1 2017-11-04 Andrew Waterman <andrew@sifive.com>
3 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
5 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
7 config/i386/i386.c (choose_basereg): Use optional scratch
8 register and add assertion.
9 (x86_emit_outlined_ms2sysv_save): Use scratch register when
10 needed, and don't allocate stack.
11 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
12 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
13 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
15 2017-11-03 Jeff Law <law@redhat.com>
17 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
18 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
19 to probe at the start of a noreturn function.
21 2017-11-03 Jakub Jelinek <jakub@redhat.com>
23 PR tree-optimization/78821
24 * gimple-ssa-store-merging.c: Update the file comment.
25 (MAX_STORE_ALIAS_CHECKS): Define.
26 (struct store_operand_info): New type.
27 (store_operand_info::store_operand_info): New constructor.
28 (struct store_immediate_info): Add rhs_code and ops data members.
29 (store_immediate_info::store_immediate_info): Add rhscode, op0r
30 and op1r arguments to the ctor, initialize corresponding data members.
31 (struct merged_store_group): Add load_align_base and load_align
33 (merged_store_group::merged_store_group): Initialize them.
34 (merged_store_group::do_merge): Update them.
35 (merged_store_group::apply_stores): Pick the constant for
36 encode_tree_to_bitpos from one of the two operands, or skip
37 encode_tree_to_bitpos if neither operand is a constant.
38 (class pass_store_merging): Add process_store method decl. Remove
39 bool argument from terminate_all_aliasing_chains method decl.
40 (pass_store_merging::terminate_all_aliasing_chains): Remove
41 var_offset_p argument and corresponding handling.
42 (stmts_may_clobber_ref_p): New function.
43 (compatible_load_p): New function.
44 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
45 if there is overlap and rhs_code is not INTEGER_CST. For
46 non-overlapping stores terminate group if rhs is not mergeable.
47 (get_alias_type_for_stmts): Change first argument from
48 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
49 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
50 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
52 (get_location_for_stmts): Change first argument from
53 auto_vec<gimple *> & to vec<gimple *> &.
54 (struct split_store): Remove orig_stmts data member, add orig_stores.
55 (split_store::split_store): Create orig_stores rather than orig_stmts.
56 (find_constituent_stmts): Renamed to ...
57 (find_constituent_stores): ... this. Change second argument from
58 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
59 to info structures rather than the statements.
60 (split_group): Rename ALLOW_UNALIGNED argument to
61 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
62 it. Adjust find_constituent_stores caller.
63 (imm_store_chain_info::output_merged_store): Handle rhs_code other
64 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
65 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
66 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
67 (mem_valid_for_store_merging): New function.
68 (handled_load): New function.
69 (pass_store_merging::process_store): New method.
70 (pass_store_merging::execute): Use process_store method. Adjust
71 terminate_all_aliasing_chains caller.
73 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
75 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
76 Return true for more constants, symbols and label references.
77 (aarch64_valid_floating_const): Remove unused function.
79 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
82 * config/aarch64/aarch64.c (aarch64_layout_frame):
83 Undo forcing of LR at bottom of frame.
85 2017-11-03 Jeff Law <law@redhat.com>
88 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
89 for int_registers_saved.
91 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
92 extracted from tree-ssa-dom.c.
93 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
94 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
95 (record_equivalences_from_incoming_edge): Add additional argument
96 to single_pred_edge_ignoring_loop_edges call.
97 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
98 (uncprop_dom_walker::before_dom_children): Add additional argument
99 to single_pred_edge_ignoring_loop_edges call.
100 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
101 single_pred_edge_ignoring_loop_edges rather than open coding.
102 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
104 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
106 * match.pd (-(-A)): Rewrite.
108 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
110 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
111 (rs6000_emit_int_cmove): New declaration.
112 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
113 (rs6000_emit_sISEL): Delete.
114 (rs6000_emit_int_cmove): Make non-static.
115 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
116 instead of rs6000_emit_sISEL.
118 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
120 * asan.c (create_cond_insert_point): Maintain profile.
121 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
123 * basic-block.h (struct basic_block_def): Remove frequency.
124 (EDGE_FREQUENCY): Use to_frequency
125 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
127 (find_traces): Update to use to_frequency.
128 (find_traces_1_round): Likewise; use only IPA counts.
129 (bb_to_key): Likewise.
130 (connect_traces): Use IPA counts only.
131 (copy_bb_p): Update to use to_frequency.
132 (fix_up_crossing_landing_pad): Likewise.
133 (sanitize_hot_paths): Likewise.
134 * bt-load.c (basic_block_freq): Likewise.
135 * cfg.c (init_flow): Set count_max to uninitialized.
136 (check_bb_profile): Remove frequencies; check counts.
137 (dump_bb_info): Do not dump frequencies.
138 (update_bb_profile_for_threading): Update counts only.
139 (scale_bbs_frequencies_int): Likewise.
140 (MAX_SAFE_MULTIPLIER): Remove.
141 (scale_bbs_frequencies_gcov_type): Update counts only.
142 (scale_bbs_frequencies_profile_count): Update counts only.
143 (scale_bbs_frequencies): Update counts only.
144 * cfg.h (struct control_flow_graph): Add count-max.
145 (update_bb_profile_for_threading): Update prototype.
146 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
147 (find_many_sub_basic_blocks): Likewise.
148 * cfgcleanup.c (try_forward_edges): Likewise.
149 (try_crossjump_to_edge): Likewise.
150 * cfgexpand.c (expand_gimple_cond): Likewise.
151 (expand_gimple_tailcall): Likewise.
152 (construct_init_block): Likewise.
153 (construct_exit_block): Likewise.
154 * cfghooks.c (verify_flow_info): Check consistency of counts.
155 (dump_bb_for_graph): Do not dump frequencies.
156 (split_block_1): Do not update frequencies.
157 (split_edge): Do not update frequencies.
158 (make_forwarder_block): Do not update frequencies.
159 (duplicate_block): Do not update frequencies.
160 (account_profile_record): Do not update frequencies.
161 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
162 for global heuristics.
163 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
164 (expected_loop_iterations_unbounded): Use counts only.
165 * cfgloopmanip.c (scale_loop_profile): Simplify.
166 (create_empty_loop_on_edge): Simplify
168 (duplicate_loop_to_header_edge): Simplify
169 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
170 (update_br_prob_note): Take care of removing note when profile
172 (relink_block_chain): Do not dump frequency.
173 (rtl_account_profile_record): Use to_frequency.
174 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
175 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
176 (cgraph_update_edges_for_call_stmt_node): Likewise.
177 (cgraph_edge::verify_count_and_frequency): Update.
178 (cgraph_node::verify_node): Temporarily disable frequency verification.
179 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
181 (cgraph_edge::rebuild_edges): Convert to ipa counts.
182 * cgraphunit.c (init_lowered_empty_function): Do not initialize
184 (cgraph_node::expand_thunk): Update profile.
185 * except.c (dw2_build_landing_pads): Do not update frequency.
186 * final.c (compute_alignments): Use to_frequency.
187 (dump_basic_block_info): Do not dump frequency.
188 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
189 (dump_gimple_bb_header): Do not dump frequency.
190 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
192 * gimple-streamer-in.c (input_bb): Do not stream frequency.
193 * gimple-streamer-out.c (output_bb): Do not stream frequency.
194 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
195 (init_before_recovery): Do not update frequency.
196 (sched_create_recovery_edges): Do not update frequency.
197 * hsa-gen.c (convert_switch_statements): Do not update frequency.
198 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
199 (ipa_cp_c_finalize): Set max_count to uninitialized.
200 * ipa-fnsummary.c (get_minimal_bb): Use counts.
201 (param_change_prob): Use counts.
202 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
204 * ipa-split.c (consider_split): Use to_frequency.
205 (split_function): Use to_frequency.
206 * ira-build.c (loop_compare_func): Likewise.
207 (mark_loops_for_removal): Likewise.
208 (mark_all_loops_for_removal): Likewise.
209 * loop-doloop.c (doloop_modify): Do not update frequency.
210 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
212 * lto-streamer-in.c (input_function): Update count_max.
213 * omp-expand.c (expand_omp_taskreg): Update count_max.
214 * omp-simd-clone.c (simd_clone_adjust): Update profile.
215 * predict.c (maybe_hot_frequency_p): Use to_frequency.
216 (maybe_hot_count_p): Use ipa counts only.
217 (maybe_hot_bb_p): Simplify.
218 (maybe_hot_edge_p): Simplify.
219 (probably_never_executed): Do not take frequency argument.
220 (probably_never_executed_bb_p): Do not pass frequency.
221 (probably_never_executed_edge_p): Likewise.
222 (combine_predictions_for_bb): Check that profile is nonzero.
223 (propagate_freq): Do not set frequency.
224 (drop_profile): Simplify.
225 (counts_to_freqs): Simplify.
226 (expensive_function_p): Use to_frequency.
227 (propagate_unlikely_bbs_forward): Simplify.
228 (determine_unlikely_bbs): Simplify.
229 (estimate_bb_frequencies): Add hack to silence graphite issues.
230 (compute_function_frequency): Use ipa counts.
231 (pass_profile::execute): Update.
232 (rebuild_frequencies): Use counts only.
233 (force_edge_cold): Use counts only.
234 * profile-count.c (profile_count::dump): Dump new count types.
235 (profile_count::differs_from_p): Check compatiblity.
236 (profile_count::to_frequency): New function.
237 (profile_count::to_cgraph_frequency): New function.
238 * profile-count.h (struct function): Declare.
239 (enum profile_quality): Add profile_guessed_local and
240 profile_guessed_global0.
241 (class profile_proability): Decrease number of bits to 29;
242 update from_reg_br_prob_note and to_reg_br_prob_note.
243 (class profile_count: Update comment; decrease number of bits
244 to 61. Check compatibility.
245 (profile_count::compatible_p): New private member function.
246 (profile_count::ipa_p): New member function.
247 (profile_count::operator<): Handle global zero correctly.
248 (profile_count::operator>): Handle global zero correctly.
249 (profile_count::operator<=): Handle global zero correctly.
250 (profile_count::operator>=): Handle global zero correctly.
251 (profile_count::nonzero_p): New member function.
252 (profile_count::force_nonzero): New member function.
253 (profile_count::max): New member function.
254 (profile_count::apply_scale): Handle IPA scalling.
255 (profile_count::guessed_local): New member function.
256 (profile_count::global0): New member function.
257 (profile_count::ipa): New member function.
258 (profile_count::to_frequency): Declare.
259 (profile_count::to_cgraph_frequency): Declare.
260 * profile.c (OVERLAP_BASE): Delete.
261 (compute_frequency_overlap): Delete.
262 (compute_branch_probabilities): Do not use compute_frequency_overlap.
263 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
264 * sched-ebb.c (rank): Use counts only.
265 * shrink-wrap.c (handle_simple_exit): Use counts only.
266 (try_shrink_wrapping): Use counts only.
267 (place_prologue_for_one_component): Use counts only.
268 * tracer.c (find_best_predecessor): Use to_frequency.
269 (find_trace): Use to_frequency.
270 (tail_duplicate): Use to_frequency.
271 * trans-mem.c (expand_transaction): Do not update frequency.
272 * tree-call-cdce.c: Do not update frequency.
273 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
274 (gimple_merge_blocks): Likewise.
275 (gimple_split_edge): Likewise.
276 (gimple_duplicate_sese_region): Likewise.
277 (gimple_duplicate_sese_tail): Likewise.
278 (move_sese_region_to_fn): Likewise.
279 (gimple_account_profile_record): Likewise.
280 (insert_cond_bb): Likewise.
281 * tree-complex.c (expand_complex_div_wide): Likewise.
282 * tree-eh.c (lower_resx): Update profile.
283 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
285 (initialize_cfun): Do not initialize frequencies
286 (freqs_to_counts): Delete.
287 (copy_cfg_body): Ignore count parameter.
289 (expand_call_inline): Update count_max.
290 (optimize_inline_calls): Update count_max.
291 (tree_function_versioning): Update count_max.
292 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
293 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
295 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
296 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
297 (try_peel_loop): Likewise.
298 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
300 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
301 (tree_transform_and_unroll_loop): Do not use frequencies
302 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
303 Use reliable prediction only.
304 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
305 * tree-ssa-sink.c (select_best_block): Use to_frequency.
306 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
308 * tree-ssa-threadupdate.c (create_block_for_threading): Do
310 (any_remaining_duplicated_blocks): Likewise.
311 (update_profile): Likewise.
312 (estimated_freqs_path): Delete.
313 (freqs_to_counts_path): Delete.
314 (clear_counts_path): Delete.
315 (ssa_fix_duplicate_block_edges): Likewise.
316 (duplicate_thread_path): Likewise.
317 * tree-switch-conversion.c (gen_inbound_check): Use counts.
318 * tree-tailcall.c (decrease_profile): Do not update frequency.
319 (eliminate_tail_call): Likewise.
320 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
321 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
322 (optimize_mask_stores): Likewise.
323 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
324 * ubsan.c (ubsan_expand_null_ifn): Update profile.
325 (ubsan_expand_ptr_ifn): Update profile.
326 * value-prof.c (gimple_ic): Simplify.
327 * value-prof.h (gimple_ic): Update prototype.
328 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
329 * ipa-inline.c (compute_uninlined_call_time): Be sure that
331 (want_inline_self_recursive_call_p): Likewise.
332 (resolve_noninline_speculation): Only cummulate defined counts.
333 (inline_small_functions): Use nonzero_p.
334 (ipa_inline): Do not access freed node.
336 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
338 * config/aarch64/aarch64.c (aarch64_override_options_internal):
339 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
341 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
343 * config/riscv/riscv.c (riscv_legitimize_move): Handle
344 non-legitimate address.
346 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
348 * config/rs6000/rs6000.md (*lt0_disi): Delete.
349 (*lt0_<mode>di, *lt0_<mode>si): New.
351 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
353 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
356 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
357 Jim Wilson <jim.wilson@linaro.org>
359 * config/aarch64/aarch64-cores.def (saphira): New CPU.
360 * config/aarch64/aarch64-tune.md: Regenerated.
361 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
362 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
364 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
366 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
367 (arc_expand_prologue): Restore blink for millicode.
368 * config/arc/linux.h (LINK_EH_SPEC): Defined.
370 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
373 * config/i386/i386.c (ix86_vector_duplicate_value): Use
374 gen_vec_duplicate after forcing the scalar into a register.
376 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
378 * combine (try_combine): Print the insns input to try_combine to the
381 2017-11-02 Steve Ellcey <sellcey@cavium.com>
384 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
385 Remove second argument from aarch64_process_target_attr call.
386 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
388 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
390 (aarch64_handle_attr_arch): Remove second argument.
391 (aarch64_handle_attr_cpu): Ditto.
392 (aarch64_handle_attr_tune): Ditto.
393 (aarch64_handle_attr_isa_flags): Ditto.
394 (aarch64_process_one_target_attr): Ditto.
395 (aarch64_process_target_attr): Ditto.
396 (aarch64_option_valid_attribute_p): Remove second argument.
397 on aarch64_process_target_attr call.
399 2017-11-02 David Malcolm <dmalcolm@redhat.com>
401 * diagnostic.c: Include "selftest-diagnostic.h".
402 (selftest::assert_location_text): New function.
403 (selftest::test_diagnostic_get_location_text): New function.
404 (selftest::diagnostic_c_tests): Call it.
406 2017-11-02 David Malcolm <dmalcolm@redhat.com>
408 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
409 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
410 (class selftest::test_diagnostic_context): Move to...
411 * selftest-diagnostic.c: New file.
412 * selftest-diagnostic.h: New file.
414 2017-11-02 James Bowman <james.bowman@ftdichip.com>
416 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
417 offset range for FT32B.
418 * config/ft32/ft32.h: option "mcompress" enables relaxation.
419 * config/ft32/ft32.md: Add TARGET_NOPM.
420 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
421 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
423 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
425 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
427 2017-11-02 Jeff Law <law@redhat.com>
429 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
430 virtual keyword on FINAL OVERRIDE members.
432 * tree-ssa-propagate.h (ssa_propagation_engine): Group
433 virtuals together. Add virtual destructor.
434 (substitute_and_fold_engine): Similarly.
436 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
438 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
440 2017-11-02 Richard Biener <rguenther@suse.de>
442 PR tree-optimization/82795
443 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
445 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
447 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
449 (gcc_GAS_CHECK_FEATURE): Remove.
450 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
451 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
452 * configure: Regenerate.
454 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
456 * config/arc/arc.c (hwloop_optimize): Account for empty
459 2017-11-02 Richard Biener <rguenther@suse.de>
462 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
463 Truncate ARRAY_REF index and element size.
465 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
467 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
469 2017-11-01 Jeff Law <law@redhat.com>
471 * tree-ssa-ccp.c (ccp_folder): New class derived from
472 substitute_and_fold_engine.
473 (ccp_folder::get_value): New member function.
474 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
475 (ccp_fold_stmt): Remove prototype.
476 (ccp_finalize): Call substitute_and_fold from the ccp_class.
477 * tree-ssa-copy.c (copy_folder): New class derived from
478 substitute_and_fold_engine.
479 (copy_folder::get_value): Renamed from get_value.
480 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
481 * tree-vrp.c (vrp_folder): New class derived from
482 substitute_and_fold_engine.
483 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
484 (vrp_folder::get_value): New member function.
485 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
486 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
487 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
488 provide a class interface to folder/substitute routines.
489 (ssa_prop_fold_stmt_fn): Remove typedef.
490 (ssa_prop_get_value_fn): Likewise.
491 (subsitute_and_fold): Remove prototype.
492 (replace_uses_in): Likewise.
493 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
494 Renamed from replace_uses_in. Call the virtual member function
495 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
496 (substitute_and_fold_dom_walker): Remove initialization of
497 data member entries for calbacks. Add substitute_and_fold_engine
498 member and initialize it.
499 (substitute_and_fold_dom_walker::before_dom_children0: Use the
500 member functions for get_value, replace_phi_args_in c
501 replace_uses_in, and fold_stmt calls.
502 (substitute_and_fold_engine::substitute_and_fold): Renamed from
503 substitute_and_fold. Remove assert. Update ctor call.
505 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
506 (ssa_prop_visit_phi_fn): Likewise.
507 (class ssa_propagation_engine): New class to provide an interface
509 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
511 (ssa_prop_visit_phi): Likewise.
512 (ssa_propagation_engine::simulate_stmt): Moved into class.
513 Call visit_phi/visit_stmt from the class rather than via
514 file scoped static variables.
515 (ssa_propagation_engine::simulate_block): Moved into class.
516 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
517 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
518 set file scoped statics for the visit_stmt/visit_phi callbacks.
519 * tree-complex.c (complex_propagate): New class derived from
520 ssa_propagation_engine.
521 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
522 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
523 (tree_lower_complex): Call ssa_propagate via the complex_propagate
525 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
526 ssa_propagation_engine.
527 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
528 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
529 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
530 * tree-ssa-copy.c (copy_prop): New class derived from
531 ssa_propagation_engine.
532 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
533 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
534 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
535 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
536 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
537 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
538 (execute_vrp): Call ssa_propagate from the vrp_prop class.
540 2017-11-01 Jakub Jelinek <jakub@redhat.com>
542 PR rtl-optimization/82778
543 PR rtl-optimization/82597
544 * compare-elim.c (struct comparison): Add in_a_setter field.
545 (find_comparison_dom_walker::before_dom_children): Remove killed
546 bitmap and df_simulate_find_defs call, instead walk the defs.
547 Compute last_setter and initialize in_a_setter. Merge definitions
548 with first initialization for a few variables.
549 (try_validate_parallel): Use insn_invalid_p instead of
550 recog_memoized. Return insn rather than just the pattern.
551 (try_merge_compare): Fix up comment. Don't uselessly test if
552 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
554 (execute_compare_elim_after_reload): Remove df_chain_add_problem
557 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
558 Alan Hayward <alan.hayward@arm.com>
559 David Sherwood <david.sherwood@arm.com>
561 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
562 aarch64_hard_regno_nregs to get the number of registers
565 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
566 Alan Hayward <alan.hayward@arm.com>
567 David Sherwood <david.sherwood@arm.com>
569 * config/aarch64/constraints.md (Upl): Rename to...
571 * config/aarch64/aarch64.md
572 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
575 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
576 Alan Hayward <alan.hayward@arm.com>
577 David Sherwood <david.sherwood@arm.com>
579 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
580 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
583 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
584 Alan Hayward <alan.hayward@arm.com>
585 David Sherwood <david.sherwood@arm.com>
587 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
588 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
589 (aarch64_evpc_dup): Generate rtl direcly, rather than using
591 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
593 * config/aarch64/iterators.md: Add a comment above the permute
594 unspecs to say that they are generated directly by
595 aarch64_expand_vec_perm_const.
596 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
598 2017-11-01 Nathan Sidwell <nathan@acm.org>
600 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
602 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
604 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
605 and medany code models, and describe what they do.
607 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
609 Revert accidental duplicate:
611 * combine.c (can_change_dest_mode): Reject changes in
612 REGMODE_NATURAL_SIZE.
614 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
616 PR rtl-optimization/64682
617 PR rtl-optimization/69567
618 PR rtl-optimization/69737
619 PR rtl-optimization/82683
620 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
621 register mentioned in the note, drop the note, unless it came from I3,
622 in which case it should go to I3 again.
624 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
626 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
627 and return false if not.
628 (clear_bytes_written_by, live_bytes_read): Update accordingly.
630 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
632 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
633 range is known to be empty.
635 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
636 Alan Hayward <alan.hayward@arm.com>
637 David Sherwood <david.sherwood@arm.com>
639 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
640 and CONST_VECTOR_NUNITS instead of computing the number of units from
641 the byte sizes of the vector and element.
642 (simplify_binary_operation_1): Likewise.
643 (simplify_const_binary_operation): Likewise.
644 (simplify_ternary_operation): Likewise.
646 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
647 Alan Hayward <alan.hayward@arm.com>
648 David Sherwood <david.sherwood@arm.com>
650 * var-tracking.c (INT_MEM_OFFSET): Replace with...
651 (int_mem_offset): ...this new function.
652 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
653 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
654 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
657 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
658 Alan Hayward <alan.hayward@arm.com>
659 David Sherwood <david.sherwood@arm.com>
661 * lower-subreg.c (interesting_mode_p): New function.
662 (compute_costs, find_decomposable_subregs, decompose_register)
663 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
664 (resolve_clobber, dump_choices): Use it.
666 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
667 Alan Hayward <alan.hayward@arm.com>
668 David Sherwood <david.sherwood@arm.com>
670 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
672 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
673 Alan Hayward <alan.hayward@arm.com>
674 David Sherwood <david.sherwood@arm.com>
676 * alias.c (find_base_value, find_base_term): Only process integer
677 truncations. Check the precision rather than the size.
679 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
680 Alan Hayward <alan.hayward@arm.com>
681 David Sherwood <david.sherwood@arm.com>
683 * machmode.h (is_narrower_int_mode): New function
684 * optabs.c (expand_float, expand_fix): Use it.
685 * dwarf2out.c (rotate_loc_descriptor): Likewise.
687 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
688 Alan Hayward <alan.hayward@arm.com>
689 David Sherwood <david.sherwood@arm.com>
691 * rtl.h (narrower_subreg_mode): New function.
692 * ira-color.c (update_costs_from_allocno): Use it.
694 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
695 Alan Hayward <alan.hayward@arm.com>
696 David Sherwood <david.sherwood@arm.com>
698 * optabs-query.h (convert_optab_p): New function, split out from...
699 (convert_optab_handler): ...here.
700 (widening_optab_handler): Delete.
701 (find_widening_optab_handler): Remove permit_non_widening parameter.
702 (find_widening_optab_handler_and_mode): Likewise. Provide an
703 override that operates on mode class wrappers.
704 * optabs-query.c (widening_optab_handler): Delete.
705 (find_widening_optab_handler_and_mode): Remove permit_non_widening
706 parameter. Assert that the two modes are the same class and that
707 the "from" mode is narrower than the "to" mode. Use
708 convert_optab_handler instead of widening_optab_handler.
709 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
710 instead of widening_optab_handler.
711 * expr.c (expand_expr_real_2): Update calls to
712 find_widening_optab_handler.
713 * optabs.c (expand_widen_pattern_expr): Likewise.
714 (expand_binop_directly): Take the insn_code as a parameter.
715 (expand_binop): Only call find_widening_optab_handler for
716 conversion optabs; use optab_handler otherwise. Update calls
717 to find_widening_optab_handler and expand_binop_directly.
718 Use convert_optab_handler instead of widening_optab_handler.
719 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
720 find_widening_optab_handler and use scalar_mode rather than
722 (convert_plusminus_to_widen): Likewise.
724 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
725 Alan Hayward <alan.hayward@arm.com>
726 David Sherwood <david.sherwood@arm.com>
728 * machmode.h (fixed_size_mode): New class.
729 * rtl.h (get_pool_mode): Return fixed_size_mode.
730 * gengtype.c (main): Add fixed_size_mode.
731 * target.def (get_raw_result_mode): Return a fixed_size_mode.
732 (get_raw_arg_mode): Likewise.
733 * doc/tm.texi: Regenerate.
734 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
735 * targhooks.c (default_get_reg_raw_mode): Likewise.
736 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
737 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
738 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
739 (msp430_get_raw_result_mode): Likewise.
740 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
741 * dbxout.c (dbxout_parms): Require fixed-size modes.
742 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
743 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
744 * omp-low.c (lower_oacc_reductions): Likewise.
745 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
746 (simplify_subreg): Update accordingly.
747 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
748 (force_const_mem): Update accordingly. Return NULL_RTX for modes
749 that aren't fixed-size.
750 (get_pool_mode): Return a fixed_size_mode.
751 (output_constant_pool_2): Take a fixed_size_mode.
753 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
754 Alan Hayward <alan.hayward@arm.com>
755 David Sherwood <david.sherwood@arm.com>
757 * doc/rtl.texi (vec_series): Document.
758 (const): Say that the operand can be a vec_series.
759 * rtl.def (VEC_SERIES): New rtx code.
760 * rtl.h (const_vec_series_p_1): Declare.
761 (const_vec_series_p): New function.
762 * emit-rtl.h (gen_const_vec_series): Declare.
763 (gen_vec_series): Likewise.
764 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
765 (gen_vec_series): Likewise.
766 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
767 * simplify-rtx.c (simplify_unary_operation): Handle negations
769 (simplify_binary_operation_series): New function.
770 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
771 (test_vector_ops_series): New function.
772 (test_vector_ops): Call it.
773 * config/powerpcspe/altivec.md (altivec_lvsl): Use
774 gen_const_vec_series.
775 (altivec_lvsr): Likewise.
776 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
778 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
779 Alan Hayward <alan.hayward@arm.com>
780 David Sherwood <david.sherwood@arm.com>
782 * doc/rtl.texi (const): Update description of address constants.
783 Say that vector constants are allowed too.
784 * common.md (E, F): Use CONSTANT_P instead of checking for
786 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
787 checking for CONST_VECTOR.
788 * expmed.c (make_tree): Use build_vector_from_val for a CONST
790 * expr.c (expand_expr_real_2): Check for vector modes instead
791 of checking for CONST_VECTOR.
792 * rtl.h (const_vec_p): New function.
793 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
794 (unwrap_const_vec_duplicate): Handle them here too.
796 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
797 David Malcolm <dmalcolm@redhat.com>
798 Alan Hayward <alan.hayward@arm.com>
799 David Sherwood <david.sherwood@arm.com>
801 * rtl.h (vec_duplicate_p): New function.
802 * selftest-rtl.c (assert_rtx_eq_at): New function.
803 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
804 (assert_rtx_eq_at): Declare.
805 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
806 * selftest-run-tests.c (selftest::run_tests): Call it.
807 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
808 (simplify_unary_operation_1): Recursively handle vector duplicates.
809 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
811 (simplify_subreg): Handle subregs of vector duplicates.
812 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
813 (selftest::simplify_rtx_c_tests): New functions.
815 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
816 Alan Hayward <alan.hayward@arm.com>
817 David Sherwood <david.sherwood@arm.com>
819 * emit-rtl.h (gen_const_vec_duplicate): Declare.
820 (gen_vec_duplicate): Likewise.
821 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
823 (gen_const_vector): ...here.
824 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
825 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
826 whose elements are all equal.
827 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
828 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
829 (simplify_relational_operation): Likewise.
830 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
832 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
833 (aarch64_expand_vector_init): Likewise.
834 * config/arm/arm.c (neon_vdup_constant): Likewise.
835 (neon_expand_vector_init): Likewise.
836 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
837 (arm_block_set_unaligned_vect): Likewise.
838 (arm_block_set_aligned_vect): Likewise.
839 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
840 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
841 (expand_vec_perm_even_odd_pack): Likewise.
842 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
843 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
844 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
845 gen_const_vec_duplicate.
846 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
847 * config/mips/mips.c (mips_gen_const_int_vector): Use
848 gen_const_vec_duplicate.
849 (mips_expand_vector_init): Use CONST0_RTX.
850 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
851 (define_split): Use gen_const_vec_duplicate.
852 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
853 (define_split): Use gen_const_vec_duplicate.
854 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
855 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
856 * config/spu/spu.c (spu_const): Likewise.
858 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
859 Alan Hayward <alan.hayward@arm.com>
860 David Sherwood <david.sherwood@arm.com>
862 * combine.c (can_change_dest_mode): Reject changes in
863 REGMODE_NATURAL_SIZE.
865 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
867 * configure.ac (--enable-libssp): New.
868 (gcc_cv_libc_provides_ssp): Check for explicit setting before
869 trying to determine target-specific default. Adjust indentation.
870 * configure: Regenerated.
871 * doc/install.texi (Configuration): Expand --disable-libssp
874 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
876 config/i386/i386.c (ix86_expand_epilogue): Correct stack
879 2017-10-31 Martin Jambor <mjambor@suse.cz>
882 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
884 2017-10-31 David Malcolm <dmalcolm@redhat.com>
886 * auto-profile.c (autofdo_source_profile::read): Use
887 UNKNOWN_LOCATION rather than 0.
888 * diagnostic-core.h (warning_at_rich_loc): Rename to...
889 (warning_at): ...this overload.
890 (warning_at_rich_loc_n): Rename to...
891 (warning_n): ...this overload.
892 (error_at_rich_loc): Rename to...
893 (error_at): ...this overload.
894 (pedwarn_at_rich_loc): Rename to...
895 (pedwarn): ...this overload.
896 (permerror_at_rich_loc): Rename to...
897 (permerror): ...this overload.
898 (inform_at_rich_loc): Rename to...
899 (inform): ...this overload.
900 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
901 (diagnostic_n_impl_richloc): Rename to...
902 (diagnostic_n_impl): ...this rich_location *-based decl.
903 (inform_at_rich_loc): Rename to...
904 (inform): ...this, and add an assertion.
905 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
906 (warning_at_rich_loc): Rename to...
907 (warning_at): ...this, and add an assertion.
908 (warning_at_rich_loc_n): Rename to...
909 (warning_n): ...this, and add an assertion.
910 (warning_n): Update location_t-based implementation for removal of
911 location_t-based diagnostic_n_impl.
912 (pedwarn_at_rich_loc): Rename to...
913 (pedwarn): ...this, and add an assertion.
914 (permerror_at_rich_loc): Rename to...
915 (permerror): ...this, and add an assertion.
916 (error_n): Update for removal of location_t-based diagnostic_n_impl.
917 (error_at_rich_loc): Rename to...
918 (error_at): ...this, and add an assertion.
919 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
920 (driver::do_spec_on_infiles): Likewise.
921 * substring-locations.c (format_warning_va): Update for renaming
922 of inform_at_rich_loc.
924 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
926 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
927 _Float<N>X built-in functions so that the variant without the
928 "__builtin_" prefix is only enabled for the GNU C and Objective C
929 languages when they are in non-strict ANSI/ISO mode.
930 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
931 * target.def (floatn_builtin_p): Add a target hook to control
932 whether _Float<N> and _Float<N>X built-in functions without the
933 "__builtin_" prefix are enabled, and return true for C and
934 Objective C in the default hook. Include langhooks.h in
936 * targhooks.h (default_floatn_builtin_p): Likewise.
937 * targhooks.c (default_floatn_builtin_p): Likewise.
938 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
939 floatn_builtin_p target hook.
940 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
942 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
943 Eric Botcazou <ebotcazou@adacore.com>
945 PR rtl-optimization/81803
946 * lra-constraints.c (curr_insn_transform): Also reload the whole
947 register for a strict subreg no wider than a word if this is for
948 a WORD_REGISTER_OPERATIONS target.
950 2017-10-31 Jason Merrill <jason@redhat.com>
952 * gdbinit.in: Skip over inlines from timevar.h.
954 2017-10-31 Martin Liska <mliska@suse.cz>
956 * doc/gcov.texi: Document new option.
957 * gcov.c (print_usage): Likewise print it.
958 (process_args): Support the argument.
959 (format_count): New function.
960 (format_gcov): Use the function.
962 2017-10-31 Martin Liska <mliska@suse.cz>
964 * gcov.c (struct name_map): do not use typedef.
965 Define operator== and operator<.
966 (name_search): Remove.
968 (main): Do not allocate names.
969 (process_file): Add vertical space.
970 (generate_results): Use std::find.
971 (release_structures): Do not release memory.
972 (find_source): Use std::find.
974 2017-10-31 Martin Liska <mliska@suse.cz>
976 * gcov.c (struct line_info): Remove it's typedef.
977 (line_info::line_info): Add proper ctor.
978 (line_info::has_block): Do not use a typedef.
979 (struct source_info): Do not use typedef.
981 (get_cycles_count): Likewise.
982 (output_intermediate_file): Iterate via vector iterator.
983 (add_line_counts): Use std::vector methods.
984 (accumulate_line_counts): Likewise.
985 (output_lines): Likewise.
987 2017-10-31 Martin Liska <mliska@suse.cz>
989 * gcov.c (struct source_info): Remove typedef.
990 (source_info::source_info): Add proper ctor.
991 (accumulate_line_counts): Use struct, not it's typedef.
992 (output_gcov_file): Likewise.
993 (output_lines): Likewise.
994 (main): Do not allocate an array.
995 (output_intermediate_file): Use size of vector container.
996 (process_file): Resize the vector.
997 (generate_results): Do not preallocate, use newly added vector
999 (release_structures): Do not release sources.
1000 (find_source): Use vector methods.
1001 (add_line_counts): Do not use typedef.
1003 2017-10-31 Martin Liska <mliska@suse.cz>
1005 * doc/gcov.texi: Document that.
1006 * gcov.c (add_line_counts): Mark lines with a non-executed
1008 (output_line_beginning): Handle such lines.
1009 (output_lines): Pass new argument.
1010 (output_intermediate_file): Print it in intermediate format.
1012 2017-10-31 Martin Liska <mliska@suse.cz>
1014 * color-macros.h: New file.
1015 * diagnostic-color.c: Factor out color related to macros to
1017 * doc/gcov.texi: Document -k option.
1018 * gcov.c (INCLUDE_STRING): Include string.h.
1019 (print_usage): Add -k option.
1020 (process_args): Parse it.
1021 (pad_count_string): New function.
1022 (output_line_beginning): Likewise.
1023 (DEFAULT_LINE_START): New macro.
1024 (output_lines): Support color output.
1026 2017-10-31 Martin Liska <mliska@suse.cz>
1028 PR gcov-profile/82633
1029 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
1030 their interaction with GCOV infrastructure.
1031 * configure.ac: Add -fkeep-{inline,static}-functions to
1033 * configure: Regenerate.
1035 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
1038 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
1040 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
1043 * config/rs6000/rs6000.md (allocate_stack): Force update interval
1044 into a register if it does not fit into an immediate offset field.
1046 2017-10-31 Olivier Hainque <hainque@adacore.com>
1048 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
1050 2017-10-31 Julia Koval <julia.koval@intel.com>
1052 * config.gcc: Add gfniintrin.h.
1053 * config/i386/gfniintrin.h: New.
1054 * config/i386/i386-builtin-types.def
1055 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1056 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1057 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1058 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1059 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1060 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
1061 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
1062 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
1063 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
1064 V64QI_FTYPE_V64QI_V64QI_INT): New types.
1065 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1066 * config/i386/immintrin.h: Include gfniintrin.h.
1067 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
1069 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
1071 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
1073 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1075 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
1076 (arm_ashldi3_1bit): Remove pattern.
1077 (ashrdi3): Remove shift by 1 expansion.
1078 (arm_ashrdi3_1bit): Remove pattern.
1079 (lshrdi3): Remove shift by 1 expansion.
1080 (arm_lshrdi3_1bit): Remove pattern.
1081 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
1082 cost of ashldi3 by 1.
1083 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
1084 (<shift>di3_neon): Likewise.
1086 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
1088 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
1089 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
1090 and (*aarch64_simd_mov<VQ:mode>).
1091 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
1092 pattern alternative.
1093 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
1094 attributes to match pattern alternative.
1096 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
1098 * config.gcc (powerpc*-*-*): Add emmintrin.h.
1099 * config/rs6000/emmintrin.h: New file.
1100 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
1102 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1104 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
1105 * (movdi_vfp_cortexa8): Remove pattern.
1107 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1109 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
1111 (Specific, alpha*-dec-osf5.1): Remove.
1112 (Specific, mips-sgi-irix5): Remove.
1113 (Specific, mips-sgi-irix6): Remove.
1115 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1118 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
1119 arguments to clear_bit_region_be.
1121 2017-10-30 Jim Wilson <wilson@tuliptree.org>
1123 * gimplify.c: Include memmodel.h.
1125 2017-10-30 Martin Jambor <mjambor@suse.cz>
1127 * omp-grid.c (grid_attempt_target_gridification): Also insert a
1128 condition whether loop should be executed at all.
1130 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
1132 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1133 gimple folding of vec_madd() intrinsics.
1134 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
1135 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
1136 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
1138 2017-10-30 Richard Biener <rguenther@suse.de>
1140 PR tree-optimization/82762
1142 2017-10-23 Richard Biener <rguenther@suse.de>
1144 PR tree-optimization/82129
1146 2017-08-01 Richard Biener <rguenther@suse.de>
1148 PR tree-optimization/81181
1149 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
1150 (compute_antic): ... end of iteration here.
1152 2017-10-30 Joseph Myers <joseph@codesourcery.com>
1154 * doc/invoke.texi (C Dialect Options): Document -std=c17,
1155 -std=iso9899:2017 and -std=gnu17.
1156 * doc/standards.texi (C Language): Document C17 support.
1157 * doc/cpp.texi (Overview): Mention -std=c17.
1158 (Standard Predefined Macros): Document C11 and C17 values of
1159 __STDC_VERSION__. Do not refer to C99 support as incomplete.
1160 * doc/extend.texi (Inline): Do not list individual options for
1161 standards newer than C99.
1162 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
1164 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
1167 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
1169 * asan.c (asan_finish_file): Align asan globals array by shadow
1172 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1175 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
1176 (struct store_immediate_info): Add bitregion_start and bitregion_end
1178 (store_immediate_info::store_immediate_info): Add brs and bre
1179 arguments and initialize bitregion_{start,end} from those.
1180 (struct merged_store_group): Add bitregion_start, bitregion_end,
1181 align_base and mask fields. Drop unnecessary struct keyword from
1182 struct store_immediate_info. Add do_merge method.
1183 (clear_bit_region_be): Use memset instead of loop storing zeros.
1184 (merged_store_group::do_merge): New method.
1185 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
1186 stores as long as the surrounding bitregions have no gaps.
1187 (merged_store_group::merge_overlapping): Use do_merge.
1188 (merged_store_group::apply_stores): Test that bitregion_{start,end}
1189 is byte aligned, rather than requiring that start and width are
1190 byte aligned. Drop unnecessary struct keyword from
1191 struct store_immediate_info. Allocate and populate also mask array.
1192 Make start of the arrays relative to bitregion_start rather than
1193 start and size them according to bitregion_{end,start} difference.
1194 (struct imm_store_chain_info): Drop unnecessary struct keyword from
1195 struct store_immediate_info.
1196 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
1197 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
1198 struct keyword from struct store_immediate_info.
1199 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
1200 between stores as long as the surrounding bitregions have no gaps.
1202 (struct split_store): Add orig non-static data member.
1203 (split_store::split_store): Initialize orig to false.
1204 (find_constituent_stmts): Return store_immediate_info *, non-NULL
1205 if there is exactly a single original stmt. Change stmts argument
1206 to pointer from reference, if NULL, don't push anything to it. Add
1207 first argument, use it to optimize skipping over orig stmts that
1208 are known to be before bitpos already. Simplify.
1209 (split_group): Return unsigned int count how many stores are or
1210 would be needed rather than a bool. Add allow_unaligned argument.
1211 Change split_stores argument from reference to pointer, if NULL,
1212 only do a dry run computing how many stores would be produced.
1213 Rewritten algorithm to use both alignment and misalign if
1214 !allow_unaligned and handle bitfield stores with gaps.
1215 (imm_store_chain_info::output_merged_store): Set start_byte_pos
1216 from bitregion_start instead of start. Compute allow_unaligned
1217 here, if true, do 2 split_group dry runs to compute which one
1218 produces fewer stores and prefer aligned if equal. Punt if
1219 new count is bigger or equal than original before emitting any
1220 statements, rather than during that. Remove no longer needed
1221 new_ssa_names tracking. Replace num_stmts with
1222 split_stores.length (). Use 32-bit stack allocated entries
1223 in split_stores auto_vec. Try to reuse original store lhs/rhs1
1224 if possible. Handle bitfields with gaps.
1225 (pass_store_merging::execute): Ignore bitsize == 0 stores.
1226 Compute bitregion_{start,end} for the stores and construct
1227 store_immediate_info with that. Formatting fixes.
1229 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
1232 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
1233 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
1235 2017-10-29 Jim Wilson <wilson@tuliptree.org>
1237 * gimplify.c: Include tm_p.h.
1239 * common.opt (gcoff): Re-add as ignored option.
1240 (gcoff1, gcoff2, gcoff3): Likewise.
1242 * Makefile.in (OBJS): Delete sdbout.o.
1243 (GTFILES): Delete $(srcdir)/sdbout.c.
1244 * debug.h: Delete sdb_debug_hooks.
1245 * final.c: Delete sdbout.h include.
1246 (final_scan_insn): Delete SDB_DEBUG check.
1247 (rest_of_clean_state): Likewise.
1248 * output.h: Delete sdb_begin_function_line.
1251 * toplev.c: Delete sdbout.h include.
1252 (process_options): Delete SDB_DEBUG check.
1253 * tree-core.h (tree_type_common): Delete pointer field of
1255 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
1256 TYPE_SYMTAB_POINTER.
1257 * tree.h (TYPE_SYMTAB_POINTER): Delete.
1258 (TYPE_SYMTAB_IS_POINTER): Delete.
1259 (TYPE_SYMTAB_IS_DIE): Renumber.
1260 * xcoffout.c: Refer to former sdbout.c file.
1261 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
1263 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
1264 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
1265 (Debugging Options): Delete -gcoff.
1266 (-gstabs): Delete SDB reference.
1268 (-gcoff@var{level}): Delete.
1269 * doc/passes.texi (Debugging information output): Delete SDB and
1270 sdbout.c references.
1271 * doc/tm.texi: Regenerate.
1272 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
1273 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
1275 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
1276 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
1277 (SDB_DEBUGGING_INFO): Delete.
1278 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
1279 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
1280 * target.def (output_source_filename): Delete COFF reference.
1282 * common.opt (gcoff): Delete.
1283 (gxcoff+): Update Negative chain.
1284 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
1286 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
1287 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
1288 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
1290 (expand_function_start): Change sdb reference to past tense.
1291 (expand_function_end): Change sdb reference to past tense.
1292 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
1293 * opts.c (debug_type_names): Delete coff entry.
1294 (common_handle_option): Delete OPT_gcoff case.
1295 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
1297 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
1298 * config/cris/cris.h: Delete SDB reference in comment.
1299 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
1300 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
1301 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
1302 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
1304 (ix86_expand_prologue): Likewise.
1305 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
1306 * config/ia64/ia64.h: Likewise.
1307 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
1308 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
1310 * config/mmix/mmix.h: Likewise.
1311 * config/nds32/nds32.c: Likewise.
1312 * config/stormy/storym16.h: Likewise.
1313 * config/visium/visium.h: Likewise.
1314 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
1316 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
1318 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
1319 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
1320 FRAME_POINTER_REGNUM point at high end of local var area.
1322 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
1324 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
1325 Move comment around. Do not reset best_edge for a copiable
1326 destination if the copy would cause a partition change.
1327 (better_edge_p): Remove redundant check.
1329 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
1331 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
1333 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
1335 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
1336 for math functions that have _Float<N> and _Float<N>X variants.
1337 (mathfn_built_in_2): Add support for math functions that have
1338 _Float<N> and _Float<N>X variants.
1339 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
1340 (expand_builtin_mathfn_ternary): Add support for fma with
1341 _Float<N> and _Float<N>X variants.
1342 (expand_builtin): Likewise.
1343 (fold_builtin_3): Likewise.
1344 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
1345 create math function _Float<N> and _Float<N>X variants as external
1347 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
1348 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
1349 the __builtin_ prefix and if not strict ansi, without the prefix.
1350 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
1351 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
1352 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
1353 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
1354 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
1355 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
1356 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
1357 function signatures for fma _Float<N> and _Float<N>X variants.
1358 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
1359 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
1360 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
1361 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
1362 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
1363 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
1364 * gencfn-macros.c (print_case_cfn): Add support for math functions
1365 that have _Float<N> and _Float<N>X variants.
1366 (print_define_operator_list): Likewise.
1367 (fltfn_suffixes): Likewise.
1369 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
1370 for math functions that have _Float<N> and _Float<N>X variants.
1371 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
1372 and _Float<N>X variants.
1373 (COPYSIGN): Likewise.
1376 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
1377 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
1379 (integer_valued_read_call_p): Likewise.
1380 * fold-const-call.c (fold_const_call_ss): Likewise.
