33562
31812 PR tree-optimization/61912
31813 PR tree-optimization/77485
31814 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
31815 and computed trims into the dump file.
31817 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
31819 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
31821 2017-01-16 Jakub Jelinek <jakub@redhat.com>
31824 * gimplify.c (gimplify_init_constructor): If want_value and
31825 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
31829 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
31830 sequence. Formatting fixes.
31831 (doloop_optimize): Formatting fixes.
31834 * gcc.c (debug_level_greater_than_spec_func): New function.
31835 (static_spec_functions): Add debug-level-gt spec function.
31836 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
31838 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
31839 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
31840 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
31841 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
31842 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
31843 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
31845 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
31847 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
31848 QImode fixups to general and mask registers only.
31850 2017-01-16 Carl Love <cel@us.ibm.com>
31852 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
31853 for built-in functions
31854 vector signed char vec_nabs (vector signed char)
31855 vector signed short vec_nabs (vector signed short)
31856 vector signed int vec_nabs (vector signed int)
31857 vector signed long long vec_nabs (vector signed long long)
31858 vector float vec_nabs (vector float)
31859 vector double vec_nabs (vector double)
31860 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
31862 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
31863 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
31864 * doc/extend.texi: Update the documentation file for the new built-in
31867 2017-01-16 Martin Sebor <msebor@redhat.com>
31869 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
31872 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31874 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
31875 UNSPEC_VSX__XXSPLTD to require special splat handling.
31877 2017-01-16 David Malcolm <dmalcolm@redhat.com>
31880 * system.h: Poison strndup.
31882 2017-01-16 Alan Modra <amodra@gmail.com>
31885 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
31888 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
31890 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
31892 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
31894 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
31895 call recog here. Assert that INSN_CODE (insn) is non-negative.
31897 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
31900 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
31902 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
31903 in the currently scheduled RTL fragment.
31905 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
31907 PR rtl-optimization/78751
31908 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
31911 2017-01-14 Jeff Law <law@redhat.com>
31913 PR tree-optimization/79090
31914 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
31915 variable length stores.
31916 (compute_trims): Delete dead assignment to *trim_tail.
31917 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
31920 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
31922 PR rtl-optimization/78626
31923 PR rtl-optimization/78727
31924 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
31925 of a block, and split such blocks after everything else is finished.
31927 2017-01-14 Alan Modra <amodra@gmail.com>
31930 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
31931 target legitimate_combined_insn.
31932 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
31933 (rs6000_legitimate_combined_insn): New function.
31934 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
31936 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
31937 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
31938 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
31940 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
31942 * doc/frontends.texi (G++ and GCC): Remove references to Java.
31944 2017-01-13 Jeff Law <law@redhat.com>
31946 PR tree-optimization/33562
31947 PR tree-optimization/61912
31948 PR tree-optimization/77485
31949 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
31951 (delete_dead_assignment): Likewise.
31952 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
31953 statement to delete_dead_call and delete_dead_assignment.
31955 2017-01-13 David Malcolm <dmalcolm@redhat.com>
31958 * substring-locations.c (format_warning_va): Strengthen case 1 so
31959 that both endpoints of the substring must be within the format
31960 range for just the substring to be printed.
31962 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
31964 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
31965 * config/i386/i386.c (ix86_target_string): Add missing options
31966 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
31967 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
31968 flags_other and ix86_target_other to flags2_other. Display unknown
31970 (ix86_valid_target_attribute_inner_p): Add missing options and
31971 reorder options by implied ISAs, as in ix86_target_string.
31973 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
31975 * hash-table.h (hash_table::too_empty_p): New function.
31976 (hash_table::expand): Use it.
31977 (hash_table::traverse): Likewise.
31978 (hash_table::empty_slot): Use sizeof (value_type) instead of
31979 sizeof (PTR) to convert bytes to elements. Shrink the table
31980 if the current size is excessive for the current number of
31983 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
31985 * ira-costs.c (record_reg_classes): Break from the inner loop
31986 early once alt_fail is known to be true. Update outer loop
31987 handling accordingly.
31989 2017-01-13 Jeff Law <law@redhat.com>
31991 * tree-ssa-dse.c (decrement_count): New function.
31992 (increment_start_addr, maybe_trim_memstar_call): Likewise.
31993 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
31994 when we know the partially dead statement is a mem* function.
31996 PR tree-optimization/61912
31997 PR tree-optimization/77485
31998 * tree-ssa-dse.c: Include expr.h.
31999 (maybe_trim_constructor_store): New function.
32000 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
32002 PR tree-optimization/33562
32003 PR tree-optimization/61912
32004 PR tree-optimization/77485
32005 * doc/invoke.texi: Document new dse-max-object-size param.
32006 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
32007 * tree-ssa-dse.c: Include params.h.
32008 (dse_store_status): New enum.
32009 (initialize_ao_ref_for_dse): New, partially extracted from
32011 (valid_ao_ref_for_dse, normalize_ref): New.
32012 (setup_live_bytes_from_ref, compute_trims): Likewise.
32013 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
32014 (maybe_trim_partially_dead_store): Likewise.
32015 (maybe_trim_complex_store): Likewise.
32016 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
32017 Track what bytes live from the original store. Return tri-state
32018 for dead, partially dead or live.
32019 (dse_dom_walker): Add constructor, destructor and new private members.
32020 (delete_dead_call, delete_dead_assignment): New extracted from
32022 (dse_optimize_stmt): Make a member of dse_dom_walker.
32023 Use initialize_ao_ref_for_dse.
32025 PR tree-optimization/33562
32026 PR tree-optimization/61912
32027 PR tree-optimization/77485
32028 * sbitmap.h (bitmap_count_bits): Prototype.
32029 (bitmap_clear_range, bitmap_set_range): Likewise.
32030 * sbitmap.c (bitmap_clear_range): New function.
32031 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
32033 2017-01-13 Martin Liska <mliska@suse.cz>
32036 * function.c (set_cfun): Add new argument force.
32037 * function.h (set_cfun): Likewise.
32038 * ipa-inline-transform.c (inline_call): Use the function when
32039 strict alising from is dropped for function we inline to.
32041 2017-01-13 Richard Biener <rguenther@suse.de>
32043 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
32044 for dumping GIMPLE INTEGER_CSTs.
32046 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32048 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
32049 to 201112L since C++17.
32051 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
32054 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
32055 if -fsanitize=kernel-address is present.
32057 2017-01-13 Richard Biener <rguenther@suse.de>
32059 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
32060 as _Literal ( type ) number in case usual suffixes do not
32061 preserve all information.
32063 2017-01-13 Richard Biener <rguenther@suse.de>
32065 PR tree-optimization/77283
32066 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
32067 and ssa-iterators.h.
32068 (is_feasible_trace): Implement a cost model based on joiner
32071 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
32074 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
32075 char or short to __float128/_Float128 directly.
32077 2017-01-12 Martin Sebor <msebor@redhat.com>
32079 to -Wformat-overflow.
32080 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
32081 (min_bytes_remaining): Same.
32082 (get_string_length): Same.
32083 (format_string): Same.
32084 (format_directive): Same.
32086 (pass_sprintf_length::handle_gimple_call): Same.
32088 2017-01-12 Jakub Jelinek <jakub@redhat.com>
32090 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
32091 info.nowrite calls with no lhs that can't throw. Return bool
32092 whether gsi_remove has been called or not.
32093 (pass_sprintf_length::handle_gimple_call): Return bool whether
32094 try_substitute_return_value called gsi_remove. Formatting fix.
32095 (pass_sprintf_length::execute): Don't use gsi_remove if
32096 handle_gimple_call returned true.
32099 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
32100 be removed due to side-effects, don't remove following barrier nor
32101 turn the successor edge into fallthru edge.
32103 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32106 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
32107 element-reversing loads and stores as not swappable.
32109 2017-01-12 Nathan Sidwell <nathan@acm.org>
32110 Nicolai Stange <nicstange@gmail.com>
32112 * combine.c (try_combine): Don't ignore result of overlap checking
32113 loop. Combine overlap & asm check into single loop.
32115 2017-01-12 Richard Biener <rguenther@suse.de>
32117 * tree-pretty-print.c (dump_generic_node): Provide -gimple
32118 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
32120 2017-01-12 Richard Biener <rguenther@suse.de>
32122 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
32123 and TS_TARGET_OPTION directly derive from TS_BASE.
32124 * tree-core.h (tree_optimization_option): Derive from tree_base.
32125 (tree_target_option): Likewise.
32127 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
32129 * config/i386/i386.c (memory_address_length): Increase len
32130 only when rip_relative_addr_p returns false.
32132 2017-01-11 Julia Koval <julia.koval@intel.com>
32134 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
32135 (OPTION_MASK_ISA_SGX_SET): New.
32136 (ix86_handle_option): Handle OPT_msgx.
32137 * config.gcc: Added sgxintrin.h.
32138 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
32139 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
32140 * config/i386/i386.c (ix86_target_string): Add -msgx.
32142 (ix86_option_override_internal): Handle new options.
32143 (ix86_valid_target_attribute_inner_p): Add sgx.
32144 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
32145 * config/i386/i386.opt: Add msgx.
32146 * config/i386/sgxintrin.h: New file.
32147 * config/i386/x86intrin.h: Add sgxintrin.h.
32149 2017-01-11 Jakub Jelinek <jakub@redhat.com>
32152 * fold-const.c (maybe_nonzero_address): Return 1 for function
32154 (tree_single_nonzero_warnv_p): Don't handle function local objects
32158 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
32161 2017-01-11 David Malcolm <dmalcolm@redhat.com>
32164 * opts.c: Include "spellcheck.h"
32165 (struct string_fragment): New struct.
32166 (struct edit_distance_traits<const string_fragment &>): New
32168 (get_closest_sanitizer_option): New function.
32169 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
32171 2017-01-11 Jakub Jelinek <jakub@redhat.com>
32173 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
32175 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
32176 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
32177 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
32178 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
32179 for initial die_offset if dwarf_split_debug_info.
32180 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
32181 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
32183 (output_skeleton_debug_sections): Formatting fix. Use
32184 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
32185 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
32187 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
32189 * config/arm/cortex-a53.md: Add bypasses for
32190 cortex_a53_r2f_cvt.
32191 (cortex_a53_r2f): Only use for transfers.
32192 (cortex_a53_f2r): Likewise.
32193 (cortex_a53_r2f_cvt): Add reservation for conversions.
32194 (cortex_a53_f2r_cvt): Likewise.
32196 2017-01-11 Tamar Christina <tamar.christina@arm.com>
32198 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
32199 to all inlined functions, change static to extern.
32201 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
32204 * config/arm/arm.c (legitimize_pic_address): Handle reference to
32206 (arm_assemble_integer): Likewise.
32208 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
32210 * config.gcc: Use new awk script to check CPU, FPU and architecture
32211 parameters for --with-... options.
32212 * config/arm/parsecpu.awk: New file
32213 * config/arm/arm-cpus.in: New file.
32214 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
32216 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
32218 * config/arm/t-arm: Update dependency rules.
32219 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
32220 of processing .def files.
32221 * config/arm/genopt.sh: Deleted.
32222 * config/arm/gentune.sh: Deleted.
32223 * config/arm/arm-cores.def: Deleted.
32224 * config/arm/arm-arches.def: Deleted.
32225 * config/arm/arm-fpus.def: Deleted.