1381 (fold_const_call_sss): Add support for copysign, fmin, and fmax
1382 _Float<N> and _Float<N>X variants.
1383 (fold_const_call_ssss): Add support for fma _Float<N> and
1384 _Float<N>X variants.
1385 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
1386 support for copysign and fma _Float<N> and _Float<N>X variants.
1387 (backprop::process_builtin_call_use): Likewise.
1388 * tree-call-cdce.c (can_test_argument_range); Add support for
1389 sqrt _Float<N> and _Float<N>X variants.
1390 (edom_only_function): Likewise.
1391 (get_no_error_domain): Likewise.
1392 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
1393 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
1394 copysign _Float<N> and _Float<N>X variants.
1395 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
1396 handled by machine independent code.
1397 (FMAF128): Likewise.
1398 * doc/cpp.texi (Common Predefined Macros): Document defining
1399 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
1400 fma _Float<N> and _Float<N>X variants.
1402 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
1405 * config/i386/i386-modes.def (CCFPU): Remove definition.
1406 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
1407 (ix86_cc_modes_compatible): Ditto.
1408 (ix86_expand_carry_flag_compare): Ditto.
1409 (ix86_expand_int_movcc): Ditto.
1410 (ix86_expand_int_addcc): Ditto.
1411 (ix86_reverse_condition): Ditto.
1412 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
1413 Return true/false for unordered/ordered fp comparisons.
1414 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
1415 (ix86_prepare_fp_compare_args): Update for rename.
1416 (ix86_expand_fp_compare): Update for rename. Generate unordered
1417 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
1418 (ix86_expand_sse_compare_and_jump): Ditto.
1419 * config/i386/predicates.md (fcmov_comparison_operator):
1420 Remove CCFPU mode handling.
1421 (ix86_comparison_operator): Ditto.
1422 (ix86_carry_flag_operator): Ditto.
1423 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
1424 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
1425 (*cmpu<mode>_cc_i387): Ditto.
1426 (FPCMP): Remove mode iterator.
1427 (unord): Remove mode attribute.
1428 (unord_subst): New define_subst transformation
1429 (unord): New define_subst attribute.
1431 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
1432 (*cmpi<unord>xf_i387): Ditto.
1433 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
1434 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
1435 using unord_subst transformation.
1436 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
1437 (round_saeonly): Also handle CCFP mode.
1438 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
1439 Remove UNSPEC_SAHF unspec handling.
1441 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
1443 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
1445 2017-10-27 Jeff Law <law@redhat.com>
1447 * gimple-ssa-sprintf.c: Include domwalk.h.
1448 (class sprintf_dom_walker): New class, derived from dom_walker.
1449 (sprintf_dom_walker::before_dom_children): New function.
1450 (struct call_info): Moved into sprintf_dom_walker class
1451 (compute_formath_length, handle_gimple_call): Likewise.
1452 (sprintf_length::execute): Call the dominator walker rather
1453 than walking the statements.
1455 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
1456 gimple statement locations.
1457 (check_array_bounds): Corresponding changes. Get the statement's
1458 location directly from wi->stmt.
1460 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
1463 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
1465 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
1467 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
1468 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
1470 2017-10-27 Jakub Jelinek <jakub@redhat.com>
1473 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
1474 * config/i386/i386.c (maybe_get_pool_constant): Removed.
1475 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
1476 maybe_get_pool_constant.
1477 * config/i386/predicates.md (zero_extended_scalar_load_operand):
1480 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1482 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
1483 2.26 caveat. Update gas and gld versions.
1484 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
1487 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1489 * cgraph.h (set_malloc_flag): Declare.
1490 * cgraph.c (set_malloc_flag_1): New function.
1491 (set_malloc_flag): Likewise.
1492 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
1493 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
1495 (read_ipa_call_summary): Add support for reading is_return_callee.
1496 (write_ipa_call_summary): Stream is_return_callee.
1497 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
1498 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
1499 ipa-prop.h, ipa-fnsummary.h.
1500 (pure_const_names): Change to static.
1501 (malloc_state_e): Define.
1502 (malloc_state_names): Define.
1503 (funct_state_d): Add field malloc_state.
1504 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
1505 (check_retval_uses): New function.
1506 (malloc_candidate_p): Likewise.
1507 (analyze_function): Add support for malloc attribute.
1508 (pure_const_write_summary): Stream malloc_state.
1509 (pure_const_read_summary): Add support for reading malloc_state.
1510 (dump_malloc_lattice): New function.
1511 (propagate_malloc): New function.
1512 (warn_function_malloc): New function.
1513 (ipa_pure_const::execute): Call propagate_malloc and
1514 ipa_free_fn_summary.
1515 (pass_local_pure_const::execute): Add support for malloc attribute.
1516 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
1517 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
1519 2017-10-27 Martin Liska <mliska@suse.cz>
1521 PR gcov-profile/82457
1522 * doc/invoke.texi: Document that one needs a non-strict ISO mode
1523 for fork-like functions to be properly instrumented.
1525 2017-10-27 Richard Biener <rguenther@suse.de>
1528 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
1529 info when we redirected EH.
1531 2017-10-26 Michael Collison <michael.collison@arm.com>
1533 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
1535 (<optab>_trunchf<GPI:mode>2: New pattern.
1536 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
1537 * config/aarch64/iterators.md (wv): New mode attribute.
1538 (vf, VF): New mode attributes.
1539 (vgp, VGP): New mode attributes.
1540 (s): Update attribute with SImode and DImode prefixes.
1542 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
1544 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
1545 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
1546 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
1547 (nios2_option_overide): Initialize it. Don't allow R0-relative
1548 addressing with PIC.
1549 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
1550 (nios2_symbolic_constant_p): Likewise.
1551 (nios2_legitimate_address_p): Likewise.
1552 (nios2_r0rel_section_name_p): New.
1553 (nios2_symbol_ref_in_r0rel_data_p): New.
1554 (nios2_emit_move_sequence): Handle r0rel_constant_p.
1555 (r0rel_constant_p): New.
1556 (nios2_print_operand_address): Handle r0rel_constant_p.
1557 (nios2_cdx_narrow_form_p): Likewise.
1558 * config/nios2/nios2.opt (mr0rel-sec=): New option.
1559 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
1560 (Nios II Options): Document -mr0rel-sec.
1562 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
1564 * config/nios2/nios2.c: Include xregex.h.
1565 (nios2_gprel_sec_regex): New.
1566 (nios2_option_overide): Initialize it. Don't allow GP-relative
1567 addressing with PIC.
1568 (nios2_small_section_name_p): Check for regex match.
1569 * config/nios2/nios2.opt (mgprel-sec=): New option.
1570 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
1571 (Nios II Options): Document -mgprel-sec.
1573 2017-10-26 Jim Wilson <wilson@tuliptree.org>
1575 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
1577 2017-10-26 Tom de Vries <tom@codesourcery.com>
1579 PR tree-optimization/82707
1580 * gimple.c (gimple_copy): Fix unsharing of
1581 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
1583 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
1585 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
1586 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
1587 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
1588 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
1589 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
1590 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
1591 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
1592 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
1593 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
1594 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
1595 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
1596 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
1597 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
1598 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
1599 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
1600 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
1601 _mm512_mask_cmpunord_ps_mask): New intrinsics.
1603 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
1605 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
1607 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
1608 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
1609 warning to rs6000.c. Remove the Undocumented flag, since it has
1611 (-mabi=ibmlongdouble): Likewise.
1612 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
1613 already set, set the default format for long double.
1614 (rs6000_debug_reg_global): Print whether long double is IBM or
1616 (rs6000_option_override_internal): Rework setting long double
1617 format. Only warn if the user is changing the long double default
1618 and they did not use -Wno-psabi.
1619 * doc/invoke.texi (PowerPC options): Update the documentation for
1620 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
1622 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
1623 Alan Hayward <alan.hayward@arm.com>
1624 David Sherwood <david.sherwood@arm.com>
1626 * rtl.h (wider_subreg_mode): New function.
1627 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
1628 rather than an unsigned int *.
1629 * ira-color.c (regno_max_ref_width): Replace with...
1630 (regno_max_ref_mode): ...this new variable.
1631 (coalesced_pseudo_reg_slot_compare): Update accordingly.
1632 Use wider_subreg_mode.
1633 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
1634 rather than an unsigned int *.
1635 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
1636 (process_alt_operands): Likewise.
1637 (invariant_p): Likewise.
1638 * lra-spills.c (assign_mem_slot): Likewise.
1639 (add_pseudo_to_slot): Likewise.
1640 * lra.c (collect_non_operand_hard_regs): Likewise.
1641 (add_regs_to_insn_regno_info): Likewise.
1642 * reload1.c (regno_max_ref_width): Replace with...
1643 (regno_max_ref_mode): ...this new variable.
1644 (reload): Update accordingly. Update call to
1645 ira_sort_regnos_for_alter_reg.
1646 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
1647 (init_eliminable_invariants): Update to use regno_max_ref_mode.
1648 (scan_paradoxical_subregs): Likewise.
1650 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
1652 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
1653 (aarch64_frame): Add emit_frame_chain boolean.
1654 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1655 Move eh_return case to aarch64_layout_frame.
1656 (aarch64_layout_frame): Initialize emit_frame_chain.
1657 (aarch64_expand_prologue): Use emit_frame_chain.
1659 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
1661 * config/aarch64/aarch64.c (aarch64_layout_frame):
1662 Ensure LR is always stored at the bottom of the callee-saves.
1663 Remove rarely used frame layout which saves callee-saves at top of
1664 frame, so the store of LR can be used as a valid probe in all cases.
1666 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
1668 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
1669 Improve unaligned TImode/TFmode base/offset split.
1671 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
1672 Alan Hayward <alan.hayward@arm.com>
1673 David Sherwood <david.sherwood@arm.com>
1675 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
1676 * combine.c (find_single_use_1): Likewise.
1677 (expand_field_assignment): Likewise.
1678 (move_deaths): Likewise.
1679 * lra-constraints.c (simplify_operand_subreg): Likewise.
1680 (curr_insn_transform): Likewise.
1681 * lra.c (collect_non_operand_hard_regs): Likewise.
1682 (add_regs_to_insn_regno_info): Likewise.
1683 * rtlanal.c (reg_referenced_p): Likewise.
1684 (covers_regno_no_parallel_p): Likewise.
1686 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
1688 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
1689 Don't print any bits outside the precision of the value.
1690 * wide-int.cc (test_printing): Add some new tests.
1692 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1694 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
1695 supports -xbrace_comment option.
1696 * configure: Regenerate.
1697 * config.in: Regenerate.
1698 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
1699 (ASM_CPU_SPEC): Use it.
1701 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
1703 * target.def (static_rtx_alignment): New hook.
1704 * targhooks.h (default_static_rtx_alignment): Declare.
1705 * targhooks.c (default_static_rtx_alignment): New function.
1706 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
1707 * doc/tm.texi: Regenerate.
1708 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
1709 instead of targetm.constant_alignment. Remove call to
1711 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
1712 (cris_preferred_mininum_alignment): New function, split out from...
1713 (cris_constant_alignment): ...here.
1714 (cris_static_rtx_alignment): New function.
1715 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
1717 (ix86_constant_alignment): ...here.
1718 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
1719 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
1720 (mmix_static_rtx_alignment): New function.
1721 * config/spu/spu.c (spu_static_rtx_alignment): New function.
1722 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
1724 2017-10-26 Tamar Christina <tamar.christina@arm.com>
1727 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
1728 Add flag_trapping_math and flag_fp_int_builtin_inexact.
1730 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
1732 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
1733 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
1734 mark as a sign-extending load.
1735 (local_pic_load_u): Define.
1737 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
1740 * fold-const.c (operand_equal_for_comparison_p): Also return true
1741 if ARG0 is a simple variant of ARG1 with narrower precision.
1742 (fold_ternary_loc): Always pass unstripped operands to the predicate.
1744 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
1746 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
1748 * i386.h (processor_costs): Add gather_static, gather_per_elt,
1749 scatter_static, scatter_per_elt.
1750 * x86-tune-costs.h: Add new cost entries.
1752 2017-10-25 Richard Biener <rguenther@suse.de>
1754 * tree-ssa-sccvn.h (vn_eliminate): Declare.
1755 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
1756 class pass_fre): Move to ...
1757 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
1758 class pass_fre): ... here and adjust for statistics.
1760 2017-10-25 Jakub Jelinek <jakub@redhat.com>
1763 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
1764 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
1765 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
1766 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
1769 2017-10-25 Richard Biener <rguenther@suse.de>
1771 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
1772 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
1773 eliminate_push_avail, eliminate_insert): Move inside...
1774 (class eliminate_dom_walker): ... this class in preparation
1776 (fini_eliminate): Remove by merging with ...
1777 (eliminate): ... this function. Adjust for class changes.
1778 (pass_pre::execute): Remove fini_eliminate call.
1779 (pass_fre::execute): Likewise.
1781 2017-10-24 Jakub Jelinek <jakub@redhat.com>
1784 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
1785 (VPERMI2, VPERMI2I): New mode iterators.
1786 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
1787 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
1789 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
1790 mode iterator. Remove 3 old define_insn patterns.
1791 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
1792 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
1793 VPERMI2 mode iterator, remove the other two expanders.
1794 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
1795 to use VPERMI2 mode iterator, add another alternative for vpermi2*
1796 instructions, remove the other two patterns.
1797 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
1798 mode iterator, remove the other two patterns.
1799 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
1800 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
1801 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
1802 and adjust argument order accordingly.
1803 (ix86_expand_vec_perm): Adjust caller.
1804 (expand_vec_perm_1): Likewise.
1805 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
1806 (expand_vec_perm_vpermt2_vpshub2): ... this.
1807 (ix86_expand_vec_perm_const_1): Adjust caller.
1808 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
1811 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
1812 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
1813 (vec_shl_<mode>): Remove unused expander.
1814 (avx512bw_<shift_insn><mode>3): New define_insn.
1815 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
1816 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
1818 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
1821 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
1824 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
1826 PR rtl-optimization/82396
1827 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
1828 (autopref_multipass_init): Simplify initialization.
1829 (autopref_rank_data): Simplify sort order.
1830 * gcc/sched-int.h (autopref_multipass_data_): Remove
1831 multi_mem_insn_p, min_offset and max_offset.
1833 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
1836 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1837 Check special value of flag_omit_frame_pointer.
1838 (aarch64_can_eliminate): Likewise.
1839 (aarch64_override_options_after_change_1): Simplify handling of
1840 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
1842 2017-10-24 Richard Biener <rguenther@suse.de>
1844 PR tree-optimization/82697
1845 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
1846 zero for conditional load and unconditional store.
1848 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
1850 * doc/install.texi: Document bootstrap-cet.
1852 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
1855 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
1856 ENDBR instruction at function entrance if function is only
1859 2017-10-24 Jakub Jelinek <jakub@redhat.com>
1862 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
1863 patterns to better describe from which operation the CF is computed.
1864 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
1865 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
1866 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
1867 is 0, use _0 suffixed expanders instead of emitting a comparison
1870 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
1872 * config/i386/i386.md(*movsf_internal, *movdf_internal):
1873 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
1875 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
1878 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
1879 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
1880 * loop-iv.c (iv_get_reaching_def): Likewise.
1881 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
1882 variable is promoted and the partition contains undefined values.
1884 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
1886 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
1888 (nios2_address_cost): Define.
1889 (nios2_legitimize_address): Recognize (exp + constant) directly.
1890 (TARGET_ADDRESS_COST): Define.
1892 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
1894 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
1895 (nios2_symbolic_memory_operand_p): Declare.
1896 (nios2_split_large_constant): Declare.
1897 (nios2_split_symbolic_memory_operand): Declare.
1898 * config/nios2/nios2.c: Adjust includes.
1899 (nios2_symbolic_constant_allowed): New.
1900 (nios2_symbolic_constant_p): New.
1901 (nios2_plus_symbolic_constant_p): New.
1902 (nios2_valid_addr_expr_p): Recognize addresses involving
1904 (nios2_legitimate_address_p): Likewise, also LO_SUM.
1905 (nios2_symbolic_memory_operand_p): New.
1906 (nios2_large_constant_p): New.
1907 (nios2_split_large_constant): New.
1908 (nios2_split_plus_large_constant): New.
1909 (nios2_split_symbolic_memory_operand): New.
1910 (nios2_legitimize_address): Code refactoring. Handle addresses
1911 involving symbolic constants.
1912 (nios2_emit_move_sequence): Likewise.
1913 (nios2_print_operand): Improve error output.
1914 (nios2_print_operand_address): Handle LO_SUM.
1915 (nios2_cdx_narrow_form_p): Likewise.
1916 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
1917 operands involving symbolic constants.
1918 (movhi_internal, movsi_internal): Likewise.
1919 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
1920 (extendhisi2, extendqi<mode>2): Likewise.
1922 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
1924 * tree-pass.h (PROP_rtl_split_insns): Define.
1925 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
1927 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
1929 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
1931 2017-10-23 Jakub Jelinek <jakub@redhat.com>
1934 * target.def (const_not_ok_for_debug_p): Default to
1935 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
1936 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
1937 * targhooks.c (default_const_not_ok_for_debug_p): New function.
1938 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
1939 which targetm.const_not_ok_for_debug_p returned true.
1940 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
1942 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
1944 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
1945 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
1946 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
1948 (ix86_const_not_ok_for_debug_p): New function.
1949 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
1950 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
1952 2017-10-23 David Malcolm <dmalcolm@redhat.com>
1955 * system.h: Conditionally include "unique-ptr.h" if
1956 INCLUDE_UNIQUE_PTR is defined.
1957 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
1958 of defining INCLUDE_UNIQUE_PTR before including "system.h".
1960 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
1962 * config/rl78/rl78.md: New define_expand "subdi3".
1964 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
1967 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
1968 DF_REF_INSN if DF_REF_INSN_INFO is false.
1970 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
1972 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
1973 xmm_move instead of sse_move.
1974 (sse_store_index): New function.
1975 (ix86_register_move_cost): Be more sensible about mismatch stall;
1976 model AVX moves correctly; make difference between sse->integer and
1978 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
1979 moves; make difference between SSE and AVX.
1980 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
1981 and zmm_move. Increase size of sse load and store tables;
1982 add unaligned load and store tables; add ssemmx_to_integer.
1983 * x86-tune-costs.h: Update all entries according to real
1984 move latencies from Agner Fog's manual and chip documentation.
1986 2017-10-23 Jakub Jelinek <jakub@redhat.com>
1989 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
1990 * config/i386/constraints.md (Wf): New constraint.
1991 * config/i386/i386.md (UNSPEC_SBB): New unspec.
1992 (cmp<dwi>_doubleword): Removed.
1993 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
1994 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
1995 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
1996 expand with cmp<dwi>_doubleword. For LTU and GEU use
1997 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
1999 * common.opt (gcolumn-info): Enable by default.
2000 * doc/invoke.texi (gcolumn-info): Document new default.
2002 2017-10-23 Richard Biener <rguenther@suse.de>
2004 PR tree-optimization/82672
2005 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
2006 Fold the stmt if we propagated into it.
2008 2017-10-23 Richard Biener <rguenther@suse.de>
2010 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
2011 (bitmap_remove_expr_from_set): ... this. All callers call this
2012 for non-constant values.
2013 (bitmap_set_subtract): Rename to...
2014 (bitmap_set_subtract_expressions): ... this. Adjust and
2016 (bitmap_set_contains_value): Remove superfluous check.
2017 (bitmap_set_replace_value): Inline into single caller ...
2018 (bitmap_value_replace_in_set): ... here and simplify.
2019 (dependent_clean): Merge into ...
2020 (clean): ... this using an overload. Adjust.
2021 (prune_clobbered_mems): Adjust.
2022 (compute_antic_aux): Likewise.
2023 (compute_partial_antic_aux): Likewise.
2025 2017-10-23 Richard Biener <rguenther@suse.de>
2027 PR tree-optimization/82129
2029 2017-08-01 Richard Biener <rguenther@suse.de>
2031 PR tree-optimization/81181
2032 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2033 (compute_antic): ... end of iteration here.
2035 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2037 * target.def (starting_frame_offset): New hook.
2038 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
2039 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
2040 * doc/tm.texi.in: Regenerate.
2041 * hooks.h (hook_hwi_void_0): Declare.
2042 * hooks.c (hook_hwi_void_0): New function.
2043 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
2044 STARTING_FRAME_OFFSET.
2045 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
2046 * reload1.c (reload): Likewise.
2047 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
2048 instead of STARTING_FRAME_OFFSET.
2049 * function.c (try_fit_stack_local): Likewise.
2050 (assign_stack_local_1): Likewise
2051 (instantiate_virtual_regs): Likewise.
2052 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
2053 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
2054 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
2055 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
2056 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
2057 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
2058 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
2059 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
2060 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
2061 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
2062 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
2063 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
2064 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
2065 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
2066 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
2067 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
2068 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
2069 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
2070 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
2071 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
2072 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
2073 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
2074 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
2075 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
2076 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
2077 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
2078 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
2079 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
2080 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
2081 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
2082 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
2083 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
2084 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
2085 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
2086 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
2087 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
2088 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
2089 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
2090 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
2091 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
2092 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
2093 * config/avr/avr.c (avr_starting_frame_offset): Make static and
2094 return a HOST_WIDE_INT.
2095 (avr_builtin_setjmp_frame_value): Use it instead of
2096 STARTING_FRAME_OFFSET.
2097 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2098 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
2099 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
2101 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2102 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
2103 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
2104 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2105 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
2106 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
2107 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2108 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
2109 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
2110 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2111 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
2112 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
2114 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2115 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
2116 * config/mips/mips.c (mips_compute_frame_info): Refer to
2117 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2118 (mips_starting_frame_offset): New function.
2119 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2120 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
2121 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
2122 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
2123 and return a HOST_WIDE_INT.
2124 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2125 (mmix_initial_elimination_offset): Refer to
2126 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2127 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
2128 * config/pa/pa.c (pa_starting_frame_offset): New function.
2129 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
2130 (pa_expand_prologue): Likewise.
2131 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2132 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
2133 !FRAME_GROWS_DOWNWARD handling to...
2134 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2135 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
2136 !FRAME_GROWS_DOWNWARD handling to...
2137 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2138 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
2139 !FRAME_GROWS_DOWNWARD handling to...
2140 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2141 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
2143 (rs6000_starting_frame_offset): New function.
2144 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
2145 !FRAME_GROWS_DOWNWARD handling to...
2146 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2147 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
2148 !FRAME_GROWS_DOWNWARD handling to...
2149 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2150 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
2151 !FRAME_GROWS_DOWNWARD handling to...
2152 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2153 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
2154 (rs6000_starting_frame_offset): New function.
2155 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
2156 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
2157 * config/vax/vax.c (vax_starting_frame_offset): New function.
2158 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
2159 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2160 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
2161 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
2162 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2163 * system.h (STARTING_FRAME_OFFSET): Poison.
2165 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2167 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
2168 SCALAR_TYPE_MODE instead of TYPE_MODE.
2170 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2171 Alan Hayward <alan.hayward@arm.com>
2172 David Sherwood <david.sherwood@arm.com>
2174 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
2176 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2177 Alan Hayward <alan.hayward@arm.com>
2178 David Sherwood <david.sherwood@arm.com>
2180 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
2182 2017-10-23 Richard Biener <rguenther@suse.de>
2184 PR tree-optimization/82129
2185 * tree-ssa-pre.c (bitmap_set_and): Remove.
2186 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
2187 canonicalizing expressions in the set to those with lowest
2188 ID rather than taking that from the first edge.
2190 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2192 * combine.c (rtx_equal_for_field_assignment_p): Use
2193 byte_lowpart_offset.
2195 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2196 Alan Hayward <alan.hayward@arm.com>
2197 David Sherwood <david.sherwood@arm.com>
2199 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
2200 to a SUBREG_PROMOTED_VAR.
2202 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2203 Alan Hayward <alan.hayward@arm.com>
2204 David Sherwood <david.sherwood@arm.com>
2206 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
2207 (expand_debug_source_expr): Likewise.
2208 * combine.c (combine_simplify_rtx): Likewise.
2209 * cse.c (fold_rtx): Likewise.
2210 * optabs.c (expand_float): Likewise.
2211 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2212 (simplify_binary_operation_1): Likewise.
2214 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2215 Alan Hayward <alan.hayward@arm.com>
2216 David Sherwood <david.sherwood@arm.com>
2218 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
2219 (record_promoted_value): Likewise.
2220 * expr.c (expand_expr_real_2): Likewise.
2221 * ree.c (update_reg_equal_equiv_notes): Likewise.
2222 (combine_set_extension): Likewise.
2223 * rtlanal.c (low_bitmask_len): Likewise.
2224 * simplify-rtx.c (neg_const_int): Likewise.
2225 (simplify_binary_operation_1): Likewise.
2227 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2228 Alan Hayward <alan.hayward@arm.com>
2229 David Sherwood <david.sherwood@arm.com>
2231 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
2232 * regcprop.c (maybe_mode_change): Likewise.
2233 * reload1.c (alter_reg): Likewise.
2235 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2237 * inchash.h (inchash::hash::add_wide_int): New function.
2238 * lto-streamer-out.c (hash_tree): Use it.
2240 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2242 * inchash.h (inchash::hash::add_wide_int): Rename to...
2243 (inchash::hash::add_hwi): ...this.
2244 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
2245 (polymorphic_call_target_hasher::hash): Likewise.
2246 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
2247 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
2248 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
2249 * lto-streamer-out.c (hash_tree): Likewise.
2250 * optc-save-gen.awk: Likewise.
2251 * tree.c (add_expr): Likewise.
2253 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
2256 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
2257 for ordered inequality comparisons even with TARGET_IEEE_FP.
2259 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
2262 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
2263 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
2264 Expand with cmp<dwi>_doubleword.
2266 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2268 * extend.texi: Add x86 specific to 'nocf_check' attribute.
2269 List CET intrinsics.
2270 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
2271 specific to -fcf-protection option.
2273 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2275 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
2276 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
2277 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
2278 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
2279 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
2280 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
2281 (extra_objs): Add cet.o for Linux/x86 targets.
2282 (tmake_file): Add i386/t-cet for Linux/x86 targets.
2283 * config/i386/cet.c: New file.
2284 * config/i386/cetintrin.h: Likewise.
2285 * config/i386/t-cet: Likewise.
2286 * config/i386/cpuid.h (bit_SHSTK): New.
2287 (bit_IBT): Likewise.
2288 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
2289 pass IBT and SHSTK bits.
2290 * config/i386/i386-builtin-types.def
2291 (VOID_FTYPE_UNSIGNED_PVOID): New.
2292 (VOID_FTYPE_UINT64_PVOID): Likewise.
2293 * config/i386/i386-builtin.def: Add CET intrinsics.
2294 * config/i386/i386-c.c (ix86_target_macros_internal): Add
2295 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
2296 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
2297 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
2299 * config/i386/i386.c (rest_of_insert_endbranch): New.
2300 (pass_data_insert_endbranch): Likewise.
2301 (pass_insert_endbranch): Likewise.
2302 (make_pass_insert_endbranch): Likewise.
2303 (ix86_notrack_prefixed_insn_p): Likewise.
2304 (ix86_target_string): Add -mibt, -mshstk flags.
2305 (ix86_option_override_internal): Add flag_cf_protection
2307 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
2308 (ix86_print_operand): Add 'notrack' prefix output.
2309 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
2310 (ix86_expand_builtin): Expand CET intrinsics.
2311 (x86_output_mi_thunk): Add 'endbranch' instruction.
2312 * config/i386/i386.h (TARGET_IBT): New.
2313 (TARGET_IBT_P): Likewise.
2314 (TARGET_SHSTK): Likewise.
2315 (TARGET_SHSTK_P): Likewise.
2316 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
2317 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
2318 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
2319 (builtin_setjmp_setup): New pattern.
2320 (builtin_longjmp): Likewise.
2321 (rdssp<mode>): Likewise.
2322 (incssp<mode>): Likewise.
2323 (saveprevssp): Likewise.
2324 (rstorssp): Likewise.
2325 (wrss<mode>): Likewise.
2326 (wruss<mode>): Likewise.
2327 (setssbsy): Likewise.
2328 (clrssbsy): Likewise.
2329 (nop_endbr): Likewise.
2330 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
2332 * config/i386/immintrin.h: Include <cetintrin.h>.
2333 * config/i386/linux-common.h
2334 (file_end_indicate_exec_stack_and_cet): New prototype.
2335 (TARGET_ASM_FILE_END): New.
2337 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
2339 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
2340 latencies instead of having separate table; make difference between
2341 integer and float costs.
2342 * i386.h (processor_costs): Remove scalar_stmt_cost,
2343 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
2344 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
2346 * x86-tune-costs.h: Remove entries which has been removed in
2347 procesor_costs from all tables; make cond_taken_branch_cost
2348 and cond_not_taken_branch_cost COST_N_INSNS based.
2350 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
2352 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
2354 2017-10-20 Jakub Jelinek <jakub@redhat.com>
2356 * config/i386/i386.md (isa): Remove fma_avx512f.
2357 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
2358 <avx512>_fmadd_<mode>_mask3<round_name>,
2359 <avx512>_fmsub_<mode>_mask<round_name>,
2360 <avx512>_fmsub_<mode>_mask3<round_name>,
2361 <avx512>_fnmadd_<mode>_mask<round_name>,
2362 <avx512>_fnmadd_<mode>_mask3<round_name>,
2363 <avx512>_fnmsub_<mode>_mask<round_name>,
2364 <avx512>_fnmsub_<mode>_mask3<round_name>,
2365 <avx512>_fmaddsub_<mode>_mask<round_name>,
2366 <avx512>_fmaddsub_<mode>_mask3<round_name>,
2367 <avx512>_fmsubadd_<mode>_mask<round_name>,
2368 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
2369 (*vec_widen_umult_even_v16si<mask_name>,
2370 *vec_widen_smult_even_v16si<mask_name>): Likewise.
2371 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
2373 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2375 * extend.texi: Add 'nocf_check' documentation.
2376 * gimple.texi: Add second parameter to
2377 gimple_build_call_from_tree.
2378 * invoke.texi: Add -fcf-protection documentation.
2379 * rtl.texi: Add REG_CALL_NOTRACK documenation.
2381 2017-10-20 Richard Biener <rguenther@suse.de>
2383 PR tree-optimization/82473
2384 * tree-vect-loop.c (vectorizable_reduction): Properly get at
2385 the largest input type.
2387 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2389 * c-attribs.c (handle_nocf_check_attribute): New function.
2390 (c_common_attribute_table): Add 'nocf_check' handling.
2391 * gimple-parser.c: Add second argument NULL to
2392 gimple_build_call_from_tree.
2393 * attrib.c (comp_type_attributes): Check nocf_check attribute.
2394 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
2396 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
2397 * common.opt: Add fcf-protection flag.
2398 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
2399 * flag-types.h: Add enum cf_protection_level.
2400 * gimple.c (gimple_build_call_from_tree): Add second parameter.
2401 Add 'nocf_check' attribute propagation to gimple call.
2402 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
2403 (gimple_build_call_from_tree): Update prototype.
2404 (gimple_call_nocf_check_p): New function.
2405 (gimple_call_set_nocf_check): Likewise.
2406 * gimplify.c: Add second argument to gimple_build_call_from_tree.
2407 * ipa-icf.c: Add nocf_check attribute in statement hash.
2408 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
2409 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
2410 * toplev.c (process_options): Add flag_cf_protection handling.
2412 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
2414 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
2416 2017-10-20 Richard Biener <rguenther@suse.de>
2418 PR tree-optimization/82603
2419 * tree-if-conv.c (predicate_mem_writes): Make sure to only
2420 remove false predicated stores.
2422 2017-10-20 Richard Biener <rguenther@suse.de>
2424 * graphite-isl-ast-to-gimple.c
2425 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
2426 Remove return value and simplify, dump copied stmt after lhs
2428 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
2429 Reduce dump verbosity.
2430 (gsi_insert_earliest): Likewise.
2431 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
2432 * graphite.c (print_global_statistics): Adjust dumping.
2433 (print_graphite_scop_statistics): Likewise.
2434 (print_graphite_statistics): Do not dump loops here.
2435 (graphite_transform_loops): But here.
2437 2017-10-20 Nicolas Roche <roche@adacore.com>
2439 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
2440 * configure: Regenerate.
2442 2017-10-20 Jakub Jelinek <jakub@redhat.com>
2445 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
2446 functions when optimizing replace GIMPLE_RETURN stmts with
2447 calls to __builtin_unreachable ().
2450 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
2451 for -fsanitize=thread link of executables.
2452 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
2453 link of executables.
2456 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
2458 (<shift_insn><mode>3<mask_name>): Change the last of the 3
2459 define_insns for logical vector shifts to use VI248_AVX512BW
2460 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
2461 condition, useless isa and prefix attributes. Change the first
2462 2 of these define_insns to ...
2463 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
2464 define_insn for avx512vl.
2465 (<shift_insn><mode>3): ... and this, new define_insn without
2466 masking for non-avx512vl.
2469 * config/i386/sse.md (*andnot<mode>3,
2470 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
2471 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
2472 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
2473 not applied use empty suffix even for TARGET_AVX512VL.
2474 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
2475 is applied, supply evex,evex or evex,evex,evex instead of just
2478 2017-10-20 Julia Koval <julia.koval@intel.com>
2480 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
2481 (OPTION_MASK_ISA_GFNI_UNSET): New.
2482 (ix86_handle_option): Handle OPT_mgfni.
2483 * config/i386/cpuid.h (bit_GFNI): New.
2484 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
2485 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
2486 * config/i386/i386.c (ix86_target_string): Add -mgfni.
2487 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
2488 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
2489 * config/i386/i386.opt: Add mgfni.
2491 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
2493 * config/msp430/msp430.c (msp430_option_override): Disable
2494 -fdelete-null-pointer-checks.
2495 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
2497 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
2499 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
2500 of x87 and SSE instructions.
2502 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
2504 * asan.c (create_cond_insert_point): Do not update edge count.
2505 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
2506 (afdo_propagate_circuit): Likewise.
2507 (afdo_calculate_branch_prob): Likewise.
2508 (afdo_annotate_cfg): Likewise.
2509 * basic-block.h (struct edge_def): Remove count.
2510 (edge_def::count): New accessor.
2511 * bb-reorder.c (rotate_loop): Update.
2512 (find_traces_1_round): Update.
2513 (connect_traces): Update.
2514 (sanitize_hot_paths): Update.
2515 * cfg.c (unchecked_make_edge): Update.
2516 (make_single_succ_edge): Update.
2517 (check_bb_profile): Update.
2518 (dump_edge_info): Update.
2519 (update_bb_profile_for_threading): Update.
2520 (scale_bbs_frequencies_int): Update.
2521 (scale_bbs_frequencies_gcov_type): Update.
2522 (scale_bbs_frequencies_profile_count): Update.
2523 (scale_bbs_frequencies): Update.
2524 * cfganal.c (connect_infinite_loops_to_exit): Update.
2525 * cfgbuild.c (compute_outgoing_frequencies): Update.
2526 (find_many_sub_basic_blocks): Update.
2527 * cfgcleanup.c (try_forward_edges): Update.
2528 (try_crossjump_to_edge): Update
2529 * cfgexpand.c (expand_gimple_cond): Update
2530 (expand_gimple_tailcall): Update
2531 (construct_exit_block): Update
2532 * cfghooks.c (verify_flow_info): Update
2533 (redirect_edge_succ_nodup): Update
2534 (split_edge): Update
2535 (make_forwarder_block): Update
2536 (duplicate_block): Update
2537 (account_profile_record): Update
2538 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
2539 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
2540 * cfgloopmanip.c (scale_loop_profile): Update.
2542 (lv_adjust_loop_entry_edge): Update.
2543 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
2544 (force_nonfallthru_and_redirect): Update.
2545 (purge_dead_edges): Update.
2546 (rtl_flow_call_edges_add): Update.
2547 * cgraphunit.c (init_lowered_empty_function): Update.
2548 (cgraph_node::expand_thunk): Update.
2549 * gimple-pretty-print.c (dump_probability): Update.
2550 (dump_edge_probability): Update.
2551 * gimple-ssa-isolate-paths.c (isolate_path): Update.
2552 * haifa-sched.c (sched_create_recovery_edges): Update.
2553 * hsa-gen.c (convert_switch_statements): Update.
2554 * ifcvt.c (dead_or_predicable): Update.
2555 * ipa-inline-transform.c (inline_transform): Update.
2556 * ipa-split.c (split_function): Update.
2557 * ipa-utils.c (ipa_merge_profiles): Update.
2558 * loop-doloop.c (add_test): Update.
2559 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
2560 * lto-streamer-in.c (input_cfg): Update.
2561 (input_function): Update.
2562 * lto-streamer-out.c (output_cfg): Update.
2563 * modulo-sched.c (sms_schedule): Update.
2564 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
2565 * predict.c (maybe_hot_edge_p): Update.
2566 (unlikely_executed_edge_p): Update.
2567 (probably_never_executed_edge_p): Update.
2568 (dump_prediction): Update.
2569 (drop_profile): Update.
2570 (propagate_unlikely_bbs_forward): Update.
2571 (determine_unlikely_bbs): Update.
2572 (force_edge_cold): Update.
2573 * profile.c (compute_branch_probabilities): Update.
2574 * reg-stack.c (better_edge): Update.
2575 * shrink-wrap.c (handle_simple_exit): Update.
2576 * tracer.c (better_p): Update.
2577 * trans-mem.c (expand_transaction): Update.
2578 (split_bb_make_tm_edge): Update.
2579 * tree-call-cdce.c: Update.
2580 * tree-cfg.c (gimple_find_sub_bbs): Update.
2581 (gimple_split_edge): Update.
2582 (gimple_duplicate_sese_region): Update.
2583 (gimple_duplicate_sese_tail): Update.
2584 (gimple_flow_call_edges_add): Update.
2585 (insert_cond_bb): Update.
2586 (execute_fixup_cfg): Update.
2587 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
2588 * tree-complex.c (expand_complex_div_wide): Update.
2589 * tree-eh.c (lower_resx): Update.
2590 (unsplit_eh): Update.
2591 (cleanup_empty_eh_move_lp): Update.
2592 * tree-inline.c (copy_edges_for_bb): Update.
2593 (freqs_to_counts): Update.
2594 (copy_cfg_body): Update.
2595 * tree-ssa-dce.c (remove_dead_stmt): Update.
2596 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
2597 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
2598 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
2599 (unloop_loops): Update.
2600 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
2601 * tree-ssa-loop-split.c (connect_loops): Update.
2602 (split_loop): Update.
2603 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
2604 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
2605 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
2606 * tree-ssa-reassoc.c (branch_fixup): Update.
2607 * tree-ssa-tail-merge.c (replace_block_by): Update.
2608 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
2609 (compute_path_counts): Update.
2610 (update_profile): Update.
2611 (recompute_probabilities): Update.
2612 (update_joiner_offpath_counts): Update.
2613 (estimated_freqs_path): Update.
2614 (freqs_to_counts_path): Update.
2615 (clear_counts_path): Update.
2616 (ssa_fix_duplicate_block_edges): Update.
2617 (duplicate_thread_path): Update.
2618 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
2619 (case_bit_test_cmp): Update.
2620 (collect_switch_conv_info): Update.
2621 (gen_inbound_check): Update.
2622 (do_jump_if_equal): Update.
2623 (emit_cmp_and_jump_insns): Update.
2624 * tree-tailcall.c (decrease_profile): Update.
2625 (eliminate_tail_call): Update.
2626 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
2627 (vect_do_peeling): Update.
2628 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
2629 * ubsan.c (ubsan_expand_null_ifn): Update.
2630 (ubsan_expand_ptr_ifn): Update.
2631 * value-prof.c (gimple_divmod_fixed_value): Update.
2632 (gimple_mod_pow2): Update.
2633 (gimple_mod_subtract): Update.
2634 (gimple_ic): Update.
2635 (gimple_stringop_fixed_value): Update.
2637 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
2640 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
2642 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
2644 PR rtl-optimization/82395
2645 * ira-color.c (allocno_priority_compare_func): Fix comparison step
2646 based on non_spilled_static_chain_regno_p.
2648 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
2650 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
2651 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
2652 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
2654 2017-10-19 Martin Sebor <msebor@redhat.com>
2656 PR tree-optimization/82596
2657 * tree.c (array_at_struct_end_p): Handle STRING_CST.
2659 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
2661 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
2662 (get_mem_refs_of_builtin_call): Likewise.
2663 * builtins.c (expand_builtin_apply): Adjust call to
2664 allocate_dynamic_stack_space.
2665 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
2666 the third argument to allocate_dynamic_stack_space, otherwise -1.
2667 (expand_builtin): Deal with all alloca variants.
2668 (is_inexpensive_builtin): Likewise.
2669 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
2670 * calls.c (special_function_p): Deal with all alloca variants.
2671 (initialize_argument_information): Adjust call to
2672 allocate_dynamic_stack_space.
2673 (expand_call): Likewise.
2674 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
2675 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
2676 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
2677 use it for the stack usage computation.
2678 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
2679 * function.c (gimplify_parameters): Call build_alloca_call_expr.
2680 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
2681 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
2682 (in_loop_p): Remove first argument and useless check.
2683 (pass_walloca::execute): Remove useless test and adjust call to above.
2684 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
2685 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
2686 (gimplify_call_expr): Deal with all alloca variants.
2687 * hsa-gen.c (gen_hsa_alloca): Likewise.
2688 (gen_hsa_insns_for_call): Likewise.
2689 * ipa-pure-const.c (special_builtin_state): Likewise.
2690 * tree-chkp.c (chkp_build_returned_bound): Likewise.
2691 * tree-object-size.c (alloc_object_size): Likewise.
2692 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
2693 (call_may_clobber_ref_p_1): Likewise.
2694 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
2695 (ccp_fold_stmt): Likewise.
2696 (optimize_stack_restore): Likewise.
2697 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
2698 (mark_all_reaching_defs_necessary_1): Likewise.
2699 (propagate_necessity): Likewise.
2700 (eliminate_unnecessary_stmts): Likewise.
2701 * tree.c (build_common_builtin_nodes): Build
2702 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
2703 (build_alloca_call_expr): New function.
2704 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
2705 (CASE_BUILT_IN_ALLOCA): Likewise.
2706 (build_alloca_call_expr): Declare.
2707 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
2709 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
2712 * dwarf2out.c (new_die_raw): New static inline function.
2713 (new_die): Use it to create the DIE.
2714 (add_AT_external_die_ref): Likewise.
2715 (clone_die): Likewise.
2716 (clone_as_declaration): Likewise.
2717 (dwarf2out_vms_debug_main_pointer): Likewise.
2718 (base_type_die): Likewise. Remove early return for corner cases.
2719 Do not call add_pubtype on the DIE here.
2720 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
2721 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
2722 typedefs for base types with DW_AT_endianity. Make sure a DIE with
2723 native order exists for base types, attach the DIE manually and call
2724 add_pubtype on it. Do not equate a reverse order DIE to the type.
2726 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
2728 * config/arm/arm.c (align_ok_ldrd_strd): New function.
2729 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
2731 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
2733 2017-10-19 Jakub Jelinek <jakub@redhat.com>
2735 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
2736 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
2737 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
2738 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
2739 * opts.c (sanitizer_opts): Add builtin.
2740 * ubsan.c (instrument_builtin): New function.
2741 (pass_ubsan::execute): Call it.
2742 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
2743 * doc/invoke.texi: Document -fsanitize=builtin.
2745 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
2746 builtins, store max (log2 (align), 0) into uchar field instead of
2747 align into uptr field.
2748 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
2749 store uchar 0 field instead of uptr 0 field.
2750 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
2751 instead of passing one address of struct with 2 locations pass
2752 two addresses of structs with 1 location each.
2753 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
2754 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
2755 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
2756 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
2757 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
2758 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
2759 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
2760 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
2762 2017-10-19 Martin Liska <mliska@suse.cz>
2765 * file-find.c (remove_prefix): Remove.
2766 * file-find.h (remove_prefix): Likewise.
2767 * gcc-ar.c: Remove smartness of lookup.
2769 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
2771 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
2772 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
2773 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
2775 2017-10-19 Jakub Jelinek <jakub@redhat.com>
2778 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
2779 (setcc + and to xor + setcc): New peephole2.
2781 2017-10-19 Tom de Vries <tom@codesourcery.com>
2783 * doc/sourcebuild.texi (Test Directives, Variants of
2784 dg-require-support): Add dg-require-stack-size.
2786 2017-10-19 Martin Liska <mliska@suse.cz>
2789 * gimplify.c (gimplify_decl_expr): Do not instrument variables
2790 that have a large alignment.
2791 (gimplify_target_expr): Likewise.
2793 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
2795 PR rtl-optimization/82602
2796 * ira.c (rtx_moveable_p): Return false for volatile asm.
2798 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
2801 * config/i386/i386-modes.def (CCGZ): New CC mode.
2802 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
2803 * config/i386/predicates.md (ix86_comparison_operator):
2805 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
2806 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
2807 with double-word subtraction.
2808 (put_condition_code): Handle CCGZmode.
2810 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
2812 * wide-int.cc (debug (const wide_int &)): New.
2813 (debug (const wide_int *)): New.
2814 (debug (const widest_int &)): New.
2815 (debug (const widest_int *)): New.
2817 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
2820 * lra-constraints.c (curr_insn_transform): Use non-input operand
2821 instead of output one for matched reload.
2823 2017-10-18 Bin Cheng <bin.cheng@arm.com>
2825 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
2826 (tree-ssa-loop-ivopts.h): New header file.
2827 (struct builtin_info): New fields.
2828 (classify_builtin_1): Compute and record base and offset parts for
2829 memset builtin partition by calling strip_offset.
2830 (offset_cmp, fuse_memset_builtins): New functions.
2831 (finalize_partitions): Fuse adjacent memset partitions by calling
2833 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
2834 Expose the interface.
2835 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
2837 2017-10-18 Bin Cheng <bin.cheng@arm.com>
2839 PR tree-optimization/82574
2840 * tree-loop-distribution.c (find_single_drs): New parameter. Check
2841 that data reference must be executed exactly once per iteration
2842 against the outermost loop in nest.
2843 (classify_partition): Update call to above function.
2845 2017-10-18 Richard Biener <rguenther@suse.de>
2847 PR tree-optimization/82591
2848 * graphite.c (graphite_transform_loops): Move code gen message
2850 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
2851 Here. Handle scop_to_isl_ast failing.
2852 (scop_to_isl_ast): Limit the number of ISL operations.
2854 2017-10-18 Richard Biener <rguenther@suse.de>
2856 * graphite-isl-ast-to-gimple.c
2857 (translate_isl_ast_to_gimple::set_rename): Simplify.
2858 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
2859 (graphite_copy_stmts_from_block): ... here.
2860 (copy_bb_and_scalar_dependences): Simplify.
2861 (add_parameters_to_ivs_params): Canonicalize.
2862 (generate_entry_out_of_ssa_copies): Simplify.
2863 * graphite-sese-to-poly.c (extract_affine_name): Simplify
2864 by passing in ISL dimension.
2865 (parameter_index_in_region_1): Rename to ...
2866 (parameter_index_in_region): ... this.
2867 (extract_affine): Adjust assert, pass down parameter index.
2868 (add_param_constraints): Use range-info when available.
2869 (build_scop_context): Adjust.
2870 * sese.c (new_sese_info): Adjust.
2871 (free_sese_info): Likewise.
2872 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
2873 Remove unused typedefs.
2874 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
2876 2017-10-18 Martin Liska <mliska@suse.cz>
2878 * combine.c (simplify_compare_const): Add gcc_fallthrough.
2880 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
2882 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
2883 (s390_sched_init): Do not reset s390_sched_state if we entered the
2884 current basic block via a fallthru edge and all others are unlikely.
2886 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
2888 * config/s390/s390.c (NUM_SIDES): New variable.
2889 (LONGRUNNING_THRESHOLD): New variable.
2890 (LATENCY_FACTOR): New variable.
2891 (s390_sched_score): Decrease score for long-running instructions on
2893 (s390_sched_variable_issue): Perform bookkeeping for long-running
2896 2017-10-18 Richard Biener <rguenther@suse.de>
2898 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
2899 Simplify with removal of the parameter rename map.
2900 (set_rename): Likewise.
2901 (should_copy_to_new_region): Likewise.
2902 (graphite_copy_stmts_from_block): Likewise.
2903 (copy_bb_and_scalar_dependences): Remove initialization of
2904 unused copied_bb_map.
2906 (copy_internal_parameters): Likewise.
2907 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
2908 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
2909 Use INTEGRAL_TYPE_P.
2910 (parameter_index_in_region_1): Rename to ...
2911 (assign_parameter_index_in_region): ... this. Assert we have
2912 a parameter we handle.
2913 (scan_tree_for_params): Adjust.
2914 * sese.h (parameter_rename_map_t): Remove.
2915 (struct sese_info_t): Remove unused parameter_rename_map and
2916 copied_bb_map members.
2917 * sese.c (new_sese_info): Adjust.
2918 (free_sese_info): Likewise.
2920 2017-10-18 Martin Liska <mliska@suse.cz>
2923 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
2924 on an abnormal edge.
2926 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
2928 * doc/invoke.texi (ffunction-sections and fdata-sections):
2931 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
2933 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
2934 the use statement can throw internally.
2936 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
2938 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
2939 any RTX present on the RHS of a SET.
2940 * compare-elim.c (try_eliminate_compare): Restore comment.
2942 2017-10-17 Jakub Jelinek <jakub@redhat.com>
2944 * langhooks.h (struct lang_hooks): Document that tree_size langhook
2945 may be also called on tcc_type nodes.
2946 * langhooks.c (lhd_tree_size): Likewise.
2948 2017-10-17 David Malcolm <dmalcolm@redhat.com>
2950 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
2951 format_warning_at_substring.
2952 (maybe_warn): Convert source_range * param to a location_t. Pass
2953 UNKNOWN_LOCATION rather than NULL to fmtwarn.
2954 (format_directive): Remove code to extract source_ranges and
2955 source_range * in favor of just a location_t.
2956 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
2958 * substring-locations.c (format_warning_va): Convert
2959 source_range * param to a location_t.
2960 (format_warning_at_substring): Likewise.
2961 * substring-locations.h (format_warning_va): Likewise.
2962 (format_warning_at_substring): Likewise.
2964 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
2966 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
2968 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
2969 and scatter/gather ops.
2971 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
2972 vec_gather_load and vec_scatter_store.
2973 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
2974 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
2975 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
2976 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
2977 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
2978 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
2980 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
2982 * reg-stack.c (compare_for_stack_reg): Add bool argument.
2983 Detect FTST instruction and handle its register pops. Only pop
2984 second operand if can_pop_second_op is true.
2985 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
2986 set can_pop_second_op to false in the compare_for_stack_reg call.
2988 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
2989 output_fp_compare for stack register operands.
2990 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
2991 instructions here. Do not emit stack register pops here. Assert
2992 that FCOMPP pops next to top stack register. Rewrite function.
2994 2017-10-17 Nathan Sidwell <nathan@acm.org>
2997 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
2998 use DECL_ASSEMBLER_NAME_RAW.
3001 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
3004 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
3005 Wilco Dijkstra <wilco.dijkstra@arm.com>
3007 * builtins.c (expand_builtin_update_setjmp_buf): Add a
3008 converstion to Pmode from the buf_addr.
3010 2017-10-17 Richard Biener <rguenther@suse.de>
3012 * graphite-dependences.c (scop_get_reads_and_writes): Change
3013 output parameters to references.
3015 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
3017 PR 71026/tree-optimization
3018 * fold-const.c (distribute_real_division): Removed.
3019 (fold_binary_loc): Remove calls to distribute_real_divison.
3021 2017-10-17 Richard Biener <rguenther@suse.de>
3023 * graphite-scop-detection.c
3024 (scop_detection::stmt_has_simple_data_refs_p): Always use
3025 the full nest as region.
3026 (try_generate_gimple_bb): Likewise.
3027 * sese.c (scalar_evolution_in_region): Simplify now that
3028 SCEV can handle instantiation in regions.
3029 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
3030 in the non-loop part of a function if requested.
3032 2017-10-17 Richard Biener <rguenther@suse.de>
3034 PR tree-optimization/82563
3035 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
3037 (graphite_regenerate_ast_isl): Call it.
3038 * graphite-scop-detection.c (build_scops): Remove entry edge split.
3040 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3042 PR tree-optimization/82549
3043 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
3044 Formatting fixes. Instead of calling make_bit_field_ref with negative
3047 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
3049 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
3050 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
3051 _mm_maskz_reduce_ss): New.
3052 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
3053 __builtin_ia32_reducess_mask): Ditto..
3054 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
3055 * config/i386/sse.md (reduces<mode>): Renamed to ...
3056 (reduces<mode><mask_scalar_name>): ... this.
3057 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
3059 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
3060 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
3062 2017-10-16 David Malcolm <dmalcolm@redhat.com>
3064 * Makefile.in (OBJS): Add unique-ptr-tests.o.
3065 * selftest-run-tests.c (selftest::run_tests): Call
3066 selftest::unique_ptr_tests_cc_tests.
3067 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
3068 * unique-ptr-tests.cc: New file.
3070 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
3073 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3075 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3076 (make_hard_regno_born, make_hard_regno_dead): Update
3077 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
3079 2017-10-16 Jeff Law <law@redhat.com>
3081 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
3083 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
3085 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
3087 2017-10-16 Olivier Hainque <hainque@adacore.com>
3089 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
3090 with_cpu if we were configured for an e500v2 target cpu name.
3092 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
3094 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
3095 * doc/invoke.texi: Document +nodsp as a valid extension for
3098 2017-10-16 Martin Liska <mliska@suse.cz>
3100 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
3101 (test_set_range): Likewise.
3102 (test_range_functions): Rename to ...
3103 (test_bit_in_range): ... this.
3104 (sbitmap_c_tests): Add new test.
3106 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3108 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
3110 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
3111 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
3113 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3115 * config/aarch64/aarch64-builtins.c
3116 (aarch64_types_quadopu_lane_qualifiers): New.
3117 (TYPES_QUADOPU_LANE): New.
3118 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
3119 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
3120 (aarch64_<sur>dot_laneq<vsi2qi>): New.
3121 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
3122 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
3123 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
3124 (Vdottype, DOTPROD): New.
3125 (sur): Add SDOT and UDOT.
3127 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3129 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
3130 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
3131 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
3133 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
3134 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
3135 Enable TARGET_DOTPROD.
3136 (cortex-a75.cortex-a55): Likewise.
3137 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
3139 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3141 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
3142 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
3144 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
3146 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
3147 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
3148 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
3149 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
3150 * config/arm/types.md (neon_dot, neon_dot_q): New.
3151 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
3153 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3155 * config/arm/arm.h (TARGET_DOTPROD): New.
3156 * config/arm/arm.c (arm_arch_dotprod): New.
3157 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
3158 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
3159 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
3160 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
3161 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
3162 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
3163 * doc/invoke.texi (armv8.2-a): Document dotprod
3165 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3167 * i386.c (ix86_vec_cost): New function.
3168 (ix86_rtx_costs): Handle vector operations better.
3169 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
3170 * x86-tune-costs.h: Add new costs to all tables.
3172 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3174 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
3176 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
3177 divsd, sqrtss and sqrtsd
3178 * x86-tune-costs.h: Add new entries to all costs.
3179 (znver1_cost): Fix to match real instruction latencies.
3181 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3182 Michael Collison <michael.collison@arm.com>
3184 * compare-elim.c: Include emit-rtl.h.
3185 (can_merge_compare_into_arith): New function.
3186 (try_validate_parallel): Likewise.
3187 (try_merge_compare): Likewise.
3188 (try_eliminate_compare): Call the above when no previous clobber
3190 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
3193 2017-10-14 Jakub Jelinek <jakub@redhat.com>
3197 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
3198 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
3202 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
3203 to be any operand_equal_p operands. For & (B - 1) require
3204 B to be power of 2. Recognize
3205 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
3207 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3210 * optabs.c (expand_memory_blockage): Fix call of
3211 targetm.have_memory_blockage.
3213 2017-10-14 Jakub Jelinek <jakub@redhat.com>
3216 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
3217 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
3218 objects to extra_objs instead of overwriting it.
3220 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3222 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
3223 Use any_fp_register_operand as operand[3] predicate. Simplify
3224 equality test for operands[2] and operands[4] memory location.
3225 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
3226 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
3227 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
3228 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
3229 any_fp_register_operand as operand[1] predicate. Simplify
3230 equality test for operands[0] and operands[3] memory location.
3231 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
3232 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
3233 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
3235 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3237 * target-insns.def: Add memory_blockage.
3238 * optabs.c (expand_memory_blockage): New function.
3239 (expand_asm_memory_barrier): Rename ...
3240 (expand_asm_memory_blockage): ... to this.
3241 (expand_mem_thread_fence): Call expand_memory_blockage
3242 instead of expand_asm_memory_barrier.
3243 (expand_mem_singnal_fence): Ditto.
3244 (expand_atomic_load): Ditto.
3245 (expand_atomic_store): Ditto.
3246 * doc/md.texi (Standard Pattern Names For Generation):
3247 Document memory_blockage instruction pattern.
3249 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
3251 * config/rl78/rl78.c (rl78_emit_libcall): New function.
3252 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
3253 * config/rl78/rl78.md: New define_expand "adddi3".
3255 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
3257 * cfghooks.c (verify_flow_info): Disable check that all probabilities
3260 2017-10-13 Jeff Law <law@redhat.com>
3262 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
3264 2017-10-13 Jakub Jelinek <jakub@redhat.com>
3267 * internal-fn.c (expand_mul_overflow): If both operands have
3268 the same highpart of -1 or 0 and the topmost bit of lowpart
3269 is different, overflow is if res <= 0 rather than res < 0.
3271 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
3273 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
3274 TARGET_P9_VECTOR code for unaligned_load case.
3276 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
3278 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
3280 2017-10-13 Nathan Sidwell <nathan@acm.org>
3282 * tree-core.h (tree_contains_struct): Make bool.
3283 * tree.c (tree_contains_struct): Likewise.
3284 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
3285 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
3286 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
3287 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
3288 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
3290 2017-10-13 Richard Biener <rguenther@suse.de>
3292 * graphite-isl-ast-to-gimple.c
3293 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
3294 parameters and dominance check.
3295 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
3296 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
3297 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
3298 Do not update SSA form here or do intermediate IL verification.
3299 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
3300 (graphite_initialize): Remove check on the number of loops in
3301 the function and inline into graphite_transform_loops.
3302 (graphite_finalize): Inline into graphite_transform_loops.
3303 (graphite_transform_loops): Perform SSA update and IL verification
3305 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
3307 2017-10-13 Richard Biener <rguenther@suse.de>
3309 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
3310 graphite_expression_type_precision): Avoid global constructor
3312 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
3313 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
3314 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
3315 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
3316 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
3317 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
3319 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
3322 * config/i386/i386.h (ix86_red_zone_size): New.
3323 * config/i386/i386.md (push peephole2s): Replace
3324 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
3326 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3327 Alan Hayward <alan.hayward@arm.com>
3328 David Sherwood <david.sherwood@arm.com>
3330 * combine.c (can_change_dest_mode): Reject changes in
3331 REGMODE_NATURAL_SIZE.
3333 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3334 Alan Hayward <alan.hayward@arm.com>
3335 David Sherwood <david.sherwood@arm.com>
3337 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
3338 (expand_debug_source_expr): Likewise.
3339 * combine.c (combine_simplify_rtx): Likewise.
3340 * cse.c (fold_rtx): Likewise.
3341 * fwprop.c (canonicalize_address): Likewise.
3342 * targhooks.c (default_shift_truncation_mask): Likewise.
3344 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3345 Alan Hayward <alan.hayward@arm.com>
3346 David Sherwood <david.sherwood@arm.com>
3348 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
3349 (widened_mode): Likewise.
3350 (expand_unop): Likewise.
3351 * ree.c (transform_ifelse): Likewise.
3352 (merge_def_and_ext): Likewise.
3353 (combine_reaching_defs): Likewise.
3354 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3356 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3357 Alan Hayward <alan.hayward@arm.com>
3358 David Sherwood <david.sherwood@arm.com>
3360 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
3361 * combine.c (gen_lowpart_for_combine): Likewise.
3362 * dwarf2out.c (rtl_for_decl_location): Likewise.
3363 * final.c (alter_subreg): Likewise.
3364 * rtlhooks.c (gen_lowpart_general): Likewise.
3365 (gen_lowpart_if_possible): Likewise.
3367 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3368 Alan Hayward <alan.hayward@arm.com>
3369 David Sherwood <david.sherwood@arm.com>
3371 * calls.c (expand_call): Use subreg_lowpart_offset.
3372 * cse.c (cse_insn): Likewise.
3373 * regcprop.c (copy_value): Likewise.
3374 (copyprop_hardreg_forward_1): Likewise.
3376 2017-10-13 Jakub Jelinek <jakub@redhat.com>
3379 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
3380 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
3381 =Q constraints to +Q and into insn condition add check
3382 that operands[0] and operands[1] are equal.
3383 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
3384 =Q constraints to +Q and into insn condition add check
3385 that operands[0] is equal to either operands[1] or operands[2].
3388 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
3389 instead of handling MINUS_EXPR twice (once for each argument),
3390 canonicalize operand order and handle just once, use rtype where
3391 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
3394 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
3395 any values of __C while still being pattern recognizable as a simple
3398 2017-10-13 Richard Biener <rguenther@suse.de>
3400 PR tree-optimization/82451
3402 2017-10-02 Richard Biener <rguenther@suse.de>
3404 PR tree-optimization/82355
3405 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
3406 a mapping for the enclosing loop but avoid generating one for
3408 (copy_bb_and_scalar_dependences): Remove premature codegen
3409 error on PHIs in blocks duplicated into multiple places.
3410 * graphite-scop-detection.c
3411 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
3412 in the region use it as loop and nest to analyze the DR in.
3413 (try_generate_gimple_bb): Likewise.
3414 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
3415 (add_loop_constraints): For blocks in a loop not in the region
3416 create a dimension with a single iteration.
3417 * sese.h (gbb_loop_at_index): Remove assert.
3419 * cfgloop.c (loop_preheader_edge): For the loop tree root
3420 return the single successor of the entry block.
3421 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3422 Reset the SCEV hashtable and niters.
3423 * graphite-scop-detection.c
3424 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
3425 assert that we only have POLYNOMIAL_CHREC that vary in loops
3426 contained in the region.
3427 (scop_detection::graphite_can_represent_expr): Adjust.
3428 (scop_detection::stmt_has_simple_data_refs_p): For loops
3429 not in the region set loop to NULL. The nest is now the
3430 entry edge to the region.
3431 (try_generate_gimple_bb): Likewise.
3432 * sese.c (scalar_evolution_in_region): Adjust for
3433 instantiate_scev change.
3434 * tree-data-ref.h (graphite_find_data_references_in_stmt):
3435 Make nest parameter the edge into the region.
3436 (create_data_ref): Likewise.
3437 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
3438 entry edge into a region and adjust instantiate_scev calls.
3439 (create_data_ref): Likewise.
3440 (graphite_find_data_references_in_stmt): Likewise.
3441 (find_data_references_in_stmt): Pass the loop preheader edge
3442 from the nest argument.
3443 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
3444 parameter the edge into the region.
3445 (instantiate_parameters): Use the loop preheader edge as entry.
3446 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
3448 (get_instantiated_value_entry): Make instantiate_below parameter
3449 the edge into the region.
3450 (instantiate_scev_name): Likewise. Adjust dominance checks,
3451 when we cannot use loop-based instantiation instantiate by
3452 walking use-def chains.
3453 (instantiate_scev_poly): Adjust.
3454 (instantiate_scev_binary): Likewise.
3455 (instantiate_scev_convert): Likewise.
3456 (instantiate_scev_not): Likewise.
3457 (instantiate_array_ref): Remove.
3458 (instantiate_scev_3): Likewise.
3459 (instantiate_scev_2): Likewise.
3460 (instantiate_scev_1): Likewise.
3461 (instantiate_scev_r): Do not blindly handle N-operand trees.
3462 Do not instantiate array-refs. Handle all constants and invariants.
3463 (instantiate_scev): Make instantiate_below parameter
3464 the edge into the region.
3465 (resolve_mixers): Use the loop preheader edge for the region
3466 parameter to instantiate_scev_r.
3467 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
3469 2017-10-13 Richard Biener <rguenther@suse.de>
3471 PR tree-optimization/82525
3472 * graphite-isl-ast-to-gimple.c
3473 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
3475 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
3476 Fail code generation when we cannot represent the isl integer.
3477 (binary_op_to_tree): Elide modulo operations that are no-ops
3478 in the type we code generate. Remove now superfluous code
3481 2017-10-13 Richard Biener <rguenther@suse.de>
3483 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
3484 (scop_detection::harmful_loop_in_region): Remove premature
3485 IV type restriction.
3486 (scop_detection::graphite_can_represent_scev): We can handle
3487 pointer IVs just fine.
3489 2017-10-13 Alan Modra <amodra@gmail.com>
3491 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
3492 "Clobbers and Scratch Registers". Add paragraph on
3493 alternative to clobbers for scratch registers and OpenBLAS
3496 2017-10-13 Alan Modra <amodra@gmail.com>
3498 * doc/extend.texi (Clobbers): Correct vax example. Delete old
3499 example of a memory input for a string of known length. Move
3500 commentary out of table. Add a number of new examples
3501 covering array memory inputs.
3503 2017-10-12 Martin Liska <mliska@suse.cz>
3505 PR tree-optimization/82493
3506 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
3507 (test_range_functions): New function.
3508 (sbitmap_c_tests): Likewise.
3509 * selftest-run-tests.c (selftest::run_tests): Run new tests.
3510 * selftest.h (sbitmap_c_tests): New function.
3512 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
3514 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
3516 * config/rs6000/amo.h: Fix spacing issue.
3518 2017-10-12 Jakub Jelinek <jakub@redhat.com>
3521 * config/i386/i386.md (*ashl<mode>3_mask_1,
3522 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
3523 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
3526 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
3528 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
3529 (profile_probability): Set max_probability
3530 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
3532 * profile-count.c (profile_probability::differs_from_p): Do not
3533 rely on max_probaiblity == 10000
3535 2017-10-12 Jeff Law <law@redhat.com>
3537 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
3540 2017-10-12 Martin Sebor <msebor@redhat.com>
3544 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
3545 (handle_alias_pairs): Call it.
3546 * common.opt (-Wattribute-alias): New option.
3547 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
3548 * doc/invoke.texi (-Wattribute-alias): Document.
3550 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
3553 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
3555 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3557 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3558 (make_hard_regno_born, make_hard_regno_dead): Update
3559 bb_killed_pseudos and bb_gen_pseudos.
3561 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
3563 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
3565 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
3567 * config/alpha/alpha.c (alpha_split_conditional_move):
3568 Use std::swap instead of manually swapping.
3569 (alpha_stdarg_optimize_hook): Ditto.
3570 (alpha_canonicalize_comparison): Ditto.
3572 2017-10-12 Bin Cheng <bin.cheng@arm.com>
3574 * tree-loop-distribution.c (struct builtin_info): New struct.
3575 (struct partition): Refactor fields into struct builtin_info.
3576 (partition_free): Free struct builtin_info.
3577 (build_size_arg_loc, build_addr_arg_loc): Delete.
3578 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
3579 information from struct builtin_info.
3580 (find_single_drs): New function refactored from classify_partition.
3581 Also moved builtin validity checks to this function.
3582 (compute_access_range, alloc_builtin): New functions.
3583 (classify_builtin_st, classify_builtin_ldst): New functions.
3584 (classify_partition): Refactor code into functions find_single_drs,
3585 classify_builtin_st and classify_builtin_ldst.
3586 (distribute_loop): Don't do runtime alias check when distributing
3588 (find_seed_stmts_for_distribution): New function.
3589 (pass_loop_distribution::execute): Refactor code finding seed
3590 stmts into above function. Support distribution for the innermost
3591 two-level loop nest. Adjust dump information.
3593 2017-10-12 Bin Cheng <bin.cheng@arm.com>
3595 * tree-loop-distribution.c: Adjust the general comment.
3596 (NUM_PARTITION_THRESHOLD): New macro.
3597 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
3598 (classify_partition): Skip builtin pattern of loop nest's inner loop.
3599 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
3600 in call to build_partition_graph.
3601 (finalize_partitions): New parameter. Make loop distribution more
3602 conservative by fusing more partitions.
3603 (distribute_loop): Don't do runtime alias check in case of loop nest
3605 (find_seed_stmts_for_distribution): New function.
3606 (prepare_perfect_loop_nest): New function.
3607 (pass_loop_distribution::execute): Refactor code finding seed stmts
3608 and loop nest into above functions. Support loop nest distribution.
3609 Adjust dump information accordingly.
3611 2017-10-12 Bin Cheng <bin.cheng@arm.com>
3613 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
3614 and set PTYPE_SEQUENTIAL for merged partition.
3616 2017-10-12 Richard Biener <rguenther@suse.de>
3618 PR tree-optimization/69728
3620 2017-09-19 Richard Biener <rguenther@suse.de>
3622 PR tree-optimization/69728
3623 * graphite-sese-to-poly.c (schedule_error): New global.
3624 (add_loop_schedule): Handle empty domain by failing the
3626 (build_original_schedule): Handle schedule_error.
3628 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
3629 domain by returning an unchanged schedule.
3631 2017-10-12 Jakub Jelinek <jakub@redhat.com>
3633 * genrecog.c (validate_pattern): For VEC_SELECT verify that
3634 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
3636 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
3638 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
3641 2017-10-12 Jakub Jelinek <jakub@redhat.com>
3644 * expr.c (store_field): Don't optimize away bitsize == 0 store
3645 from CALL_EXPR with addressable return type.
3647 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
3649 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
3650 * config/rs6000/rs6000.md (sel): Delete mode attribute.
3651 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
3652 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
3653 TARGET_ISEL instead of TARGET_ISEL<sel>.
3655 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
3657 * config/rs6000/rs6000.c
3658 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
3660 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
3662 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
3664 (reg_or_cint_operand): Fix comment.
3665 (reg_or_zero_operand): New predicate.
3666 * config/rs6000/rs6000-protos.h (output_isel): Delete.
3667 * config/rs6000/rs6000.c (output_isel): Delete.
3668 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
3669 instead of reg_or_cint_operand. Output instruction directly (not via
3671 (isel_unsigned_<mode>): Ditto.
3672 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
3673 gpc_reg_operand. Add an instruction alternative for this. Output
3674 instruction directly.
3675 (*isel_reversed_unsigned_<mode>): Ditto.
3677 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
3679 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
3680 (TARGET_CANONICALIZE_COMPARISON): Define.
3682 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
3685 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
3686 Check whether the dest is REG before adding REG_EQUIV note.
3688 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
3691 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3693 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3694 (make_hard_regno_born, make_hard_regno_dead): Update
3695 bb_killed_pseudos and bb_gen_pseudos.
3697 2017-10-11 Nathan Sidwell <nathan@acm.org>
3699 * incpath.h (enum incpath_kind): Name enum, prefix values.
3700 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
3701 * incpath.c (heads, tails): Use INC_MAX.
3702 (add_env_var_paths, add_standard_paths): Use incpath_kind.
3703 (merge_include_chains, split_quote_chain,
3704 register_include_chains): Update incpath_kind names.
3705 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
3706 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
3708 (add_framework_path, darwin_register_objc_includes): Likewise.
3709 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
3711 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
3713 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
3714 Do not use float_operator operator predicate.
3715 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
3716 * config/i386/predicates.md (float_operator): Remove predicate.
3718 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
3720 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
3721 (*jccxf_i387): Ditto.
3722 (*jcc<mode>_i387): Ditto.
3723 (*jccu<mode>_i387): Ditto.
3724 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
3725 (*jcc_*_i387 splitters): Remove.
3726 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
3727 * config/i386/i386.c (ix86_split_fp_branch): Remove.
3728 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
3731 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
3733 * profile-count.h (slow_safe_scale_64bit): New function.
3734 (safe_scale_64bit): New inline.
3735 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
3736 * profile-count.c: Include wide-int.h
3737 (slow_safe_scale_64bit): New.
3739 2017-10-11 Nathan Sidwell <nathan@acm.org>
3741 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
3742 HAS_DECL_ASSEMBLER_NAME_P.
3743 * gimple-expr.c (gimple_decl_printable_name: Check
3744 HAS_DECL_ASSEMBLER_NAME_P too.
3745 * ipa-utils.h (type_in_anonymous_namespace_p): Check
3746 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
3747 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
3748 * passes.c (rest_of_decl_compilation): Check
3749 HAS_DECL_ASSEMBLER_NAME_P too.
3750 * recog.c (verify_changes): Likewise.
3751 * tree-pretty-print.c (dump_decl_name): Likewise.
3752 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
3754 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
3755 (SET_DECL_ASSEMBLER_NAME): Use it.
3756 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
3757 (COPY_DECL_ASSEMBLER_NAME): Likewise.
3758 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
3760 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
3762 * config.gcc (i386, x86_64): Add extra objects.
3763 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
3764 (ix86_min_insn_size): Declare.
3765 (ix86_issue_rate): Declare.
3766 (ix86_adjust_cost): Declare.
3767 (ia32_multipass_dfa_lookahead): Declare.
3768 (ix86_macro_fusion_p): Declare.
3769 (ix86_macro_fusion_pair_p): Declare.
3770 (ix86_bd_has_dispatch): Declare.
3771 (ix86_bd_do_dispatch): Declare.
3772 (ix86_core2i7_init_hooks): Declare.
3773 (ix86_atom_sched_reorder): Declare.
3774 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
3775 (COSTS_N_BYTES): Move to x86-tune-costs.h.
3776 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
3777 (rip_relative_addr_p): Rename to ...
3778 (ix86_rip_relative_addr_p): ... this one; export.
3779 (memory_address_length): Update.
3780 (ix86_issue_rate): Move to x86-tune-sched.c.
3781 (ix86_flags_dependent): Move to x86-tune-sched.c.
3782 (ix86_agi_dependent): Move to x86-tune-sched.c.
3783 (exact_dependency_1): Move to x86-tune-sched.c.
3784 (exact_store_load_dependency): Move to x86-tune-sched.c.
3785 (ix86_adjust_cost): Move to x86-tune-sched.c.
3786 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
3787 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
3788 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
3789 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
3790 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
3791 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
3792 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
3793 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
3794 (min_insn_size): Rename to ...
3795 (ix86_min_insn_size): ... this one; export.
3796 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
3797 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
3798 (core2i7_first_cycle_multipass_backtrack): Move to
3799 x86-tune-sched-core.c.
3800 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
3801 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
3802 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
3803 (ix86_avoid_jump_mispredicts): Update.
3804 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
3805 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
3806 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
3807 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
3808 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
3809 (MAX_INSN): Move to ix86-tune-sched-bd.c.
3810 (MAX_IMM): Move to ix86-tune-sched-bd.c.
3811 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
3812 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
3813 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
3814 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
3815 (MAX_STORE): Move to ix86-tune-sched-bd.c.
3816 (BIG): Move to ix86-tune-sched-bd.c.
3817 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
3818 (enum insn_path): Move to ix86-tune-sched-bd.c.
3819 (get_mem_group): Move to ix86-tune-sched-bd.c.
3820 (is_cmp): Move to ix86-tune-sched-bd.c.
3821 (dispatch_violation): Move to ix86-tune-sched-bd.c.
3822 (is_branch): Move to ix86-tune-sched-bd.c.
3823 (is_prefetch): Move to ix86-tune-sched-bd.c.
3824 (init_window): Move to ix86-tune-sched-bd.c.
3825 (allocate_window): Move to ix86-tune-sched-bd.c.
3826 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
3827 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
3828 (process_end_window): Move to ix86-tune-sched-bd.c.
3829 (allocate_next_window): Move to ix86-tune-sched-bd.c.
3830 (find_constant): Move to ix86-tune-sched-bd.c.
3831 (get_num_immediates): Move to ix86-tune-sched-bd.c.
3832 (has_immediate): Move to ix86-tune-sched-bd.c.
3833 (get_insn_path): Move to ix86-tune-sched-bd.c.
3834 (get_insn_group): Move to ix86-tune-sched-bd.c.
3835 (count_num_restricted): Move to ix86-tune-sched-bd.c.
3836 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
3837 (add_insn_window): Move to ix86-tune-sched-bd.c.
3838 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
3839 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
3840 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
3841 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
3842 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
3843 (do_dispatch): Move to ix86-tune-sched-bd.c.
3844 (has_dispatch): Move to ix86-tune-sched-bd.c.
3845 * config/i386/t-i386: Add new object files.
3846 * config/i386/x86-tune-costs.h: New file.
3847 * config/i386/x86-tune-sched-atom.c: New file.
3848 * config/i386/x86-tune-sched-bd.c: New file.
3849 * config/i386/x86-tune-sched-core.c: New file.
3850 * config/i386/x86-tune-sched.c: New file.
3852 2017-10-11 Liu Hao <lh_mouse@126.com>
3854 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
3855 the generic version below instead.
3856 (should_colorize): Recognize Windows consoles as terminals
3858 * pretty-print.c [__MINGW32__] (write_all): New function.
3859 [__MINGW32__] (find_esc_head): Likewise.
3860 [__MINGW32__] (find_esc_terminator): Likewise.
3861 [__MINGW32__] (eat_esc_sequence): Likewise.
3862 [__MINGW32__] (mingw_ansi_fputs): New function that handles
3864 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
3867 2017-10-11 Richard Biener <rguenther@suse.de>
3869 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
3870 Properly call analyze_scalar_evolution with the loop of the stmt.
3872 2017-10-11 Richard Biener <rguenther@suse.de>
3874 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
3875 * tree-core.h (tree_base): Add chrec_var union member.
3876 * tree.h (CHREC_VAR): Remove.
3877 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
3878 * tree-chrec.h (build_polynomial_chrec): Adjust.
3879 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
3880 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
3882 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
3884 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
3885 * match.pd: ... here.
3886 ((T) X == (T) Y): Relax condition.
3888 2017-10-11 Bin Cheng <bin.cheng@arm.com>
3890 PR tree-optimization/82472
3891 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
3893 (break_alias_scc_partitions): Update postorder number.
3895 2017-10-11 Martin Liska <mliska@suse.cz>
3898 * opts.c (parse_no_sanitize_attribute): Do not use error_value
3900 * opts.h (parse_no_sanitize_attribute): Remove last argument.
3902 2017-10-11 Martin Liska <mliska@suse.cz>
3904 * print-rtl.c (print_insn): Move declaration of idbuf
3905 to same scope as name.
3907 2017-10-11 Martin Liska <mliska@suse.cz>
3912 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
3915 2017-10-11 Martin Liska <mliska@suse.cz>
3918 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
3921 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
3923 * config.gcc (default_gnu_indirect_function): Default to yes for
3924 arm*-*-linux* with glibc.
3926 2017-10-11 Richard Biener <rguenther@suse.de>
3928 * tree-scalar-evolution.c (get_scalar_evolution): Handle
3929 default-defs and types we do not want to analyze.
3930 (interpret_loop_phi): Replace unreachable code with an assert.
3931 (compute_scalar_evolution_in_loop): Remove and inline ...
3932 (analyze_scalar_evolution_1): ... here, replacing condition with
3933 what makes the intent clearer. Remove handling of cases
3934 get_scalar_evolution now handles.
3936 2017-10-10 Jim Wilson <wilson@tuliptree.org>
3938 PR rtl-optimization/81434
3939 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
3940 comment for main loop. In sched_group_found if, also add checks for
3941 pass and min_cost_group.
3943 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
3945 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
3946 (rs6000_insn_cost): New function.
3947 * config/rs6000/rs6000.md (cost): New attribute.
3949 2017-10-10 Jakub Jelinek <jakub@redhat.com>
3950 H.J. Lu <hongjiu.lu@intel.com>
3954 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
3955 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
3956 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
3957 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
3958 to OPTION_MASK_ISA_AVX512VL - builtins that have both
3959 OPTION_MASK_ISA_MMX and some other bit set require both
3960 mmx and the ISAs without the mmx bit.
3961 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
3962 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
3963 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
3964 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
3965 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
3966 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
3967 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
3968 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
3969 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
3970 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
3971 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
3972 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
3973 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
3974 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
3975 Add OPTION_MASK_ISA_MMX.
3977 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
3979 * config.gcc (armv7*-*-freebsd*): New target.
3980 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
3982 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
3984 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
3985 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
3988 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
3990 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
3991 a template parameter.
3992 (WIDE_INT_REF_FOR): Update accordingly.
3993 * tree.h (wi::int_traits <const_tree>): Delete.
3994 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
3995 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
3996 (wi::tree_to_wide_ref): New typedef.
3997 (wi::to_wide): New function.
3998 * calls.c (get_size_range): Use wi::to_wide when operating on
4000 * cgraph.c (cgraph_node::create_thunk): Likewise.
4001 * config/i386/i386.c (ix86_data_alignment): Likewise.
4002 (ix86_local_alignment): Likewise.
4003 * dbxout.c (stabstr_O): Likewise.
4004 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
4005 * expr.c (const_vector_from_tree): Likewise.
4006 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
4007 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
4008 (fold_negate_expr_1, int_const_binop_1, const_binop)
4009 (fold_convert_const_int_from_real, optimize_bit_field_compare)
4010 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
4011 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
4012 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
4013 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
4014 (fold_not_const, round_up_loc): Likewise.
4015 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
4016 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
4017 (alloca_call_type): Likewise.
4018 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
4019 * godump.c (go_output_typedef): Likewise.
4020 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
4021 * internal-fn.c (get_min_precision): Likewise.
4022 * ipa-cp.c (ipcp_store_vr_results): Likewise.
4023 * ipa-polymorphic-call.c
4024 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
4025 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
4026 (ipa_modify_call_arguments): Likewise.
4027 * match.pd: Likewise.
4028 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
4029 * print-tree.c (print_node_brief, print_node): Likewise.
4030 * stmt.c (expand_case): Likewise.
4031 * stor-layout.c (layout_type): Likewise.
4032 * tree-affine.c (tree_to_aff_combination): Likewise.
4033 * tree-cfg.c (group_case_labels_stmt): Likewise.
4034 * tree-data-ref.c (dr_analyze_indices): Likewise.
4035 (prune_runtime_alias_test_list): Likewise.
4036 * tree-dump.c (dequeue_and_dump): Likewise.
4037 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
4038 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
4039 * tree-pretty-print.c (dump_generic_node): Likewise.
4040 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
4041 (simple_iv_with_niters): Likewise.
4042 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
4043 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
4044 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
4045 * tree-ssa-loop-niter.c (split_to_var_and_offset)
4046 (refine_value_range_using_guard, number_of_iterations_ne_max)
4047 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
4048 (get_cst_init_from_scev, record_nonwrapping_iv)
4049 (scev_var_range_cant_overflow): Likewise.
4050 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
4051 * tree-ssa-pre.c (compute_avail): Likewise.
4052 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
4053 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
4054 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
4055 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
4056 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
4057 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
4058 (dump_case_nodes, try_switch_expansion): Likewise.
4059 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
4060 (vect_do_peeling): Likewise.
4061 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
4062 * tree-vect-stmts.c (vectorizable_load): Likewise.
4063 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
4064 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
4065 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
4066 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
4067 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
4068 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
4069 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
4070 (evrp_dom_walker::before_dom_children): Likewise.
4071 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
4072 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
4073 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
4074 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
4075 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
4076 (get_range_pos_neg): Likewise.
4077 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
4078 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
4079 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
4080 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
4081 * config/avr/avr.c (avr_fold_builtin): Likewise.
4082 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
4083 * config/msp430/msp430.c (msp430_attr): Likewise.
4084 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
4085 * config/powerpcspe/powerpcspe-c.c
4086 (altivec_resolve_overloaded_builtin): Likewise.
4087 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
4088 (rs6000_expand_ternop_builtin): Likewise.
4089 * config/rs6000/rs6000-c.c
4090 (altivec_resolve_overloaded_builtin): Likewise.
4091 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
4092 (rs6000_expand_ternop_builtin): Likewise.
4093 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
4095 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4097 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
4098 when copying loop nest with only one inner loop.
4100 2017-10-10 Richard Biener <rguenther@suse.de>
4102 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
4103 blocks if SCEV is active.
4104 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
4106 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
4107 (scev_initialize): Assert we are not yet initialized.
4109 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4111 * tree-loop-distribution.c (generate_loops_for_partition): Remove
4112 inner loop's exit stmt by making it always exit the loop, otherwise
4113 we would generate an infinite empty loop.
4115 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4117 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
4118 renaming variables in new preheader if it's deleted.
4120 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4122 * tree-loop-distribution.c (struct partition): Remove unused field
4123 loops of the structure.
4124 (partition_alloc, partition_free): Ditto.
4125 (build_rdg_partition_for_vertex): Ditto.
4127 2017-10-09 Jeff Law <law@redhat.com>
4129 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
4130 return type to match prototype and documentation.
4132 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4134 * config/rs6000/rs6000.c (processor_costs): Move to ...
4135 * config/rs6000/rs6000.h: ... here.
4136 (rs6000_cost): Declare.
4138 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
4140 * except.c (setjmp_fn): New global variable.
4141 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
4142 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
4143 if DONT_USE_BUILTIN_SETJMP is defined.
4145 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4147 * target.def (insn_cost): New hook.
4148 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
4149 * doc/tm.texi: Regenerate.
4150 * rtlanal.c (insn_cost): Use the new hook.
4152 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4154 * combine.c (combine_validate_cost): Compute the new insn_cost,
4155 not just pattern_cost.
4156 (try_combine): Adjust comment.
4158 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4160 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
4162 * combine.c (uid_insn_cost): Adjust comment.
4163 (combine_validate_cost): Adjust comment. Use pattern_cost instead
4165 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
4166 * dse.c (find_shift_sequence): Ditto.
4167 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
4168 (bb_valid_for_noce_process_p): Use pattern_cost.
4169 * rtl.h (insn_rtx_cost): Delete.
4170 (pattern_cost): New prototype.
4171 (insn_cost): New prototype.
4172 * rtlanal.c (insn_rtx_cost): Rename to...
4173 (pattern_cost): ... this.
4176 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
4178 * config/i386/i386.md (*jcc_2): Remove insn pattern.
4179 (*jcc<mode>_0_r_i387): Ditto.
4180 (*jccxf_r_i387): Ditto.
4181 (*jcc<mode>_r_i387): Ditto.
4182 (*jccu<mode>_r_i387): Ditto.
4183 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
4184 (*jcc): Rename from *jcc_1.
4186 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4188 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
4189 deferred rescans after the lvx/stvx recombination pre-pass.
4191 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
4193 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
4194 memory operation instruction support.
4195 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
4196 (rs6000-ibm-aix[789]*): Likewise.
4197 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
4198 Document new functions.
4200 2017-10-09 Richard Biener <rguenther@suse.de>
4202 PR tree-optimization/82397
4203 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
4204 equality only for semantically equal trees.
4206 2017-10-09 Richard Biener <rguenther@suse.de>
4208 PR tree-optimization/82449
4209 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
4210 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
4211 allow constant addresses.
4212 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
4215 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4217 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
4220 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4223 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
4226 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4229 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
4231 2017-10-09 Jakub Jelinek <jakub@redhat.com>
4234 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
4235 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
4237 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
4239 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
4240 (WI_BINARY_PREDICATE_RESULT): Likewise.
4241 (wi::binary_traits::operator_result): New type.
4242 (wi::binary_traits::predicate_result): Likewise.
4243 (generic_wide_int::operator~, unary generic_wide_int::operator-)
4244 (generic_wide_int::operator==, generic_wide_int::operator!=)
4245 (generic_wide_int::operator&, generic_wide_int::and_not)
4246 (generic_wide_int::operator|, generic_wide_int::or_not)
4247 (generic_wide_int::operator^, generic_wide_int::operator+
4248 (binary generic_wide_int::operator-, generic_wide_int::operator*):
4250 (operator~, unary operator-, operator==, operator!=, operator&)
4251 (operator|, operator^, operator+, binary operator-, operator*): New
4253 * expr.c (get_inner_reference): Use wi::bit_and_not.
4254 * fold-const.c (fold_binary_loc): Likewise.
4255 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
4256 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
4257 (bit_value_binop): Likewise.
4258 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
4259 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
4260 (extract_range_from_binary_expr_1): Likewise.
4261 (masked_increment): Likewise.
4262 (simplify_bit_ops_using_ranges): Likewise.
4264 2017-10-09 Martin Jambor <mjambor@suse.cz>
4267 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
4268 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
4269 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
4271 (hsa_fixup_mov_insn_type): New function.
4272 (hsa_op_with_type::get_in_type): Use it.
4273 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
4274 immediates in an assert.
4275 (hsa_op_with_type::extend_int_to_32bit): New method.
4276 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
4277 types. Convert to dest type if necessary.
4278 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
4279 (reg_for_gimple_ssa): Pass false as min32int to
4280 hsa_type_for_scalar_tree_type.
4281 (gen_hsa_addr): Fixup type when creating addresable temporary.
4282 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
4283 (gen_hsa_unary_operation): Extend operands and convert to dest type if
4284 necessary. Call hsa_fixup_mov_insn_type.
4285 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
4286 extend operands and convert to dest type if necessary.
4287 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
4288 to dest type if necessary.
4289 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
4290 if conversion nt necessary and size matches.
4291 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
4292 to dest type if necessary.
4293 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
4294 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
4296 (gen_hsa_clrsb): Likewise.
4297 (gen_hsa_ffs): Likewise.
4298 (gen_hsa_divmod): Extend operands and convert to dest type if
4300 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
4302 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
4304 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
4305 Remove empty default arguments. Use a brace block as output
4307 (conditional return): Ditto.
4309 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
4310 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
4312 (group_ending_nop): Ditto.
4313 (doloop_end): Ditto.
4314 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
4315 (splitters for those): Ditto.
4317 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
4319 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
4320 a conditional jump (and the compare for it) so that pc_rtx is the
4322 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
4323 for the deleted and renamed ctr<mode>_internal[234] patterns.
4324 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
4325 Delete second conditional return pattern.
4326 (ctr<mode>_internal2): Delete this second bdnz pattern.
4327 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
4328 (ctr<mode>_internal4): Delete this second bdz pattern.
4330 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
4332 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
4333 (always_initialized_rtx_for_ssa_name_p): New predicate.
4334 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
4335 (finish_out_of_ssa): Free new field of SA.
4336 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
4337 * tree-ssa-coalesce.c: Include tree-ssa.h.
4338 (get_parm_default_def_partitions): Remove extern keyword.
4339 (get_undefined_value_partitions): New function.
4340 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
4341 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
4343 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
4345 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
4347 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
4349 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
4351 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
4352 for targets that preffer 128bit.
4354 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
4356 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
4358 2017-10-08 Olivier Hainque <hainque@adacore.com>
4360 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
4361 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
4363 (thumb_set_return_address): Likewise.
4365 2017-10-08 Olivier Hainque <hainque@adacore.com>
4367 * common/config/arm/arm-common.c (arm_except_unwind_info):
4368 Handle DWARF2_UNWIND_INFO.
4370 2017-10-07 Michael Collison <michael.collison@arm.com>
4372 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
4375 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
4377 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
4378 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
4379 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
4380 defined, force the creation of a new block for a dispatch label.
4382 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
4384 * invoke.texi (Wsuggest-attribute=cold): Document.
4385 * common.opt (Wsuggest-attribute=cold): New
4386 * ipa-pure-const.c (warn_function_cold): New function.
4387 * predict.c (compute_function_frequency): Use it.
4388 * predict.h (warn_function_cold): Declare.