32226 * config/arm/arm-tune.md: Regenerated.
32227 * config/arm/arm-tables.opt: Regenerated.
32228 * config/arm/arm-cpu.h: New generated file.
32229 * config/arm/arm-cpu-data.h: New generated file.
32230 * config/arm/arm-cpu-cdata.h: New generated file.
32232 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
32235 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
32237 (input_varpool_node): Unpack dynamically_initialized bit.
32239 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
32241 PR rtl-optimization/79032
32242 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
32243 the alignment of the adjusted memory reference against that of MODE,
32244 instead of the alignment of the original memory reference.
32246 2017-01-11 Martin Jambor <mjambor@suse.cz>
32248 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
32250 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
32251 decorated functions.
32253 2017-01-11 Richard Biener <rguenther@suse.de>
32255 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
32256 set range/nonnull info for PHI results. Do not set it on
32257 stmts marked for removal.
32259 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
32261 * expr.c (store_field): In the bitfield case, fetch the return value
32262 from the registers before applying a single big-endian adjustment.
32263 Always do a final load for a BLKmode value not larger than a word.
32265 2017-01-10 David Malcolm <dmalcolm@redhat.com>
32268 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
32269 that we correctly handle column numbers greater than
32270 LINE_MAP_MAX_COLUMN_NUMBER.
32272 2017-01-10 Martin Sebor <msebor@redhat.com>
32274 PR middle-end/78245
32275 * gimple-ssa-sprintf.c (get_destination_size): Call
32276 {init,fini}object_sizes.
32277 * tree-object-size.c (addr_object_size): Adjust.
32278 (pass_through_call): Adjust.
32279 (pass_object_sizes::execute): Adjust.
32280 * tree-object-size.h (fini_object_sizes): Declare.
32282 2017-01-10 Martin Sebor <msebor@redhat.com>
32284 PR tree-optimization/78775
32285 * builtins.c (get_size_range): Move...
32286 * calls.c: ...to here.
32287 (alloc_max_size): Accept zero argument.
32288 (operand_signed_p): Remove.
32289 (maybe_warn_alloc_args_overflow): Call get_size_range.
32290 * calls.h (get_size_range): Declare.
32292 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
32294 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
32295 from TI's devices.csv file as of September 2016.
32296 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
32298 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
32300 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
32301 * doc/invoke.texi: Likewise.
32302 * doc/md.texi: Likewise.
32303 * doc/objc.texi: Likewise.
32305 2017-01-10 Joshua Conner <joshconner@google.com>
32307 * config/arm/fuchsia-elf.h: New file.
32308 * config/fuchsia.h: New file.
32309 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
32310 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
32312 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
32314 2016-01-10 Richard Biener <rguenther@suse.de>
32316 PR tree-optimization/79034
32317 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
32318 Propagate out degenerate PHIs in the joiner.
32320 2017-01-10 Martin Liska <mliska@suse.cz>
32322 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
32323 (sort_congruence_classes_by_decl_uid): Likewise.
32324 (sort_congruence_class_groups_by_decl_uid): Likewise.
32325 (sem_item_optimizer::merge_classes): Sort class, groups in these
32326 classes and members in the groups by DECL_UID of declarations.
32327 This would make merge operations stable.
32329 2017-01-10 Martin Liska <mliska@suse.cz>
32331 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
32332 usage of m_classes_vec.
32333 (sem_item_optimizer::~sem_item_optimizer): Likewise.
32334 (sem_item_optimizer::get_group_by_hash): Likewise.
32335 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
32336 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
32337 (sem_item_optimizer::verify_classes): Likewise.
32338 (sem_item_optimizer::process_cong_reduction): Likewise.
32339 (sem_item_optimizer::dump_cong_classes): Likewise.
32340 (sem_item_optimizer::merge_classes): Likewise.
32341 * ipa-icf.h (congruence_class_hash): Rename from
32342 congruence_class_group_hash. Remove declaration of m_classes_vec.
32344 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
32346 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
32347 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
32348 * config.gcc: Add avx512vpopcntdqintrin.h.
32349 * config/i386/avx512vpopcntdqintrin.h: New.
32350 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
32351 * config/i386/i386-builtin-types.def: Add new types.
32352 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
32353 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
32354 __builtin_ia32_vpopcountq_v8di_mask): New.
32355 * config/i386/i386-c.c (ix86_target_macros_internal): Define
32356 __AVX512VPOPCNTDQ__.
32357 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
32358 (PTA_AVX512VPOPCNTDQ): Define.
32359 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
32360 TARGET_AVX512VPOPCNTDQ_P): Define.
32361 * config/i386/i386.opt: Add mavx512vpopcntdq.
32362 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
32363 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
32365 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32367 PR middle-end/77484
32368 * predict.def (PRED_CALL): Set to 67.
32370 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
32372 * expr.c (store_field): In the bitfield case, if the value comes from
32373 a function call and is of an aggregate type returned in registers, do
32374 not modify the field mode; extract the value in all cases if the mode
32375 is BLKmode and the size is not larger than a word.
32377 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
32380 * config/i386/cpuid.h: Fix undefined behavior.
32382 2017-01-04 Jeff Law <law@redhat.com>
32384 PR tree-optimization/79007
32385 PR tree-optimization/67955
32386 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
32387 conservative for pt.null when flag_non_call_exceptions is on.
32389 2017-01-09 Jakub Jelinek <jakub@redhat.com>
32391 PR translation/79019
32392 PR translation/79020
32393 * params.def (PARAM_INLINE_MIN_SPEEDUP,
32394 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
32395 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
32397 * config/avr/avr.opt (maccumulate-args): Likewise.
32398 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
32399 * common.opt (freport-bug): Likewise.
32400 * cif-code.def (CIF_FINAL_ERROR): Likewise.