4390 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
4392 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
4395 2017-10-06 Martin Liska <mliska@suse.cz>
4397 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
4398 keyword for member functions.
4399 (struct sanopt_tree_couple): New struct.
4400 (struct sanopt_tree_couple_hash): New function.
4401 (struct sanopt_ctx): Add new hash_map.
4402 (has_dominating_ubsan_ptr_check): New function.
4403 (record_ubsan_ptr_check_stmt): Likewise.
4404 (maybe_optimize_ubsan_ptr_ifn): Likewise.
4405 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
4406 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
4408 2017-10-06 Sudakshina Das <sudi.das@arm.com>
4411 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
4412 aarch64_simd_valid_immediate on CONST_VECTORs.
4413 (aarch64_reg_or_bic_imm): Likewise.
4415 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
4417 PR rtl-optimization/82396
4418 * haifa-sched.c (ready_sort_real): Disable qsort checking.
4420 2017-10-06 Sebastian Pop <sebpop@gmail.com>
4422 * graphite-dependences.c (scop_get_reads): Move code to...
4423 (scop_get_must_writes): Move code to...
4424 (scop_get_may_writes): Move code to...
4425 (scop_get_reads_and_writes): ... here.
4426 (scop_get_dependences): Call scop_get_reads_and_writes.
4428 2017-10-06 Jakub Jelinek <jakub@redhat.com>
4430 PR tree-optimization/82434
4431 * fold-const.h (can_native_encode_type_p,
4432 can_native_encode_string_p): Remove.
4433 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
4434 don't encode anything, just return what would be otherwise returned.
4435 (native_encode_fixed, native_encode_complex, native_encode_vector):
4437 (native_encode_string): Likewise. Inline by hand
4438 can_native_encode_string_p.
4439 (can_native_encode_type_p): Remove.
4440 (can_native_encode_string_p): Remove.
4441 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
4442 STRING_CSTs using can_native_encode_string_p, test all
4443 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
4444 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
4445 argument from native_encode_expr.
4446 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
4447 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
4450 2017-10-06 Richard Biener <rguenther@suse.de>
4452 PR tree-optimization/82397
4453 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
4454 operand_equal_p but rely on data_ref_compare_tree for detecting
4456 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
4457 to match up with dr_group_sort_cmp.
4459 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4462 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
4464 * config/s390/s390-builtin-types.def: Regenerate.
4466 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4469 * config/s390/s390-builtin-types.def: Regenerate.
4470 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
4471 Change flag from B_VXE to B_VX.
4472 (s390_vec_min_dbl): Remove B_VXE flag.
4474 2017-10-06 Richard Biener <rguenther@suse.de>
4476 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
4477 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
4478 translate_isl_ast_to_gimple::is_valid_rename,
4479 translate_isl_ast_to_gimple::get_rename,
4480 translate_isl_ast_to_gimple::get_def_bb_for_const,
4481 translate_isl_ast_to_gimple::get_new_name,
4482 translate_isl_ast_to_gimple::collect_all_ssa_names,
4483 translate_isl_ast_to_gimple::copy_loop_phi_args,
4484 translate_isl_ast_to_gimple::collect_all_ssa_names,
4485 translate_isl_ast_to_gimple::copy_loop_phi_args,
4486 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
4487 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
4488 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
4489 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
4490 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
4491 translate_isl_ast_to_gimple::copy_cond_phi_args,
4492 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
4493 translate_isl_ast_to_gimple::edge_for_new_close_phis,
4494 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
4495 translate_isl_ast_to_gimple::rename_uses,
4496 translate_isl_ast_to_gimple::rename_all_uses): Remove.
4497 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
4498 (set_rename_for_each_def): Likewise.
4499 (graphite_copy_stmts_from_block): Handle debug stmt resetting
4500 here. Handle rewriting SCEV analyzable uses here.
4501 (copy_bb_and_scalar_dependences): Generate code for PHI
4503 (graphite_regenerate_ast_isl): Adjust.
4504 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
4505 (add_write, add_read): New functions.
4506 (build_cross_bb_scalars_def): Use it and simplify.
4507 (build_cross_bb_scalars_use): Likewise.
4508 (graphite_find_cross_bb_scalar_vars): Inline into...
4509 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
4510 simulating out-of-SSA. Compute liveout and add dependencies.
4511 (build_scops): Force an empty entry block.
4512 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
4514 (sese_build_liveouts): Declare.
4515 (sese_trivially_empty_bb_p): Likewise.
4516 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
4517 compute liveout and debug_liveout.
4518 (sese_bad_liveouts_use): Remove.
4519 (sese_reset_debug_liveouts_bb): Likewise.
4520 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
4521 (sese_build_liveouts): Build liveout and debug_liveout and store
4523 (new_sese_info): Adjust.
4524 (free_sese_info): Likewise.
4525 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
4526 do not build liveout here.
4527 (move_sese_in_condition): Adjust region entry.
4528 (scev_analyzable_p): Match up with chrec_apply requirements.
4529 (sese_trivially_empty_bb_p): New.
4530 * tree-into-ssa.c (get_reaching_def): Properly support generating
4531 default-defs for incremental rewrite of anonymous names.
4533 2017-10-06 Richard Biener <rguenther@suse.de>
4535 * graphite-sese-to-poly.c (extract_affine): For casts increasing
4536 precision do not perform modulo reduction.
4538 2017-10-06 Richard Biener <rguenther@suse.de>
4540 PR tree-optimization/82436
4541 * tree-vect-slp.c (vect_supported_load_permutation_p): More
4542 conservatively choose the vectorization factor when checking
4543 whether we can perform the required load permutation.
4544 (vect_transform_slp_perm_load): Assert when we may not fail.
4546 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
4548 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
4549 message for incompatible -msdata=* and -mcall-* options.
4551 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
4553 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
4554 rate for post-reload scheduling.
4556 2017-10-05 Tamar Christina <tamar.christina@arm.com>
4558 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
4560 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
4562 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
4563 to improve monte carlo in scimark.
4565 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
4567 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
4568 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
4569 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
4570 pentium4_cost, nocona_cost): Set reassociation width to 1.
4571 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
4572 width to 2 for fp operations and 1 otherwise.
4573 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
4575 (atom_cost): Set reassociation width to 2.
4576 (slm_cost, generic_cost): Set fp reassociation width
4577 to 2 and 1 otherwise.
4578 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
4579 (core_cost): Set fp reassociation width to 4 and vector to 2.
4580 (ix86_reassociation_width): Rewrite using cost table; special case
4581 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
4582 and TARGET_AVX128_OPTIMAL.
4583 * config/i386/i386.h (processor_costs): Add
4584 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
4585 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
4586 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
4587 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
4588 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
4589 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
4591 2017-10-05 Nathan Sidwell <nathan@acm.org>
4593 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
4595 2017-10-05 Tamar Christina <tamar.christina@arm.com>
4597 * config/arm/arm.c (arm_test_fpu_data): New.
4598 (arm_run_selftests): Call arm_test_fpu_data.
4600 2017-10-04 Nathan Sidwell <nathan@acm.org>
4602 * toplev.c (toplev::main): Remove excess parens on pretty_printer
4604 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
4606 2017-10-04 Sudakshina Das <sudi.das@arm.com>
4608 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
4609 check type for aarch64_simd_valid_immediate.
4610 (aarch64_output_simd_mov_immediate): Update prototype.
4611 (aarch64_simd_valid_immediate): Update prototype.
4612 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
4613 support for ORR-immediate.
4614 (and<mode>3): modified pattern to add support for BIC-immediate.
4615 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
4616 now checks for valid immediate for BIC and ORR based on new enum
4618 (aarch64_output_simd_mov_immediate): Function now used to output
4619 BIC/ORR imm as well based on new enum argument.
4620 * config/aarch64/constraints.md (Do): New vector immediate constraint.
4622 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
4623 (aarch64_reg_or_bic_imm): Likewise.
4625 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4627 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
4628 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
4630 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
4634 PR rtl-optimization/82396
4635 * haifa-sched.c (autopref_multipass_init): Simplify
4637 (autopref_rank_data): Simplify sort order.
4638 * sched-int.h (autopref_multipass_data_): Remove
4639 multi_mem_insn_p, min_offset and max_offset.
4641 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4643 * doc/sourcebuild.texi: Document vect_peeling_profitable.
4645 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4647 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
4650 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4652 * doc/sourcebuild.texi: Document vect_long_mult.
4654 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
4656 PR tree-optimization/82413
4657 * fold-const.c (build_range_check): Use widest_int when comparing
4658 the maximum ETYPE value with HIGH.
4660 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
4662 PR rtl-optimization/82396
4663 * haifa-sched.c (autopref_multipass_init): Simplify
4665 (autopref_rank_data): Simplify sort order.
4666 * sched-int.h (autopref_multipass_data_): Remove
4667 multi_mem_insn_p, min_offset and max_offset.
4669 2017-10-04 Jakub Jelinek <jakub@redhat.com>
4671 PR tree-optimization/82381
4672 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
4673 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
4676 PR tree-optimization/82374
4677 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
4678 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
4679 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
4680 current_function_decl to the new decl.
4682 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
4684 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
4685 helper macro for IEEE float128 hardware built-in functions.
4686 (SQRTF128_ODD): Add built-in functions with the round-to-odd
4688 (TRUNCF128_ODD): Likewise.
4689 (ADDF128_ODD): Likewise.
4690 (SUBF128_ODD): Likewise.
4691 (MULF128_ODD): Likewise.
4692 (DIVF128_ODD): Likewise.
4693 (FMAF128_ODD): Likewise.
4694 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
4695 UNSPEC_TRUNC_ROUND_TO_ODD.
4696 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
4697 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
4698 floating point round to odd instructions.
4699 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
4700 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
4701 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
4702 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
4703 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
4704 (trunc<mode>sf2_hw): Change the truncate with round to odd
4705 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
4706 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
4707 to odd hardware instructions.
4708 (sub<mode>3_odd): Likewise.
4709 (mul<mode>3_odd): Likewise.
4710 (div<mode>3_odd): Likewise.
4711 (sqrt<mode>2_odd): Likewise.
4712 (fma<mode>4_odd): Likewise.
4713 (fms<mode>4_odd): Likewise.
4714 (nfma<mode>4_odd): Likewise.
4715 (nfms<mode>4_odd): Likewise.
4716 (trunc<mode>df2_odd): Change the truncate with round to odd
4717 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
4719 * doc/extend.texi (PowerPC built-in functions): Update documentation
4720 for existing IEEE float128-bit built-in functions. Add built-in
4721 functions that generate the IEEE 128-bit floating point round to
4724 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
4726 PR rtl-optimization/77729
4727 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
4728 to (X&(C1&~C2))|C2 transformations.
4730 2017-10-03 Martin Jambor <mjambor@suse.cz>
4732 PR tree-optimization/82363
4733 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
4734 mismatch, mark lacc written regardless of racc.
4736 2017-10-03 Jakub Jelinek <jakub@redhat.com>
4738 PR tree-optimization/82381
4739 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
4740 stmt_to_insert nor wheather SSA_NAMEs are default defs.
4741 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
4742 fallthrough into reassoc_stmt_dominates_stmt_p.
4745 * combine.c (combine_instructions): Don't combine in unreachable
4748 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
4751 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
4752 function to not use the have_cpu variable. Do not set cpu_index,
4753 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
4755 (rs6000_valid_attribute_p): Remove duplicate initializations of
4756 old_optimize and func_optimize.
4757 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
4758 (rs6000_activate_target_options): Make global.
4759 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
4762 2017-10-02 Jakub Jelinek <jakub@redhat.com>
4764 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
4765 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
4766 Set *poffset to 0 and *psize and *pmax_size to -1 if
4767 *poffset + *psize overflows in HOST_WIDE_INT.
4769 PR tree-optimization/82387
4770 PR tree-optimization/82388
4771 PR tree-optimization/82389
4772 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
4773 instead of live_bytes non-NULL.
4775 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
4778 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
4781 2017-10-02 Richard Biener <rguenther@suse.de>
4783 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
4784 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
4785 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
4787 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
4789 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
4790 requested precision matches the type's.
4791 * calls.c (alloc_max_size): Calculate the new candidate size as
4792 a widest_int and use wi::to_widest when comparing it with the
4793 current candidate size.
4794 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
4795 zero rather than integer_zero_node.
4796 * match.pd: Check for a no-op conversion before using wi::add
4797 rather than after. Use tree_to_uhwi when summing small shift
4798 counts into an unsigned int.
4800 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
4801 Alan Hayward <alan.hayward@arm.com>
4802 David Sherwood <david.sherwood@arm.com>
4805 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
4806 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
4807 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
4808 POINTER_AND_FP_REGS.
4810 2017-10-02 Richard Biener <rguenther@suse.de>
4812 PR tree-optimization/82355
4813 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
4814 a mapping for the enclosing loop but avoid generating one for
4816 (copy_bb_and_scalar_dependences): Remove premature codegen
4817 error on PHIs in blocks duplicated into multiple places.
4818 * graphite-scop-detection.c
4819 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
4820 in the region use it as loop and nest to analyze the DR in.
4821 (try_generate_gimple_bb): Likewise.
4822 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
4823 (add_loop_constraints): For blocks in a loop not in the region
4824 create a dimension with a single iteration.
4825 * sese.h (gbb_loop_at_index): Remove assert.
4827 2017-10-01 Kevin Buettner <kevinb@redhat.com>
4829 * omp-expand.c (adjust_context_scope): New function.
4830 (expand_parallel_call): Call adjust_context_scope.
4832 2017-10-01 Jeff Law <law@redhat.com>
4834 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
4835 dom_opt_dom_walker class with direct access to private members.
4836 Add comments. Call test_for_singularity.
4837 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
4838 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
4839 m_dummy_cond anymore.
4840 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
4842 (pass_dominator:execute): Build the dummy_cond here and pass it
4843 to the dom_opt_dom_walker ctor.
4844 (test_for_singularity): New function.
4846 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
4847 Maya Rashish <coypu@sdf.org>
4849 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
4850 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
4851 (alpha*-*-netbsd*) Use nbsd_tm_file.
4852 (arm*-*-netbsdelf*) Likewise.
4853 (i[34567]86-*-netbsdelf*) Likewise.
4854 (x86_64-*-netbsd*) Likewise.
4855 (mips*-*-netbsd*) Likewise.
4856 (powerpc-*-netbsd*) Likewise.
4857 (sh*-*-netbsd*) Likewise.
4858 (sparc-*-netbsdelf*) Likewise.
4859 (sparc64-*-netbsd*) Likewise.
4860 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
4862 (vax-*-netbsdelf*) Likewise.
4863 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
4864 (UINT_FAST8_TYPE) Likewise.
4865 (INT_FAST16_TYPE) Check CHAR_FAST16.
4866 (UINT_FAST16_TYPE) Likewise.
4868 2017-09-30 Jakub Jelinek <jakub@redhat.com>
4871 * config/i386/i386.md
4872 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
4873 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
4874 *divmodsi4_zext_2): New define_insn_and_split.
4875 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
4876 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
4877 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
4878 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
4879 New define_insn_and_split.
4880 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
4881 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
4882 operands[1] having DImode when mode is SImode.
4884 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
4885 always SImode for DIV and MOD in REG_EQUAL notes.
4887 2017-09-29 Yury Gribov <tetra2005@gmail.com>
4890 * match.pd: Fix handling of NaNs in pattern.
4892 2017-09-29 Jeff Law <law@redhat.com>
4894 * sbitmap.c (bitmap_bit_in_range_p): New function.
4895 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
4896 * tree-ssa-dse.c (live_bytes_read): New function.
4897 (dse_classify_store): Ignore reads of dead bytes.
4899 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
4900 typos and whitespace errors.
4901 * config/i386/predicates.md (address_no_seg_operand): Likewise.
4902 * config/s390/s390.c (s390_emit_prologue): Likewise.
4904 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
4907 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
4908 with a symbol for LRA.
4910 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
4912 PR rtl-optimization/82338
4913 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
4915 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
4917 * genmodes.c (calc_wider_mode): Suppress qsort macro.
4918 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
4919 (qsort_chk): Declare.
4920 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
4921 (qsort_chk): New function.
4923 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4925 PR tree-optimization/82337
4926 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
4927 phi definition if the PHI result appears in an abnormal PHI.
4928 (find_basis_for_base_expr): Don't record a basis if the LHS of the
4929 basis appears in an abnormal PHI.
4931 2017-09-29 Richard Biener <rguenther@suse.de>
4933 * graphite-isl-ast-to-gimple.c
4934 (translate_isl_ast_to_gimple::set_codegen_error): New function.
4935 (binary_op_to_tree): Use it.
4936 (get_rename_from_scev): Likewise.
4937 (copy_loop_phi_nodes): Likewise.
4938 (copy_bb_and_scalar_dependences): Likewise.
4939 (translate_pending_phi_nodes): Likewise.
4941 2017-09-29 Jakub Jelinek <jakub@redhat.com>
4944 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
4945 for movabsq $(i32 << shift), r64.
4947 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
4949 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
4950 index when encoding %esp as %rsp to avoid 0x67 prefix.
4952 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
4954 * config/i386/i386.md (*movsf_internal, *movdf_internal):
4955 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
4957 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
4959 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
4960 Extensions with more than 16 double VFP registers.
4961 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
4962 to_clear_mask and all code related to it. Replace the remaining
4963 entry by a sbitmap and adapt code accordingly.
4965 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
4967 * brig-builtins.def: Change pure attributes to const.
4969 2017-09-28 Joseph Myers <joseph@codesourcery.com>
4971 * config.gcc (default_gnu_indirect_function): Default to yes for
4972 sparc*-*-linux* with glibc.
4974 2017-09-28 Joseph Myers <joseph@codesourcery.com>
4976 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
4977 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
4978 when creating .init_array and .fini_array sections with priority
4981 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
4984 * config/aarch64/aarch64.c
4985 (aarch64_builtin_support_vector_misalignment): Always return false
4986 when misalignment is unknown.
4988 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
4990 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
4991 this function to return false if the definition used by the swap
4992 instruction is artificial, or if the memory address from which the
4993 constant value is loaded is not represented by a base address held
4994 in a register or if the base address register is a frame or stack
4995 pointer. Additionally, return false if the base address of the
4996 loaded constant is a SYMBOL_REF but is not considered to be a
4998 (replace_swapped_load_constant): New function.
4999 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
5000 loaded constant vector with a load of a swapped constant vector.
5002 2017-09-27 Carl Love <cel@us.ibm.com>
5004 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
5005 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
5006 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
5009 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
5011 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
5012 first, always call autopref_rank_data otherwise.
5014 2017-09-27 Richard Biener <rguenther@suse.de>
5016 * graphite-scop-detection.c (find_scop_parameters): Move
5017 loop bound handling ...
5018 (gather_bbs::before_dom_children): ... here, avoiding the need
5019 to build scop_info->loop_nest.
5020 (record_loop_in_sese): Remove.
5021 * sese.h (sese_info_t::loop_nest): Remove.
5022 * sese.c (new_sese_info): Do not allocate loop_nest.
5023 (free_sese_info): Do not free loop_nest.
5025 2017-09-27 Jakub Jelinek <jakub@redhat.com>
5028 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
5029 lhs from calls if the lhs has addressable type.
5031 2017-09-27 Richard Biener <rguenther@suse.de>
5033 * graphite.h (scop::max_alias_set): New member.
5034 * graphite-scop-detection.c: Remove references to non-existing
5035 --param in comments.
5036 (build_alias_sets): Record the maximum alias set used for drs.
5037 (build_scops): Support zero as unlimited for
5038 --param graphite-max-arrays-per-scop.
5039 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
5041 (build_poly_sr_1): ... here. Compute alias set based on the
5042 maximum alias set used for drs rather than
5043 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
5045 2017-09-27 Richard Biener <rguenther@suse.de>
5047 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
5048 --param loop-block-tile-size=0 to disable tiling.
5050 2017-09-27 Richard Biener <rguenther@suse.de>
5052 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
5053 (graphite-max-nb-scop-params): Document special value zero.
5054 * domwalk.h (dom_walker::STOP): New symbolical constant.
5055 (dom_walker::dom_walker): Add optional parameter for bb to
5057 (dom_walker::~dom_walker): Declare.
5058 (dom_walker::before_dom_children): Document STOP return value.
5059 (dom_walker::m_user_bb_to_rpo): New member.
5060 (dom_walker::m_bb_to_rpo): Likewise.
5061 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
5062 mapping here if not provided by the user.
5063 (dom_walker::~dom_walker): Free bb to RPO mapping if not
5064 provided by the user.
5065 (dom_walker::STOP): Define.
5066 (dom_walker::walk): Do not compute bb to RPO mapping here.
5067 Support STOP return value from before_dom_children to stop
5069 * graphite-optimize-isl.c (optimize_isl): If the schedule
5070 is the same still generate code if -fgraphite-identity
5071 or -floop-parallelize-all are given.
5072 * graphite-scop-detection.c: Include cfganal.h.
5073 (gather_bbs::gather_bbs): Get and pass through bb to RPO
5075 (gather_bbs::before_dom_children): Return STOP for BBs
5077 (build_scops): Compute bb to RPO mapping and pass it to
5078 the domwalk. Treat --param graphite-max-nb-scop-params=0
5079 as not limiting the number of params.
5080 * graphite.c (graphite_initialize): Remove limit on the
5081 number of basic-blocks in a function.
5082 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
5083 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
5084 default value of 10.
5086 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5088 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
5089 Adjust code to eliminate needing to do the shift right 32-bits
5090 operation after XSCVDPSPN.
5092 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5094 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
5095 ((X / Y) != 0 -> X >= Y): Likewise.
5097 2017-09-26 Carl Love <cel@us.ibm.com>
5099 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
5100 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
5101 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
5102 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
5103 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
5104 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
5105 definitions and overloading.
5106 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
5107 statement for P9V_BUILTIN_XST_LEN_R.
5108 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
5109 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
5110 define_expand and define_insn for the instructions and builtins.
5111 * doc/extend.texi: Update the built-in documentation file for the new
5113 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
5114 define_insn for the instructions
5116 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
5119 * gcc/config/netbsd-protos.h: New file.
5120 * gcc/config/netbsd.c: New file.
5121 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
5122 * gcc/config/t-netbsd: New file.
5123 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
5124 (tmake_file) Add t-netbsd.
5125 (extra_objs) Add netbsd.o.
5127 2017-09-26 Janus Weil <janus@gcc.gnu.org>
5131 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
5133 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5135 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
5136 sign extension from a vector register to a GPR by doing a 32-bit
5137 direct move and then an EXTSW.
5138 (extendsi<mode>2 splitter): Likewise.
5139 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
5140 right or vector extract after doing XSCVDPSPN. Use
5141 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
5143 (movdi_from_sf_zero_ext): Likewise.
5144 (reload_gpr_from_vsxsf): Likewise.
5145 (p8_mfvsrd_4_disf): Delete, no longer used.
5146 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
5147 then needing to move the SFmode to a GPR to use the XSCVDPSP
5148 instruction instead of FRSP and XSCVDPSPN.
5149 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
5150 it is adjacent to the other XSCVSPDP insns.
5151 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
5152 SFmode to be in traditional Altivec registers.
5153 (vsx_xscvdpspn): Eliminate useless alternative constraint.
5154 (vsx_xscvspdpn): Likewise.
5155 (vsx_xscvspdpn_scalar): Likewise.
5157 2017-09-26 Martin Jambor <mjambor@suse.cz>
5159 * tree-sra.c (compare_access_positions): Put integral types first,
5160 stabilize sorting of integral types, remove conditions putting
5161 non-full-precision integers last.
5162 (sort_and_splice_var_accesses): Disable scalarization if a
5163 non-integert would be represented by a non-full-precision integer.
5165 2017-09-26 Joseph Myers <joseph@codesourcery.com>
5167 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
5168 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
5169 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
5170 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
5171 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
5172 conditionals inside the function instead of around it. Call
5173 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
5174 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
5176 2017-09-26 Richard Biener <rguenther@suse.de>
5178 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
5180 (scop_detection::build_scop_breadth): ... this. Removed.
5181 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
5182 (scop_detection::harmful_stmt_in_bb): Likewise.
5183 (scop_detection::graphite_can_represent_stmt): Likewise.
5184 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
5185 (scop_detection::can_represent_loop): Remove recursion, fold in ...
5186 (scop_detection::can_represent_loop_1): ... this. Removed.
5187 (scop_detection::harmful_loop_in_region): Simplify after inlining
5188 the above and remove more quadraticness.
5189 (build_scops): Adjust.
5190 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
5193 2017-09-26 Jakub Jelinek <jakub@redhat.com>
5196 * config/i386/i386.c (ix86_print_operand_address_as): Only test
5197 REGNO (base) == SP_REG if base is a REG.
5200 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
5201 if it is different SSA_NAME.
5202 (optimize_range_tests_cmp_bitwise): New function.
5203 (optimize_range_tests): Call it.
5205 2017-09-26 Richard Biener <rguenther@suse.de>
5207 PR tree-optimization/82321
5208 * graphite.c (canonicalize_loop_closed_ssa): Properly check
5209 for the def being inside the loop.
5211 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5213 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
5215 * config/s390/s390-builtins.def: Fix constraint on op4.
5217 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5219 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
5220 independent expanders.
5221 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
5222 ("vec_ordered", "vec_unordered"): New expanders.
5224 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5226 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
5229 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5231 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
5232 vec_unpacks_lo_v16qi.
5233 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
5235 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5237 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
5238 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
5239 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
5241 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5243 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
5245 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
5247 ("*vec_slb<mode>"): New insn pattern.
5248 ("vec_shr_<mode>"): New expander.
5249 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
5250 and force the shift count operand to V16QImode.
5251 ("vec_srb<mode>"): Set shift count mode to V16QI.
5253 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5255 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
5256 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
5257 ("vec_widen_smult_hi_<mode>"): New expander definitions.
5259 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
5262 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
5264 2017-09-26 Richard Biener <rguenther@suse.de>
5266 PR tree-optimization/82320
5267 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
5270 2017-09-25 Jeff Law <law@redhat.com>
5272 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
5273 prototype for new argument.
5274 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
5275 mostly extracted from rs6000_emit_allocate_stack.
5276 (rs6000_emit_probe_stack_range_stack_clash): New function.
5277 (rs6000_emit_allocate_stack): Call
5278 rs6000_emit_probe_stack_range_stack_clash as needed.
5279 (rs6000_emit_probe_stack_range): Add additional argument
5280 to call to gen_probe_stack_range{si,di}.
5281 (output_probe_stack_range): New.
5282 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
5283 (output_probe_stack_range_stack_clash): New.
5284 (rs6000_emit_prologue): Emit notes into dump file as requested.
5285 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
5286 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
5287 Add additional operand and pass it to output_probe_stack_range.
5289 2017-09-25 Bin Cheng <bin.cheng@arm.com>
5291 PR tree-optimization/82163
5292 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
5293 (checking_verify_loop_closed_ssa): New parameter.
5294 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
5295 (check_loop_closed_ssa_stmt): Delete.
5296 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
5297 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
5298 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
5301 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
5303 * brig-builtins.def: Treat HSAIL barrier builtins as
5304 setjmp/longjump style functions.
5306 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
5308 * target.def (constant_alignment): New hook.
5309 * defaults.h (CONSTANT_ALIGNMENT): Delete.
5310 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
5311 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
5312 * doc/tm.texi: Regenerate.
5313 * targhooks.h (default_constant_alignment): Declare.
5314 (constant_alignment_word_strings): Likewise.
5315 * targhooks.c (default_constant_alignment): New function.
5316 (constant_alignment_word_strings): Likewise.
5317 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
5318 instead of CONSTANT_ALIGNMENT.
5319 * varasm.c (align_variable, get_variable_align, build_constant_desc)
5320 (force_const_mem): Likewise.
5321 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
5322 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
5323 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
5324 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5325 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
5327 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
5328 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5329 constant_alignment_word_strings.
5330 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
5331 (CONSTANT_ALIGNMENT): Likewise.
5332 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5333 (arm_constant_alignment): New function.
5334 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
5335 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5336 constant_alignment_word_strings.
5337 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
5338 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5339 constant_alignment_word_strings.
5340 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
5341 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5342 (cris_constant_alignment): New function.
5343 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
5344 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5345 (epiphany_constant_alignment): New function.
5346 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
5347 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5348 constant_alignment_word_strings.
5349 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
5350 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5351 constant_alignment_word_strings.
5352 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
5353 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5354 constant_alignment_word_strings.
5355 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
5356 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
5357 * config/i386/i386.c (ix86_constant_alignment): Make static.
5358 Use the same interface as the target hook.
5359 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5360 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
5361 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5362 constant_alignment_word_strings.
5363 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
5364 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
5365 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5366 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
5367 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5368 constant_alignment_word_strings.
5369 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
5370 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5371 constant_alignment_word_strings.
5372 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
5373 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5374 constant_alignment_word_strings.
5375 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
5376 * config/microblaze/microblaze.c (microblaze_constant_alignment):
5378 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5379 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
5380 * config/mips/mips.c (mips_constant_alignment): New function.
5381 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5382 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
5383 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
5384 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5385 (mmix_constant_alignment): Make static. Use the same interface
5387 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
5388 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5389 constant_alignment_word_strings.
5390 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
5391 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5392 constant_alignment_word_strings.
5393 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
5394 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5395 constant_alignment_word_strings.
5396 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
5397 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5398 (rs6000_constant_alignment): New function.
5399 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
5400 * config/riscv/riscv.c (riscv_constant_alignment): New function.
5401 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5402 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
5403 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5404 (rs6000_constant_alignment): New function.
5405 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
5406 * config/s390/s390.c (s390_constant_alignment): New function.
5407 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5408 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
5409 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5410 constant_alignment_word_strings.
5411 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
5412 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5413 (sparc_constant_alignment): New function.
5414 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
5415 * config/spu/spu.c (spu_constant_alignment): New function.
5416 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5417 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
5418 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5419 constant_alignment_word_strings.
5420 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
5421 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5422 constant_alignment_word_strings.
5423 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
5424 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5425 constant_alignment_word_strings.
5426 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
5427 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5428 (visium_constant_alignment): New function.
5429 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
5430 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5431 (xtensa_constant_alignment): New function.
5432 * system.h (CONSTANT_ALIGNMENT): Poison.
5434 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
5436 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
5437 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
5438 (rs6000_builtin_valid_without_lhs): New helper function.
5439 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5440 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
5442 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
5444 * target.h (vec_perm_indices): Use unsigned short rather than
5446 (auto_vec_perm_indices): Likewise.
5447 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
5448 Use unsigned int rather than unsigned char.
5449 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
5451 2017-09-25 Richard Biener <rguenther@suse.de>
5453 * cfgloop.h (sort_sibling_loops): Declare.
5454 * cfgloop.c (sort_sibling_loops_cmp): New helper.
5455 (sort_sibling_loops): New function sorting the sibling loop list
5457 * graphite.c (graphite_transform_loops): Sort sibling loops.
5459 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
5461 * target.def (vec_perm_const_ok): Change sel parameter to
5463 * optabs-query.c (can_vec_perm_p): Update accordingly.
5464 * doc/tm.texi: Regenerate.
5465 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
5466 auto_vec_perm_indices and remove separate nelt field.
5467 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
5468 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
5469 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
5470 (aarch64_expand_vec_perm_const): Update accordingly.
5471 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
5472 to vec_perm_indices.
5473 * config/arm/arm.c (expand_vec_perm_d): Change perm to
5474 auto_vec_perm_indices and remove separate nelt field.
5475 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
5476 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
5477 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
5479 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
5480 to vec_perm_indices.
5481 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
5482 sel to vec_perm_indices.
5483 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
5484 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
5485 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
5487 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
5490 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
5493 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
5494 on the FUNCTION_DECL function context if it has a DIE that is a
5497 2017-09-25 Richard Biener <rguenther@suse.de>
5499 PR tree-optimization/82285
5500 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
5503 2017-09-25 Tom de Vries <tom@codesourcery.com>
5507 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
5510 2017-09-25 Richard Biener <rguenther@suse.de>
5512 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
5513 ISL errors other than isl_error_quota happen. Dump if the
5514 schedule is the same.
5515 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
5516 errors instead of aborting inside ISL.
5518 2017-09-25 Iain Sandoe <iain@codesourcery.com>
5521 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
5522 of libgcc_eh for m64.
5523 * config/i386/darwin64.h: Likewise.
5525 2017-09-25 Richard Biener <rguenther@suse.de>
5528 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
5529 attribute for incomplete types nor twice for complete ones.
5531 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
5534 * config/i386/i386.c (ix86_print_operand_address_as): Encode
5535 %esp as %rsp to avoid 0x67 prefix if there is no index or base
5538 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
5541 * config/i386/i386.opt (mprefer-avx256): Use
5542 ix86_target_flags variable.
5543 * config/i386/i386.c (ix86_target_string): Move
5544 -mprefer-avx256 to flag2_opts.
5546 2017-09-22 Jakub Jelinek <jakub@redhat.com>
5549 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
5550 and x != -1 | y != -1 into (x & y) != -1.
5552 2017-09-22 Steve Ellcey <sellcey@cavium.com>
5554 * config.gcc: Add new case statement to set
5555 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
5556 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
5557 s390x-*-linux* case statements. Added aarch64 to the list of
5558 supported architectures.
5560 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
5562 PR tree-optimization/82289
5563 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
5564 STMT_VINFO_RELEVANT_P.
5566 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
5567 Alan Hayward <alan.hayward@arm.com>
5568 David Sherwood <david.sherwood@arm.com>
5570 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
5571 for VR_RANGE only; don't allow VR_ANTI_RANGE.
5572 (extract_range_from_binary_expr_1): Don't call
5573 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
5575 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
5576 Alan Hayward <alan.hayward@arm.com>
5577 David Sherwood <david.sherwood@arm.com>
5579 * target.def (preferred_vector_alignment): New hook.
5580 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
5582 * doc/tm.texi: Regenerate.
5583 * targhooks.h (default_preferred_vector_alignment): Declare.
5584 * targhooks.c (default_preferred_vector_alignment): New function.
5585 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
5587 (DR_TARGET_ALIGNMENT): New macro.
5588 (aligned_access_p): Update commentary.
5589 (vect_known_alignment_in_bytes): New function.
5590 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
5592 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
5593 Calculate the misalignment based on the target alignment rather than
5595 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
5596 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
5597 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
5598 the target alignment, rather than masking the element misalignment
5599 with the number of elements in a vector. Also use the target
5600 alignment when calculating the maximum number of peels.
5601 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
5602 instead of TYPE_ALIGN_UNIT.
5603 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
5604 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
5605 (vect_create_addr_base_for_vector_ref): Update call accordingly.
5606 (vect_create_data_ref_ptr): Likewise.
5607 (vect_setup_realignment): Realign by ANDing with
5608 -DR_TARGET_MISALIGNMENT.
5609 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
5610 the number of peels based on DR_TARGET_ALIGNMENT.
5611 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
5612 with the guaranteed alignment boundary when deciding whether
5614 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
5615 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
5616 (ensure_base_align): Remove stmt_info parameter. Get the
5617 target base alignment from DR_TARGET_ALIGNMENT.
5618 (vectorizable_store): Update call accordingly. Interpret
5619 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
5621 (vectorizable_load): Likewise.
5623 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
5624 Alan Hayward <alan.hayward@arm.com>
5625 David Sherwood <david.sherwood@arm.com>
5627 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
5628 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
5629 (vect_enhance_data_refs_alignment): Likewise.
5631 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
5633 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
5634 error. Only quit immediately if parsing is complete.
5635 (BEGIN): Initialize fatal_err and parse_done.
5636 (begin fpu, end fpu): Check number of arguments.
5637 (begin arch, end arch): Likewise.
5638 (begin cpu, end cpu): Likewise.
5639 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
5640 (optalias): Likewise.
5642 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
5644 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
5645 * config/arm/arm-isa.h: Delete. Move definitions to ...
5646 * arm-cpus.in: ... here. Use new feature and fgroup values.
5647 * config/arm/arm.c (arm_option_override): Use lower case for feature
5649 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
5650 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
5651 * config/arm/parsecpu.awk (END): Add new command 'isa'.
5653 (print_isa_bits_for): New function.
5654 (gen_isa): New function.
5655 (gen_comm_data): Use print_isa_bits_for.
5656 (define feature): New keyword.
5657 (define fgroup): New keyword.
5658 * config/arm/t-arm (TM_H): Remove.
5659 (GTM_H): Add arm-isa.h.
5660 (arm-isa.h): Add rule to generate file.
5661 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
5662 case for feature bit names.
5664 2017-09-22 Richard Biener <rguenther@suse.de>
5666 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
5668 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
5669 print of no dependency loops ...
5670 * graphite.c (graphite_transform_loops): ... here.
5671 (canonicalize_loop_closed_ssa_form): Work from inner to outer
5673 (same_close_phi_node, remove_duplicate_close_phi,
5674 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
5675 (canonicalize_loop_closed_ssa): ... here and simplify.
5676 * graphite-optimize-isl.c: Include tree-vectorizer.h.
5677 (optimize_isl): Use dump_printf_loc to tell when we stopped
5678 optimizing because of an ISL timeout.
5680 2017-09-22 Richard Biener <rguenther@suse.de>
5682 PR tree-optimization/82291
5683 * tree-if-conv.c (predicate_mem_writes): Make sure to
5684 remove writes in blocks predicated with false.
5686 2017-09-22 Richard Biener <rguenther@suse.de>
5688 * sese.c: Include cfganal.h.
5689 (if_region_set_false_region): Remove.
5690 (create_if_region_on_edge): Likewise.
5691 (move_sese_in_condition): Re-implement without destroying
5694 2017-09-22 Richard Biener <rguenther@suse.de>
5696 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
5697 Verify both BBs contain loop PHI nodes before dispatching to
5699 (graphite_regenerate_ast_isl): Do not recompute dominators,
5700 do not verify three times. Restructure for clarity.
5701 * graphite-scop-detection.c (same_close_phi_node,
5702 remove_duplicate_close_phi, make_close_phi_nodes_unique,
5703 defined_in_loop_p, canonicalize_loop_closed_ssa,
5704 canonicalize_loop_closed_ssa_form): Simplify, remove excess
5705 checking and SSA rewrite, move to ...
5706 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
5707 (graphite_initialize): Do not pass in ctx, do not reset the
5708 SCEV cache, compute only dominators.
5709 (graphite_transform_loops): Allocate ISL ctx after
5710 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
5711 Maintain post-dominators only around build_scops.
5712 * sese.c (if_region_set_false_region): Make static. Free
5713 and recompute dominators.
5714 (move_sese_in_condition): Assert we don't get called with
5715 post-dominators computed.
5716 * sese.h (if_region_set_false_region): Remove.
5718 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
5720 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
5721 mode attribute for TARGET_AVX512VL.
5723 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
5725 * config/i386/i386.opt (mprefer-avx256): New option.
5726 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
5728 (ix86_preferred_simd_mode): Return 256-bit AVX modes
5729 for TARGET_PREFER_AVX256.
5730 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
5732 2017-09-21 Jeff Law <law@redhat.com>
5734 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
5735 Fix dump output if the only stack space is for pushed registers.
5737 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5739 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
5742 2017-09-21 Martin Sebor <msebor@redhat.com>
5745 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
5746 code (in C++) or code that triggers warnings.
5748 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
5750 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
5752 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
5754 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
5755 * sched-rgn.c: Ditto.
5756 * sel-sched-ir.c: Ditto.
5758 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
5760 * toplev.h (set_random_seed): Adjust return type.
5761 * toplev.c (init_local_tick): Move eager initialization of random_seed
5762 to get_random_seed. Adjust comment.
5763 (init_random_seed): Inline to get_random_seed, delete.
5764 (get_random_seed): Initialize random_seed lazily.
5765 (set_random_seed): Do not return previous value.
5766 (print_switch_value): Do not call get_random_seed.
5768 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
5770 * cgraph.c (delete_function_version): New, broken out from...
5771 (cgraph_node::delete_function_version): ...here. Rename to
5772 cgraph_node::delete_function_version_by_decl. Update all uses.
5773 (cgraph_node::remove): Call delete_function_version.
5775 2017-09-21 Jakub Jelinek <jakub@redhat.com>
5778 * tree-inline.c (expand_call_inline): Emit clobber stmts for
5779 VAR_DECLs to which addressable non-volatile parameters are mapped
5780 and for id->retvar after the return value assignment. Clear
5781 id->retval and id->retbnd after inlining.
5783 2017-09-21 Richard Biener <rguenther@suse.de>
5785 PR tree-optimization/82276
5786 PR tree-optimization/82244
5787 * tree-vrp.c (build_assert_expr_for): Set
5788 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
5790 (remove_range_assertions): Revert earlier change.
5792 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
5795 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
5797 2017-09-21 Richard Biener <rguenther@suse.de>
5799 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5800 Restore valid IL after code generation errors.
5801 * graphite.c (graphite_transform_loops): Diagnose code
5802 generation issues as MSG_MISSED_OPTIMIZATION and continue
5803 with processing SCOPs.
5805 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5806 Alan Hayward <alan.hayward@arm.com>
5807 David Sherwood <david.sherwood@arm.com>
5809 * calls.c (compute_argument_addresses): Use simplify_gen_binary
5810 rather than choosing between plus_constant and gen_rtx_<CODE>.
5811 * expr.c (emit_push_insn): Likewise.
5812 (expand_expr_real_2): Likewise.
5814 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5815 Alan Hayward <alan.hayward@arm.com>
5816 David Sherwood <david.sherwood@arm.com>
5818 * loop-unroll.c (split_iv): Call copy_rtx on the step.
5820 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5821 Alan Hayward <alan.hayward@arm.com>
5822 David Sherwood <david.sherwood@arm.com>
5824 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
5825 calling tree_to_uhwi.
5827 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5828 Alan Hayward <alan.hayward@arm.com>
5829 David Sherwood <david.sherwood@arm.com>
5831 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
5832 INTEGER_CST rather than a negative test for ADDR_EXPR.
5834 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5835 Alan Hayward <alan.hayward@arm.com>
5836 David Sherwood <david.sherwood@arm.com>
5838 * tree-vrp.c (extract_range_from_binary_expr_1): Check
5839 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
5841 2017-09-21 Richard Biener <rguenther@suse.de>
5843 PR tree-optimization/71351
5844 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
5845 graphite_create_new_loop_guard): Remove, fold remaining parts
5847 (translate_isl_ast_node_for): ... here and simplify.
5849 2017-09-21 Jakub Jelinek <jakub@redhat.com>
5852 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
5853 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
5854 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
5855 alternative always use QI mode, for -Os imov (=R,R) alternative
5856 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
5859 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5860 Jeff Law <law@redhat.com>
5862 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
5863 (allocate_stack_space): New function, partially extracted from
5865 (s390_emit_prologue): Track offset to most recent stack probe.
5866 Code to allocate space moved into allocate_stack_space.
5867 Dump actions when no stack is allocated.
5868 (s390_prologue_plus_offset): New function.
5869 (s390_emit_stack_probe): Likewise.
5871 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
5873 * common.opt (Wa, Wl, Wp, g, gz=): Add
5875 (gno-column-info): Remove.
5876 (gcolumn-info): Drop RejectNegative.
5878 (gno-record-gcc-switches): Remove.
5879 (grecord-gcc-switches): Drop RejectNegative.
5880 (gno-split-dwarf): Remove.
5881 (gsplit-dwarf): Drop RejectNegative.
5882 (gno-strict-dwarf): Remove.
5883 (gstrict-dwarf): Drop RejectNegative.
5884 * config/darwin.opt (gfull, gused): Add RejectNegative.
5885 * dwarf2out.c (gen_producer_string): Drop
5886 gno-record-gcc-switches handler.
5887 * optc-gen.awk: Add g to prefixes with negative forms.
5888 * opts-common.c (remapping_prefix_p): New.
5889 (find_opt): Check it.
5890 (generate_canonical_option): Test g prefix.
5891 (option_map): Add -gno- mapping.
5892 (add_misspelling_candidates): Check remapping_prefix_p.
5894 2017-09-20 Jeff Law <law@redhat.com>
5896 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
5897 thinko in stack clash protection support.
5899 * explow.c (compute_stack_clash_protection_loop_data): Use
5900 CONST_INT_P instead of explicit test. Verify object is a
5901 CONST_INT_P before looking at INTVAL.
5902 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
5903 instead of explicit test.
5905 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
5908 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
5909 address instead of to r1 and r11.
5911 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
5913 * config.gcc: Support "knm".
5914 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
5915 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
5917 * config/i386/i386.c (m_KNM): Define.
5918 (processor_target_table): Add "knm".
5920 (ix86_option_override_internal): Add "knm".
5921 (ix86_issue_rate): Add PROCESSOR_KNM.
5922 (ix86_adjust_cost): Ditto.
5923 (ia32_multipass_dfa_lookahead): Ditto.
5924 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
5925 (fold_builtin_cpu): Add M_INTEL_KNM.
5926 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
5927 (processor_type): Add PROCESSOR_KNM.
5928 * config/i386/x86-tune.def: Add m_KNM.
5929 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
5931 2017-09-20 Richard Biener <rguenther@suse.de>
5933 PR tree-optimization/80213
5934 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
5935 are allowed in empty BBs as well.
5936 (canonicalize_loop_closed_ssa): Also look for other complex
5938 (scop_detection::get_sese): Include the loop-closed PHI block
5940 (scop_detection::merge_sese): Remove code adding extra blocks.
5941 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
5942 (build_scops): Assert the final returned scop is invalid.
5944 2017-09-20 Richard Biener <rguenther@suse.de>
5946 PR tree-optimization/82264
5947 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
5949 (vn_phi_lookup): Likewise.
5950 (vn_phi_insert): Likewise.
5952 2017-09-20 Jakub Jelinek <jakub@redhat.com>
5954 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
5955 that fits into uhwi or shwi, add DW_AT_const_value regardless
5956 of early_dwarf without going through RTL, using add_AT_unsigned
5959 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
5960 (DEBUG_LTO_ABBREV_SECTION): Likewise.
5961 (DEBUG_LTO_MACINFO_SECTION): Likewise.
5962 (DEBUG_MACRO_SECTION): Likewise.
5963 (DEBUG_LTO_MACRO_SECTION): Likewise.
5964 (DEBUG_STR_DWO_SECTION): Likewise.
5965 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
5966 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
5967 (DEBUG_LTO_DWO_LINE_SECTION): Define.
5968 (DEBUG_LTO_LINE_STR_SECTION): Define.
5969 (init_sections_and_labels): Initialize debug_line_str_section
5970 variable. Initialize debug_loc_section for -gdwarf-5 to
5971 DEBUG_LOCLISTS_SECTION. Formatting fixes.
5973 2017-09-20 Richard Biener <rguenther@suse.de>
5975 * graphite-sese-to-poly.c (extract_affine): Properly handle
5976 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
5978 2017-09-20 Richard Biener <rguenther@suse.de>
5980 PR tree-optimization/81373
5981 * graphite-scop-detection.c (build_cross_bb_scalars_def):
5982 Force SESE live-out defs to be handled even if they are
5985 2017-09-19 Jeff Law <law@redhat.com>
5987 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
5988 nothing for stack adjustments with REG_STACK_CHECK.
5989 * sched-deps.c (parse_add_or_inc): Reject insns with
5990 REG_STACK_CHECK from dependency breaking.
5991 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
5992 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
5993 * reg-notes.def (STACK_CHECK): New note.
5995 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
5996 (ix86_expand_prologue): Dump stack clash info as needed.
5997 Call ix86_adjust_stack_and_probe_stack_clash as needed.
5999 * function.c (dump_stack_clash_frame_info): New function.
6000 * function.h (dump_stack_clash_frame_info): Prototype.
6001 (enum stack_clash_probes): New enum.
6003 * config/alpha/alpha.c (alpha_expand_prologue): Also check
6004 flag_stack_clash_protection.
6005 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
6006 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
6007 (arm_frame_pointer_required): Likewise.
6008 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
6009 (ia64_expand_prologue): Likewise.
6010 * config/mips/mips.c (mips_expand_prologue): Likewise.
6011 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
6012 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6013 (sparc_flat_expand_prologue): Likewise.
6014 * config/spu/spu.c (spu_expand_prologue): Likewise.
6016 * explow.c: Include "params.h".
6017 (anti_adjust_stack_and_probe_stack_clash): New function.
6018 (get_stack_check_protect): Likewise.
6019 (compute_stack_clash_protection_loop_data): Likewise.
6020 (emit_stack_clash_protection_loop_start): Likewise.
6021 (emit_stack_clash_protection_loop_end): Likewise.
6022 (allocate_dynamic_stack_space): Use get_stack_check_protect.
6023 Use anti_adjust_stack_and_probe_stack_clash.
6024 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
6025 (emit_stack_clash_protection_loop_start): Likewise.
6026 (emit_stack_clash_protection_loop_end): Likewise.
6027 * rtl.h (get_stack_check_protect): Prototype.
6028 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
6029 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
6030 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
6032 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
6034 * doc/tm.texi: Rebuilt.
6035 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
6036 get_stack_check_protect.
6037 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
6038 * config/arm/arm.c (arm_expand_prologue): Likewise.
6039 (arm_frame_pointer_required): Likewise.
6040 * config/i386/i386.c (ix86_expand_prologue): Likewise.
6041 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
6042 * config/mips/mips.c (mips_expand_prologue): Likewise.
6043 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
6044 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
6045 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6046 (sparc_flat_expand_prologue): Likewise.
6048 * common.opt (-fstack-clash-protection): New option.
6049 * flag-types.h (enum stack_check_type): Note difference between
6050 -fstack-check= and -fstack-clash-protection.
6051 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
6052 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
6053 * toplev.c (process_options): Issue warnings/errors for cases
6054 not handled with -fstack-clash-protection.
6055 * doc/invoke.texi (-fstack-clash-protection): Document new option.
6056 (-fstack-check): Note additional problem with -fstack-check=generic.
6057 Note that -fstack-check is primarily for Ada and refer users
6058 to -fstack-clash-protection for stack-clash-protection.
6059 Document new params for stack clash protection.
6061 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
6063 * config/i386/i386.c (ix86_split_long_move): Do not handle
6064 address used for LEA in a special way.
6066 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
6068 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
6070 2017-09-19 Martin Sebor <msebor@redhat.com>
6073 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
6074 of incompatible types.
6076 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
6078 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6079 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
6080 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6081 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
6083 2017-09-19 Richard Biener <rguenther@suse.de>
6085 PR tree-optimization/82244
6086 * tree-vrp.c (remove_range_assertions): Do not propagate
6087 a constant to abnormals but replace the assert with a copy.
6089 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
6091 PR rtl-optimization/57878
6092 PR rtl-optimization/68988
6093 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
6094 avoidance test involving non_reload_pseudos. Move frequency test
6095 below the general fragmentation avoidance test.
6097 2017-09-19 Richard Biener <rguenther@suse.de>
6099 PR tree-optimization/69728
6100 * graphite-sese-to-poly.c (schedule_error): New global.
6101 (add_loop_schedule): Handle empty domain by failing the
6103 (build_original_schedule): Handle schedule_error.
6105 2017-09-19 Richard Biener <rguenther@suse.de>
6107 * graphite-scop-detection.c (scop_detection::can_represent_loop):
6108 Do not iterate to sibling loops but only to siblings of inner
6111 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
6114 * config/m68k/m68k.md (moveq feeding equality comparison): Check
6115 that the registers are different.
6117 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
6119 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
6120 to processor_model and "amdfam17h" to arch_names_table.
6121 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
6123 2017-09-18 Jakub Jelinek <jakub@redhat.com>
6126 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
6128 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6129 Alan Hayward <alan.hayward@arm.com>
6130 David Sherwood <david.sherwood@arm.com>
6132 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
6134 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
6136 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
6137 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
6138 vect_schedule_slp_instance.
6139 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
6140 Update call to vect_slp_analyze_node_operations. Simplify return
6142 (vect_slp_analyze_bb_1): Update call accordingly.
6143 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
6144 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
6145 (vect_schedule_slp): Update call accordingly.
6147 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6148 Alan Hayward <alan.hayward@arm.com>
6149 David Sherwood <david.sherwood@arm.com>
6151 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
6152 with types that aren't in fact scalar.
6154 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6156 * tree-vect-slp.c (vect_record_max_nunits): New function,
6158 (vect_build_slp_tree_1): ...here.
6159 (vect_build_slp_tree_2): Call it for phis too.
6161 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6163 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
6164 to vect_get_vec_def_for_operand when getting the mask operand.
6166 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6167 Alan Hayward <alan.hayward@arm.com>
6168 David Sherwood <david.sherwood@arm.com>
6170 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
6173 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6174 Alan Hayward <alan.hayward@arm.com>
6175 David Sherwood <david.sherwood@arm.com>
6177 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
6178 calculation for vector booleans.
6180 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6181 Alan Hayward <alan.hayward@arm.com>
6182 David Sherwood <david.sherwood@arm.com>
6184 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
6186 (vect_transform_stmt): ...here.
6187 (vect_analyze_stmt): Use it instead of calling
6188 vectorizable_live_operation directly.
6190 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
6192 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
6193 non-SIMT targets in acc vector loops.
6195 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
6197 * configure.ac: Add arc and check if assembler supports gdwarf2.
6198 * configure: Regenerate.
6200 2017-09-18 Richard Biener <rguenther@suse.de>
6202 PR tree-optimization/82220
6203 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
6204 epilogue niters from the min_profitable_iters compute.
6206 2017-09-18 Jakub Jelinek <jakub@redhat.com>
6209 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
6210 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
6211 (ix86_init_pic_reg): Revert 2017-09-01 changes.
6213 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
6216 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
6217 switching to a new text section.
6219 2017-09-18 Richard Biener <rguenther@suse.de>
6221 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
6223 (build_alias_set): Reject aliases with no access function.
6225 2017-09-18 Richard Biener <rguenther@suse.de>
6227 PR tree-optimization/79622
6228 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
6230 (build_cross_bb_scalars_use): Likewise.
6232 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
6234 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
6236 2017-09-18 Alan Modra <amodra@gmail.com>
6239 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
6240 stack_pointer_rtx for count 0. Update comments. Break up
6241 large rtl expression.
6243 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
6245 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
6246 Increase to 20 bytes.
6247 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
6248 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
6249 or avx version of the stub.
6251 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
6254 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
6255 compute the minimum stack alignment. Also update preferred stack
6256 boundary for leaf functions.
6258 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
6260 PR tree-optimization/82228
6261 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
6264 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
6266 * common/config/nds32/nds32-common.c
6267 (nds32_option_optimization_table): Refine formatting.
6268 (nds32_option_optimization_table): Use -fsched-pressure and
6269 -fomit-frame-pointer for specific optimization level.
6271 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
6273 * config/nds32/nds32.c: Refine formatting and comments.
6274 * config/nds32/nds32.h: Likewise.
6275 * config/nds32/nds32.md: Likewise.
6276 * config/nds32/nds32-cost.c: Likewise.
6277 * config/nds32/nds32-isr.c: Likewise.
6278 * config/nds32/nds32-md-auxiliary.c: Likewise.
6279 * config/nds32/nds32-multiple.md: Likewise.
6280 * config/nds32/nds32-predicates.c: Likewise.
6282 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
6283 Jakub Jelinek <jakub@redhat.com>
6285 Add support for -std=c++2a.
6286 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
6288 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
6290 2017-09-15 Steve Ellcey <sellcey@cavium.com>
6293 * doc/extend.texi (Common Function Attributes): Add
6294 references to ARM, AArch64, and S/390 specific attributes.
6295 (Function Specific Option Pragmas): Add AArch64 and S/390
6296 to list of back ends that support the target pragma.
6298 2017-09-15 Nathan Sidwell <nathan@acm.org>
6300 * doc/standards.texi: Fix C++17 description. Update URLs for
6303 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
6305 * common.opt (Wcast-align=strict): New warning option.
6306 * doc/invoke.texi: Document -Wcast-align=strict.
6308 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
6310 * cgraph.h (cgraph_thunk_info): Add comments.
6311 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
6312 assert for VIRTUAL_* arguments stricter.
6314 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
6316 PR tree-optimization/71026
6317 * match.pd: Move RDIV patterns from fold-const.c
6318 * fold-const.c (distribute_real_division): Removed.
6319 (fold_binary_loc): Remove calls to distribute_real_divison.
6321 2017-09-15 Jakub Jelinek <jakub@redhat.com>
6323 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
6324 c++1z and gnu++1z as deprecated. Change other references to
6325 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
6326 Change -Wc++1z-compat to -Wc++17-compat.
6327 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
6328 * dwarf2out.c (highest_c_language): Handle C++17.
6329 (gen_compile_unit_die): Likewise.
6331 2017-09-15 Jakub Jelinek <jakub@redhat.com>
6333 PR rtl-optimization/82192
6334 * combine.c (make_extraction): Don't look through non-paradoxical
6335 SUBREGs or TRUNCATE if pos + len is or might be bigger than
6338 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6339 Alan Hayward <alan.hayward@arm.com>
6340 David Sherwood <david.sherwood@arm.com>
6342 * target.def (function_arg_offset): New hook.
6343 * targhooks.h (default_function_arg_offset): Declare.
6344 * targhooks.c (default_function_arg_offset): New function.
6345 * function.c (locate_and_pad_parm): Use
6346 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
6347 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
6348 (TARGET_FUNCTION_ARG_OFFSET): ...this.
6349 * doc/tm.texi: Regenerate.
6350 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
6351 * config/spu/spu.c (spu_function_arg_offset): New function.
6352 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
6353 * system.h (FUNCTION_ARG_OFFSET): Poison.
6355 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6356 Alan Hayard <alan.hayward@arm.com>
6357 David Sherwood <david.sherwood@arm.com>
6359 * target.def (truly_noop_truncation): New hook.
6360 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
6361 than TRULY_NOOP_TRUNCATION.
6362 * hooks.h (hook_bool_uint_uint_true): Declare.
6363 * hooks.c (hook_bool_uint_uint_true): New function.
6364 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
6365 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
6366 * doc/tm.texi: Regenerate.
6367 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
6368 rather than TRULY_NOOP_TRUNCATION in comments.
6369 (simplify_comparison): Likewise.
6370 (record_truncated_value): Likewise.
6371 * expmed.c (extract_bit_field_1): Likewise.
6372 (extract_split_bit_field): Likewise.
6373 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
6374 instead of TRULY_NOOP_TRUNCATION.
6375 * function.c (assign_parm_setup_block): Likewise.
6376 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
6377 * rtlhooks.c: Include target.h.
6378 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
6379 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
6380 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
6381 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
6382 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
6383 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
6384 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
6385 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
6386 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
6387 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
6388 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
6389 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
6390 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
6391 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
6392 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
6393 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
6394 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
6395 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
6396 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
6397 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
6398 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
6399 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
6400 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
6401 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
6402 * config/mips/mips.c (mips_truly_noop_truncation): New function.
6403 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
6404 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
6405 rather than TRULY_NOOP_TRUNCATION in comments.
6406 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
6407 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
6408 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
6409 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
6410 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
6411 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
6412 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
6413 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
6414 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
6415 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
6416 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
6417 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
6418 rather than TRULY_NOOP_TRUNCATION in comments.
6419 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
6420 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
6421 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
6422 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
6423 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
6424 TRULY_NOOP_TRUNCATION condition.
6425 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
6426 (TRULY_NOOP_TRUNCATION): Delete.
6427 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
6428 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
6429 * config/spu/spu.c (spu_truly_noop_truncation): New function.
6430 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
6431 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
6432 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
6433 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
6434 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
6435 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
6436 rather than TRULY_NOOP_TRUNCATION in comments.
6437 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
6438 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
6439 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
6440 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
6441 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
6442 * system.h (TRULY_NOOP_TRUNCATION): Poison.
6444 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
6447 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
6448 (*cmp_ior): Likewise.
6449 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
6450 (*ior_scc_scc_cmp): Likewise.
6451 (*and_scc_scc): Likewise.
6452 (*and_scc_scc_cmp): Likewise.
6454 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6455 Alan Hayard <alan.hayward@arm.com>
6456 David Sherwood <david.sherwood@arm.com>
6458 * target.def (can_change_mode_class): New hook.
6459 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
6460 (hard_regno_nregs): Likewise.
6461 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
6462 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
6463 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
6464 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
6465 (LOAD_EXTEND_OP): Update accordingly.
6466 * doc/tm.texi: Regenerate.
6467 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
6468 CANNOT_CHANGE_MODE_CLASS.
6469 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
6470 (REG_CAN_CHANGE_MODE_P): ...this new macro.
6471 * combine.c (simplify_set): Update accordingly.
6472 * emit-rtl.c (validate_subreg): Likewise.
6473 * recog.c (general_operand): Likewise.
6474 * regcprop.c (mode_change_ok): Likewise.
6475 * reload1.c (choose_reload_regs): Likewise.
6476 (inherit_piecemeal_p): Likewise.
6477 * rtlanal.c (simplify_subreg_regno): Likewise.
6478 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
6479 instead of CANNOT_CHANGE_MODE_CLASS.
6480 (reload_cse_simplify_operands): Likewise.
6481 * reload.c (push_reload): Use targetm.can_change_mode_class
6482 instead of CANNOT_CHANGE_MODE_CLASS.
6483 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
6484 REG_CANNOT_CHANGE_MODE_P.
6485 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6486 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
6487 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6488 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6489 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6490 (arm_can_change_mode_class): New function.
6491 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
6492 than CANNOT_CHANGE_MODE_CLASS in comments.
6493 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6494 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
6495 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
6496 (ix86_can_change_mode_class): ...this new function, inverting the
6497 sense of the return value.
6498 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6499 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6500 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6501 (ia64_can_change_mode_class): New function.
6502 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6503 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
6504 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
6505 (m32c_can_change_mode_class): ...this new function, inverting the
6506 sense of the return value.
6507 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6508 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6509 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
6510 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
6511 (mips_can_change_mode_class): ...this new function, inverting the
6512 sense of the return value.
6513 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6514 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6515 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6516 (msp430_can_change_mode_class): New function.
6517 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6518 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
6519 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6520 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6521 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6522 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
6523 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6524 (pa_cannot_change_mode_class): Replace with...
6525 (pa_can_change_mode_class): ...this new function, inverting the
6526 sense of the return value.
6527 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
6528 than CANNOT_CHANGE_MODE_CLASS in comments.
6529 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6530 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
6531 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6532 (pdp11_cannot_change_mode_class): Replace with...
6533 (pdp11_can_change_mode_class): ...this new function, inverting the
6534 sense of the return value.
6535 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6536 * config/powerpcspe/powerpcspe-protos.h
6537 (rs6000_cannot_change_mode_class_ptr): Delete.
6538 * config/powerpcspe/powerpcspe.c
6539 (rs6000_cannot_change_mode_class_ptr): Delete.
6540 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6541 (rs6000_option_override_internal): Assign to
6542 targetm.can_change_mode_class instead of
6543 rs6000_cannot_change_mode_class_ptr.
6544 (rs6000_cannot_change_mode_class): Replace with...
6545 (rs6000_can_change_mode_class): ...this new function, inverting the
6546 sense of the return value.
6547 (rs6000_debug_cannot_change_mode_class): Replace with...
6548 (rs6000_debug_can_change_mode_class): ...this new function.
6549 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6550 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
6551 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6552 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6553 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
6555 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
6556 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6557 (rs6000_option_override_internal): Assign to
6558 targetm.can_change_mode_class instead of
6559 rs6000_cannot_change_mode_class_ptr.
6560 (rs6000_cannot_change_mode_class): Replace with...
6561 (rs6000_can_change_mode_class): ...this new function, inverting the
6562 sense of the return value.
6563 (rs6000_debug_cannot_change_mode_class): Replace with...
6564 (rs6000_debug_can_change_mode_class): ...this new function.
6565 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6566 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
6567 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
6568 (s390_can_change_mode_class): ...this new function, inverting the
6569 sense of the return value.
6570 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6571 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6572 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
6573 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6574 (sh_cannot_change_mode_class): Replace with...
6575 (sh_can_change_mode_class): ...this new function, inverting the
6576 sense of the return value.
6577 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6578 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6579 (sparc_can_change_mode_class): New function.
6580 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6581 * config/spu/spu.c (spu_can_change_mode_class): New function.
6582 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6583 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6584 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6585 (visium_can_change_mode_class): New function.
6586 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
6588 2017-09-15 Richard Biener <rguenther@suse.de>
6590 PR tree-optimization/82217
6591 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
6592 but not undefined case.
6594 2017-09-15 Jakub Jelinek <jakub@redhat.com>
6597 * postreload.c (reload_cse_simplify_operands): Skip
6598 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
6600 2017-09-15 Richard Biener <rguenther@suse.de>
6602 PR tree-optimization/68823
6603 * graphite-scop-detection.c (build_alias_set): If we have a
6604 possible dependence check whether we can handle them by just
6605 looking at the DRs DR_ACCESS_FNs.
6606 (build_scops): If build_alias_set fails, fail the SCOP.
6608 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
6610 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
6611 to support float128 built-in functions that require the ISA 3.0
6613 (BU_FLOAT128_3_HW): Likewise.
6614 (SQRTF128): Add support for the IEEE 128-bit square root and fma
6616 (FMAF128): Likewise.
6618 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
6619 support for built-in functions that need the ISA 3.0 IEEE 128-bit
6620 floating point instructions.
6621 (rs6000_invalid_builtin): Likewise.
6622 (rs6000_builtin_mask_names): Likewise.
6623 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
6624 (RS6000_BTM_FLOAT128_HW): Likewise.
6625 (RS6000_BTM_COMMON): Likewise.
6626 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
6628 * doc/extend.texi (RS/6000 built-in functions): Document the
6629 IEEE 128-bit floating point square root and fused multiply-add
6632 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
6634 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
6635 reg (r2) isn't in the set of registers defined in the prologue.
6637 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6638 Alan Hayward <alan.hayward@arm.com>
6639 David Sherwood <david.sherwood@arm.com>
6641 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
6642 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
6643 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
6644 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
6645 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
6647 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
6648 max_vectorization_factor.
6649 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
6651 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6652 Alan Hayward <alan.hayward@arm.com>
6653 David Sherwood <david.sherwood@arm.com>
6655 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
6656 (vect_worthwhile_without_simd_p): Declare.
6657 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
6658 (vectorizable_reduction): Use it.
6659 * tree-vect-stmts.c (vectorizable_shift): Likewise.
6660 (vectorizable_operation): Likewise.
6662 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6663 Alan Hayward <alan.hayward@arm.com>
6664 David Sherwood <david.sherwood@arm.com>
6666 * tree-vectorizer.h (vect_get_num_copies): New function.
6667 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
6668 * tree-vect-loop.c (vectorizable_reduction): Likewise.
6669 (vectorizable_induction): Likewise.
6670 (vectorizable_live_operation): Likewise.
6671 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
6672 (vectorizable_bswap): Likewise.
6673 (vectorizable_call): Likewise.
6674 (vectorizable_conversion): Likewise.
6675 (vectorizable_assignment): Likewise.
6676 (vectorizable_shift): Likewise.
6677 (vectorizable_operation): Likewise.
6678 (vectorizable_store): Likewise.
6679 (vectorizable_load): Likewise.
6680 (vectorizable_condition): Likewise.
6681 (vectorizable_comparison): Likewise.
6682 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
6684 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6685 Alan Hayward <alan.hayward@arm.com>
6686 David Sherwood <david.sherwood@arm.com>
6688 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
6689 of vect_init_vector.
6691 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6692 Alan Hayward <alan.hayward@arm.com>
6693 David Sherwood <david.sherwood@arm.com>
6695 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
6696 an inline wrapper that provides a location.
6697 (gimple_build_vector): Likewise.
6698 * gimple-fold.c (gimple_build_vector_from_val): New function.
6699 (gimple_build_vector): Likewise.
6700 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
6701 functions to build the initial value. Always return a gimple value.
6702 (get_initial_defs_for_reduction): Likewise. Only compute
6704 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
6705 vect_init_vector on the results from get_initial_def(s)_for_reduction.
6706 (vectorizable_induction): Use gimple_build_vector rather than
6709 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6710 Alan Hayward <alan.hayward@arm.com>
6711 David Sherwood <david.sherwood@arm.com>
6713 * target.h (vec_perm_indices): New typedef.
6714 (auto_vec_perm_indices): Likewise.
6715 * optabs-query.h: Include target.h
6716 (can_vec_perm_p): Take a vec_perm_indices *.
6717 * optabs-query.c (can_vec_perm_p): Likewise.
6718 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
6719 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
6720 * tree-vect-generic.c (lower_vec_perm): Likewise.
6721 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
6722 (vect_grouped_load_supported): Likewise.
6723 (vect_shift_permute_load_chain): Likewise.
6724 (vect_permute_store_chain): Use auto_vec_perm_indices.
6725 (vect_permute_load_chain): Likewise.
6726 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
6727 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
6728 Update uses of can_vec_perm_p.
6729 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
6730 mode with a number of elements. Take a vec_perm_indices *.
6731 (vect_create_epilog_for_reduction): Update accordingly.
6732 Use auto_vec_perm_indices.
6733 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
6734 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
6735 (vect_transform_slp_perm_load): Likewise.
6736 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
6737 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
6738 (vect_gen_perm_mask_checked): Likewise.
6739 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
6740 (vect_gen_perm_mask_checked): Likewise.
6741 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
6742 (vectorizable_store): Likewise.
6743 (vectorizable_load): Likewise.
6744 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
6745 (vectorizable_bswap): Likewise.
6747 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6748 Alan Hayward <alan.hayward@arm.com>
6749 David Sherwood <david.sherwood@arm.com>
6751 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
6752 * tree.c (build_vector): Likewise.
6753 (build_vector_from_ctor): Update accordingly.
6754 (build_vector_from_val): Likewise.
6755 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
6756 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
6757 * tree-vect-generic.c (add_rshift): Likewise.
6758 (expand_vector_divmod): Likewise.
6759 (optimize_vector_constructor): Likewise.
6760 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
6761 (vect_transform_slp_perm_load): Likewise.
6762 (vect_schedule_slp_instance): Likewise.
6763 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
6764 (vectorizable_call): Likewise.
6765 (vect_gen_perm_mask_any): Likewise. Add elements in order.
6766 * expmed.c (make_tree): Likewise.
6767 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
6768 a vector passed to build_vector.
6769 (fold_convert_const): Likewise.
6770 (exact_inverse): Likewise.
6771 (fold_ternary_loc): Likewise.
6772 (fold_relational_const): Likewise.
6773 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
6774 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
6775 (const_unop): Likewise. Store the reduction accumulator in a
6776 variable rather than an array.
6777 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
6778 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
6779 the new vector, rather than constructing it after the input arrays.
6780 (native_interpret_vector): Use auto_vec<tree> when building
6781 a vector passed to build_vector. Add elements in order.
6782 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
6783 auto_vec<tree> when building a vector passed to build_vector.
6784 (vect_create_epilog_for_reduction): Likewise.
6785 (vectorizable_induction): Likewise.
6786 (get_initial_def_for_reduction): Likewise. Fix indentation of
6788 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
6790 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
6791 passed to build_vector.
6793 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6794 Alan Hayward <alan.hayward@arm.com>
6795 David Sherwood <david.sherwood@arm.com>
6797 * tree-core.h (tree_base::u): Add an "nelts" field.
6798 (tree_vector): Use VECTOR_CST_NELTS as the length.
6799 * tree.c (tree_size): Likewise.
6800 (make_vector): Initialize VECTOR_CST_NELTS.
6801 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
6802 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
6803 TYPE_VECTOR_SUBPARTS.
6804 * expr.c (const_vector_mask_from_tree): Consistently use "units"
6805 as the number of units, setting it from VECTOR_CST_NELTS.
6806 (const_vector_from_tree): Likewise.
6807 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
6808 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
6809 (fold_negate_expr_1): Likewise.
6810 (fold_convert_const): Likewise.
6811 (const_binop): Likewise. Differentiate the number of output and
6813 (const_unop): Likewise.
6814 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
6815 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
6816 in cases that did the opposite.
6818 2017-09-14 Richard Biener <rguenther@suse.de>
6820 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
6823 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
6825 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
6827 2017-09-14 Jakub Jelinek <jakub@redhat.com>
6830 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
6831 if and where to split a bb, except for splitting before debug insn
6832 sequences followed by non-label real insn. Delete debug insns
6833 in between basic blocks.
6835 * combine.c (make_compound_operation_int): Formatting fixes.
6837 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
6838 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6839 * config/netbsd.h (LINK_EH_SPEC): Likewise.
6840 * config/sol2.h (LINK_EH_SPEC): Likewise.
6841 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6842 * config/s390/linux.h (LINK_SPEC): Likewise.
6843 * config/freebsd.h (LINK_EH_SPEC): Likewise.
6844 * config/openbsd.h (LINK_EH_SPEC): Likewise.
6845 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6846 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
6847 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
6848 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6849 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
6850 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
6852 2017-09-13 Jakub Jelinek <jakub@redhat.com>
6854 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
6856 (ENDFILE_LINUX_SPEC): Likewise.
6857 (LINK_EH_SPEC): Likewise.
6858 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
6859 (LINK_OS_LINUX_SPEC32): Likewise.
6860 (LINK_OS_LINUX_SPEC64): Likewise.
6861 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
6862 (LINK_OS_LINUX_SPEC): Likewise.
6864 2017-09-13 Martin Liska <mliska@suse.cz>
6867 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
6868 CASE_HIGH is NULL_TREE.
6870 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
6871 Alan Hayward <alan.hayward@arm.com>
6872 David Sherwood <david.sherwood@arm.com>
6874 * target.def (secondary_memory_needed): New hook.
6875 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
6876 instead of SECONDARY_MEMORY_NEEDED.
6877 (secondary_memory_needed_mode): Likewise.
6878 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
6879 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
6880 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
6881 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
6882 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
6883 * doc/tm.texi: Regenerate.
6884 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
6885 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
6886 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6887 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
6888 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
6889 * config/i386/i386.c (inline_secondary_memory_needed): Put the
6890 mode argument first and change the reg_class arguments to reg_class_t.
6891 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
6892 Make static. Update the call to inline_secondary_memory_needed.
6893 (ix86_register_move_cost): Update the call to
6894 inline_secondary_memory_needed.
6895 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6896 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
6898 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
6899 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
6901 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
6902 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
6903 * config/mips/mips.c (mips_secondary_memory_needed): Make static
6904 and match hook interface. Add comment from mips.h.
6905 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6906 * config/mmix/mmix.md (truncdfsf2): Refer to
6907 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
6909 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
6910 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
6911 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6912 (pa_secondary_memory_needed): New function.
6913 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
6914 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
6915 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6916 (pdp11_secondary_memory_needed): Make static and match hook interface.
6917 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
6918 * config/powerpcspe/powerpcspe-protos.h
6919 (rs6000_secondary_memory_needed_ptr): Delete.
6920 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
6922 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6923 (rs6000_option_override_internal): Assign to
6924 targetm.secondary_memory_needed rather than
6925 rs6000_secondary_memory_needed_ptr.
6926 (rs6000_secondary_memory_needed): Match hook interface.
6927 (rs6000_debug_secondary_memory_needed): Likewise.
6928 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
6929 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
6930 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
6931 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6932 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
6933 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
6935 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
6936 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6937 (rs6000_option_override_internal): Assign to
6938 targetm.secondary_memory_needed rather than
6939 rs6000_secondary_memory_needed_ptr.
6940 (rs6000_secondary_memory_needed): Match hook interface.
6941 (rs6000_debug_secondary_memory_needed): Likewise.
6942 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
6943 * config/s390/s390.c (s390_secondary_memory_needed): New function.
6944 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6945 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
6946 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6947 (sparc_secondary_memory_needed): New function.
6948 * lra-constraints.c (check_and_process_move): Refer to
6949 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
6951 (curr_insn_transform): Likewise.
6952 (process_alt_operands): Use targetm.secondary_memory_needed
6953 instead of TARGET_SECONDARY_MEMORY_NEEDED.
6954 (check_secondary_memory_needed_p): Likewise.
6955 (choose_split_class): Likewise.
6956 * reload.c: Unconditionally include code that was previously
6957 conditional on SECONDARY_MEMORY_NEEDED.
6958 (push_secondary_reload): Use targetm.secondary_memory_needed
6959 instead of TARGET_SECONDARY_MEMORY_NEEDED.
6960 (push_reload): Likewise.
6961 * reload1.c: Unconditionally include code that was previously
6962 conditional on SECONDARY_MEMORY_NEEDED.
6963 (choose_reload_regs): Use targetm.secondary_memory_needed
6964 instead of TARGET_SECONDARY_MEMORY_NEEDED.
6965 (gen_reload): Likewise.
6966 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
6968 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
6969 Alan Hayward <alan.hayward@arm.com>
6970 David Sherwood <david.sherwood@arm.com>
6972 * target.def (secondary_memory_needed_mode): New hook:
6973 * targhooks.c (default_secondary_memory_needed_mode): Declare.
6974 * targhooks.h (default_secondary_memory_needed_mode): New function.
6975 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
6976 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
6977 * doc/tm.texi: Regenerate.
6978 * lra-constraints.c (check_and_process_move): Use
6979 targetm.secondary_memory_needed_mode instead of
6980 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
6981 (curr_insn_transform): Likewise.
6982 * reload.c (get_secondary_mem): Likewise.
6983 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
6984 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
6986 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
6987 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
6988 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
6989 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
6990 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
6992 * config/powerpcspe/powerpcspe-protos.h
6993 (rs6000_secondary_memory_needed_mode): Delete.
6994 * config/powerpcspe/powerpcspe.c
6995 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
6996 (rs6000_secondary_memory_needed_mode): Make static.
6997 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
6998 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
7000 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7002 (rs6000_secondary_memory_needed_mode): Make static.
7003 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7004 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
7005 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7006 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7007 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7009 (sparc_secondary_memory_needed_mode): New function.
7010 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
7012 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
7014 * config/aarch64/constraints.md (Umq): New constraint.
7015 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
7017 (mov<mode>): Update condition.
7019 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7021 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
7022 when bitposition is the same.
7024 2017-09-13 Richard Biener <rguenther@suse.de>
7026 * dwarf2out.c (output_die_symbol): Remove.
7027 (output_die): Do not output a DIEs symbol.
7029 2017-09-13 Richard Biener <rguenther@suse.de>
7032 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
7033 default-def to avoid breaking iterator update with the weird
7034 interaction with cgraph_update_edges_for_call_stmt_node.
7036 2017-09-13 Richard Biener <rguenther@suse.de>
7038 * tree-cfg.c (verify_gimple_assign_binary): Add verification
7039 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
7040 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
7041 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
7043 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
7045 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
7046 Disable pc relative literal load irrespective of
7047 TARGET_FIX_ERR_A53_84341 for default.
7049 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
7051 * config/sparc/sparc.c (output_return): Output the source location of
7052 the insn in the delay slot, if any.
7053 (output_sibcall): Likewise.
7055 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
7058 * common.opt (-static-pie): New alias.
7059 (shared): Negate static-pie.
7060 (-no-pie): Update help text.
7062 (static-pie): New option.
7063 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
7064 -static-pie support.
7065 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
7066 (LINK_EH_SPEC): Likewise.
7067 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7068 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7069 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7070 * gcc.c (LINK_COMMAND_SPEC): Likewise.
7071 (init_gcc_specs): Likewise.
7072 (init_spec): Likewise.
7073 (display_help): Update help message for -pie.
7074 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
7077 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
7079 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
7080 (movdi_aarch64): Likewise.
7081 (movti_aarch64): Likewise.
7083 2017-09-12 Simon Wright <simon@pushface.org>
7086 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
7087 calculation of the minor version, always output as 0.
7089 2017-09-12 Jakub Jelinek <jakub@redhat.com>
7092 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
7093 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
7094 on it early, rather than manual conversion late. For
7095 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
7096 instead of performing manual conversion.
7098 2017-09-12 Carl Love <cel@us.ibm.com>
7100 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
7101 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
7103 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7104 VMULOSW): Add definitions.
7105 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7106 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7107 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7108 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7109 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7111 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7113 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
7115 (movti_aarch64): Likewise.
7116 (movdf_aarch64): Likewise.
7117 (movtf_aarch64): Likewise.
7118 (load_pairdi): Likewise.
7119 (store_pairdi): Likewise.
7120 (load_pairdf): Likewise.
7121 (store_pairdf): Likewise.
7122 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
7123 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
7124 (ldr_got_small_<mode>): Likewise.
7125 (ldr_got_small_28k_<mode>): Likewise.
7126 (ldr_got_tiny): Likewise.
7127 * config/aarch64/iterators.md (ldst_sz): New.
7128 (ldpstp_sz): Likewise.
7129 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
7131 (thunderx_load): Split load_8 to load_16.
7132 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
7134 (thunderx2t99_storepair_basic): Split store_8 to store_16.
7135 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
7136 (xgene1_store_pair): Split store_8 to store_16.
7137 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
7138 (falkor_st_0_st_sd): Split store_8 to store_16.
7140 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7142 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
7143 and store1/2/3/4 to store_4/8/12/16.
7144 * config/aarch64/aarch64.md: Update for rename.
7145 * config/arm/arm.md: Likewise.: Likewise.
7146 * config/arm/arm.c: Likewise.
7147 * config/arm/thumb1.md: Likewise.
7148 * config/arm/thumb2.md: Likewise.
7149 * config/arm/vfp.md: Likewise.
7150 * config/arm/arm-generic.md: Likewise.
7151 * config/arm/arm1020e.md: Likewise.
7152 * config/arm/arm1026ejs.md: Likewise.
7153 * config/arm/arm1136jfs.md: Likewise.
7154 * config/arm/arm926ejs.md: Likewise.
7155 * config/arm/cortex-a15.md: Likewise.
7156 * config/arm/cortex-a17.md: Likewise.
7157 * config/arm/cortex-a5.md: Likewise.
7158 * config/arm/cortex-a53.md: Likewise.
7159 * config/arm/cortex-a57.md: Likewise.
7160 * config/arm/cortex-a7.md: Likewise.
7161 * config/arm/cortex-a8.md: Likewise.
7162 * config/arm/cortex-a9.md: Likewise.
7163 * config/arm/cortex-m4.md: Likewise.
7164 * config/arm/cortex-m7.md: Likewise.
7165 * config/arm/cortex-r4.md: Likewise.
7166 * config/arm/exynos-m1.md: Likewise.
7167 * config/arm/fa526.md: Likewise.
7168 * config/arm/fa606te.md: Likewise.
7169 * config/arm/fa626te.md: Likewise.
7170 * config/arm/fa726te.md: Likewise.
7171 * config/arm/fmp626.md: Likewise.
7172 * config/arm/iwmmxt.md: Likewise.
7173 * config/arm/ldmstm.md: Likewise.
7174 * config/arm/marvell-pj4.md: Likewise.
7175 * config/arm/xgene1.md: Likewise.
7176 * config/aarch64/thunderx.md: Likewise.
7177 * config/aarch64/thunderx2t99.md: Likewise.
7178 * config/aarch64/falkor.md: Likewise.
7180 2017-09-12 Martin Liska <mliska@suse.cz>
7182 * attribs.c (private_lookup_attribute): New function.
7183 * attribs.h (private_lookup_attribute): Declared here.
7184 (lookup_attribute): Called from this place.
7186 2017-09-12 Richard Biener <rguenther@suse.de>
7188 PR tree-optimization/82157
7189 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
7190 stmts with side-effects.
7192 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7193 Alan Hayward <alan.hayward@arm.com>
7194 David Sherwood <david.sherwood@arm.com>
7196 * target.def (hard_regno_nregs): New hook.
7197 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
7198 * targhooks.h (default_hard_regno_nregs): Declare.
7199 * targhooks.c (default_hard_regno_nregs): New function.
7200 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
7201 (TARGET_HARD_REGNO_NREGS): ...this hook.
7202 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
7203 (CLASS_MAX_NREGS): Likewise.
7204 * doc/tm.texi: Regenerate.
7205 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
7206 instead of HARD_REGNO_NREGS.
7207 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
7208 HARD_REGNO_NREGS in the comment.
7209 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
7210 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
7211 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
7212 Return an unsigned int.
7213 (TARGET_HARD_REGNO_NREGS): Redefine.
7214 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
7215 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
7216 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
7217 (arc_hard_regno_nregs): New function.
7218 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
7219 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
7220 (arm_hard_regno_nregs): New function.
7221 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
7222 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
7223 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
7224 (TARGET_HARD_REGNO_NREGS): Redefine.
7225 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
7226 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
7227 (HARD_REGNO_NREGS): Delete.
7228 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
7229 (cr16_hard_regno_nregs): New function.
7230 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
7231 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
7232 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
7233 (cris_hard_regno_nregs): New function.
7234 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
7235 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
7236 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
7237 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
7238 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
7239 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
7240 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
7241 (frv_hard_regno_nregs): Make static. Take and return an
7243 (frv_class_max_nregs): Remove outdated copy of documentation.
7244 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
7245 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
7246 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
7247 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
7248 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
7249 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
7250 (TARGET_HARD_REGNO_NREGS): Redefine.
7251 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
7252 (CLASS_MAX_NREGS): Update comment.
7253 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
7254 (ia64_hard_regno_nregs): New function.
7255 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
7256 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
7257 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
7258 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
7259 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
7261 (m32c_hard_regno_nregs): Likewise. Make static.
7262 (TARGET_HARD_REGNO_NREGS): Redefine.
7263 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
7264 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
7265 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
7266 (m68k_hard_regno_nregs): New function.
7267 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
7268 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
7269 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
7270 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
7271 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
7272 Take and return an unsigned int.
7273 (TARGET_HARD_REGNO_NREGS): Redefine.
7274 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
7275 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
7276 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
7277 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
7278 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
7279 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
7280 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
7281 (msp430_hard_regno_nregs): Make static. Take and return an
7283 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
7284 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
7285 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
7286 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
7287 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
7288 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
7289 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
7290 (TARGET_HARD_REGNO_NREGS): Redefine.
7291 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
7292 (PA_HARD_REGNO_NREGS): ...this.
7293 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
7294 (PA_HARD_REGNO_NREGS): ...this.
7295 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
7296 (pa_hard_regno_nregs): New function.
7297 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
7298 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
7299 (pdp11_hard_regno_nregs): New function.
7300 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
7301 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
7302 (rs6000_hard_regno_nregs_hook): New function.
7303 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
7304 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
7305 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
7306 Take and return an unsigned int. Move earlier in file.
7307 (TARGET_HARD_REGNO_NREGS): Redefine.
7308 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
7309 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
7310 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
7311 (rl78_hard_regno_nregs): Make static. Take and return an
7313 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
7314 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
7315 (rs6000_hard_regno_nregs_hook): New function.
7316 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
7317 * config/rx/rx.c (rx_hard_regno_nregs): New function.
7318 (TARGET_HARD_REGNO_NREGS): Redefine.
7319 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
7320 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
7321 instead of HARD_REGNO_NREGS.
7322 (s390_hard_regno_nregs): New function.
7323 (s390_hard_regno_mode_ok): Add comment from s390.h.
7324 (TARGET_HARD_REGNO_NREGS): Redefine.
7325 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
7326 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
7327 (sh_hard_regno_nregs): New function.
7328 (sh_pass_in_reg_p): Use it.
7329 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
7330 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
7331 (sparc_hard_regno_nregs): New function.
7332 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
7333 * config/spu/spu.c (spu_hard_regno_nregs): New function.
7334 (spu_function_arg_advance): Use it, supplying a valid register number.
7335 (TARGET_HARD_REGNO_NREGS): Redefine.
7336 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
7337 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
7338 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
7339 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
7340 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
7341 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
7342 (CLASS_MAX_NREGS): Remove copy of old documentation.
7343 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
7344 (visium_hard_regno_nregs): New function.
7345 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
7346 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
7347 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
7348 xtensa_hard_regno_nregs): New function.
7349 * system.h (HARD_REGNO_NREGS): Poison.
7351 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7353 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
7354 hard_regno_nregs instead of HARD_REGNO_NREGS.
7355 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
7356 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
7357 (c6x_expand_epilogue): Likewise.
7358 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
7359 (frv_read_iacc_argument): Likewise.
7360 * config/sh/sh.c: Include regs.h.
7361 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
7362 (regs_used): Likewise.
7363 (output_stack_adjust): Likewise.
7364 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
7365 * expmed.c: Include regs.h.
7366 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
7367 * ree.c: Include regs.h.
7368 (combine_reaching_defs): Use hard_regno_nregs instead of
7370 (add_removable_extension): Likewise.
7372 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7374 * regs.h (hard_regno_nregs): Turn into a function.
7375 (end_hard_regno): Update accordingly.
7376 * caller-save.c (setup_save_areas): Likewise.
7377 (save_call_clobbered_regs): Likewise.
7378 (replace_reg_with_saved_mem): Likewise.
7379 (insert_restore): Likewise.
7380 (insert_save): Likewise.
7381 * combine.c (can_change_dest_mode): Likewise.
7382 (move_deaths): Likewise.
7383 (distribute_notes): Likewise.
7384 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
7385 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
7386 (rs6000_split_multireg_move): Likewise.
7387 (rs6000_register_move_cost): Likewise.
7388 (rs6000_memory_move_cost): Likewise.
7389 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
7390 (rs6000_split_multireg_move): Likewise.
7391 (rs6000_register_move_cost): Likewise.
7392 (rs6000_memory_move_cost): Likewise.
7393 * cselib.c (cselib_reset_table): Likewise.
7394 (cselib_lookup_1): Likewise.
7395 * emit-rtl.c (set_mode_and_regno): Likewise.
7396 * function.c (aggregate_value_p): Likewise.
7397 * ira-color.c (setup_profitable_hard_regs): Likewise.
7398 (check_hard_reg_p): Likewise.
7399 (calculate_saved_nregs): Likewise.
7400 (assign_hard_reg): Likewise.
7401 (improve_allocation): Likewise.
7402 (calculate_spill_cost): Likewise.
7403 * ira-emit.c (modify_move_list): Likewise.
7404 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
7405 (ira_hard_reg_in_set_p): Likewise.
7406 * ira.c (setup_reg_mode_hard_regset): Likewise.
7407 (clarify_prohibited_class_mode_regs): Likewise.
7408 (check_allocation): Likewise.
7409 * lra-assigns.c (find_hard_regno_for_1): Likewise.
7410 (lra_setup_reg_renumber): Likewise.
7411 (setup_try_hard_regno_pseudos): Likewise.
7412 (spill_for): Likewise.
7413 (assign_hard_regno): Likewise.
7414 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
7415 * lra-constraints.c (in_class_p): Likewise.
7416 (lra_constraint_offset): Likewise.
7417 (simplify_operand_subreg): Likewise.
7418 (lra_constraints): Likewise.
7419 (split_reg): Likewise.
7420 (split_if_necessary): Likewise.
7421 (invariant_p): Likewise.
7422 (inherit_in_ebb): Likewise.
7423 * lra-lives.c (process_bb_lives): Likewise.
7424 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
7425 (get_hard_regs): Likewise.
7426 (do_remat): Likewise.
7427 * lra-spills.c (assign_spill_hard_regs): Likewise.
7428 * mode-switching.c (create_pre_exit): Likewise.
7429 * postreload.c (reload_combine_recognize_pattern): Likewise.
7430 * recog.c (peep2_find_free_register): Likewise.
7431 * regcprop.c (kill_value_regno): Likewise.
7432 (set_value_regno): Likewise.
7433 (copy_value): Likewise.
7434 (maybe_mode_change): Likewise.
7435 (find_oldest_value_reg): Likewise.
7436 (copyprop_hardreg_forward_1): Likewise.
7437 * regrename.c (check_new_reg_p): Likewise.
7438 (regrename_do_replace): Likewise.
7439 * reload.c (push_reload): Likewise.
7440 (combine_reloads): Likewise.