32401 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
32402 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
32403 translatable string.
32404 * config/i386/i386.c (function_value_32): Likewise.
32405 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
32406 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
32408 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
32409 * common/config/msp430/msp430-common.c (msp430_handle_option):
32411 * symtab.c (symtab_node::verify_base): Likewise.
32412 * opts.c (set_debug_level): Likewise.
32413 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
32414 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
32415 missing whitespace to translatable strings.
32416 * config/avr/avr.md (bswapsi2): Fix typo in comment.
32417 * config/sh/superh.h: Likewise.
32418 * config/i386/xopintrin.h: Likewise.
32419 * config/i386/znver1.md: Likewise.
32420 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
32421 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
32422 * double-int.h (struct double_int): Likewise.
32423 * double-int.c (div_and_round_double): Likewise.
32424 * wide-int.cc: Likewise.
32425 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
32426 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
32427 * cfgcleanup.c (crossjumps_occured): Renamed to ...
32428 (crossjumps_occurred): ... this.
32429 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
32432 PR tree-optimization/78899
32433 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
32434 returning bool return struct loop *, NULL for failure and the new
32436 (versionable_outer_loop_p): Don't version outer loop if it has
32437 dont_vectorized bit set.
32438 (tree_if_conversion): When versioning outer loop, ensure
32439 tree_if_conversion is performed also on the inner loop of the
32440 non-vectorizable outer loop copy.
32441 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
32442 LOOP_VECTORIZED in inner loop of the scalar outer loop and
32443 prevent vectorization of it.
32444 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
32445 the outer loop vectorization of the non-scalar version is attempted
32446 before vectorization of the inner loop in scalar version. If
32447 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
32448 vectorization of its inner loop.
32449 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
32450 has 2 inner loops, rename also on edges from bb whose single pred
32451 is outer_loop->header. Fix typo in function comment.
32453 2017-01-09 Martin Sebor <msebor@redhat.com>
32456 * asan.c (asan_emit_stack_protection): Increase local buffer size
32457 to avoid snprintf truncation warning.
32459 2017-01-09 Andrew Pinski <apinski@cavium.com>
32461 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
32462 to reference thunderx2t99 for the tuning structure
32463 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
32465 (thunderx2t99_extra_costs): This.
32466 * config/aarch64/aarch64-tune.md: Regenerate.
32467 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
32468 (vulcan_addrcost_table): This.
32469 (vulcan_regmove_cost): Rename to ...
32470 (thunderx2t99_regmove_cost): This.
32471 (vulcan_vector_cost): Rename to ...
32472 (thunderx2t99_vector_cost): this.
32473 (vulcan_branch_cost): Rename to ...
32474 (thunderx2t99_branch_cost): This.
32475 (vulcan_tunings): Rename to ...
32476 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
32477 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
32479 2017-01-09 Martin Jambor <mjambor@suse.cz>
32483 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
32484 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
32485 (propagate_vr_accross_jump_function): Use the above function for all
32486 value range computations for pass-through jump functions and type
32487 converasion from explicit value range values.
32488 (ipcp_propagate_stage): Do not attempt to deduce types of formal
32489 parameters from TYPE_ARG_TYPES.
32490 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
32491 (ipa_write_node_info): Stream type of the actual argument.
32492 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
32494 2017-01-09 Martin Liska <mliska@suse.cz>
32497 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
32498 (lookup_compiler): Do not show error message with have_E.
32500 2017-01-09 Jakub Jelinek <jakub@redhat.com>
32502 PR tree-optimization/78938
32503 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
32504 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
32505 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
32506 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
32509 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32511 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
32514 2017-01-09 Richard Biener <rguenther@suse.de>
32516 PR tree-optimization/78997
32517 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
32518 name condition properly.
32520 2017-01-09 Richard Biener <rguenther@suse.de>
32523 * dwarf2out.c (is_cxx): New overload with context.
32524 (is_naming_typedef_decl): Use it.
32526 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
32528 * invoke.texi (Option Summary): Correct spacing in option lists
32529 and add line breaks to fix over-long lines.
32531 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
32533 PR middle-end/17660
32535 * extend.texi (Common Variable Attributes): Add xref to GCC
32536 Internals manual to explain mode attribute keywords.
32538 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
32541 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
32542 and Preprocessor Options.
32543 (Options for Linking): Document -pthread here....
32544 (RS/6000 and PowerPC Options): ...not here.
32545 (Solaris 2 Options): ...or here.
32546 * doc/cppopts.texi: Document -pthread.
32548 2017-01-08 Martin Sebor <msebor@redhat.com>
32550 PR middle-end/77708
32551 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
32552 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
32553 New member functions.
32554 (format_directive): Used them.
32556 (pass_sprintf_length::handle_gimple_call): Same.
32557 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
32558 to avoid truncation for any argument.
32559 (extract_affine_mul): Same.
32560 * tree.c (get_file_function_name): Same.
32562 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32564 PR middle-end/77484
32565 * predict.def (PRED_INDIR_CALL): Set to 86.
32567 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
32569 PR preprocessor/54124
32570 * doc/cppopts.texi: Reformat -d subtable to list the full name
32571 of the options. Add cross-reference to the docs for the general
32572 compiler -d options.
32573 * doc/invoke.texi (Developer Options): Add cross-reference to the
32574 preprocessor-specific -d option documentation.
32576 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
32578 PR preprocessor/13498
32579 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
32580 redudant material, and reflect new command-line options.
32581 (System Headers): Likewise.
32583 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
32585 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
32586 -isystem, and -idirafter. Copy-edit.
32587 * doc/cppopts.texi: Copy-edit. Remove contradiction about
32588 default for -ftrack-macro-expansion. Delete obsolete and
32589 badly-formatted implementation details about -fdebug-cpp output.