7441 (find_dummy_reload): Likewise.
7442 (operands_match_p): Likewise.
7443 (find_reloads): Likewise.
7444 (find_equiv_reg): Likewise.
7445 (reload_adjust_reg_for_mode): Likewise.
7446 * reload1.c (count_pseudo): Likewise.
7447 (count_spilled_pseudo): Likewise.
7448 (find_reg): Likewise.
7449 (clear_reload_reg_in_use): Likewise.
7450 (free_for_value_p): Likewise.
7451 (allocate_reload_reg): Likewise.
7452 (choose_reload_regs): Likewise.
7453 (reload_adjust_reg_for_temp): Likewise.
7454 (emit_reload_insns): Likewise.
7455 (delete_output_reload): Likewise.
7456 * rtlanal.c (subreg_get_info): Likewise.
7457 * sched-deps.c (sched_analyze_reg): Likewise.
7458 * sel-sched.c (init_regs_for_mode): Likewise.
7459 (mark_unavailable_hard_regs): Likewise.
7460 (choose_best_reg_1): Likewise.
7461 (verify_target_availability): Likewise.
7462 * valtrack.c (dead_debug_insert_temp): Likewise.
7463 * var-tracking.c (track_loc_p): Likewise.
7464 (emit_note_insn_var_location): Likewise.
7465 * varasm.c (make_decl_rtl): Likewise.
7466 * reginfo.c (choose_hard_reg_mode): Likewise.
7467 (init_reg_modes_target): Refer directly to
7468 this_target_regs->x_hard_regno_nregs.
7470 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7472 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
7473 instead of hard_regno_nregs.
7475 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7477 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
7478 end_hard_regno instead of hard_regno_nregs.
7479 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
7480 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
7481 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
7482 * ira-color.c (improve_allocation): Likewise.
7483 * lra-assigns.c (find_hard_regno_for_1): Likewise.
7484 * lra-lives.c (mark_regno_live): Likewise.
7485 (mark_regno_dead): Likewise.
7486 * lra-remat.c (operand_to_remat): Likewise.
7487 * lra.c (collect_non_operand_hard_regs): Likewise.
7488 * postreload.c (reload_combine_note_store): Likewise.
7489 (move2add_valid_value_p): Likewise.
7490 * reload.c (regno_clobbered_p): Likewise.
7492 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7494 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
7496 * config/v850/v850.c (v850_reorg): Likewise.
7497 * reload.c (refers_to_regno_for_reload_p): Likewise.
7498 (find_equiv_reg): Likewise.
7499 * reload1.c (reload_reg_reaches_end_p): Likewise.
7501 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7503 * caller-save.c (add_used_regs): Use REG_NREGS instead of
7505 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
7506 * config/arm/arm.c (output_move_neon): Likewise.
7507 (arm_attr_length_move_neon): Likewise.
7508 (neon_split_vcombine): Likewise.
7509 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
7510 (c6x_mark_reg_written): Likewise.
7511 (c6x_dwarf_register_span): Likewise.
7512 * config/i386/i386.c (ix86_save_reg): Likewise.
7513 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
7514 (rws_access_reg): Likewise.
7515 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
7516 * mode-switching.c (create_pre_exit): Likewise.
7517 * ree.c (combine_reaching_defs): Likewise.
7518 (add_removable_extension): Likewise.
7519 * regcprop.c (find_oldest_value_reg): Likewise.
7520 (copyprop_hardreg_forward_1): Likewise.
7521 * reload.c (reload_inner_reg_of_subreg): Likewise.
7522 (push_reload): Likewise.
7523 (combine_reloads): Likewise.
7524 (find_dummy_reload): Likewise.
7525 (reload_adjust_reg_for_mode): Likewise.
7526 * reload1.c (find_reload_regs): Likewise.
7527 (forget_old_reloads_1): Likewise.
7528 (reload_reg_free_for_value_p): Likewise.
7529 (reload_adjust_reg_for_temp): Likewise.
7530 (emit_reload_insns): Likewise.
7531 (delete_output_reload): Likewise.
7532 * sel-sched.c (choose_best_reg_1): Likewise.
7533 (choose_best_pseudo_reg): Likewise.
7535 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7536 Alan Hayward <alan.hayward@arm.com>
7537 David Sherwood <david.sherwood@arm.com>
7539 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
7540 * target.def (slow_unaligned_access): New hook.
7541 * targhooks.h (default_slow_unaligned_access): Declare.
7542 * targhooks.c (default_slow_unaligned_access): New function.
7543 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
7544 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
7545 * doc/tm.texi: Regenerate.
7546 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
7547 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
7548 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
7550 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
7551 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
7553 (rs6000_slow_unaligned_access): New function.
7554 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
7555 (expand_block_compare): Likewise.
7556 (expand_strn_compare): Likewise.
7557 (rs6000_rtx_costs): Likewise.
7558 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
7559 (riscv_slow_unaligned_access): Likewise.
7560 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
7561 (riscv_slow_unaligned_access_p): ...this and make static.
7562 (riscv_option_override): Update accordingly.
7563 (riscv_slow_unaligned_access): New function.
7564 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
7565 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
7566 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
7567 (rs6000_slow_unaligned_access): New function.
7568 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
7569 (rs6000_rtx_costs): Likewise.
7570 * config/rs6000/rs6000-string.c (expand_block_compare)
7571 (expand_strn_compare): Use targetm.slow_unaligned_access instead
7572 of SLOW_UNALIGNED_ACCESS.
7573 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
7574 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
7575 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
7576 of SLOW_UNALIGNED_ACCESS.
7577 * expmed.c (simple_mem_bitfield_p): Likewise.
7578 * expr.c (alignment_for_piecewise_move): Likewise.
7579 (emit_group_load_1): Likewise.
7580 (emit_group_store): Likewise.
7581 (copy_blkmode_from_reg): Likewise.
7582 (emit_push_insn): Likewise.
7583 (expand_assignment): Likewise.
7584 (store_field): Likewise.
7585 (expand_expr_real_1): Likewise.
7586 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
7587 * lra-constraints.c (simplify_operand_subreg): Likewise.
7588 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
7589 * gimple-ssa-store-merging.c: Likewise in block comment at start
7591 * tree-ssa-strlen.c: Include target.h.
7592 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
7593 of SLOW_UNALIGNED_ACCESS.
7594 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
7596 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7598 PR rtl-optimization/82185
7599 * expmed.c (emit_store_flag_int): Only test tem if it has been
7602 2017-09-12 Richard Biener <rguenther@suse.de>
7605 * match.pd ((FTYPE) N CMP CST): Fix typo.
7607 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
7609 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
7611 (mips_near_type_p): Add 'short_call' attribute as a synonym
7613 * doc/extend.texi (short_call): Document new function attribute.
7615 2017-09-12 Jakub Jelinek <jakub@redhat.com>
7618 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
7619 assertion check that in the condition.
7620 (get_atomic_generic_size): Likewise. Before testing if parameter
7621 has pointer type, if it has array type, call for C++
7622 default_conversion to perform array-to-pointer conversion.
7624 2017-09-12 Richard Biener <rguenther@suse.de>
7626 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
7627 for operations we cannot scalarize.
7629 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
7631 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
7632 vectors heap vectors. Clean up comments.
7633 Make visited_bbs a reference.
7634 (profitable_jump_thread_path): Make GC
7635 vectors heap vectors. Clean up comments.
7637 (convert_and_register_jump_thread_path): Make GC vectors heap
7639 (check_subpath_and_update_thread_path): Same. Clean up comments.
7640 Make visited_bbs a reference.
7641 (handle_phi): Abstract common code to to
7642 register_jump_thread_path_if_profitable.
7643 Rename VAR_BB to DEF_BB.
7645 Make GC vectors heap vectors.
7646 Make visited_bbs a reference.
7647 (handle_assignment): Same.
7648 (register_jump_thread_path_if_profitable): New.
7649 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
7651 Make GC vectors heap vectors. Clean up comments.
7652 Make visited_bbs a reference.
7653 (find_jump_threads_backwards): Make visited_bbs live in the stack.
7654 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
7657 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
7660 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
7661 words of E_DImode object are reachable by xtensa_uimm8x4 access.
7663 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
7665 Revert r251800 and r251799.
7667 2017-09-11 Martin Jambor <mjambor@suse.cz>
7670 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
7671 arguments in advance.
7672 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
7673 use it to find predecessor edges.
7674 (naive_outof_ssa): Collect vector of predecessors.
7676 2017-09-08 Jason Merrill <jason@redhat.com>
7678 PR c++/70029 - ICE with ref-qualifier and -flto
7679 * langhooks.h (struct lang_hooks_for_types): Add
7680 copy_lang_qualifiers.
7681 * attribs.c (build_type_attribute_qual_variant): Use it.
7682 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
7684 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
7685 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
7687 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
7690 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
7691 (*mulsi3_sp64): New instruction.
7692 (mulsi3): New expander.
7694 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
7696 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
7698 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7700 * sancov.c: Include memmodel.h.
7702 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
7705 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
7708 2017-09-07 Carl Love <cel@us.ibm.com>
7710 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
7711 the sldi instruction.
7713 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
7715 * sancov.c: Include tm_p.h.
7717 2017-09-07 Jakub Jelinek <jakub@redhat.com>
7720 * output.h (switch_to_other_text_partition): New declaration.
7721 * varasm.c (switch_to_other_text_partition): New function.
7722 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
7723 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
7724 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
7725 to the other text partition before emitting LCL label and switch back
7726 after emitting the word after it.
7728 2017-09-07 Richard Biener <rguenther@suse.de>
7730 * passes.def (pass_split_crit_edges): Remove instance before PRE.
7731 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
7732 critical edges here, after loop init.
7733 (pass_data_pre): Remove PROP_no_crit_edges flags.
7734 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
7735 for valueization of call args to avoid leaking VN_TOP.
7736 (visit_use): Assert we do not visit default defs.
7737 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
7738 Use error_mark_node to more easily detect leaking VN_TOP.
7739 All default-defs are varying, not VN_TOP. Mark them visited.
7740 (run_scc_vn): Make code match comment.
7742 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
7744 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
7745 OPTION_MASK_FLOAT128_KEYWORD.
7746 (POWERPC_MASKS): Likewise.
7747 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
7748 support for the -mfloat128-type option, and make -mfloat128
7749 default on PowerPC Linux systems. Define or undefine
7750 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
7751 Define __float128 to be __ieee128 if IEEE 128-bit support is
7752 enabled, or undefine it.
7753 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
7754 Delete defining __FLOAT128_TYPE__.
7755 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
7756 -mfloat128-type option and make -mfloat128 default on PowerPC
7758 (TARGET_FLOAT128_TYPE): Likewise.
7759 (-mfloat128-type): Likewise.
7760 * config/rs6000/rs6000.c (rs6000_option_override_internal):
7761 Delete the -mfloat128-type option and make -mfloat128 default on
7762 PowerPC Linux systems. Always use __ieee128 to be the keyword for
7763 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
7764 128-bit floating point is enabled. Change tests from using
7765 -mfloat128-type to -mfloat128.
7766 (rs6000_mangle_type): Use the correct mangling for the __float128
7767 type even if normal long double is restricted to 64-bits.
7768 (floatn_mode): Enable the _Float128 type by default on VSX Linux
7770 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
7771 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
7772 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
7774 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
7775 documentation for -mfloat128.
7777 2017-09-06 Olivier Hainque <hainque@adacore.com>
7779 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
7781 2017-09-06 Wish Wu <wishwu007@gmail.com>
7782 Jakub Jelinek <jakub@redhat.com>
7784 * asan.c (initialize_sanitizer_builtins): Add
7785 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
7786 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
7787 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
7788 BT_FN_VOID_UINT64_PTR variables.
7789 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
7790 (BT_FN_VOID_UINT16_UINT16): Likewise.
7791 (BT_FN_VOID_UINT32_UINT32): Likewise.
7792 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
7793 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
7794 (BT_FN_VOID_UINT64_PTR): Likewise.
7795 * common.opt (flag_sanitize_coverage): New variable.
7796 (fsanitize-coverage=trace-pc): Remove.
7797 (fsanitize-coverage=): Add.
7798 * flag-types.h (enum sanitize_coverage_code): New enum.
7799 * fold-const.c (fold_range_test): Disable non-short-circuit
7800 optimization if flag_sanitize_coverage.
7801 (fold_truth_andor): Likewise.
7802 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
7803 * opts.c (COVERAGE_SANITIZER_OPT): Define.
7804 (coverage_sanitizer_opts): New array.
7805 (get_closest_sanitizer_option): Add OPTS argument, handle also
7806 OPT_fsanitize_coverage_.
7807 (parse_sanitizer_options): Adjusted to also handle
7808 OPT_fsanitize_coverage_.
7809 (common_handle_option): Add OPT_fsanitize_coverage_.
7810 * sancov.c (instrument_comparison, instrument_switch): New function.
7811 (sancov_pass): Add trace-cmp support.
7812 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
7813 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
7814 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
7815 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
7816 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
7817 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
7818 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
7819 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
7820 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
7821 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
7823 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
7825 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
7826 error. Only quit immediately if parsing is complete.
7827 (BEGIN): Initialize fatal_err and parse_done.
7828 (begin fpu, end fpu): Check number of arguments.
7829 (begin arch, end arch): Likewise.
7830 (begin cpu, end cpu): Likewise.
7831 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
7832 (optalias): Likewise.
7834 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
7836 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
7837 * config/arm/arm-isa.h: Delete. Move definitions to ...
7838 * arm-cpus.in: ... here. Use new feature and fgroup values.
7839 * config/arm/arm.c (arm_option_override): Use lower case for feature
7841 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
7842 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
7843 * config/arm/parsecpu.awk (END): Add new command 'isa'.
7845 (print_isa_bits_for): New function.
7846 (gen_isa): New function.
7847 (gen_comm_data): Use print_isa_bits_for.
7848 (define feature): New keyword.
7849 (define fgroup): New keyword.
7850 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
7851 (arm-isa.h): Add rule to generate file.
7852 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
7853 case for feature bit names.
7855 2017-09-06 Richard Biener <rguenther@suse.de>
7857 * tree-ssa-pre.c (NECESSARY): Remove.
7858 (create_expression_by_pieces): Do not touch pass-local flags.
7859 (insert_into_preds_of_block): Likewise.
7860 (do_pre_regular_insertion): Likewise.
7861 (eliminate_insert): Likewise.
7862 (eliminate_dom_walker::before_dom_children): Likewise.
7863 (fini_eliminate): Do not look at inserted_exprs.
7864 (mark_operand_necessary): Remove.
7865 (remove_dead_inserted_code): Replace with simple work-list
7866 algorithm based on inserted_exprs and SSA uses.
7867 (pass_pre::execute): Re-order fini_eliminate and
7868 remove_dead_inserted_code.
7870 2017-09-06 Olivier Hainque <hainque@adacore.com>
7872 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
7873 for VxWorks 7. Adjust surrounding comments.
7875 2017-09-06 Richard Biener <rguenther@suse.de>
7877 * gimple-ssa-strength-reduction.c
7878 (find_candidates_dom_walker::before_dom_children): Also allow
7881 2017-09-06 Richard Biener <rguenther@suse.de>
7883 PR tree-optimization/82108
7884 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
7885 for gap in the non-permutation SLP case.
7887 2017-09-06 Martin Jambor <mjambor@suse.cz>
7889 PR tree-optimization/82078
7890 * tree-sra.c (sort_and_splice_var_accesses): Move call to
7891 add_access_to_work_queue...
7892 (build_accesses_from_assign): ...here.
7893 (propagate_all_subaccesses): Make sure racc is the group
7894 representative, if there is one.
7896 2017-09-06 Jakub Jelinek <jakub@redhat.com>
7899 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
7902 2017-09-06 Richard Biener <rguenther@suse.de>
7904 * gimple-ssa-strength-reduction.c
7905 (find_candidates_dom_walker::before_doom_children): Use a
7906 type and not a mode check.
7908 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
7911 * config/arm/predicates.md (arm_general_adddi_operand): Create new
7913 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
7915 2017-09-05 Jeff Law <law@redhat.com>
7917 PR tree-optimization/64910
7918 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
7919 cases where we have 3 or more operands.
7921 2017-09-05 Jakub Jelinek <jakub@redhat.com>
7924 * omp-low.c (lower_omp_for): Recompute tree invariant if
7925 gimple_omp_for_initial/final is ADDR_EXPR.
7928 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
7929 into gimple val before gimplification fo the COND_EXPR.
7931 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
7933 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
7934 REGION_COPY argument.
7935 (thread_through_all_blocks): Remove unused argument to
7936 duplicate_thread_path.
7938 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7939 Alan Hayward <alan.hayward@arm.com>
7940 David Sherwood <david.sherwood@arm.com>
7942 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
7943 Take a scalar_mode rather than a machine_mode.
7944 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
7945 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
7946 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
7947 (aarch64_gen_adjusted_ldpstp): Likewise.
7948 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
7950 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7951 Alan Hayward <alan.hayward@arm.com>
7952 David Sherwood <david.sherwood@arm.com>
7954 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
7955 Take a scalar_int_mode instead of a machine_mode.
7956 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
7957 (aarch64_output_scalar_simd_mov_immediate): Likewise.
7958 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
7959 (aarch64_simd_attr_length_rglist): Delete.
7960 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
7961 a scalar_int_mode instead of a machine_mode.
7962 (aarch64_add_offset): Likewise.
7963 (aarch64_internal_mov_immediate): Likewise
7964 (aarch64_add_constant_internal): Likewise.
7965 (aarch64_add_constant): Likewise.
7966 (aarch64_movw_imm): Likewise.
7967 (aarch64_rtx_arith_op_extract_p): Likewise.
7968 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
7969 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
7970 Remove assert that the mode isn't a vector.
7971 (aarch64_output_scalar_simd_mov_immediate): Likewise.
7972 (aarch64_expand_mov_immediate): Update calls after above changes.
7973 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
7974 (aarch64_and_bitmask_imm): Check for scalar integer modes.
7975 (aarch64_move_imm): Likewise.
7976 (aarch64_can_const_movi_rtx_p): Likewise.
7977 (aarch64_strip_extend): Likewise.
7978 (aarch64_extr_rtx_p): Likewise.
7979 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
7980 a CONST_INT when the mode parameter is VOIDmode.
7981 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
7983 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7985 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
7986 * stor-layout.c (bitwise_mode_for_mode): Likewise.
7987 (bitwise_type_for_mode): Update accordingly.
7989 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7991 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
7992 * stor-layout.c (mode_for_size_tree): Likewise.
7993 (mode_for_array): Update accordingly.
7994 (layout_decl): Likewise.
7995 (compute_record_mode): Likewise. Only set the mode once.
7997 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7999 * target.def (get_mask_mode): Change return type to opt_mode.
8001 * doc/tm.texi: Regenerate.
8002 * targhooks.h (default_get_mask_mode): Return an opt_mode.
8003 * targhooks.c (default_get_mask_mode): Likewise.
8004 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
8005 * optabs-query.c (can_vec_mask_load_store_p): Update use of
8006 targetm.get_mask_mode.
8007 * tree.c (build_truth_vector_type): Likewise.
8009 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8011 * machmode.h (mode_for_vector): Return an opt_mode.
8012 * stor-layout.c (mode_for_vector): Likewise.
8013 (mode_for_int_vector): Update accordingly.
8014 (layout_type): Likewise.
8015 * config/i386/i386.c (emit_memmov): Likewise.
8016 (ix86_expand_set_or_movmem): Likewise.
8017 (ix86_expand_vector_init): Likewise.
8018 (ix86_get_mask_mode): Likewise.
8019 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
8021 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
8022 * expmed.c (extract_bit_field_1): Likewise.
8023 * expr.c (expand_expr_real_2): Likewise.
8024 * optabs-query.c (can_vec_perm_p): Likewise.
8025 (can_vec_mask_load_store_p): Likewise.
8026 * optabs.c (expand_vec_perm): Likewise.
8027 * targhooks.c (default_get_mask_mode): Likewise.
8028 * tree-vect-stmts.c (vectorizable_store): Likewise.
8029 (vectorizable_load): Likewise.
8030 (get_vectype_for_scalar_type_and_size): Likewise.
8032 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8034 * machmode.h (mode_for_int_vector): New function.
8035 * stor-layout.c (mode_for_int_vector): Likewise.
8036 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
8037 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
8038 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
8039 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
8040 (s390_expand_vcond): Likewise.
8042 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8044 * machmode.h (opt_machine_mode): New type.
8045 (opt_mode<T>): Allow construction from anything that can be
8047 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
8048 (mode_for_size): Return an opt_machine_mode.
8049 * stor-layout.c (mode_for_size): Likewise.
8050 (mode_for_size_tree): Update call accordingly.
8051 (bitwise_mode_for_mode): Likewise.
8052 (make_fract_type): Likewise.
8053 (make_accum_type): Likewise.
8054 * caller-save.c (replace_reg_with_saved_mem): Update call
8056 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8057 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8058 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8059 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8060 * expmed.c (extract_bit_field_1): Likewise.
8061 * reload.c (get_secondary_mem): Likewise.
8062 * varasm.c (assemble_integer): Likewise.
8063 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
8066 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8068 * machmode.h (decimal_float_mode_for_size): New function.
8069 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
8070 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
8071 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
8072 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
8073 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
8074 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
8076 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8078 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
8079 (get_builtin_sync_mode): Likewise.
8080 (expand_ifn_atomic_compare_exchange): Likewise.
8081 (expand_builtin_atomic_clear): Likewise.
8082 (expand_builtin_atomic_test_and_set): Likewise.
8083 (fold_builtin_atomic_always_lock_free): Likewise.
8084 * calls.c (compute_argument_addresses): Likewise.
8085 (emit_library_call_value_1): Likewise.
8086 (store_one_arg): Likewise.
8087 * combine.c (combine_instructions): Likewise.
8088 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
8089 * config/arm/arm.c (arm_function_value): Likewise.
8090 (aapcs_allocate_return_reg): Likewise.
8091 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
8092 * config/i386/i386.c (construct_container): Likewise.
8093 (ix86_gimplify_va_arg): Likewise.
8094 (ix86_expand_sse_cmp): Likewise.
8095 (emit_memmov): Likewise.
8096 (emit_memset): Likewise.
8097 (expand_small_movmem_or_setmem): Likewise.
8098 (ix86_expand_pextr): Likewise.
8099 (ix86_expand_pinsr): Likewise.
8100 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
8101 * config/microblaze/microblaze.c (microblaze_block_move_straight):
8103 * config/mips/mips.c (mips_function_value_1) Likewise.
8104 (mips_block_move_straight): Likewise.
8105 (mips_expand_ins_as_unaligned_store): Likewise.
8106 * config/powerpcspe/powerpcspe.c
8107 (rs6000_darwin64_record_arg_advance_flush): Likewise.
8108 (rs6000_darwin64_record_arg_flush): Likewise.
8109 * config/rs6000/rs6000.c
8110 (rs6000_darwin64_record_arg_advance_flush): Likewise.
8111 (rs6000_darwin64_record_arg_flush): Likewise.
8112 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
8113 (sparc_function_value_1): Likewise.
8114 * config/spu/spu.c (adjust_operand): Likewise.
8115 (spu_emit_branch_or_set): Likewise.
8116 (arith_immediate_p): Likewise.
8117 * emit-rtl.c (gen_lowpart_common): Likewise.
8118 * expr.c (expand_expr_real_1): Likewise.
8119 * function.c (assign_parm_setup_block): Likewise.
8120 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8121 * reload1.c (alter_reg): Likewise.
8122 * stor-layout.c (mode_for_vector): Likewise.
8123 (layout_type): Likewise.
8125 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8127 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
8128 (spu_convert_move): Likewise.
8129 * lower-subreg.c (resolve_simple_move): Likewise.
8131 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8134 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
8135 define_insn to a define_expand.
8136 (altivec_vsum2sws_direct): New define_insn.
8137 (altivec_vsumsws): Convert from a define_insn to a define_expand.
8139 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8141 * config/arm/arm.c (arm_option_params_internal): Improve setting of
8144 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
8149 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
8150 realign stack if stack alignment needed is less than incoming
8153 2017-09-05 Marek Polacek <polacek@redhat.com>
8156 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
8159 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8161 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
8163 2017-09-05 Richard Biener <rguenther@suse.de>
8165 PR tree-optimization/82084
8166 * fold-const.c (can_native_encode_string_p): Handle wide characters.
8168 2017-09-05 Richard Biener <rguenther@suse.de>
8170 PR tree-optimization/82102
8171 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
8173 2017-09-05 Martin Liska <mliska@suse.cz>
8175 PR tree-optimization/82032
8176 * tree-cfg.c (generate_range_test): New function.
8177 * tree-cfg.h (generate_range_test): Declared here.
8178 * tree-cfgcleanup.c (convert_single_case_switch): New function.
8179 (cleanup_control_expr_graph): Use it.
8180 * tree-switch-conversion.c (try_switch_expansion): Remove
8182 (emit_case_nodes): Use generate_range_test.
8184 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
8187 * config/i386/i386.md (*<btsc><mode>_mask): Add
8188 TARGET_USE_BT to insn constraint.
8189 (*btr<mode>_mask): Ditto.
8191 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
8193 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
8194 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
8196 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
8199 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
8200 TARGET_NEON and TARGET_IWMMXT.
8201 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
8202 TARGET_NEON and TARGET_IWMMXT.
8203 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
8205 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
8207 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
8208 (ix86_rewrite_tls_address): Ditto.
8209 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
8210 (ix86_rewrite_tls_address_1): Ditto.
8211 (ix86_rewrite_tls_address): Ditto.
8212 * config/i386/predicates.md (tls_address_pattern): New predicate.
8213 * config/i386/i386.md (TLS address splitter): New splitter.
8215 2017-09-04 Richard Biener <rguenther@suse.de>
8217 PR tree-optimization/82084
8218 * fold-const.h (can_native_encode_string_p): Declare.
8219 * fold-const.c (can_native_encode_string_p): Factor out from ...
8220 (native_encode_string): ... here.
8221 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
8222 vectorizing stores from constants we later cannot handle.
8224 2017-09-04 Marek Polacek <polacek@redhat.com>
8227 * doc/invoke.texi: Update -Wtautological-compare documentation.
8229 2017-09-04 Jeff Law <law@redhat.com>
8231 PR tree-optimization/64910
8232 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
8233 swap the first and last operand if the last is a constant.
8235 2017-09-04 Marek Polacek <polacek@redhat.com>
8238 * convert.c (do_narrow): When sanitizing signed integer overflows,
8239 bail out for signed types.
8240 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
8242 2017-09-04 Richard Biener <rguenther@suse.de>
8244 PR tree-optimization/82060
8245 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
8246 Move devirtualization after stmt folding and before EH/AB/noreturn
8247 cleanup to get the stmt refs canonicalized. Use a bool instead
8248 of gimple_modified_p since that doesn't work for NOPs. Schedule
8249 NOPs generated by folding for removal.
8251 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8252 Alan Hayward <alan.hayward@arm.com>
8253 David Sherwood <david.sherwood@arm.com>
8255 * coretypes.h (pad_direction): New enum.
8256 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
8257 (FUNCTION_ARG_PADDING): Likewise.
8258 * target.def (function_arg_padding): New hook.
8259 * targhooks.h (default_function_arg_padding): Declare.
8260 * targhooks.c (default_function_arg_padding): New function.
8261 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
8262 (TARGET_FUNCTION_ARG_PADDING): ...this.
8263 * doc/tm.texi: Regenerate.
8264 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
8265 instead of direction.
8266 (compute_argument_addresses): Likewise.
8267 (load_register_parameters): Likewise.
8268 (emit_library_call_value_1): Likewise.
8269 (store_one_arg): Use targetm.calls.function_arg_padding instead
8270 of FUNCTION_ARG_PADDING.
8271 (must_pass_in_stack_var_size_or_pad): Likewise.
8272 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
8273 (emit_group_store): Likewise.
8274 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
8275 instead of FUNCTION_ARG_PADDING.
8276 (emit_push_insn): Likewise, and propagate enum change throughout
8278 * function.h (direction): Delete.
8279 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
8281 * function.c (assign_parm_find_stack_rtl): Likewise.
8282 (assign_parm_setup_block_p): Likewise.
8283 (assign_parm_setup_block): Likewise.
8284 (gimplify_parameters): Likewise.
8285 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
8286 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
8288 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
8289 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8290 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
8291 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
8292 (aarch64_function_arg_padding): ...this new function.
8293 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
8294 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8295 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
8296 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8297 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
8298 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8299 (arm_pad_arg_upward): Replace with...
8300 (arm_function_arg_padding): ...this new function.
8301 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
8303 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
8304 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
8305 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8306 (ia64_hpux_function_arg_padding): Replace with...
8307 (ia64_function_arg_padding): ...this new function. Use pad_direction
8308 instead of direction. Check for TARGET_HPUX.
8309 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
8310 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8311 (iq2000_function_arg_padding): New function.
8312 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
8313 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
8314 (mips_function_arg_padding): ...this new function.
8315 (mips_pad_reg_upward): Update accordingly.
8316 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8317 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
8318 targetm.calls.function_arg_padding.
8319 (FUNCTION_ARG_PADDING): Delete.
8320 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8321 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
8322 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8323 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
8324 (nios2_block_reg_padding): Return pad_direction instead of direction.
8325 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
8326 instead of direction.
8327 (nios2_function_arg_padding): Likewise. Make static.
8328 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8329 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
8330 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
8331 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
8332 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8333 (pa_function_arg_padding): Make static. Return pad_direction instead
8335 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
8336 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8337 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
8338 instead of direction. Use targetm.calls.function_arg_padding.
8339 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
8340 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
8341 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
8342 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
8343 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
8345 (function_arg_padding): Rename to...
8346 (rs6000_function_arg_padding): ...this. Make static. Return
8347 pad_direction instead of direction.
8348 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
8349 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
8350 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8351 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
8352 instead of direction. Use targetm.calls.function_arg_padding.
8353 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
8354 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
8355 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
8356 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
8357 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8358 (function_arg_padding): Rename to...
8359 (rs6000_function_arg_padding): ...this. Make static. Return
8360 pad_direction instead of direction.
8361 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
8362 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
8363 * config/s390/s390.c (s390_function_arg_padding): New function.
8364 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8365 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
8366 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
8367 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8368 (function_arg_padding): Rename to...
8369 (sparc_function_arg_padding): ...this. Make static. Return
8370 pad_direction instead of direction.
8371 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
8372 * config/spu/spu.c (spu_function_arg_padding): New function.
8373 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8374 * system.h (FUNCTION_ARG_PADDING): Poison.
8376 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8377 Alan Hayward <alan.hayward@arm.com>
8378 David Sherwood <david.sherwood@arm.com>
8380 * target.def (modes_tieable_p): New hook.
8381 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
8382 (TARGET_MODES_TIEABLE_P): ...this.
8383 * doc/tm.texi.in: Regenerate.
8384 * hooks.h (hook_bool_mode_mode_true): Declare.
8385 * hooks.c (hook_bool_mode_mode_true): New function.
8386 * combine.c (subst): Use targetm.modes_tieable_p instead of
8388 * dse.c (find_shift_sequence): Likewise.
8389 * expmed.c (extract_low_bits): Likewise.
8390 * lower-subreg.c: Include target.h.
8391 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
8393 * rtlanal.c (rtx_cost): Likewise.
8394 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
8395 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
8396 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
8397 (TARGET_MODES_TIEABLE_P): Redefine.
8398 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
8399 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
8400 (TARGET_MODES_TIEABLE_P): Redefine.
8401 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
8402 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
8403 (arc_modes_tieable_p): New function.
8404 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
8405 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
8406 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
8407 (arm_modes_tieable_p): Make static.
8408 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
8409 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
8410 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
8411 (TARGET_MODES_TIEABLE_P): Redefine.
8412 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
8413 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
8414 (TARGET_MODES_TIEABLE_P): Redefine.
8415 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
8416 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
8417 (cr16_modes_tieable_p): New function.
8418 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
8419 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
8420 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
8421 (TRULY_NOOP_TRUNCATION): Update comment.
8422 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
8423 (TRULY_NOOP_TRUNCATION): Update comment.
8424 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
8425 (frv_modes_tieable_p): New function.
8426 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
8427 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
8428 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
8429 (TARGET_MODES_TIEABLE_P): Redefine.
8430 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
8431 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
8432 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
8433 (TARGET_MODES_TIEABLE_P): Redefine.
8434 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
8435 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
8436 (ia64_modes_tieable_p): New function.
8437 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
8438 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
8439 (iq2000_modes_tieable_p): New function.
8440 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
8441 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
8442 (lm32_modes_tieable_p): New function.
8443 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
8444 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
8445 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
8446 (TARGET_MODES_TIEABLE_P): Redefine.
8447 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
8448 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
8449 (m32r_modes_tieable_p): New function.
8450 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
8451 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
8452 (m68k_modes_tieable_p): New function.
8453 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
8454 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
8455 (mcore_modes_tieable_p): New function.
8456 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
8457 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
8459 (TARGET_MODES_TIEABLE_P): Redefine.
8460 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
8461 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
8462 * config/mips/mips.c (mips_modes_tieable_p): Make static.
8463 (TARGET_MODES_TIEABLE_P): Redefine.
8464 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
8465 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
8466 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
8467 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
8468 (mn10300_modes_tieable_p): ...this and make static.
8469 (TARGET_MODES_TIEABLE_P): Redefine.
8470 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
8471 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
8472 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
8473 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
8474 (msp430_modes_tieable_p): Make static.
8475 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
8476 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
8477 (TARGET_MODES_TIEABLE_P): Redefine.
8478 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
8479 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
8480 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
8481 (TARGET_MODES_TIEABLE_P): Redefine.
8482 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
8483 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
8484 * config/pa/pa.c (pa_modes_tieable_p): Make static.
8485 (TARGET_MODES_TIEABLE_P): Redefine.
8486 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
8487 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
8488 (pdp11_modes_tieable_p): New function.
8489 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
8490 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
8491 (rs6000_modes_tieable_p): New function.
8492 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
8493 * config/powerpcspe/powerpcspe.md: Update comment.
8494 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
8495 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
8496 (TARGET_MODES_TIEABLE_P): Redefine.
8497 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
8498 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
8499 (rl78_modes_tieable_p): New function.
8500 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
8501 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
8502 (rs6000_modes_tieable_p): New function.
8503 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
8504 * config/rs6000/rs6000.md: Update comment.
8505 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
8506 * config/rx/rx.c (rx_modes_tieable_p): New function.
8507 (TARGET_MODES_TIEABLE_P): Redefine.
8508 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
8509 * config/s390/s390.c (s390_modes_tieable_p): New function.
8510 (TARGET_MODES_TIEABLE_P): Redefine.
8511 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
8512 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
8513 (sh_modes_tieable_p): New function.
8514 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
8515 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
8516 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
8517 (sparc_modes_tieable_p): Make static.
8518 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
8519 * config/spu/spu.c (spu_modes_tieable_p): New function.
8520 (TARGET_MODES_TIEABLE_P): Redefine.
8521 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
8522 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
8523 (TARGET_MODES_TIEABLE_P): Redefine.
8524 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
8525 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
8526 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
8527 * config/v850/v850.c (v850_modes_tieable_p): New function.
8528 (TARGET_MODES_TIEABLE_P): Redefine.
8529 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
8530 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
8531 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
8532 (visium_modes_tieable_p): New function.
8533 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
8534 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
8535 (xtensa_modes_tieable_p): New function.
8536 * system.h (MODES_TIEABLE_P): Poison.
8538 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8539 Alan Hayward <alan.hayward@arm.com>
8540 David Sherwood <david.sherwood@arm.com>
8542 * target.def (hard_regno_mode_ok): New hook.
8543 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
8544 (TARGET_HARD_REGNO_MODE_OK): ...this.
8545 * doc/tm.texi.in: Regenerate.
8546 * hooks.h (hook_bool_uint_mode_true): Declare.
8547 * hooks.c (hook_bool_uint_mode_true): New function.
8548 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
8550 * genpreds.c (write_insn_preds_c): Add an include of target.h.
8551 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
8552 instead of HARD_REGNO_MODE_OK.
8553 * caller-save.c: Include target.h.
8554 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
8556 * combine.c (can_combine_p): Likewise.
8557 (combinable_i3pat): Likewise.
8558 (can_change_dest_mode): Likewise.
8559 * expr.c (init_expr_target): Likewise.
8560 (convert_move): Likewise.
8561 (convert_modes): Likewise.
8562 * ira.c (setup_prohibited_class_mode_regs): Likewise.
8563 (setup_prohibited_mode_move_regs): Likewise.
8564 * ira.h (target_ira): Likewise.
8565 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8566 * lra-constraints.c (process_alt_operands): Likewise.
8567 (split_reg): Likewise.
8568 * recog.c (peep2_find_free_register): Likewise.
8569 * ree.c (combine_reaching_defs): Likewise.
8570 * regcprop.c (maybe_mode_change): Likewise.
8571 * reginfo.c (init_reg_sets_1): Likewise.
8572 (choose_hard_reg_mode): Likewise.
8573 (simplifiable_subregs): Likewise.
8574 * regrename.c (check_new_reg_p): Likewise.
8575 * reload.c (find_valid_class): Likewise.
8576 (find_valid_class_1): Likewise.
8577 (reload_inner_reg_of_subreg): Likewise.
8578 (push_reload): Likewise.
8579 (combine_reloads): Likewise.
8580 (find_dummy_reload): Likewise.
8581 (find_reloads): Likewise.
8582 * reload1.c (find_reg): Likewise.
8583 (set_reload_reg): Likewise.
8584 (allocate_reload_reg): Likewise.
8585 (choose_reload_regs): Likewise.
8586 (reload_adjust_reg_for_temp): Likewise.
8587 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
8588 (simplify_subreg_regno): Likewise.
8589 * sel-sched.c (init_regs_for_mode): Likewise.
8590 * varasm.c (make_decl_rtl): Likewise.
8591 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
8592 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
8594 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
8595 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
8596 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8597 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
8598 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
8599 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8600 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
8601 (arc_mode_class): Delete.
8602 (HARD_REGNO_MODE_OK): Delete.
8603 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8604 (arc_hard_regno_mode_ok): Rename old array to...
8605 (arc_hard_regno_mode_ok_modes): ...this.
8606 (arc_conditional_register_usage): Update accordingly.
8607 (arc_mode_class): Make static.
8608 (arc_hard_regno_mode_ok): New function.
8609 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
8610 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
8611 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8612 (arm_hard_regno_mode_ok): Make static.
8613 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
8615 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
8616 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
8617 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
8619 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8620 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
8621 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
8622 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
8623 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
8624 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8625 * config/bfin/predicates.md (valid_reg_operand): Use
8626 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
8627 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
8628 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
8629 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8630 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
8631 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
8632 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8633 (cr16_hard_regno_mode_ok): Make static and return a bool.
8634 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
8635 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8636 (cris_hard_regno_mode_ok): New function.
8637 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
8638 (epiphany_mode_class): Delete.
8639 (HARD_REGNO_MODE_OK): Delete.
8640 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
8641 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8642 (hard_regno_mode_ok): Rename to...
8643 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
8644 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
8645 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
8647 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
8648 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
8649 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8650 (frv_hard_regno_mode_ok): Make static and return a bool.
8651 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
8653 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
8654 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
8655 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
8656 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
8658 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8659 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
8660 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
8661 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
8663 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8664 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
8665 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8666 (ia64_hard_regno_mode_ok): New function.
8667 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
8668 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8669 (iq2000_hard_regno_mode_ok): New function.
8670 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
8671 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8672 (lm32_hard_regno_mode_ok): New function.
8673 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
8674 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
8675 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
8676 instead of HARD_REGNO_MODE_OK.
8677 (m32c_hard_regno_ok): Rename to...
8678 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
8679 (m32c_cannot_change_mode_class): Update accordingly.
8680 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8681 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
8682 (m32r_mode_class): Delete.
8683 (HARD_REGNO_MODE_OK): Delete.
8684 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8685 (m32r_hard_regno_mode_ok): Rename to...
8686 (m32r_hard_regno_modes): ...this.
8687 (m32r_mode_class): Make static.
8688 (m32r_hard_regno_mode_ok): New function.
8689 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
8690 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
8691 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8692 (m68k_hard_regno_mode_ok): Make static.
8693 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
8694 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8695 (mcore_hard_regno_mode_ok): New function.
8696 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
8697 (HARD_REGNO_MODE_OK): Delete.
8698 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
8700 (microblaze_hard_regno_mode_ok_p): ...this and make static.
8701 (microblaze_hard_regno_mode_ok): New function.
8702 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8703 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
8704 (mips_hard_regno_mode_ok): Delete.
8705 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
8706 (mips_hard_regno_mode_ok_p): ...this and make static.
8707 (mips_hard_regno_mode_ok_p): Rename to...
8708 (mips_hard_regno_mode_ok_uncached): ...this.
8709 (mips_hard_regno_mode_ok): New function.
8710 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
8711 of HARD_REGNO_MODE_OK.
8712 (mips_option_override): Update after above name changes.
8713 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8714 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
8715 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
8716 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
8717 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
8718 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8719 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
8720 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
8721 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
8722 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8723 (msp430_hard_regno_mode_ok): Make static and return a bool.
8724 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
8725 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
8726 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
8728 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8729 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
8730 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
8731 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
8732 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
8733 (PA_HARD_REGNO_MODE_OK): ...this
8734 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
8735 (PA_HARD_REGNO_MODE_OK): ...this.
8736 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8737 (pa_hard_regno_mode_ok): New function.
8738 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
8739 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8740 (pdp11_hard_regno_mode_ok): New function.
8741 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
8742 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
8744 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
8746 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8747 (rs6000_hard_regno_mode_ok): Rename to...
8748 (rs6000_hard_regno_mode_ok_uncached): ...this.
8749 (rs6000_init_hard_regno_mode_ok): Update accordingly.
8750 (rs6000_hard_regno_mode_ok): New function.
8751 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
8752 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
8753 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
8754 (riscv_hard_regno_mode_ok): ...this and make static.
8755 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8756 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
8757 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
8758 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8759 (rl78_hard_regno_mode_ok): Make static and return bool.
8760 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
8761 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
8763 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
8764 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8765 (rs6000_hard_regno_mode_ok): Rename to...
8766 (rs6000_hard_regno_mode_ok_uncached): ...this.
8767 (rs6000_init_hard_regno_mode_ok): Update accordingly.
8768 (rs6000_hard_regno_mode_ok): New function.
8769 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
8770 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
8771 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8772 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
8773 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
8774 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
8775 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8776 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
8777 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
8778 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8779 (sh_hard_regno_mode_ok): Make static.
8780 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
8781 instead of HARD_REGNO_MODE_OK.
8782 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
8783 (sparc_mode_class): Delete.
8784 (HARD_REGNO_MODE_OK): Delete.
8785 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8786 (hard_regno_mode_classes): Make static.
8787 (sparc_mode_class): Likewise.
8788 (sparc_hard_regno_mode_ok): New function.
8789 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
8790 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
8791 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
8793 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8794 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
8795 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
8796 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
8797 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
8798 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8799 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
8800 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
8801 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8802 (visium_hard_regno_mode_ok): New function.
8803 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
8804 instead of HARD_REGNO_MODE_OK.
8805 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
8806 (HARD_REGNO_MODE_OK): Delete.
8807 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
8808 (xtensa_hard_regno_mode_ok_p): ...this and make static.
8809 (xtensa_option_override): Update accordingly.
8810 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8811 (xtensa_hard_regno_mode_ok): New function.
8812 * system.h (HARD_REGNO_MODE_OK): Poison.
8814 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8815 Alan Hayward <alan.hayward@arm.com>
8816 David Sherwood <david.sherwood@arm.com>
8818 * target.def (hard_regno_call_part_clobbered): New hook.
8819 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
8820 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
8821 * doc/tm.texi: Regenerate.
8822 * hooks.h (hook_bool_uint_mode_false): Declare.
8823 * hooks.c (hook_bool_uint_mode_false): New function.
8824 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8825 * cselib.c (cselib_process_insn): Use
8826 targetm.hard_regno_call_part_clobbered instead of
8827 HARD_REGNO_CALL_PART_CLOBBERED.
8828 * ira-conflicts.c (ira_build_conflicts): Likewise.
8829 * ira-costs.c (ira_tune_allocno_costs): Likewise.
8830 * lra-constraints.c (need_for_call_save_p): Likewise.
8831 * lra-lives.c: Include target.h.
8832 (check_pseudos_live_through_calls): Use
8833 targetm.hard_regno_call_part_clobbered instead of
8834 HARD_REGNO_CALL_PART_CLOBBERED.
8835 * regcprop.c: Include target.h.
8836 (copyprop_hardreg_forward_1): Use
8837 targetm.hard_regno_call_part_clobbered instead of
8838 HARD_REGNO_CALL_PART_CLOBBERED.
8839 * reginfo.c (choose_hard_reg_mode): Likewise.
8840 * regrename.c (check_new_reg_p): Likewise.
8841 * reload.c (find_equiv_reg): Likewise.
8842 * reload1.c (emit_reload_insns): Likewise.
8843 * sched-deps.c (deps_analyze_insn): Likewise.
8844 * sel-sched.c (init_regs_for_mode): Likewise.
8845 (mark_unavailable_hard_regs): Likewise.
8846 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
8847 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8848 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
8850 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8851 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8852 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
8854 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
8856 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8857 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8858 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
8860 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8861 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8862 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
8864 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8865 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
8867 * config/powerpcspe/powerpcspe.c
8868 (rs6000_hard_regno_call_part_clobbered): New function.
8869 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8870 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8871 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
8873 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8874 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8875 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
8877 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8878 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8879 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
8881 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8882 Alan Hayward <alan.hayward@arm.com>
8883 David Sherwood <david.sherwood@arm.com>
8885 * rtl.h (subreg_memory_offset): Declare.
8886 * emit-rtl.c (subreg_memory_offset): New function.
8887 * expmed.c (store_bit_field_1): Use it.