32590 * doc/cppwarnopts.texi: Copy-edit.
32592 2017-01-07 David Malcolm <dmalcolm@redhat.com>
32595 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
32596 that the transition from a max line width >= 1<<10 to narrower
32597 lines works correctly.
32599 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
32601 * doc/options.texi (PerFunction): New.
32602 * opt-functions.awk (switch_flags): Map both Optimization and
32603 PerFunction to CL_OPTIMIZATION.
32604 * opth-gen.awk: Test for PerFunction flag along with
32606 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
32607 it only when the latter is present. Skip those that don't in
32608 the hash function generator.
32609 * common.opt (fvar-tracking): Mark as PerFunction instead of
32611 (fvar-tracking-assignments): Likewise.
32612 (fvar-tracking-assignments-toggle): Likewise.
32613 (fvar-tracking-uninit): Likewise.
32615 2017-01-07 Jakub Jelinek <jakub@redhat.com>
32617 PR translation/79018
32618 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
32621 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
32624 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
32625 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
32626 TARGET_LONG_JUMP_TABLE_OFFSETS.
32627 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
32628 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
32629 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
32630 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
32631 * config/m68k/m68k.md (tablejump expander): Likewise.
32632 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
32633 TARGET_LONG_JUMP_TABLE_OFFSETS.
32634 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
32635 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
32637 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
32638 David Holsgrove <david.holsgrove@xilinx.com>
32640 * common/config/microblaze/microblaze-common.c
32641 (TARGET_EXCEPT_UNWIND_INFO): Remove.
32642 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
32644 * config/microblaze/microblaze.c (microblaze_must_save_register)
32645 (microblaze_expand_epilogue, microblaze_return_addr): Handle
32647 (microblaze_eh_return): New function.
32648 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
32649 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
32650 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
32651 * config/microblaze/microblaze.md (eh_return): New pattern.
32653 2017-01-06 Jakub Jelinek <jakub@redhat.com>
32655 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
32656 GCC_DIAGNOSTIC_STRINGIFY): Define.
32658 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
32660 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32662 * config/arm/arm.md (<mcrr>): New.
32664 * config/arm/arm.c (arm_arch5te): New.
32665 (arm_option_override): Set arm_arch5te.
32666 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
32668 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
32669 (arm_mcrr_qualifiers): ... this. New.
32670 (MRRC_QUALIFIERS): Define to...
32671 (arm_mrrc_qualifiers): ... this. New.
32672 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
32674 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
32675 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
32676 (MRRCI, mrrc, MRRC): New.
32677 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
32678 VUNSPEC_MRRC2): New.
32680 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32682 * config/arm/arm.md (<mcr>): New.
32684 * config/arm/arm.c (arm_coproc_builtin_available): Add
32685 support for mcr, mrc, mcr2 and mrc2.
32686 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
32687 (arm_mcr_qualifiers): ... this. New.
32688 (MRC_QUALIFIERS): Define to ...
32689 (arm_mrc_qualifiers): ... this. New.
32690 (MCR_QUALIFIERS): Define to ...
32691 (arm_mcr_qualifiers): ... this. New.
32692 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
32694 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
32695 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
32696 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
32697 VUNSPEC_MRC2): New.
32699 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32701 * config/arm/arm.md (*ldc): New.
32705 * config/arm/arm.c (arm_coproc_builtin_available): Add
32706 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
32707 (arm_coproc_ldc_stc_legitimate_address): New.
32708 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
32709 'qualifier_const_pointer'.
32710 (LDC_QUALIFIERS): Define to...
32711 (arm_ldc_qualifiers): ... this. New.
32712 (STC_QUALIFIERS): Define to...
32713 (arm_stc_qualifiers): ... this. New.
32714 * config/arm/arm-protos.h
32715 (arm_coproc_ldc_stc_legitimate_address): New.
32716 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
32717 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
32718 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
32719 stc2, stcl, stc2l): New.
32720 * config/arm/constraints.md (Uz): New.
32721 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
32722 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
32723 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
32724 VUNSPEC_STC2L): New.
32726 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32728 * config/arm/arm.md (<cdp>): New.
32729 * config/arm/arm.c (neon_const_bounds): Rename this ...
32730 (arm_const_bounds): ... this.
32731 (arm_coproc_builtin_available): New.
32732 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
32733 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
32734 (CDP_QUALIFIERS): Define to...
32735 (arm_cdp_qualifiers): ... this. New.
32737 (arm_expand_builtin_args): Add case for 6 arguments.
32738 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
32739 (arm_const_bounds): ... this.
32740 (arm_coproc_builtin_available): New.
32741 * config/arm/arm_acle.h (__arm_cdp): New.
32743 * config/arm/arm_acle_builtins.def (cdp): New.
32745 * config/arm/iterators.md (CDPI,CDP,cdp): New.
32746 * config/arm/neon.md: Rename all 'neon_const_bounds' to
32747 'arm_const_bounds'.
32748 * config/arm/types.md (coproc): New.
32749 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
32750 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
32751 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
32752 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
32754 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32756 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
32757 (UBINOP_QUALIFIERS): New.
32759 (acle_builtin_data): New. Change comment.
32760 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
32761 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
32762 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
32763 arm_acle_builtins.def.
32764 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
32765 (arm_init_acle_builtins): New.
32766 (CRC32_BUILTIN): Remove.
32767 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
32768 crc32cb, crc32ch and crc32cw.
32769 (arm_init_crc32_builtins): Remove.
32770 (arm_init_builtins): Use arm_init_acle_builtins rather
32771 than arm_init_crc32_builtins.
32772 (arm_expand_acle_builtin): New.
32773 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
32774 * config/arm/arm_acle_builtins.def: New.
32776 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32778 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
32779 (arm_builtin_datum): ... this.
32780 (arm_init_neon_builtin): Rename to ...
32781 (arm_init_builtin): ... this. Add a new parameters PREFIX
32782 and USE_SIG_IN_NAME.
32783 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
32784 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
32785 'arm_builtin_datum'.
32786 (arm_init_vfp_builtins): Likewise.
32787 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
32788 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
32789 (arm_expand_neon_args): Rename to ...
32790 (arm_expand_builtin_args): ... this. Rename builtin_arg
32791 enum values and differentiate between ARG_BUILTIN_MEMORY
32792 and ARG_BUILTIN_NEON_MEMORY.
32793 (arm_expand_neon_builtin_1): Rename to ...
32794 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
32795 values, arm_expand_builtin_args and add bool parameter NEON.
32796 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
32797 (arm_expand_vfp_builtin): Likewise.
32798 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
32800 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32802 PR middle-end/77484
32803 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
32804 * predict.c (tree_estimate_probability_bb): Reverse direction of
32805 polymorphic call predictor.
32807 2017-01-06 David Malcolm <dmalcolm@redhat.com>
32809 * passes.c (execute_one_pass): Split out pass-skipping logic into...
32810 (determine_pass_name_match): ...this new function and...
32811 (should_skip_pass_p): ...this new function.
32813 2017-01-06 Nathan Sidwell <nathan@acm.org>
32815 * ipa-visibility.c (function_and_variable_visibility): Reformat
32816 comments and long lines. Remove extrneous if.
32817 * symtab.c (symtab_node::make_decl_local): Fix code format.
32818 (symtab_node::set_section_for_node): Fix comment typo.
32820 2017-01-06 Martin Liska <mliska@suse.cz>
32823 * lra-constraints.c: Rename invariant to lra_invariant.
32824 * predict.c (set_even_probabilities): Initialize e to NULL.
32826 2017-01-05 Martin Sebor <msebor@redhat.com>
32828 PR tree-optimization/78910
32829 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
32830 (format_integer): Correct off-by-one error in the handling
32831 of precision with negative numbers in signed conversions..
32833 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
32835 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
32837 2017-01-05 Jakub Jelinek <jakub@redhat.com>
32839 PR tree-optimization/71016
32840 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
32841 factor_out_conditional_conversion. Formatting fix.
32842 (factor_out_conditional_conversion): Add cond_stmt argument.
32843 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
32844 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
32847 2017-01-05 David Malcolm <dmalcolm@redhat.com>
32849 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
32850 read-rtl-function.o, and selftest-rtl.o.
32851 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
32852 (selftest::aarch64_test_loading_full_dump): New function.
32853 (selftest::aarch64_run_selftests): New function.
32854 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
32855 selftest::aarch64_run_selftests.
32856 * config/i386/i386.c
32857 (selftest::ix86_test_loading_dump_fragment_1): New function.
32858 (selftest::ix86_test_loading_call_insn): New function.
32859 (selftest::ix86_test_loading_full_dump): New function.
32860 (selftest::ix86_test_loading_unspec): New function.
32861 (selftest::ix86_run_selftests): Call the new functions.
32862 * emit-rtl.c (maybe_set_max_label_num): New function.
32863 * emit-rtl.h (maybe_set_max_label_num): New decl.
32864 * function.c (instantiate_decls): Guard call to
32865 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
32866 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
32868 * gensupport.c (gen_reader::gen_reader): Pass "false"
32869 for new "compact" param of rtx_reader.
32870 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
32871 rather than an empty string for NULL strings.
32872 * read-md.c: Potentially include config.h rather than bconfig.h.
32873 Wrap include of errors.h with #ifdef GENERATOR_FILE.
32874 (have_error): New global, copied from errors.c.
32875 (md_reader::read_name): Rename to...
32876 (md_reader::read_name_1): ...this, adding "out_loc" param,
32877 and converting "missing name or number" to returning false, rather
32879 (md_reader::read_name): Reimplement in terms of read_name_1.
32880 (md_reader::read_name_or_nil): New function.
32881 (md_reader::read_string): Handle "(nil)" by returning NULL.
32882 (md_reader::md_reader): Add new param "compact".
32883 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
32884 (md_reader::read_file): New method.
32885 * read-md.h (md_reader::md_reader): Add new param "compact".
32886 (md_reader::read_file): New method.
32887 (md_reader::is_compact): New accessor.
32888 (md_reader::read_name): Convert return type from void to file_location.
32889 (md_reader::read_name_or_nil): New decl.
32890 (md_reader::read_name_1): New decl.
32891 (md_reader::m_compact): New field.
32892 (noop_reader::noop_reader): Pass "false" for new "compact" param
32894 (rtx_reader::rtx_reader): Add new "compact" param.
32895 (rtx_reader::read_rtx_operand): Make virtual and convert return
32896 type from void to rtx.
32897 (rtx_reader::read_until): New decl.
32898 (rtx_reader::handle_any_trailing_information): New virtual function.
32899 (rtx_reader::postprocess): New virtual function.
32900 (rtx_reader::finalize_string): New virtual function.
32901 (rtx_reader::m_in_call_function_usage): New field.
32902 (rtx_reader::m_reuse_rtx_by_id): New field.
32903 * read-rtl-function.c: New file.
32904 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
32905 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
32906 (selftest::verify_three_block_rtl_cfg): New decl.
32907 * read-rtl-function.h: New file.
32908 * read-rtl.c: Potentially include config.h rather than bconfig.h.
32909 For host, include function.h, memmodel.h, and emit-rtl.h.
32910 (one_time_initialization): New function.
32911 (struct compact_insn_name): New struct.
32912 (compact_insn_names): New array.
32913 (find_code): Handle insn codes in compact dumps.
32914 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
32915 (bind_subst_iter_and_attr): Likewise.