8888 * expr.c (undefined_operand_subword_p): Likewise.
8889 * simplify-rtx.c (simplify_subreg): Likewise.
8891 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
8893 PR rtl-optimization/57448
8896 * optabs.c (expand_atomic_load): Place compiler memory barriers if
8897 using atomic_load pattern.
8898 (expand_atomic_store): Likewise.
8900 2017-09-04 Jakub Jelinek <jakub@redhat.com>
8903 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
8904 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
8905 handling. Use replace_call_with_value with NULL instead of
8906 gsi_replace, unlink_stmt_vdef and release_defs.
8908 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
8911 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
8913 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8916 * rtl.h (emit_library_call_value_1): Declare.
8917 (emit_library_call): Replace declaration with a series of overloads.
8918 Remove the parameter count argument.
8919 (emit_library_call_value): Likewise.
8920 * calls.c (emit_library_call_value_1): Make global. Replace varargs
8921 with an "rtx_mode_t *".
8922 (emit_library_call_value): Delete.
8923 (emit_library_call): Likewise.
8924 * asan.c (asan_emit_stack_protection): Update calls accordingly.
8925 (asan_emit_allocas_unpoison): Likewise.
8926 * builtins.c (expand_builtin_powi): Likewise.
8927 (expand_asan_emit_allocas_unpoison): Likewise.
8928 * cfgexpand.c (expand_main_function): Likewise.
8929 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
8930 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
8931 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
8932 * config/arm/arm.c (arm_trampoline_init): Likewise.
8933 (arm_call_tls_get_addr): Likewise.
8934 (arm_expand_divmod_libfunc): Likewise.
8935 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
8936 (smulsi3_highpart): Likewise.
8937 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
8938 (c6x_expand_compare): Likewise.
8939 (c6x_expand_movmem): Likewise.
8940 * config/frv/frv.c (frv_trampoline_init): Likewise.
8941 * config/i386/i386.c (ix86_trampoline_init): Likewise.
8942 (ix86_expand_divmod_libfunc): Likewise.
8943 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
8944 (ia64_expand_compare): Likewise.
8945 (ia64_profile_hook): Likewise.
8946 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
8947 (nonlocal_goto): Likewise.
8948 (restore_stack_nonlocal): Likewise.
8949 * config/m32r/m32r.c (block_move_call): Likewise.
8950 (m32r_trampoline_init): Likewise.
8951 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
8952 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
8953 (m68k_call_m68k_read_tp): Likewise.
8954 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
8955 (microblaze_expand_divide): Likewise.
8956 * config/mips/mips.h (mips_args): Likewise.
8957 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
8958 (MIPS_ICACHE_SYNC): Likewise.
8959 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
8960 (nios2_trampoline_init): Likewise.
8961 * config/pa/pa.c (hppa_tls_call): Likewise.
8962 (pa_trampoline_init): Likewise.
8963 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
8964 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
8965 (expand_strn_compare): Likewise.
8966 (rs6000_generate_compare): Likewise.
8967 (rs6000_expand_float128_convert): Likewise.
8968 (output_profile_hook): Likewise.
8969 (rs6000_trampoline_init): Likewise.
8970 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
8971 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
8972 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
8973 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
8974 (rs6000_generate_compare): Likewise.
8975 (rs6000_expand_float128_convert): Likewise.
8976 (output_profile_hook): Likewise.
8977 (rs6000_trampoline_init): Likewise.
8978 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
8979 * config/sh/sh.c (sh_trampoline_init): Likewise.
8980 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
8981 (sparc_emit_float_lib_cmp): Likewise.
8982 (sparc32_initialize_trampoline): Likewise.
8983 (sparc64_initialize_trampoline): Likewise.
8984 (sparc_profile_hook): Likewise.
8985 * config/spu/spu.c (ea_load_store): Likewise.
8986 * config/spu/spu.md (floatunssidf2): Likewise.
8987 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
8988 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
8989 * config/visium/visium.c (expand_block_move_4): Likewise.
8990 (expand_block_move_2): Likewise.
8991 (expand_block_move_1): Likewise.
8992 (expand_block_set_4): Likewise.
8993 (expand_block_set_2): Likewise.
8994 (expand_block_set_1): Likewise.
8995 (visium_trampoline_init): Likewise.
8996 (visium_profile_hook): Likewise.
8997 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
8998 (xtensa_setup_frame_addresses): Likewise.
8999 (xtensa_trampoline_init): Likewise.
9000 * except.c (sjlj_emit_function_enter): Likewise.
9001 (sjlj_emit_function_exit): Likewise.
9002 * explow.c (allocate_dynamic_stack_space): Likewise.
9003 (probe_stack_range): Likewise.
9004 * expr.c (convert_mode_scalar): Likewise.
9005 * optabs.c (expand_binop): Likewise.
9006 (expand_twoval_binop_libfunc): Likewise.
9007 (expand_unop): Likewise.
9008 (prepare_cmp_insn): Likewise.
9009 (prepare_float_lib_cmp): Likewise.
9010 (expand_float): Likewise.
9011 (expand_fix): Likewise.
9012 (expand_fixed_convert): Likewise.
9013 (maybe_emit_sync_lock_test_and_set): Likewise.
9014 (expand_atomic_compare_and_swap): Likewise.
9015 (expand_mem_thread_fence): Likewise.
9016 (expand_atomic_fetch_op): Likewise.
9018 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
9020 * doc/generic.texi (OpenACC): Adjust URL.
9021 * doc/invoke.texi (C Dialect Options): Ditto.
9023 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
9025 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
9026 predicate for operand 1. Add (m,<S>) constraint.
9027 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
9028 Prevent memory operand 1 with register operand 2.
9030 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
9032 PR rtl-optimization/82024
9033 * combine.c (try_combine): If the combination result is a PARALLEL,
9034 and we only need to retain the SET in there that would be placed
9035 at I2, check that we can place that at I3 instead, before doing so.
9037 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9040 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
9042 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
9043 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
9046 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
9047 Jeff Law <law@redhat.com>
9049 * varasm.c (bss_initializer_p): Do not put constants into .bss
9050 (categorize_decl_for_section): Handle bss_initializer_p returning
9051 false when DECL_INITIAL is NULL.
9053 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9056 * config/s390/s390.c (s390_can_inline_p): New function.
9058 2017-09-01 Jeff Law <law@redhat.com>
9060 PR tree-optimization/82052
9061 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
9062 Always initialize the returned slot after a hash table miss
9063 when INSERT is true.
9065 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
9067 * config/s390/s390.md (mem_signal_fence): Remove.
9068 * doc/md.texi (mem_signal_fence): Remove.
9069 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
9071 * target-insns.def (mem_signal_fence): Remove.
9073 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9076 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
9079 * asan.c (create_odr_indicator): Strip name encoding from assembler
9080 name before appending it after __odr_asan_.
9082 2017-09-01 Martin Liska <mliska@suse.cz>
9084 PR tree-optimization/82059
9085 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
9086 frequency only when an edge is redirected.
9088 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9090 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
9091 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
9092 (arc_conditional_register_usage): Remove ARC600 lp_count
9094 (arc_file_start): Emit Tag_ARC_CPU_variation.
9095 (arc_can_use_doloop_p): New conditions to use ZOLs.
9096 (hwloop_fail): New function.
9097 (hwloop_optimize): Likewise.
9098 (hwloop_pattern_reg): Likewise.
9099 (arc_doloop_hooks): New struct, to be used with reorg_loops.
9100 (arc_reorg_loops): New function, calls reorg_loops.
9101 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
9102 (arc600_corereg_hazard): Remove ZOL checking, case handled by
9104 (arc_loop_hazard): Remove function, functionality moved into
9106 (arc_hazard): Remove arc_loop_hazard call.
9107 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
9108 into hwloop_optimize.
9109 (arc_label_align): Remove ZOL handling.
9110 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
9111 * config/arc/arc.md (doloop_begin): Remove pattern.
9112 (doloop_begin_i): Likewise.
9113 (doloop_end_i): Likewise.
9114 (doloop_fallback): Likewise.
9115 (doloop_fallback_m): Likewise.
9116 (doloop_end): Reimplement expand.
9117 (arc_lp): New pattern for LP instruction.
9118 (loop_end): New pattern.
9119 (loop_fail): Likewise.
9120 (decrement_and_branch_until_zero): Likewise.
9121 * config/arc/arc.opt (mlpc-width): New option.
9122 * doc/invoke.texi (mlpc-width): Document option.
9124 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9126 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
9127 (arc_ccfsm_advance): Fix checking for delay slots.
9128 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
9130 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9132 * config/arc/arc.md (movqi_insn): Add stores to save constant long
9134 (movhi_insn): Update store instruction constraint which are saving
9135 6-bit short immediates.
9136 (movsi_insn): Consider also short scaled load operations.
9137 (zero_extendhisi2_i): Use Usd constraint instead of T.
9138 (extendhisi2_i): Add q constraint.
9139 (arc_clzsi2): Add type and length attributes.
9140 (arc_ctzsi2): Likewise.
9141 * config/arc/constraints.md (Usc): Update constraint, the
9142 assembler can parse two relocations for a single instruction.
9144 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9146 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
9147 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
9149 2017-08-31 Olivier Hainque <hainque@adacore.com>
9151 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
9152 match as powerpc-wrs-vxworks*.
9154 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
9156 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
9157 register constraint for by-element operand.
9158 (aarch64_mls_elt_merge<mode>): Likewise.
9160 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9162 * config/arc/arc.c (arc_can_follow_jump): Check for short
9165 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9167 * config.gcc: Use g.opt for arc.
9168 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
9169 functionality moved to ...
9170 (legitimate_scaled_address_p): New function, ...here.
9171 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
9172 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
9173 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
9175 (arc_override_options): Handle G option.
9176 (arc_output_pic_addr_const): Correct function definition.
9177 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
9178 (arc_decl_anon_ns_mem_p): Delete.
9179 (arc_in_small_data_p): Overhaul this function to take into
9180 consideration the value given via G option.
9181 (arc_rewrite_small_data_1): Renamed and corrected old
9182 arc_rewrite_small_data function.
9183 (arc_rewrite_small_data): New function.
9184 (small_data_pattern): Don't use pic_offset_table_rtx.
9185 * config/arc/arc.h (CC1_SPEC): Recognize G option.
9186 * config/arc/simdext.md (movmisalignv2hi): Use
9187 prepare_move_operands function.
9189 (movmisalign*): Likewise.
9190 * doc/invoke.texi (ARC options): Document -G option.
9192 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9194 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
9196 * config/arc/arc.c (arc_print_operand): Output scalled address for
9197 sdata whenever is possible.
9198 (arc_in_small_data_p): Allow sdata for 64bit datum when double
9199 load/stores are available.
9200 (compact_sda_memory_operand): Check for the alignment required by
9201 code density instructions.
9202 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
9204 * config/arc/constraints.md (Usd): Update constraint.
9205 (Us0): New constraint.
9206 (Usc): Update constraint.
9208 2017-08-31 Richard Biener <rguenther@suse.de>
9211 * dwarf2out.c (dwarf2out_early_global_decl): Process each
9214 2017-08-31 Tamar Christina <tamar.christina@arm.com>
9216 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
9217 Resize type_signature.
9219 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9220 Alan Hayward <alan.hayward@arm.com>
9221 David Sherwood <david.sherwood@arm.com>
9223 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
9224 subregs whose inner modes can be stored in GPRs.
9225 (aarch64_classify_index): Likewise.
9227 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9228 Alan Hayward <alan.hayward@arm.com>
9229 David Sherwood <david.sherwood@arm.com>
9231 * config/aarch64/iterators.md (V_cmp_result): Rename to...
9232 (V_INT_EQUIV): ...this.
9233 (v_cmp_result): Rename to...
9234 (v_int_equiv): ...this.
9235 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
9236 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
9237 (copysign<mode>3): Likewise.
9238 (aarch64_simd_bsl<mode>_internal): Likewise.
9239 (aarch64_simd_bsl<mode>): Likewise.
9240 (vec_cmp<mode><mode>): Likewise.
9241 (vcond<mode><mode>): Likewise.
9242 (vcond<v_cmp_mixed><mode>): Likewise.
9243 (vcondu<mode><v_cmp_mixed>): Likewise.
9244 (aarch64_cm<optab><mode>): Likewise.
9245 (aarch64_cmtst<mode>): Likewise.
9246 (aarch64_fac<optab><mode>): Likewise.
9247 (vec_perm_const<mode>): Likewise.
9248 (vcond_mask_<mode><v_cmp_result>): Rename to...
9249 (vcond_mask_<mode><v_int_equiv>): ...this.
9250 (vec_cmp<mode><v_cmp_result>): Rename to...
9251 (vec_cmp<mode><v_int_equiv>): ...this.
9253 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9254 Alan Hayward <alan.hayward@arm.com>
9255 David Sherwood <david.sherwood@arm.com>
9257 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
9259 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
9260 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
9261 (UNSPEC_LD4_DREG): New unspecs.
9262 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
9263 (aarch64_ld2<mode>_dreg_be): Replace with...
9264 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
9266 (aarch64_ld3<mode>_dreg_le)
9267 (aarch64_ld3<mode>_dreg_be): Replace with...
9268 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
9270 (aarch64_ld4<mode>_dreg_le)
9271 (aarch64_ld4<mode>_dreg_be): Replace with...
9272 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
9275 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9277 PR tree-optimization/81987
9278 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
9279 insert an initializer in a location not dominated by the stride
9282 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
9284 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
9285 on the entire header of the finally block in the fallthru case.
9287 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
9289 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
9291 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
9293 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
9294 rs6000_emit_move_from_cr and call renamed function.
9295 (rs6000_emit_prologue): Call renamed functions.
9296 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
9297 movesi_from_cr, remove volatile CRs.
9299 2017-08-30 Jon Beniston <jon@beniston.com>
9300 Richard Biener <rguenther@suse.de>
9302 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
9303 of VECTOR_MODE_P check.
9304 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
9305 element vector types.
9307 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9309 * df.h (df_read_modify_subreg_p): Remove in favor of...
9310 * rtl.h (read_modify_subreg_p): ...this new function. Take a
9311 const_rtx instead of an rtx.
9312 * cprop.c (local_cprop_find_used_regs): Update accordingly.
9313 * df-problems.c (df_word_lr_mark_ref): Likewise.
9314 * ira-lives.c (mark_pseudo_reg_live): Likewise.
9315 (mark_pseudo_reg_dead): Likewise.
9316 (mark_ref_dead): Likewise.
9317 * reginfo.c (init_subregs_of_mode): Likewise.
9318 * sched-deps.c (sched_analyze_1): Likewise.
9319 * df-scan.c (df_def_record_1): Likewise.
9320 (df_uses_record): Likewise.
9321 (df_read_modify_subreg_p): Remove in favor of...
9322 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
9323 const_rtx instead of an rtx.
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 * rtl.h (partial_subreg_p): New function.
9330 * caller-save.c (save_call_clobbered_regs): Use it.
9331 * calls.c (expand_call): Likewise.
9332 * combine.c (combinable_i3pat): Likewise.
9333 (simplify_set): Likewise.
9334 (make_extraction): Likewise.
9335 (make_compound_operation_int): Likewise.
9336 (gen_lowpart_or_truncate): Likewise.
9337 (force_to_mode): Likewise.
9338 (make_field_assignment): Likewise.
9339 (reg_truncated_to_mode): Likewise.
9340 (record_truncated_value): Likewise.
9341 (move_deaths): Likewise.
9342 * cse.c (record_jump_cond): Likewise.
9343 (cse_insn): Likewise.
9344 * cselib.c (cselib_lookup_1): Likewise.
9345 * expmed.c (extract_bit_field_using_extv): Likewise.
9346 * function.c (assign_parm_setup_reg): Likewise.
9347 * ifcvt.c (noce_convert_multiple_sets): Likewise.
9348 * ira-build.c (create_insn_allocnos): Likewise.
9349 * lra-coalesce.c (merge_pseudos): Likewise.
9350 * lra-constraints.c (match_reload): Likewise.
9351 (simplify_operand_subreg): Likewise.
9352 (curr_insn_transform): Likewise.
9353 * lra-lives.c (process_bb_lives): Likewise.
9354 * lra.c (new_insn_reg): Likewise.
9355 (lra_substitute_pseudo): Likewise.
9356 * regcprop.c (mode_change_ok): Likewise.
9357 (maybe_mode_change): Likewise.
9358 (copyprop_hardreg_forward_1): Likewise.
9359 * reload.c (push_reload): Likewise.
9360 (find_reloads): Likewise.
9361 (find_reloads_subreg_address): Likewise.
9362 * reload1.c (alter_reg): Likewise.
9363 (eliminate_regs_1): Likewise.
9364 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9366 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
9368 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
9369 back to if statements, including unpack.
9371 2017-08-30 Martin Liska <mliska@suse.cz>
9374 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
9376 (func_checker::compare_asm_inputs_outputs): ... this function.
9377 (func_checker::compare_gimple_asm): Use the function to compare
9378 also ASM constrains.
9379 * ipa-icf-gimple.h: Rename the function.
9381 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9382 Alan Hayward <alan.hayward@arm.com>
9383 David Sherwood <david.sherwood@arm.com>
9385 * coretypes.h (complex_mode): New type.
9386 * gdbhooks.py (build_pretty_printer): Handle it.
9387 * machmode.h (complex_mode): New class.
9388 (complex_mode::includes_p): New function.
9389 (is_complex_int_mode): Likewise.
9390 (is_complex_float_mode): Likewise.
9391 * genmodes.c (get_mode_class): Handle complex mode classes.
9392 * function.c (expand_function_end): Use is_complex_int_mode.
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 * coretypes.h (scalar_mode_pod): New typedef.
9399 * gdbhooks.py (build_pretty_printer): Handle it.
9400 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
9401 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
9402 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
9403 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
9406 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9407 Alan Hayward <alan.hayward@arm.com>
9408 David Sherwood <david.sherwood@arm.com>
9410 * machmode.h (mode_for_vector): Take a scalar_mode instead
9412 * stor-layout.c (mode_for_vector): Likewise.
9413 * explow.c (promote_mode): Use as_a <scalar_mode>.
9414 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
9416 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9417 Alan Hayward <alan.hayward@arm.com>
9418 David Sherwood <david.sherwood@arm.com>
9420 * target.def (preferred_simd_mode): Take a scalar_mode
9421 instead of a machine_mode.
9422 * targhooks.h (default_preferred_simd_mode): Likewise.
9423 * targhooks.c (default_preferred_simd_mode): Likewise.
9424 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
9425 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
9426 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
9427 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
9428 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
9429 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
9430 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
9431 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
9433 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
9434 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
9435 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
9436 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
9437 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
9438 * doc/tm.texi: Regenerate.
9439 * optabs-query.c (can_vec_mask_load_store_p): Return false for
9442 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9443 Alan Hayward <alan.hayward@arm.com>
9444 David Sherwood <david.sherwood@arm.com>
9446 * target.def (scalar_mode_supported_p): Take a scalar_mode
9447 instead of a machine_mode.
9448 * targhooks.h (default_scalar_mode_supported_p): Likewise.
9449 * targhooks.c (default_scalar_mode_supported_p): Likewise.
9450 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
9451 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
9452 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
9453 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
9454 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
9455 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
9456 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
9457 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
9458 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
9459 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
9460 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
9461 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
9463 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
9464 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
9465 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
9466 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
9467 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
9469 * doc/tm.texi: Regenerate.
9471 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9472 Alan Hayward <alan.hayward@arm.com>
9473 David Sherwood <david.sherwood@arm.com>
9475 * coretypes.h (opt_scalar_mode): New typedef.
9476 * gdbhooks.py (build_pretty_printers): Handle it.
9477 * machmode.h (mode_iterator::get_2xwider): Add overload for
9479 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
9481 * expr.c (convert_mode_scalar): Likewise.
9482 * omp-low.c (omp_clause_aligned_alignment): Likewise.
9483 * optabs.c (expand_float): Likewise.
9484 (expand_fix): Likewise.
9485 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
9487 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9488 Alan Hayward <alan.hayward@arm.com>
9489 David Sherwood <david.sherwood@arm.com>
9491 * optabs.c (expand_float): Explicitly check for scalars before
9492 using a branching expansion.
9493 (expand_fix): Likewise.
9495 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9496 Alan Hayward <alan.hayward@arm.com>
9497 David Sherwood <david.sherwood@arm.com>
9499 * expr.c (convert_mode): Split scalar handling out into...
9500 (convert_mode_scalar): ...this new function. Treat the modes
9503 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9504 Alan Hayward <alan.hayward@arm.com>
9505 David Sherwood <david.sherwood@arm.com>
9507 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
9509 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
9511 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9512 Alan Hayward <alan.hayward@arm.com>
9513 David Sherwood <david.sherwood@arm.com>
9515 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
9516 rather than a machine_mode.
9517 (fixed_from_string): Likewise.
9518 (fixed_convert): Likewise.
9519 (fixed_convert_from_int): Likewise.
9520 (fixed_convert_from_real): Likewise.
9521 (real_convert_from_fixed): Likewise.
9522 * fixed-value.c (fixed_from_double_int): Likewise.
9523 (fixed_from_string): Likewise.
9524 (fixed_convert): Likewise.
9525 (fixed_convert_from_int): Likewise.
9526 (fixed_convert_from_real): Likewise.
9527 (real_convert_from_fixed): Likewise.
9528 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
9530 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9531 Alan Hayward <alan.hayward@arm.com>
9532 David Sherwood <david.sherwood@arm.com>
9534 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
9535 of separate mode class checks. Do not allow vector modes here.
9536 (immed_wide_int_const): Use as_a <scalar_mode>.
9537 * explow.c (trunc_int_for_mode): Likewise.
9538 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
9539 (wi::shwi): Likewise.
9540 (wi::min_value): Likewise.
9541 (wi::max_value): Likewise.
9542 * dwarf2out.c (loc_descriptor): Likewise.
9543 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
9546 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9547 Alan Hayward <alan.hayward@arm.com>
9548 David Sherwood <david.sherwood@arm.com>
9550 * tree.h (SCALAR_TYPE_MODE): New macro.
9551 * expr.c (expand_expr_addr_expr_1): Use it.
9552 (expand_expr_real_2): Likewise.
9553 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
9554 (fold_convert_const_fixed_from_int): Likewise.
9555 (fold_convert_const_fixed_from_real): Likewise.
9556 (native_encode_fixed): Likewise
9557 (native_encode_complex): Likewise
9558 (native_encode_vector): Likewise.
9559 (native_interpret_fixed): Likewise.
9560 (native_interpret_real): Likewise.
9561 (native_interpret_complex): Likewise.
9562 (native_interpret_vector): Likewise.
9563 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
9564 (simd_clone_adjust_argument_types): Likewise.
9565 (simd_clone_init_simd_arrays): Likewise.
9566 (simd_clone_adjust): Likewise.
9567 * stor-layout.c (layout_type): Likewise.
9568 * tree.c (build_minus_one_cst): Likewise.
9569 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
9570 * tree-inline.c (estimate_move_cost): Likewise.
9571 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
9572 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
9573 (vectorizable_reduction): Likewise.
9574 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
9575 (vect_recog_mixed_size_cond_pattern): Likewise.
9576 (check_bool_pattern): Likewise.
9577 (adjust_bool_pattern): Likewise.
9578 (search_type_for_mask_1): Likewise.
9579 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
9580 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
9581 (vectorizable_load): Likewise.
9582 (vectorizable_store): Likewise.
9583 * ubsan.c (ubsan_encode_value): Likewise.
9584 * varasm.c (output_constant): Likewise.
9586 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9587 Alan Hayward <alan.hayward@arm.com>
9588 David Sherwood <david.sherwood@arm.com>
9590 * coretypes.h (scalar_mode): New class.
9591 * machmode.h (scalar_mode): Likewise.
9592 (scalar_mode::includes_p): New function.
9593 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
9594 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
9595 * genmodes.c (get_mode_class): Handle remaining scalar modes.
9596 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
9597 * expmed.c (store_bit_field_1): Likewise.
9598 (extract_bit_field_1): Likewise.
9599 * expr.c (write_complex_part): Likewise.
9600 (read_complex_part): Likewise.
9601 (emit_move_complex_push): Likewise.
9602 (expand_expr_real_2): Likewise.
9603 * function.c (assign_parm_setup_reg): Likewise.
9604 (assign_parms_unsplit_complex): Likewise.
9605 * optabs.c (expand_binop): Likewise.
9606 * rtlanal.c (subreg_get_info): Likewise.
9607 * simplify-rtx.c (simplify_immed_subreg): Likewise.
9608 * varasm.c (output_constant_pool_2): Likewise.
9610 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9611 Alan Hayward <alan.hayward@arm.com>
9612 David Sherwood <david.sherwood@arm.com>
9614 * expmed.c (extract_high_half): Use scalar_int_mode and remove
9616 (expmed_mult_highpart_optab): Likewise.
9617 (expmed_mult_highpart): Likewise.
9619 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9620 Alan Hayward <alan.hayward@arm.com>
9621 David Sherwood <david.sherwood@arm.com>
9623 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
9624 instead of a machine_mode.
9625 (builtin_memset_read_str): Likewise.
9626 * builtins.c (c_readstr): Likewise.
9627 (builtin_memcpy_read_str): Likewise.
9628 (builtin_strncpy_read_str): Likewise.
9629 (builtin_memset_read_str): Likewise.
9630 (builtin_memset_gen_str): Likewise.
9631 (expand_builtin_signbit): Use scalar_int_mode for local variables.
9632 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
9633 instead of a machine_mode.
9634 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
9636 (make_extraction): Likewise.
9637 (try_widen_shift_mode): Take and return scalar_int_modes instead
9639 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
9640 a scalar_int_mode instead of a machine_mode.
9641 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
9642 (avr_addr_space_pointer_mode): Likewise.
9643 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
9644 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
9645 (msp430_unwind_word_mode): Likewise.
9646 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
9647 (spu_addr_space_pointer_mode): Likewise.
9648 (spu_addr_space_address_mode): Likewise.
9649 (spu_libgcc_cmp_return_mode): Likewise.
9650 (spu_libgcc_shift_count_mode): Likewise.
9651 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
9652 (rl78_addr_space_pointer_mode): Likewise.
9653 (fl78_unwind_word_mode): Likewise.
9654 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
9656 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
9657 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
9658 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
9659 (mips_valid_pointer_mode): Likewise.
9660 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
9661 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
9662 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
9664 (ft32_addr_space_address_mode): Likewise.
9665 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
9666 scalar_int_mode instead of a machine_mode.
9667 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
9669 (m32c_addr_space_address_mode): Likewise.
9670 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
9671 (rs6000_eh_return_filter_mode): Likewise.
9672 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
9673 (rs6000_eh_return_filter_mode): Likewise.
9674 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
9675 (s390_libgcc_shift_count_mode): Likewise.
9676 (s390_unwind_word_mode): Likewise.
9677 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
9679 * target.def (mode_rep_extended): Likewise.
9680 (valid_pointer_mode): Likewise.
9681 (addr_space.valid_pointer_mode): Likewise.
9682 (eh_return_filter_mode): Return a scalar_int_mode rather than
9684 (libgcc_cmp_return_mode): Likewise.
9685 (libgcc_shift_count_mode): Likewise.
9686 (unwind_word_mode): Likewise.
9687 (addr_space.pointer_mode): Likewise.
9688 (addr_space.address_mode): Likewise.
9689 * doc/tm.texi: Regenerate.
9690 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
9692 (do_jump): Use scalar_int_mode for local variables.
9693 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
9694 rather than a machine_mode.
9695 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
9696 (scompare_loc_descriptor_wide): Likewise.
9697 (scompare_loc_descriptor_narrow): Likewise.
9698 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
9700 * except.c (sjlj_emit_dispatch_table): Likewise.
9701 (expand_builtin_eh_copy_values): Likewise.
9702 * explow.c (convert_memory_address_addr_space_1): Likewise.
9703 Take a scalar_int_mode rather than a machine_mode.
9704 (convert_memory_address_addr_space): Take a scalar_int_mode rather
9705 than a machine_mode.
9706 (memory_address_addr_space): Use scalar_int_mode for local variables.
9707 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
9708 rather than a machine_mode.
9709 * expmed.c (mask_rtx): Likewise.
9710 (init_expmed_one_conv): Likewise.
9711 (expand_mult_highpart_adjust): Likewise.
9712 (extract_high_half): Likewise.
9713 (expmed_mult_highpart_optab): Likewise.
9714 (expmed_mult_highpart): Likewise.
9715 (expand_smod_pow2): Likewise.
9716 (expand_sdiv_pow2): Likewise.
9717 (emit_store_flag_int): Likewise.
9718 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
9720 (extract_low_bits): Likewise.
9721 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
9723 * expr.c (pieces_addr::adjust): Likewise.
9724 (can_store_by_pieces): Likewise.
9725 (store_by_pieces): Likewise.
9726 (clear_by_pieces_1): Likewise.
9727 (expand_expr_addr_expr_1): Likewise.
9728 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
9729 (expand_expr_real_1): Likewise.
9730 (try_casesi): Likewise.
9731 * final.c (shorten_branches): Likewise.
9732 * fold-const.c (fold_convert_const_int_from_fixed): Change the
9733 type of "mode" to machine_mode.
9734 * internal-fn.c (expand_arith_overflow_result_store): Take a
9735 scalar_int_mode rather than a machine_mode.
9736 (expand_mul_overflow): Use scalar_int_mode for local variables.
9737 * loop-doloop.c (doloop_modify): Likewise.
9738 (doloop_optimize): Likewise.
9739 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
9740 than a machine_mode.
9741 (expand_doubleword_shift_condmove): Likewise.
9742 (expand_doubleword_shift): Likewise.
9743 (expand_doubleword_clz): Likewise.
9744 (expand_doubleword_popcount): Likewise.
9745 (expand_doubleword_parity): Likewise.
9746 (expand_absneg_bit): Use scalar_int_mode for local variables.
9747 (prepare_float_lib_cmp): Likewise.
9748 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
9749 rather than a machine_mode.
9750 (convert_memory_address_addr_space): Likewise.
9751 (get_mode_bounds): Likewise.
9752 (get_address_mode): Return a scalar_int_mode rather than a
9754 * rtlanal.c (get_address_mode): Likewise.
9755 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
9756 than a machine_mode.
9757 * targhooks.c (default_mode_rep_extended): Likewise.
9758 (default_valid_pointer_mode): Likewise.
9759 (default_addr_space_valid_pointer_mode): Likewise.
9760 (default_eh_return_filter_mode): Return a scalar_int_mode rather
9761 than a machine_mode.
9762 (default_libgcc_cmp_return_mode): Likewise.
9763 (default_libgcc_shift_count_mode): Likewise.
9764 (default_unwind_word_mode): Likewise.
9765 (default_addr_space_pointer_mode): Likewise.
9766 (default_addr_space_address_mode): Likewise.
9767 * targhooks.h (default_eh_return_filter_mode): Likewise.
9768 (default_libgcc_cmp_return_mode): Likewise.
9769 (default_libgcc_shift_count_mode): Likewise.
9770 (default_unwind_word_mode): Likewise.
9771 (default_addr_space_pointer_mode): Likewise.
9772 (default_addr_space_address_mode): Likewise.
9773 (default_mode_rep_extended): Take a scalar_int_mode rather than
9775 (default_valid_pointer_mode): Likewise.
9776 (default_addr_space_valid_pointer_mode): Likewise.
9777 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
9779 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
9780 rather than a machine_mode.
9781 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
9782 for local variables.
9783 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
9784 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
9785 than a machine_mode.
9787 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9788 Alan Hayward <alan.hayward@arm.com>
9789 David Sherwood <david.sherwood@arm.com>
9791 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
9792 the mode argument to scalar_int_mode.
9793 (do_jump_by_parts_zero_rtx): Likewise.
9794 (do_jump_by_parts_equality_rtx): Likewise.
9795 (do_jump_by_parts_greater): Take a mode argument.
9796 (do_jump_by_parts_equality): Likewise.
9797 (do_jump_1): Update calls accordingly.
9799 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9800 Alan Hayward <alan.hayward@arm.com>
9801 David Sherwood <david.sherwood@arm.com>
9803 * is-a.h (safe_dyn_cast): New function.
9804 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
9805 (jump_table_for_label): Likewise.
9806 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
9807 instead of an rtx_insn *.
9808 (shorten_branches): Use dyn_cast instead of LABEL_P and
9809 JUMP_TABLE_DATA_P. Use jump_table_for_label and
9810 rtx_jump_table_data::get_data_mode.
9811 (final_scan_insn): Likewise.
9813 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9814 Alan Hayward <alan.hayward@arm.com>
9815 David Sherwood <david.sherwood@arm.com>
9817 * combine.c (try_combine): Use is_a <scalar_int_mode> when
9818 trying to combine a full-register integer set with a subreg
9821 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9822 Alan Hayward <alan.hayward@arm.com>
9823 David Sherwood <david.sherwood@arm.com>
9825 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
9826 that is always either address_mode or pointer_mode.
9828 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9829 Alan Hayward <alan.hayward@arm.com>
9830 David Sherwood <david.sherwood@arm.com>
9832 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
9833 when the two are known to be equal.
9835 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9836 Alan Hayward <alan.hayward@arm.com>
9837 David Sherwood <david.sherwood@arm.com>
9839 * simplify-rtx.c (simplify_const_unary_operation): Use
9840 is_a <scalar_int_mode> instead of checking for a nonzero
9841 precision. Forcibly convert op_mode to a scalar_int_mode
9842 in that case. More clearly differentiate the operand and
9843 result modes and use the former when deciding what the value
9844 of a count-bits operation should be. Use is_int_mode instead
9845 of checking for a MODE_INT. Remove redundant check for whether
9846 this mode has a zero precision.
9848 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9849 Alan Hayward <alan.hayward@arm.com>
9850 David Sherwood <david.sherwood@arm.com>
9852 * optabs.c (widen_leading): Change the type of the mode argument
9853 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
9854 (widen_bswap): Likewise.
9855 (expand_parity): Likewise.
9856 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
9857 (expand_ffs): Likewise.
9858 (epand_unop): Check for scalar integer modes before calling the
9861 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9862 Alan Hayward <alan.hayward@arm.com>
9863 David Sherwood <david.sherwood@arm.com>
9865 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
9867 (expand_expr_real_1): Update call accordingly.
9869 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9870 Alan Hayward <alan.hayward@arm.com>
9871 David Sherwood <david.sherwood@arm.com>
9873 * expmed.c (store_bit_field_using_insv): Add op0_mode and
9874 value_mode arguments. Use scalar_int_mode internally.
9875 (store_bit_field_1): Rename the new integer mode from imode
9876 to op0_mode and use it instead of GET_MODE (op0). Update calls
9877 to store_split_bit_field, store_bit_field_using_insv and
9878 store_fixed_bit_field.
9879 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
9880 Use scalar_int_mode internally. Use a bit count rather than a mode
9881 when calculating the largest bit size for get_best_mode.
9882 Update calls to store_split_bit_field and store_fixed_bit_field_1.
9883 (store_fixed_bit_field_1): Add mode and value_mode arguments.
9884 Remove assertion that OP0 has a scalar integer mode.
9885 (store_split_bit_field): Add op0_mode and value_mode arguments.
9886 Update calls to extract_fixed_bit_field.
9887 (extract_bit_field_using_extv): Add an op0_mode argument.
9888 Use scalar_int_mode internally.
9889 (extract_bit_field_1): Rename the new integer mode from imode to
9890 op0_mode and use it instead of GET_MODE (op0). Update calls to
9891 extract_split_bit_field, extract_bit_field_using_extv and
9892 extract_fixed_bit_field.
9893 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
9894 to extract_split_bit_field and extract_fixed_bit_field_1.
9895 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
9896 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
9898 (extract_split_bit_field): Add an op0_mode argument. Update call
9899 to extract_fixed_bit_field.
9901 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9902 Alan Hayward <alan.hayward@arm.com>
9903 David Sherwood <david.sherwood@arm.com>
9905 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
9906 * explow.c (hard_function_value): Likewise.
9907 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
9908 convert_to_mode call outside the loop.
9909 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
9910 for the mode iterator. Require the mode specified by max_pieces
9912 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
9914 (copy_blkmode_to_reg): Likewise.
9915 (set_storage_via_setmem): Likewise.
9916 * optabs.c (prepare_cmp_insn): Likewise.
9917 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
9918 * stor-layout.c (finish_bitfield_representative): Likewise.
9920 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9921 Alan Hayward <alan.hayward@arm.com>
9922 David Sherwood <david.sherwood@arm.com>
9924 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
9925 * expr.c (convert_move): Use them.
9926 (convert_modes): Likewise.
9927 (store_expr_with_bounds): Likewise.
9929 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9930 Alan Hayward <alan.hayward@arm.com>
9931 David Sherwood <david.sherwood@arm.com>
9933 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
9934 parameter for the mode of "x". Remove the "known_x", "known_mode"
9935 and "known_ret" arguments. Change the type of the mode argument
9937 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
9938 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
9939 (reg_num_sign_bit_copies_for_combine): Likewise.
9940 * rtlanal.c (nonzero_bits1): Likewise.
9941 (num_sign_bit_copies1): Likewise.
9942 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
9943 (reg_num_sign_bit_copies_general): Likewise.
9944 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
9945 (reg_nonzero_bits_general): Likewise.
9947 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9948 Alan Hayward <alan.hayward@arm.com>
9949 David Sherwood <david.sherwood@arm.com>
9951 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
9952 than in subroutines. Return 1 for non-integer modes.
9953 (cached_num_sign_bit_copies): Change the type of the mode parameter
9955 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
9956 classes. Handle CONST_INT_P first and then check whether X also
9957 has a scalar integer mode. Check the same thing for inner registers
9958 of a SUBREG and for values that are being extended or truncated.
9960 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9961 Alan Hayward <alan.hayward@arm.com>
9962 David Sherwood <david.sherwood@arm.com>
9964 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
9965 in subroutines. Return the mode mask for non-integer modes.
9966 (cached_nonzero_bits): Change the type of the mode parameter
9968 (nonzero_bits1): Likewise. Remove early exit for other mode
9969 classes. Handle CONST_INT_P first and then check whether X
9970 also has a scalar integer mode.
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 * expr.c (widest_int_mode_for_size): Make the comment match the code.
9977 Return a scalar_int_mode and assert that the size is greater than
9979 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
9980 (op_by_pieces_d::op_by_pieces_d): Likewise.
9981 (op_by_pieces_d::run): Likewise.
9982 (can_store_by_pieces): Likewise.
9984 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9985 Alan Hayward <alan.hayward@arm.com>
9986 David Sherwood <david.sherwood@arm.com>
9988 * combine.c (extract_left_shift): Add a mode argument and update
9990 (make_compound_operation_int): Change the type of the mode parameter
9991 to scalar_int_mode and update the call to extract_left_shift.
9993 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9994 Alan Hayward <alan.hayward@arm.com>
9995 David Sherwood <david.sherwood@arm.com>
9997 * combine.c (simplify_and_const_int): Change the type of the mode
9998 parameter to scalar_int_mode.
9999 (simplify_and_const_int_1): Likewise. Update recursive call.
10001 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10002 Alan Hayward <alan.hayward@arm.com>
10003 David Sherwood <david.sherwood@arm.com>
10005 * combine.c (simplify_compare_const): Check that the mode is a
10006 scalar_int_mode (rather than VOIDmode) before testing its
10008 (simplify_comparison): Move COMPARISON_P handling out of the
10009 loop and restrict the latter part of the loop to scalar_int_modes.
10010 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
10011 and when considering SUBREG_REGs. Use is_int_mode instead of
10012 checking GET_MODE_CLASS against MODE_INT.
10014 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10015 Alan Hayward <alan.hayward@arm.com>
10016 David Sherwood <david.sherwood@arm.com>
10018 * combine.c (try_widen_shift_mode): Move check for equal modes to...
10019 (simplify_shift_const_1): ...here. Use scalar_int_mode for
10020 shift_unit_mode and for modes involved in scalar shifts.
10022 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10023 Alan Hayward <alan.hayward@arm.com>
10024 David Sherwood <david.sherwood@arm.com>
10026 * combine.c (force_int_to_mode): New function, split out from...
10027 (force_to_mode): ...here. Keep xmode up-to-date and use it
10028 instead of GET_MODE (x).
10030 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10031 Alan Hayward <alan.hayward@arm.com>
10032 David Sherwood <david.sherwood@arm.com>
10034 * optabs-query.h (extraction_insn::struct_mode): Change type to
10035 opt_scalar_int_mode and update comment.
10036 (extraction_insn::field_mode): Change type to scalar_int_mode.
10037 (extraction_insn::pos_mode): Likewise.
10038 * combine.c (make_extraction): Update accordingly.
10039 * optabs-query.c (get_traditional_extraction_insn): Likewise.
10040 (get_optab_extraction_insn): Likewise.
10041 * recog.c (simplify_while_replacing): Likewise.
10042 * expmed.c (narrow_bit_field_mem): Change the type of the mode
10043 parameter to opt_scalar_int_mode.
10045 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10046 Alan Hayward <alan.hayward@arm.com>
10047 David Sherwood <david.sherwood@arm.com>
10049 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
10050 to a scalar_int_mode instead of a machine_mode.
10051 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
10052 (get_best_mode): Return a boolean and use a pointer argument to store
10053 the selected mode. Replace the limit mode parameter with a bit limit.
10054 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
10055 for the values returned by bit_field_mode_iterator::next_mode.
10056 (store_bit_field): Update call to get_best_mode.
10057 (store_fixed_bit_field): Likewise.
10058 (extract_fixed_bit_field): Likewise.
10059 * expr.c (optimize_bitfield_assignment_op): Likewise.
10060 * fold-const.c (optimize_bit_field_compare): Likewise.
10061 (fold_truth_andor_1): Likewise.
10062 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
10063 Update for new type of m_mode.
10064 (get_best_mode): As above.
10066 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10067 Alan Hayward <alan.hayward@arm.com>
10068 David Sherwood <david.sherwood@arm.com>
10070 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
10071 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
10072 (store_bit_field): Check is_a <scalar_int_mode> before calling
10073 strict_volatile_bitfield_p.
10074 (extract_bit_field): Likewise.
10076 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10077 Alan Hayward <alan.hayward@arm.com>
10078 David Sherwood <david.sherwood@arm.com>
10080 * target.def (cstore_mode): Return a scalar_int_mode.
10081 * doc/tm.texi: Regenerate.
10082 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
10083 * targhooks.h (default_cstore_mode): Likewise.
10084 * targhooks.c (default_cstore_mode): Likewise, using a forced
10086 * expmed.c (emit_cstore): Expect the target of the cstore to be
10089 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10090 Alan Hayward <alan.hayward@arm.com>
10091 David Sherwood <david.sherwood@arm.com>
10093 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
10095 (niter_desc): Likewise mode.
10096 (iv_analyze): Add a mode parameter.
10098 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
10099 and change its type to scalar_int_mode.
10100 * loop-iv.c: Update commentary at head of file.
10101 (iv_constant): Pass the mode paraeter before the rtx it describes
10102 and change its type to scalar_int_mode. Remove VOIDmode handling.
10103 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
10104 (iv_extend): Likewise.
10105 (shorten_into_mode): Likewise.
10106 (iv_add): Use scalar_int_mode.
10107 (iv_mult): Likewise.
10108 (iv_shift): Likewise.
10109 (canonicalize_iv_subregs): Likewise.
10110 (get_biv_step_1): Pass the outer_mode parameter before the rtx
10111 it describes and change its mode to scalar_int_mode. Also change
10112 the type of the returned inner_mode to scalar_int_mode.
10113 (get_biv_step): Likewise, turning outer_mode from a pointer
10114 into a direct parameter. Update call to get_biv_step_1.
10115 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
10116 iv_constant and get_biv_step.
10117 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
10118 and change its type to scalar_int_mode. Don't initialise iv->mode
10119 to VOIDmode and remove later checks for its still being VOIDmode.
10120 Update calls to iv_analyze_op and iv_analyze_expr. Check
10121 is_a <scalar_int_mode> when changing the mode under consideration.
10122 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
10123 Update call to iv_analyze_expr.
10124 (iv_analyze_op): Add a mode parameter. Reject subregs whose
10125 inner register is not also a scalar_int_mode. Update call to
10127 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
10128 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
10129 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
10130 separate mode class checks. Update calls to iv_analyze. Remove
10131 fix-up of VOIDmodes after iv_analyze_biv.
10132 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
10133 don't have a scalar_int_mode. Update call to biv_p.
10135 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10136 Alan Hayward <alan.hayward@arm.com>
10137 David Sherwood <david.sherwood@arm.com>
10139 * cfgexpand.c (convert_debug_memory_address): Use
10140 as_a <scalar_int_mode>.
10141 * combine.c (expand_compound_operation): Likewise.
10142 (make_extraction): Likewise.
10143 (change_zero_ext): Likewise.
10144 (simplify_comparison): Likewise.
10145 * cse.c (cse_insn): Likewise.
10146 * dwarf2out.c (minmax_loc_descriptor): Likewise.
10147 (mem_loc_descriptor): Likewise.
10148 (loc_descriptor): Likewise.
10149 * expmed.c (init_expmed_one_mode): Likewise.
10150 (synth_mult): Likewise.
10151 (emit_store_flag_1): Likewise.
10152 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
10153 of a comparison with size.
10154 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
10155 (reduce_to_bit_field_precision): Likewise.
10156 * function.c (expand_function_end): Likewise.
10157 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
10158 * loop-doloop.c (doloop_modify): Likewise.
10159 * optabs.c (expand_binop): Likewise.
10160 (expand_unop): Likewise.
10161 (expand_copysign_absneg): Likewise.
10162 (prepare_cmp_insn): Likewise.
10163 (maybe_legitimize_operand): Likewise.
10164 * recog.c (const_scalar_int_operand): Likewise.
10165 * rtlanal.c (get_address_mode): Likewise.
10166 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10167 (simplify_cond_clz_ctz): Likewise.
10168 * tree-nested.c (get_nl_goto_field): Likewise.