32916 (add_condition_to_string): Likewise.
32917 (add_condition_to_rtx): Likewise.
32918 (apply_attribute_uses): Likewise.
32919 (add_current_iterators): Likewise.
32920 (apply_iterators): Likewise.
32921 (initialize_iterators): Guard usage of apply_subst_iterator with
32922 #ifdef GENERATOR_FILE.
32923 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
32924 (md_reader::read_mapping): Likewise.
32925 (add_define_attr_for_define_subst): Likewise.
32926 (add_define_subst_attr): Likewise.
32927 (read_subst_mapping): Likewise.
32928 (check_code_iterator): Likewise.
32929 (rtx_reader::read_rtx): Likewise. Move one-time initialization
32931 (one_time_initialization): New function.
32932 (rtx_reader::read_until): New method.
32933 (read_flags): New function.
32934 (parse_reg_note_name): New function.
32935 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
32936 Handle reuse_rtx ids.
32937 Wrap iterator lookup within #ifdef GENERATOR_FILE.
32938 Add parsing support for RTL dumps, mirroring the special-cases in
32939 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
32940 values, and calling handle_any_trailing_information.
32941 (rtx_reader::read_rtx_operand): Convert return type from void
32942 to rtx, returning return_rtx. Handle case 'e'. Call
32943 finalize_string on XSTR and XTMPL fields.
32944 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
32945 "(nil)" values were omitted. Call the postprocess vfunc on the
32947 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
32948 class ctor. Initialize m_in_call_function_usage. Call
32949 one_time_initialization.
32950 * rtl-tests.c (selftest::test_uncond_jump): Call
32951 set_new_first_and_last_insn.
32952 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
32953 * selftest-rtl.c: New file.
32954 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
32955 (selftest::get_insn_by_uid): New decl.
32956 * selftest-run-tests.c (selftest::run_tests): Call
32957 read_rtl_function_c_tests.
32958 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
32959 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
32962 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
32964 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
32965 operands in a special way. Assert that pos+len <= mode precision.
32967 2017-01-05 Jakub Jelinek <jakub@redhat.com>
32969 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
32970 3 argument Alias with unlimited for the negative form.
32971 (fno-vect-cost-model): Removed.
32973 2017-01-05 Martin Liska <mliska@suse.cz>
32975 * hsa-gen.c (gen_hsa_divmod): New function.
32976 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
32978 2017-01-05 Martin Liska <mliska@suse.cz>
32981 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
32984 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32986 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
32987 small constant length operands.
32989 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32991 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
32992 between loop iterations.
32994 2017-01-05 Martin Liska <mliska@suse.cz>
32997 * gimplify.c (gimplify_decl_expr): Compare to
32998 asan_poisoned_variables instread of checking flags.
32999 (gimplify_target_expr): Likewise.
33000 (gimplify_expr): Likewise.
33001 (gimplify_function_tree): Conditionally initialize
33002 asan_poisoned_variables.
33004 2017-01-04 Jeff Law <law@redhat.com>
33006 PR tree-optimizatin/78812
33007 * rtl.h (contains_mem_rtx_p): Prototype.
33008 * ifcvt.c (containts_mem_rtx_p): Move from here to...
33009 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
33010 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
33011 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
33012 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
33014 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33016 * input.c (assert_char_at_range): Default-initialize actual_range.
33018 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33020 * df-scan.c (df_ref_create_structure): Make regno unsigned,
33021 to match the caller.
33023 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33025 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
33026 insns after final jump in test to emit dummy move.
33028 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33030 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
33031 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
33033 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33035 * multiple_target.c (create_dispatcher_calls): Init e_next.
33036 * tree-ssa-loop-split.c (split_loop): Init border.
33037 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
33040 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
33045 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
33046 (altivec_register_operand): Do not return true if the operand
33047 contains a SUBREG mixing SImode and SFmode.
33048 (vsx_register_operand): Likewise.
33049 (vsx_reg_sfsubreg_ok): New predicate.
33050 (vfloat_operand): Do not return true if the operand contains a
33051 SUBREG mixing SImode and SFmode.
33052 (vint_operand): Likewise.
33053 (vlogical_operand): Likewise.
33054 (gpc_reg_operand): Likewise.
33055 (int_reg_operand): Likewise.
33056 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
33057 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
33058 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
33060 (rs6000_emit_move_si_sf_subreg): New helper function.
33061 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
33062 fixup SUBREGs involving SImode and SFmode.
33063 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
33064 numbers for the new peephole2 optimization.
33065 (peephole2 for SFmode unions): New peephole2 to optimize cases in
33066 the GLIBC math library that do AND/IOR/XOR operations on single
33067 precision floating point.
33068 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
33069 target macros to say whether we need to avoid SUBREGs mixing
33071 (TARGET_ALLOW_SF_SUBREG): Likewise.
33072 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
33073 (UNSPEC_SI_FROM_SF): Likewise.
33074 (iorxor): Change spacing.
33075 (and_ior_xor): New iterator for AND, IOR, and XOR.
33076 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
33077 (movdi_from_sf_zero_ext): Likewise.
33078 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
33079 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
33080 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
33081 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
33082 (fms<mode>4): Likewise.
33083 (fnma<mode>4): Likewise.
33084 (fnms<mode>4): Likewise.
33085 (nfma<mode>4): Likewise.
33086 (nfms<mode>4): Likewise.
33088 2017-01-04 Marek Polacek <polacek@redhat.com>
33091 * doc/invoke.texi: Document -Wpointer-compare.
33093 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33095 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
33098 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
33099 descriptions for -gdwarf-5 and emit them as uleb128 instead of
33102 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
33105 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
33106 documentation of the powerpc_popcntb_ok attribute.
33107 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
33108 code to issue warning messages if a requested CPU configuration is
33109 not supported by the binary (assembler and loader) toolchain.
33110 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
33111 made to define a built-in function that has been disabled.
33112 (paired_init_builtins): Add assertion to prevent ICE if attempt is
33113 made to define a built-in function that has been disabled.
33114 (altivec_init_builtins): Add comment explaining why definition
33115 of the DST built-in functions is not preceded by an assertion
33116 check. Add assertions to prevent ICE if attempts are made to
33117 define an altivec predicate or an abs* built-in function that has
33119 (htm_init_builtins): Add comment explaining why definition of the
33120 htm built-in functions is not preceded by an assertion check.
33122 2017-01-04 Jeff Law <law@redhat.com>
33124 PR tree-optimizatin/67955
33125 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
33126 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
33127 the points-to solution does not include pt_null. Use DECL_PT_UID
33130 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
33132 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
33133 Use gen_int_mode instead of gen_lopwart for const_int operands.
33135 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33137 PR tree-optimization/71563
33138 * match.pd: Simplify X << Y into X if Y is known to be 0 or
33139 out of range value - has low bits known to be zero.
33141 2017-01-04 Alan Modra <amodra@gmail.com>
33143 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
33144 * configure: Regenerate.
33145 * config.in: Regenerate.
33147 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33150 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
33151 a substring of the message, but strcmp with the whole message. Ifdef
33152 ENABLE_NLS, translate the message first using dgettext.
33154 2017-01-03 Jeff Law <law@redhat.com>
33156 PR tree-optimizatin/78856
33157 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
33158 (mark_threaded_blocks): Remove code to truncate thread paths that
33159 cross multiple loop headers. Instead invalidate the cached loop
33160 iteration information and handle case of a thread path walking
33161 into an irreducible region.
33163 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
33166 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
33167 assertions. Add support for doing the signbit if the IEEE 128-bit
33168 floating point value is in a GPR.
33169 * config/rs6000/rs6000.md (Fsignbit): Delete.
33170 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
33171 Update the length attribute if the value is in a GPR.
33172 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
33173 the sign or zero extension instruction, since the value is always 0/1.
33174 (signbit<mode>2_dm2): Delete using <Fsignbit>.
33177 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
33178 extracting SImode to a GPR register so that we can generate a
33179 store, limit the vector to be in a traditional Altivec register
33180 for the vextuwrx instruction.
33182 2017-01-03 Ian Lance Taylor <iant@google.com>
33184 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
33186 2017-01-03 Martin Sebor <msebor@redhat.com>
33188 PR tree-optimization/78696
33189 * gimple-ssa-sprintf.c (format_floating): Correct handling of
33190 precision. Use MPFR for %f for greater fidelity. Correct handling
33192 (pass_sprintf_length::compute_format_length): Set width and precision
33193 specified by asrerisk to void_node for vararg functions.
33194 (try_substitute_return_value): Adjust dump output.
33196 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
33198 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
33200 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
33202 * doc/invoke.texi (SPARC options): Document -mlra as the default.
33203 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
33204 -mlra/-mno-lra was passed to the compiler.
33206 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
33208 PR rtl-optimization/65618
33209 * emit-rtl.c (try_split): Move initialization of "before" and
33210 "after" to just before the call to emit_insn_after_setloc.
33212 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
33214 * doc/md.texi (Standard Names): Remove reference to Java frontend.
33216 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
33218 * dwarf2out.c (gen_enumeration_type_die): When
33219 -gno-strict-dwarf, add a DW_AT_encoding attribute.
33221 2017-01-03 Jakub Jelinek <jakub@redhat.com>
33223 PR tree-optimization/78965
33224 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
33225 Change first argument from const call_info & to call_info &. For %n
33226 set info.nowrite to false.
33228 PR middle-end/78901
33229 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
33230 possibly throwing calls.
33232 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
33233 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
33234 and fns handling, rather than in a separate case SSA_NAME.
33236 2017-01-02 Jeff Law <law@redhat.com>
33238 * config/darwin-driver.c (darwin_driver_init): Const-correctness
33239 fixes for first_period and second_period variables.
33241 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
33244 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
33245 (*insvqi_1): New insn pattern.
33246 (*insvqi_1_mem_rex64): Ditto.
33247 (*insvqi_2): Ditto.
33248 (*insvqi_3): Rename from *insvqi.
33250 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
33252 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
33254 * doc/cfg.texi (Edges): Remove reference to Java.
33255 (Maintaining the CFG): Ditto.
33257 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33259 PR middle-end/77674
33260 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
33261 transparent aliases.
33263 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33265 PR middle-end/77484
33266 * predict.def (PRED_CALL): Update hitrate.
33267 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
33268 * predict.c (tree_estimate_probability_bb): Split CALL predictor
33269 into direct/indirect/polymorphic variants.
33271 2017-01-01 Jakub Jelinek <jakub@redhat.com>
33273 Update copyright years.
33275 * gcc.c (process_command): Update copyright notice dates.
33276 * gcov-dump.c (print_version): Ditto.
33277 * gcov.c (print_version): Ditto.
33278 * gcov-tool.c (print_version): Ditto.
33279 * gengtype.c (create_file): Ditto.
33280 * doc/cpp.texi: Bump @copying's copyright year.
33281 * doc/cppinternals.texi: Ditto.
33282 * doc/gcc.texi: Ditto.
33283 * doc/gccint.texi: Ditto.
33284 * doc/gcov.texi: Ditto.
33285 * doc/install.texi: Ditto.
33286 * doc/invoke.texi: Ditto.
33288 Copyright (C) 2017 Free Software Foundation, Inc.
33290 Copying and distribution of this file, with or without modification,
33291 are permitted in any medium without royalty provided the copyright
33292 notice and this notice are preserved.