10169 * tree.c (build_vector_type_for_mode): Likewise.
10170 * var-tracking.c (use_narrower_mode): Likewise.
10172 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10173 Alan Hayward <alan.hayward@arm.com>
10174 David Sherwood <david.sherwood@arm.com>
10176 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
10177 * builtins.c (expand_builtin_signbit): Use it.
10178 * cfgexpand.c (expand_debug_expr): Likewise.
10179 * dojump.c (do_jump): Likewise.
10180 (do_compare_and_jump): Likewise.
10181 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
10182 * expmed.c (make_tree): Likewise.
10183 * expr.c (expand_expr_real_2): Likewise.
10184 (expand_expr_real_1): Likewise.
10185 (try_casesi): Likewise.
10186 * fold-const-call.c (fold_const_call_ss): Likewise.
10187 * fold-const.c (unextend): Likewise.
10188 (extract_muldiv_1): Likewise.
10189 (fold_single_bit_test): Likewise.
10190 (native_encode_int): Likewise.
10191 (native_encode_string): Likewise.
10192 (native_interpret_int): Likewise.
10193 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
10194 * internal-fn.c (expand_addsub_overflow): Likewise.
10195 (expand_neg_overflow): Likewise.
10196 (expand_mul_overflow): Likewise.
10197 (expand_arith_overflow): Likewise.
10198 * match.pd: Likewise.
10199 * stor-layout.c (layout_type): Likewise.
10200 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10201 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
10202 * tree-ssanames.c (get_range_info): Likewise.
10203 * tree-switch-conversion.c (array_value_type) Likewise.
10204 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
10205 (vect_recog_divmod_pattern): Likewise.
10206 (vect_recog_mixed_size_cond_pattern): Likewise.
10207 * tree-vrp.c (extract_range_basic): Likewise.
10208 (simplify_float_conversion_using_ranges): Likewise.
10209 * tree.c (int_fits_type_p): Likewise.
10210 * ubsan.c (instrument_bool_enum_load): Likewise.
10211 * varasm.c (mergeable_string_section): Likewise.
10212 (narrowing_initializer_constant_valid_p): Likewise.
10213 (output_constant): Likewise.
10215 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10216 Alan Hayward <alan.hayward@arm.com>
10217 David Sherwood <david.sherwood@arm.com>
10219 * machmode.h (NARROWEST_INT_MODE): New macro.
10220 * expr.c (alignment_for_piecewise_move): Use it instead of
10221 GET_CLASS_NARROWEST_MODE (MODE_INT).
10222 (push_block): Likewise.
10223 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
10225 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10227 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10228 Alan Hayward <alan.hayward@arm.com>
10229 David Sherwood <david.sherwood@arm.com>
10231 * postreload.c (move2add_valid_value_p): Change the type of the
10232 mode parameter to scalar_int_mode.
10233 (move2add_use_add2_insn): Add a mode parameter and use it instead
10235 (move2add_use_add3_insn): Likewise.
10236 (reload_cse_move2add): Update accordingly.
10238 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10239 Alan Hayward <alan.hayward@arm.com>
10240 David Sherwood <david.sherwood@arm.com>
10242 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
10244 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
10245 * optabs.c (expand_unop): Likewise.
10247 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10248 Alan Hayward <alan.hayward@arm.com>
10249 David Sherwood <david.sherwood@arm.com>
10251 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
10252 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
10253 (popcount_loc_descriptor): Likewise.
10254 (bswap_loc_descriptor): Likewise.
10255 (rotate_loc_descriptor): Likewise.
10256 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
10257 calling the functions above.
10259 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10260 Alan Hayward <alan.hayward@arm.com>
10261 David Sherwood <david.sherwood@arm.com>
10263 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
10265 (try_combine): Likewise.
10266 (simplify_if_then_else): Likewise.
10267 * cse.c (cse_insn): Likewise.
10268 * dwarf2out.c (mem_loc_descriptor): Likewise.
10269 * emit-rtl.c (gen_lowpart_common): Likewise.
10270 * simplify-rtx.c (simplify_truncation): Likewise.
10271 (simplify_binary_operation_1): Likewise.
10272 (simplify_const_relational_operation): Likewise.
10273 (simplify_ternary_operation): Likewise.
10274 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
10276 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10277 Alan Hayward <alan.hayward@arm.com>
10278 David Sherwood <david.sherwood@arm.com>
10280 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
10281 * reload.c (push_reload): Likewise.
10282 (find_reloads): Likewise.
10284 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10285 Alan Hayward <alan.hayward@arm.com>
10286 David Sherwood <david.sherwood@arm.com>
10288 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
10289 (make_compound_operation_int): Likewise.
10290 (change_zero_ext): Likewise.
10291 * expr.c (convert_move): Likewise.
10292 (convert_modes): Likewise.
10293 * fwprop.c (forward_propagate_subreg): Likewise.
10294 * loop-iv.c (get_biv_step_1): Likewise.
10295 * optabs.c (widen_operand): Likewise.
10296 * postreload.c (move2add_valid_value_p): Likewise.
10297 * recog.c (simplify_while_replacing): Likewise.
10298 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10299 (simplify_binary_operation_1): Likewise. Remove redundant
10300 mode equality check.
10302 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10303 Alan Hayward <alan.hayward@arm.com>
10304 David Sherwood <david.sherwood@arm.com>
10306 * combine.c (combine_simplify_rtx): Add checks for
10307 is_a <scalar_int_mode>.
10308 (simplify_if_then_else): Likewise.
10309 (make_field_assignment): Likewise.
10310 (simplify_comparison): Likewise.
10311 * ifcvt.c (noce_try_bitop): Likewise.
10312 * loop-invariant.c (canonicalize_address_mult): Likewise.
10313 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10315 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10316 Alan Hayward <alan.hayward@arm.com>
10317 David Sherwood <david.sherwood@arm.com>
10319 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
10320 is_a <scalar_int_mode> instead of != BLKmode.
10322 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10323 Alan Hayward <alan.hayward@arm.com>
10324 David Sherwood <david.sherwood@arm.com>
10326 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
10327 instead of != VOIDmode.
10328 * combine.c (if_then_else_cond): Likewise.
10329 (change_zero_ext): Likewise.
10330 * dwarf2out.c (mem_loc_descriptor): Likewise.
10331 (loc_descriptor): Likewise.
10332 * rtlanal.c (canonicalize_condition): Likewise.
10333 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
10335 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10336 Alan Hayward <alan.hayward@arm.com>
10337 David Sherwood <david.sherwood@arm.com>
10339 * simplify-rtx.c (simplify_binary_operation_1): Use
10340 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
10342 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10343 Alan Hayward <alan.hayward@arm.com>
10344 David Sherwood <david.sherwood@arm.com>
10346 * wide-int.h (int_traits<unsigned char>) New class.
10347 (int_traits<unsigned short>) Likewise.
10348 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
10349 Use GET_MODE_UNIT_PRECISION and remove redundant test for
10351 * combine.c (set_nonzero_bits_and_sign_copies): Use
10352 is_a <scalar_int_mode>.
10353 (find_split_point): Likewise.
10354 (combine_simplify_rtx): Likewise.
10355 (simplify_logical): Likewise.
10356 (expand_compound_operation): Likewise.
10357 (expand_field_assignment): Likewise.
10358 (make_compound_operation): Likewise.
10359 (extended_count): Likewise.
10360 (change_zero_ext): Likewise.
10361 (simplify_comparison): Likewise.
10362 * dwarf2out.c (scompare_loc_descriptor): Likewise.
10363 (ucompare_loc_descriptor): Likewise.
10364 (minmax_loc_descriptor): Likewise.
10365 (mem_loc_descriptor): Likewise.
10366 (loc_descriptor): Likewise.
10367 * expmed.c (init_expmed_one_mode): Likewise.
10368 * lra-constraints.c (lra_constraint_offset): Likewise.
10369 * optabs.c (prepare_libcall_arg): Likewise.
10370 * postreload.c (move2add_note_store): Likewise.
10371 * reload.c (operands_match_p): Likewise.
10372 * rtl.h (load_extend_op): Likewise.
10373 * rtlhooks.c (gen_lowpart_general): Likewise.
10374 * simplify-rtx.c (simplify_truncation): Likewise.
10375 (simplify_unary_operation_1): Likewise.
10376 (simplify_binary_operation_1): Likewise.
10377 (simplify_const_binary_operation): Likewise.
10378 (simplify_const_relational_operation): Likewise.
10379 (simplify_subreg): Likewise.
10380 * stor-layout.c (bitwise_mode_for_mode): Likewise.
10381 * var-tracking.c (adjust_mems): Likewise.
10382 (prepare_call_arguments): Likewise.
10384 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10385 Alan Hayward <alan.hayward@arm.com>
10386 David Sherwood <david.sherwood@arm.com>
10388 * machmode.h (is_int_mode): New fuction.
10389 * combine.c (find_split_point): Use it.
10390 (combine_simplify_rtx): Likewise.
10391 (simplify_if_then_else): Likewise.
10392 (simplify_set): Likewise.
10393 (simplify_shift_const_1): Likewise.
10394 (simplify_comparison): Likewise.
10395 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
10396 * cse.c (notreg_cost): Likewise.
10397 (cse_insn): Likewise.
10398 * cselib.c (cselib_lookup_1): Likewise.
10399 * dojump.c (do_jump_1): Likewise.
10400 (do_compare_rtx_and_jump): Likewise.
10401 * dse.c (get_call_args): Likewise.
10402 * dwarf2out.c (rtl_for_decl_init): Likewise.
10403 (native_encode_initializer): Likewise.
10404 * expmed.c (emit_store_flag_1): Likewise.
10405 (emit_store_flag): Likewise.
10406 * expr.c (convert_modes): Likewise.
10407 (store_field): Likewise.
10408 (expand_expr_real_1): Likewise.
10409 * fold-const.c (fold_read_from_constant_string): Likewise.
10410 * gimple-ssa-sprintf.c (get_format_string): Likewise.
10411 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
10412 * optabs.c (expand_binop): Likewise.
10413 (expand_unop): Likewise.
10414 (expand_abs_nojump): Likewise.
10415 (expand_one_cmpl_abs_nojump): Likewise.
10416 * simplify-rtx.c (mode_signbit_p): Likewise.
10417 (val_signbit_p): Likewise.
10418 (val_signbit_known_set_p): Likewise.
10419 (val_signbit_known_clear_p): Likewise.
10420 (simplify_relational_operation_1): Likewise.
10421 * tree.c (vector_type_mode): Likewise.
10423 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10424 Alan Hayward <alan.hayward@arm.com>
10425 David Sherwood <david.sherwood@arm.com>
10427 * machmode.h (smallest_mode_for_size): Fix formatting.
10428 (smallest_int_mode_for_size): New function.
10429 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
10430 instead of smallest_mode_for_size.
10431 * combine.c (make_extraction): Likewise.
10432 * config/arc/arc.c (arc_expand_movmem): Likewise.
10433 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
10434 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
10435 * config/s390/s390.c (s390_expand_insv): Likewise.
10436 * config/sparc/sparc.c (assign_int_registers): Likewise.
10437 * config/spu/spu.c (spu_function_value): Likewise.
10438 (spu_function_arg): Likewise.
10439 * coverage.c (get_gcov_type): Likewise.
10440 (get_gcov_unsigned_t): Likewise.
10441 * dse.c (find_shift_sequence): Likewise.
10442 * expmed.c (store_bit_field_1): Likewise.
10443 * expr.c (convert_move): Likewise.
10444 (store_field): Likewise.
10445 * internal-fn.c (expand_arith_overflow): Likewise.
10446 * optabs-query.c (get_best_extraction_insn): Likewise.
10447 * optabs.c (expand_twoval_binop_libfunc): Likewise.
10448 * stor-layout.c (layout_type): Likewise.
10449 (initialize_sizetypes): Likewise.
10450 * targhooks.c (default_get_mask_mode): Likewise.
10451 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
10453 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10454 Alan Hayward <alan.hayward@arm.com>
10455 David Sherwood <david.sherwood@arm.com>
10457 * machmode.h (opt_mode::else_blk): New function.
10458 (int_mode_for_mode): Declare.
10459 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
10460 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
10462 * cfgexpand.c (expand_debug_expr): Likewise.
10463 * combine.c (gen_lowpart_or_truncate): Likewise.
10464 (gen_lowpart_for_combine): Likewise.
10465 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
10466 * config/avr/avr.c (avr_to_int_mode): Likewise.
10467 (avr_out_plus_1): Likewise.
10468 (avr_out_plus): Likewise.
10469 (avr_out_round): Likewise.
10470 * config/i386/i386.c (ix86_split_to_parts): Likewise.
10471 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
10472 (s390_expand_vcond): Likewise.
10473 * config/spu/spu.c (spu_split_immediate): Likewise.
10474 (spu_expand_mov): Likewise.
10475 * dse.c (get_stored_val): Likewise.
10476 * expmed.c (store_bit_field_1): Likewise.
10477 (convert_extracted_bit_field): Use int_mode_for_mode instead of
10479 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
10480 (extract_low_bits): Likewise.
10481 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
10482 handling rather than repeating the check.
10483 (emit_group_store): Likewise.
10484 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
10485 * optabs.c (expand_absneg_bit): Likewise.
10486 (expand_copysign_absneg): Likewise.
10487 (expand_copysign_bit): Likewise.
10488 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
10489 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
10490 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
10491 * var-tracking.c (prepare_call_arguments): Likewise.
10492 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
10493 int_mode_for_mode instead of mode_for_size.
10494 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
10496 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10497 Alan Hayward <alan.hayward@arm.com>
10498 David Sherwood <david.sherwood@arm.com>
10500 * machmode.h (int_mode_for_size): New function.
10501 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
10502 instead of mode_for_size.
10503 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
10505 * combine.c (expand_field_assignment): Use int_mode_for_size
10506 instead of mode_for_size.
10507 (make_extraction): Likewise.
10508 (simplify_shift_const_1): Likewise.
10509 (simplify_comparison): Likewise.
10510 * dojump.c (do_jump): Likewise.
10511 * dwarf2out.c (mem_loc_descriptor): Likewise.
10512 * emit-rtl.c (init_derived_machine_modes): Likewise.
10513 * expmed.c (flip_storage_order): Likewise.
10514 (convert_extracted_bit_field): Likewise.
10515 * expr.c (copy_blkmode_from_reg): Likewise.
10516 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
10517 * internal-fn.c (expand_mul_overflow): Likewise.
10518 * lower-subreg.c (simple_move): Likewise.
10519 * optabs-libfuncs.c (init_optabs): Likewise.
10520 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10521 * tree.c (vector_type_mode): Likewise.
10522 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
10523 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
10524 * tree-vect-generic.c (expand_vector_parallel): Likewise.
10525 * tree-vect-stmts.c (vectorizable_load): Likewise.
10526 (vectorizable_store): Likewise.
10528 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10529 Alan Hayward <alan.hayward@arm.com>
10530 David Sherwood <david.sherwood@arm.com>
10532 * coretypes.h (pod_mode): New type.
10533 (scalar_int_mode_pod): New typedef.
10534 * machmode.h (pod_mode): New class.
10535 (int_n_data_t::m): Change type to scalar_int_mode_pod.
10536 * genmodes.c (emit_mode_int_n): Update accordingly.
10537 * lower-subreg.h (target_lower_subreg): Change type to
10538 scalar_int_mode_pod.
10539 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
10540 scalar_int_mode_pod.
10542 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10543 Alan Hayward <alan.hayward@arm.com>
10544 David Sherwood <david.sherwood@arm.com>
10546 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
10547 machine_mode to scalar_int_mode.
10548 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
10549 (rs6000_option_override_internal): Remove cast to int.
10550 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
10551 machine_mode to scalar_int_mode.
10552 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
10553 (rs6000_option_override_internal): Remove cast to int.
10554 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
10555 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
10557 * config/s390/s390.c (s390_expand_builtin): Likewise.
10558 * coretypes.h (scalar_int_mode): New type.
10559 (opt_scalar_int_mode): New typedef.
10560 * machmode.h (scalar_int_mode): New class.
10561 (scalar_int_mode::includes_p): New function.
10562 (byte_mode): Change type to scalar_int_mode.
10563 (word_mode): Likewise.
10564 (ptr_mode): Likewise.
10565 * emit-rtl.c (byte_mode): Likewise.
10566 (word_mode): Likewise.
10567 (ptr_mode): Likewise.
10568 (init_derived_machine_modes): Update accordingly.
10569 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
10570 and MODE_PARTIAL_INT.
10571 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
10572 opt_scalar_int_mode.
10574 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10575 Alan Hayward <alan.hayward@arm.com>
10576 David Sherwood <david.sherwood@arm.com>
10578 * target.def (libgcc_floating_mode_supported_p): Take a
10580 * doc/tm.texi: Regenerate.
10581 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
10583 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
10584 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
10587 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10588 Alan Hayward <alan.hayward@arm.com>
10589 David Sherwood <david.sherwood@arm.com>
10591 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
10592 * doc/tm.texi: Regenerate.
10593 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
10594 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
10595 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
10596 * targhooks.h (default_floatn_mode): Likewise.
10597 * targhooks.c (default_floatn_mode): Likewise.
10598 * tree.c (build_common_tree_nodes): Update accordingly.
10600 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10601 Alan Hayward <alan.hayward@arm.com>
10602 David Sherwood <david.sherwood@arm.com>
10604 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
10605 (mode_iterator::iterate_p): Likewise.
10606 (mode_iterator::get_wider): Likewise.
10607 * expr.c (init_expr_target): Use opt_scalar_float_mode.
10609 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10610 Alan Hayward <alan.hayward@arm.com>
10611 David Sherwood <david.sherwood@arm.com>
10613 * coretypes.h (opt_scalar_float_mode): New typedef.
10614 * machmode.h (float_mode_for_size): New function.
10615 * emit-rtl.c (double_mode): Delete.
10616 (init_emit_once): Use float_mode_for_size.
10617 * stor-layout.c (layout_type): Likewise.
10618 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
10620 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10621 Alan Hayward <alan.hayward@arm.com>
10622 David Sherwood <david.sherwood@arm.com>
10624 * output.h (assemble_real): Take a scalar_float_mode.
10625 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
10626 * config/arm/arm.md (consttable_4): Likewise.
10627 (consttable_8): Likewise.
10628 (consttable_16): Likewise.
10629 * config/mips/mips.md (consttable_float): Likewise.
10630 * config/s390/s390.c (s390_output_pool_entry): Likewise.
10631 * varasm.c (assemble_real): Take a scalar_float_mode.
10632 (output_constant_pool_2): Update accordingly.
10633 (output_constant): Likewise.
10635 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10636 Alan Hayward <alan.hayward@arm.com>
10637 David Sherwood <david.sherwood@arm.com>
10639 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
10640 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
10641 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
10642 (native_encode_real): Likewise.
10643 (native_interpret_real): Likewise.
10644 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
10645 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10647 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10648 Alan Hayward <alan.hayward@arm.com>
10649 David Sherwood <david.sherwood@arm.com>
10651 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
10652 <scalar_float_mode>. Simplify.
10653 (gen_extend_conv_libfunc): Likewise.
10655 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10656 Alan Hayward <alan.hayward@arm.com>
10657 David Sherwood <david.sherwood@arm.com>
10659 * coretypes.h (scalar_float_mode): New type.
10660 * machmode.h (mode_traits::from_int): Use machine_mode if
10661 USE_ENUM_MODES is defined.
10662 (is_a): New function.
10664 (dyn_cast): Likewise.
10665 (scalar_float_mode): New class.
10666 (scalar_float_mode::includes_p): New function.
10667 (is_float_mode): Likewise.
10668 * gdbhooks.py (MachineModePrinter): New class.
10669 (build_pretty_printer): Use it for scalar_float_mode.
10670 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
10671 (format_helper::format_helper): Turn into a template.
10672 * genmodes.c (get_mode_class): New function.
10673 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
10674 or machine_mode if none.
10675 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
10676 as_a <scalar_float_mode>.
10677 * dwarf2out.c (mem_loc_descriptor): Likewise.
10678 (insert_float): Likewise.
10679 (add_const_value_attribute): Likewise.
10680 * simplify-rtx.c (simplify_immed_subreg): Likewise.
10681 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
10682 (expand_unop): Update accordingly.
10683 (expand_abs_nojump): Likewise.
10684 (expand_copysign_absneg): Take a scalar_float_mode.
10685 (expand_copysign_bit): Likewise.
10686 (expand_copysign): Update accordingly.
10688 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10689 Alan Hayward <alan.hayward@arm.com>
10690 David Sherwood <david.sherwood@arm.com>
10692 * coretypes.h (opt_mode): New class.
10693 * machmode.h (opt_mode): Likewise.
10694 (opt_mode::else_void): New function.
10695 (opt_mode::require): Likewise.
10696 (opt_mode::exists): Likewise.
10697 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
10698 (GET_MODE_2XWIDER_MODE): Likewise.
10699 (mode_iterator::get_wider): Update accordingly.
10700 (mode_iterator::get_2xwider): Likewise.
10701 (mode_iterator::get_known_wider): Likewise, turning into a template.
10702 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
10703 forcing a wider mode to exist.
10704 * config/cr16/cr16.h (LONG_REG_P): Likewise.
10705 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
10706 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
10707 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
10708 * lower-subreg.c (init_lower_subreg): Likewise.
10709 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
10710 on the final iteration.
10711 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
10712 a wider mode exists before asking for a move pattern.
10713 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
10714 forcing a wider mode to exist.
10715 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
10716 returning false if no such mode exists.
10717 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
10718 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
10719 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
10720 Avoid checking for a MODE_INT if we already know the mode is not a
10722 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
10723 forcing a wider mode to exist.
10724 (expmed_mult_highpart_optab): Likewise.
10725 (expmed_mult_highpart): Likewise.
10726 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
10728 * lto-streamer-in.c (lto_input_mode_table): Likewise.
10729 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
10730 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
10731 * internal-fn.c (expand_mul_overflow): Update use of
10732 GET_MODE_2XWIDER_MODE.
10733 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10734 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
10735 GET_MODE_WIDER_MODE.
10736 (convert_plusminus_to_widen): Likewise.
10737 * tree-switch-conversion.c (array_value_type): Likewise.
10738 * var-tracking.c (emit_note_insn_var_location): Likewise.
10739 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10740 Return false inside rather than outside the loop if no wider mode
10742 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
10743 and GET_MODE_2XWIDER_MODE
10744 (can_compare_p): Use else_void.
10745 * gdbhooks.py (OptMachineModePrinter): New class.
10746 (build_pretty_printer): Use it for opt_mode.
10748 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10749 Alan Hayward <alan.hayward@arm.com>
10750 David Sherwood <david.sherwood@arm.com>
10752 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
10753 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
10755 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10756 Alan Hayward <alan.hayward@arm.com>
10757 David Sherwood <david.sherwood@arm.com>
10759 * machmode.h (mode_traits): New structure.
10760 (get_narrowest_mode): New function.
10761 (mode_iterator::start): Likewise.
10762 (mode_iterator::iterate_p): Likewise.
10763 (mode_iterator::get_wider): Likewise.
10764 (mode_iterator::get_known_wider): Likewise.
10765 (mode_iterator::get_2xwider): Likewise.
10766 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
10767 (FOR_EACH_MODE): Likewise.
10768 (FOR_EACH_MODE_FROM): Likewise.
10769 (FOR_EACH_MODE_UNTIL): Likewise.
10770 (FOR_EACH_WIDER_MODE): Likewise.
10771 (FOR_EACH_2XWIDER_MODE): Likewise.
10772 * builtins.c (expand_builtin_strlen): Use new mode iterators.
10773 * combine.c (simplify_comparison): Likewise
10774 * config/i386/i386.c (type_natural_mode): Likewise.
10775 * cse.c (cse_insn): Likewise.
10776 * dse.c (find_shift_sequence): Likewise.
10777 * emit-rtl.c (init_derived_machine_modes): Likewise.
10778 (init_emit_once): Likewise.
10779 * explow.c (hard_function_value): Likewise.
10780 * expmed.c (extract_fixed_bit_field_1): Likewise.
10781 (extract_bit_field_1): Likewise.
10782 (expand_divmod): Likewise.
10783 (emit_store_flag_1): Likewise.
10784 * expr.c (init_expr_target): Likewise.
10785 (convert_move): Likewise.
10786 (alignment_for_piecewise_move): Likewise.
10787 (widest_int_mode_for_size): Likewise.
10788 (emit_block_move_via_movmem): Likewise.
10789 (copy_blkmode_to_reg): Likewise.
10790 (set_storage_via_setmem): Likewise.
10791 (compress_float_constant): Likewise.
10792 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10793 * optabs-query.c (get_best_extraction_insn): Likewise.
10794 * optabs.c (expand_binop): Likewise.
10795 (expand_twoval_unop): Likewise.
10796 (expand_twoval_binop): Likewise.
10797 (widen_leading): Likewise.
10798 (widen_bswap): Likewise.
10799 (expand_parity): Likewise.
10800 (expand_unop): Likewise.
10801 (prepare_cmp_insn): Likewise.
10802 (prepare_float_lib_cmp): Likewise.
10803 (expand_float): Likewise.
10804 (expand_fix): Likewise.
10805 (expand_sfix_optab): Likewise.
10806 * postreload.c (move2add_use_add2_insn): Likewise.
10807 * reg-stack.c (reg_to_stack): Likewise.
10808 * reginfo.c (choose_hard_reg_mode): Likewise.
10809 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
10810 * stor-layout.c (mode_for_size): Likewise.
10811 (smallest_mode_for_size): Likewise.
10812 (mode_for_vector): Likewise.
10813 (finish_bitfield_representative): Likewise.
10814 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
10815 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
10816 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10817 * var-tracking.c (prepare_call_arguments): Likewise.
10819 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10820 Alan Hayward <alan.hayward@arm.com>
10821 David Sherwood <david.sherwood@arm.com>
10823 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
10824 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
10825 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
10826 * machmode.h (mode_size): Move earlier in file.
10827 (mode_precision): Likewise.
10828 (mode_inner): Likewise.
10829 (mode_nunits): Likewise.
10830 (mode_unit_size): Likewise.
10831 (unit_unit_precision): Likewise.
10832 (mode_wider): Likewise.
10833 (mode_2xwider): Likewise.
10834 (machine_mode): New class.
10835 (mode_to_bytes): New function.
10836 (mode_to_bits): Likewise.
10837 (mode_to_precision): Likewise.
10838 (mode_to_inner): Likewise.
10839 (mode_to_unit_size): Likewise.
10840 (mode_to_unit_precision): Likewise.
10841 (mode_to_nunits): Likewise.
10842 (GET_MODE_SIZE): Use mode_to_bytes.
10843 (GET_MODE_BITSIZE): Use mode_to_bits.
10844 (GET_MODE_PRECISION): Use mode_to_precision.
10845 (GET_MODE_INNER): Use mode_to_inner.
10846 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
10847 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
10848 (GET_MODE_NUNITS): Use mode_to_nunits.
10849 * system.h (ALWAYS_INLINE): New macro.
10850 * config/powerpcspe/powerpcspe-c.c
10851 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
10852 int for arg1_mode and arg2_mode.
10854 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10855 Alan Hayward <alan.hayward@arm.com>
10856 David Sherwood <david.sherwood@arm.com>
10858 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
10859 Prefix mode names with E_ in case statements.
10860 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10861 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
10862 (aarch64_split_simd_move): Likewise.
10863 (aarch64_gen_storewb_pair): Likewise.
10864 (aarch64_gen_loadwb_pair): Likewise.
10865 (aarch64_gen_store_pair): Likewise.
10866 (aarch64_gen_load_pair): Likewise.
10867 (aarch64_get_condition_code_1): Likewise.
10868 (aarch64_constant_pool_reload_icode): Likewise.
10869 (get_rsqrte_type): Likewise.
10870 (get_rsqrts_type): Likewise.
10871 (get_recpe_type): Likewise.
10872 (get_recps_type): Likewise.
10873 (aarch64_gimplify_va_arg_expr): Likewise.
10874 (aarch64_simd_container_mode): Likewise.
10875 (aarch64_emit_load_exclusive): Likewise.
10876 (aarch64_emit_store_exclusive): Likewise.
10877 (aarch64_expand_compare_and_swap): Likewise.
10878 (aarch64_gen_atomic_cas): Likewise.
10879 (aarch64_emit_bic): Likewise.
10880 (aarch64_emit_atomic_swap): Likewise.
10881 (aarch64_emit_atomic_load_op): Likewise.
10882 (aarch64_evpc_trn): Likewise.
10883 (aarch64_evpc_uzp): Likewise.
10884 (aarch64_evpc_zip): Likewise.
10885 (aarch64_evpc_ext): Likewise.
10886 (aarch64_evpc_rev): Likewise.
10887 (aarch64_evpc_dup): Likewise.
10888 (aarch64_gen_ccmp_first): Likewise.
10889 (aarch64_gen_ccmp_next): Likewise.
10890 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
10891 (alpha_emit_xfloating_libcall): Likewise.
10892 (emit_insxl): Likewise.
10893 (alpha_arg_type): Likewise.
10894 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
10895 (arc_preferred_simd_mode): Likewise.
10896 (arc_secondary_reload): Likewise.
10897 (get_arc_condition_code): Likewise.
10898 (arc_print_operand): Likewise.
10899 (arc_legitimate_constant_p): Likewise.
10900 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10901 * config/arc/arc.md (casesi_load): Likewise.
10902 (casesi_compact_jump): Likewise.
10903 * config/arc/predicates.md (proper_comparison_operator): Likewise.
10904 (cc_use_register): Likewise.
10905 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10906 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
10907 (arm_init_iwmmxt_builtins): Likewise.
10908 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
10909 (neon_expand_vector_init): Likewise.
10910 (arm_attr_length_move_neon): Likewise.
10911 (maybe_get_arm_condition_code): Likewise.
10912 (arm_emit_vector_const): Likewise.
10913 (arm_preferred_simd_mode): Likewise.
10914 (arm_output_iwmmxt_tinsr): Likewise.
10915 (thumb1_output_casesi): Likewise.
10916 (thumb2_output_casesi): Likewise.
10917 (arm_emit_load_exclusive): Likewise.
10918 (arm_emit_store_exclusive): Likewise.
10919 (arm_expand_compare_and_swap): Likewise.
10920 (arm_evpc_neon_vuzp): Likewise.
10921 (arm_evpc_neon_vzip): Likewise.
10922 (arm_evpc_neon_vrev): Likewise.
10923 (arm_evpc_neon_vtrn): Likewise.
10924 (arm_evpc_neon_vext): Likewise.
10925 (arm_validize_comparison): Likewise.
10926 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
10927 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
10928 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
10929 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
10930 (c6x_preferred_simd_mode): Likewise.
10931 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
10932 (epiphany_rtx_costs): Likewise.
10933 * config/epiphany/predicates.md (proper_comparison_operator):
10935 * config/frv/frv.c (condexec_memory_operand): Likewise.
10936 (frv_emit_move): Likewise.
10937 (output_move_single): Likewise.
10938 (output_condmove_single): Likewise.
10939 (frv_hard_regno_mode_ok): Likewise.
10940 (frv_matching_accg_mode): Likewise.
10941 * config/h8300/h8300.c (split_adds_subs): Likewise.
10942 (h8300_rtx_costs): Likewise.
10943 (h8300_print_operand): Likewise.
10944 (compute_mov_length): Likewise.
10945 (output_logical_op): Likewise.
10946 (compute_logical_op_length): Likewise.
10947 (compute_logical_op_cc): Likewise.
10948 (h8300_shift_needs_scratch_p): Likewise.
10949 (output_a_shift): Likewise.
10950 (compute_a_shift_length): Likewise.
10951 (compute_a_shift_cc): Likewise.
10952 (expand_a_rotate): Likewise.
10953 (output_a_rotate): Likewise.
10954 * config/i386/i386.c (classify_argument): Likewise.
10955 (function_arg_advance_32): Likewise.
10956 (function_arg_32): Likewise.
10957 (function_arg_64): Likewise.
10958 (function_value_64): Likewise.
10959 (ix86_gimplify_va_arg): Likewise.
10960 (ix86_legitimate_constant_p): Likewise.
10961 (put_condition_code): Likewise.
10962 (split_double_mode): Likewise.
10963 (ix86_avx256_split_vector_move_misalign): Likewise.
10964 (ix86_expand_vector_logical_operator): Likewise.
10965 (ix86_split_idivmod): Likewise.
10966 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
10967 (ix86_build_const_vector): Likewise.
10968 (ix86_build_signbit_mask): Likewise.
10969 (ix86_match_ccmode): Likewise.
10970 (ix86_cc_modes_compatible): Likewise.
10971 (ix86_expand_branch): Likewise.
10972 (ix86_expand_sse_cmp): Likewise.
10973 (ix86_expand_sse_movcc): Likewise.
10974 (ix86_expand_int_sse_cmp): Likewise.
10975 (ix86_expand_vec_perm_vpermi2): Likewise.
10976 (ix86_expand_vec_perm): Likewise.
10977 (ix86_expand_sse_unpack): Likewise.
10978 (ix86_expand_int_addcc): Likewise.
10979 (ix86_split_to_parts): Likewise.
10980 (ix86_vectorize_builtin_gather): Likewise.
10981 (ix86_vectorize_builtin_scatter): Likewise.
10982 (avx_vpermilp_parallel): Likewise.
10983 (inline_memory_move_cost): Likewise.
10984 (ix86_tieable_integer_mode_p): Likewise.
10985 (x86_maybe_negate_const_int): Likewise.
10986 (ix86_expand_vector_init_duplicate): Likewise.
10987 (ix86_expand_vector_init_one_nonzero): Likewise.
10988 (ix86_expand_vector_init_one_var): Likewise.
10989 (ix86_expand_vector_init_concat): Likewise.
10990 (ix86_expand_vector_init_interleave): Likewise.
10991 (ix86_expand_vector_init_general): Likewise.
10992 (ix86_expand_vector_set): Likewise.
10993 (ix86_expand_vector_extract): Likewise.
10994 (emit_reduc_half): Likewise.
10995 (ix86_emit_i387_round): Likewise.
10996 (ix86_mangle_type): Likewise.
10997 (ix86_expand_round_sse4): Likewise.
10998 (expand_vec_perm_blend): Likewise.
10999 (canonicalize_vector_int_perm): Likewise.
11000 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
11001 (expand_vec_perm_1): Likewise.
11002 (expand_vec_perm_interleave3): Likewise.
11003 (expand_vec_perm_even_odd_pack): Likewise.
11004 (expand_vec_perm_even_odd_1): Likewise.
11005 (expand_vec_perm_broadcast_1): Likewise.
11006 (ix86_vectorize_vec_perm_const_ok): Likewise.
11007 (ix86_expand_vecop_qihi): Likewise.
11008 (ix86_expand_mul_widen_hilo): Likewise.
11009 (ix86_expand_sse2_abs): Likewise.
11010 (ix86_expand_pextr): Likewise.
11011 (ix86_expand_pinsr): Likewise.
11012 (ix86_preferred_simd_mode): Likewise.
11013 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
11014 * config/i386/sse.md (*andnot<mode>3): Likewise.
11015 (<mask_codefor><code><mode>3<mask_name>): Likewise.
11016 (*<code><mode>3): Likewise.
11017 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
11018 (ia64_expand_atomic_op): Likewise.
11019 (ia64_arg_type): Likewise.
11020 (ia64_mode_to_int): Likewise.
11021 (ia64_scalar_mode_supported_p): Likewise.
11022 (ia64_vector_mode_supported_p): Likewise.
11023 (expand_vec_perm_broadcast): Likewise.
11024 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
11025 (iq2000_function_arg_advance): Likewise.
11026 (iq2000_function_arg): Likewise.
11027 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
11028 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
11029 (m68k_libcall_value): Likewise.
11030 (m68k_function_value): Likewise.
11031 (sched_attr_op_type): Likewise.
11032 * config/mcore/mcore.c (mcore_output_move): Likewise.
11033 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
11035 (microblaze_function_arg): Likewise.
11036 * config/mips/mips.c (mips16_build_call_stub): Likewise.
11037 (mips_print_operand): Likewise.
11038 (mips_mode_ok_for_mov_fmt_p): Likewise.
11039 (mips_vector_mode_supported_p): Likewise.
11040 (mips_preferred_simd_mode): Likewise.
11041 (mips_expand_vpc_loongson_even_odd): Likewise.
11042 (mips_expand_vec_unpack): Likewise.
11043 (mips_expand_vi_broadcast): Likewise.
11044 (mips_expand_vector_init): Likewise.
11045 (mips_expand_vec_reduc): Likewise.
11046 (mips_expand_msa_cmp): Likewise.
11047 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
11048 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
11049 (cc_flags_for_mode): Likewise.
11050 * config/msp430/msp430.c (msp430_print_operand): Likewise.
11051 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
11052 (nds32_output_casesi_pc_relative): Likewise.
11053 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11054 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
11055 (nvptx_gen_unpack): Likewise.
11056 (nvptx_gen_pack): Likewise.
11057 (nvptx_gen_shuffle): Likewise.
11058 (nvptx_gen_wcast): Likewise.
11059 (nvptx_preferred_simd_mode): Likewise.
11060 * config/pa/pa.c (pa_secondary_reload): Likewise.
11061 * config/pa/predicates.md (base14_operand): Likewise.
11062 * config/powerpcspe/powerpcspe-c.c
11063 (altivec_resolve_overloaded_builtin): Likewise.
11064 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
11066 (rs6000_preferred_simd_mode): Likewise.
11067 (output_vec_const_move): Likewise.
11068 (rs6000_expand_vector_extract): Likewise.
11069 (rs6000_split_vec_extract_var): Likewise.
11070 (reg_offset_addressing_ok_p): Likewise.
11071 (rs6000_legitimate_offset_address_p): Likewise.
11072 (rs6000_legitimize_address): Likewise.
11073 (rs6000_emit_set_const): Likewise.
11074 (rs6000_const_vec): Likewise.
11075 (rs6000_emit_move): Likewise.
11076 (spe_build_register_parallel): Likewise.
11077 (rs6000_darwin64_record_arg_recurse): Likewise.
11078 (swap_selector_for_mode): Likewise.
11079 (spe_init_builtins): Likewise.
11080 (paired_init_builtins): Likewise.
11081 (altivec_init_builtins): Likewise.
11082 (do_load_for_compare): Likewise.
11083 (rs6000_generate_compare): Likewise.
11084 (rs6000_expand_float128_convert): Likewise.
11085 (emit_load_locked): Likewise.
11086 (emit_store_conditional): Likewise.
11087 (rs6000_output_function_epilogue): Likewise.
11088 (rs6000_handle_altivec_attribute): Likewise.
11089 (rs6000_function_value): Likewise.
11090 (emit_fusion_gpr_load): Likewise.
11091 (emit_fusion_p9_load): Likewise.
11092 (emit_fusion_p9_store): Likewise.
11093 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
11094 (fusion_gpr_mem_load): Likewise.
11095 (fusion_addis_mem_combo_load): Likewise.
11096 (fusion_addis_mem_combo_store): Likewise.
11097 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
11098 (fusion_gpr_mem_load): Likewise.
11099 (fusion_addis_mem_combo_load): Likewise.
11100 (fusion_addis_mem_combo_store): Likewise.
11101 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11103 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
11104 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
11105 (rs6000_preferred_simd_mode): Likewise.
11106 (output_vec_const_move): Likewise.
11107 (rs6000_expand_vector_extract): Likewise.
11108 (rs6000_split_vec_extract_var): Likewise.
11109 (reg_offset_addressing_ok_p): Likewise.
11110 (rs6000_legitimate_offset_address_p): Likewise.
11111 (rs6000_legitimize_address): Likewise.
11112 (rs6000_emit_set_const): Likewise.
11113 (rs6000_const_vec): Likewise.
11114 (rs6000_emit_move): Likewise.
11115 (rs6000_darwin64_record_arg_recurse): Likewise.
11116 (swap_selector_for_mode): Likewise.
11117 (paired_init_builtins): Likewise.
11118 (altivec_init_builtins): Likewise.
11119 (rs6000_expand_float128_convert): Likewise.
11120 (emit_load_locked): Likewise.
11121 (emit_store_conditional): Likewise.
11122 (rs6000_output_function_epilogue): Likewise.
11123 (rs6000_handle_altivec_attribute): Likewise.
11124 (rs6000_function_value): Likewise.
11125 (emit_fusion_gpr_load): Likewise.
11126 (emit_fusion_p9_load): Likewise.
11127 (emit_fusion_p9_store): Likewise.
11128 * config/rx/rx.c (rx_gen_move_template): Likewise.
11129 (flags_from_mode): Likewise.
11130 * config/s390/predicates.md (s390_alc_comparison): Likewise.
11131 (s390_slb_comparison): Likewise.
11132 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
11133 (s390_vector_mode_supported_p): Likewise.
11134 (s390_cc_modes_compatible): Likewise.
11135 (s390_match_ccmode_set): Likewise.
11136 (s390_canonicalize_comparison): Likewise.
11137 (s390_emit_compare_and_swap): Likewise.
11138 (s390_branch_condition_mask): Likewise.
11139 (s390_rtx_costs): Likewise.
11140 (s390_secondary_reload): Likewise.
11141 (__SECONDARY_RELOAD_CASE): Likewise.
11142 (s390_expand_cs): Likewise.
11143 (s390_preferred_simd_mode): Likewise.
11144 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
11145 * config/sh/sh.c (sh_print_operand): Likewise.
11146 (dump_table): Likewise.
11147 (sh_secondary_reload): Likewise.
11148 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11149 * config/sh/sh.md (casesi_worker_1): Likewise.
11150 (casesi_worker_2): Likewise.
11151 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
11152 (fcc_comparison_operator): Likewise.
11153 * config/sparc/sparc.c (sparc_expand_move): Likewise.
11154 (emit_soft_tfmode_cvt): Likewise.
11155 (sparc_preferred_simd_mode): Likewise.
11156 (output_cbranch): Likewise.
11157 (sparc_print_operand): Likewise.
11158 (sparc_expand_vec_perm_bmask): Likewise.
11159 (vector_init_bshuffle): Likewise.
11160 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11161 (spu_vector_mode_supported_p): Likewise.
11162 (spu_expand_insv): Likewise.
11163 (spu_emit_branch_or_set): Likewise.
11164 (spu_handle_vector_attribute): Likewise.
11165 (spu_builtin_splats): Likewise.
11166 (spu_builtin_extract): Likewise.
11167 (spu_builtin_promote): Likewise.
11168 (spu_expand_sign_extend): Likewise.
11169 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11170 (tilegx_simd_int): Likewise.
11171 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
11172 (tilepro_simd_int): Likewise.
11173 * config/v850/v850.c (const_double_split): Likewise.
11174 (v850_print_operand): Likewise.
11175 (ep_memory_offset): Likewise.
11176 * config/vax/vax.c (vax_rtx_costs): Likewise.
11177 (vax_output_int_move): Likewise.
11178 (vax_output_int_add): Likewise.
11179 (vax_output_int_subtract): Likewise.
11180 * config/visium/predicates.md (visium_branch_operator): Likewise.
11181 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
11182 (visium_print_operand_address): Likewise.
11183 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11184 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
11185 (xtensa_expand_conditional_branch): Likewise.
11186 (xtensa_copy_incoming_a7): Likewise.
11187 (xtensa_output_literal): Likewise.
11188 * dfp.c (decimal_real_maxval): Likewise.
11189 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11191 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11192 Alan Hayward <alan.hayward@arm.com>
11193 David Sherwood <david.sherwood@arm.com>
11195 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
11196 (mode_nunits_inline): Likewise.
11197 (mode_inner_inline): Likewise.
11198 (mode_unit_size_inline): Likewise.
11199 (mode_unit_precision_inline): Likewise.
11200 (emit_insn_modes_h): Likewise. Also emit a #define of the
11202 (emit_mode_wider): Add an E_ prefix to mode names.
11203 (emit_mode_complex): Likewise.
11204 (emit_mode_inner): Likewise.
11205 (emit_mode_adjustments): Likewise.
11206 (emit_mode_int_n): Likewise.
11207 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
11208 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
11209 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
11210 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
11211 (CRC32_BUILTIN, ENTRY): Likewise.
11212 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
11213 (aarch64_pop_regs): Likewise.
11214 (aarch64_process_components): Likewise.
11215 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
11216 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
11217 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
11218 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
11219 * config/arm/arm.c (arm_init_libfuncs): Likewise.
11220 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
11222 * config/i386/i386-builtin.def (pcmpestr): Likewise.
11223 (pcmpistr): Likewise.
11224 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
11225 * config/mmix/mmix.c (mmix_output_condition): Likewise.
11226 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
11228 * config/rl78/rl78.c (mduc_regs): Likewise.
11229 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
11230 (htm_expand_builtin): Likewise.
11231 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
11232 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
11233 * config/xtensa/xtensa.c (print_operand): Likewise.
11234 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
11235 (NUM_MODE_VECTOR_INT): Likewise.
11236 * genoutput.c (null_operand): Likewise.
11237 (output_operand_data): Likewise.
11238 * genrecog.c (print_parameter_value): Likewise.
11239 * lra.c (debug_operand_data): Likewise.
11241 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11243 * dwarf2out.c (scompare_loc_descriptor_wide)
11244 (scompare_loc_descriptor_narrow): New functions, split out from...
11245 (scompare_loc_descriptor): ...here.
11246 * expmed.c (emit_store_flag_int): New function, split out from...
11247 (emit_store_flag): ...here.
11249 2017-08-30 Richard Biener <rguenther@suse.de>
11251 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
11252 (dwarf2out_early_finish): Move setting of AT_pubnames from
11253 early debug output to early finish.
11255 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11257 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
11258 and -mdata-region to the assembler.
11260 2017-08-30 Richard Biener <rguenther@suse.de>
11262 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
11264 (gen_subprogram_die): Add DW_AT_object_pointer only early.
11265 (dwarf2out_early_global_decl): Only generate a DIE for the
11266 abstract origin if it doesn't already exist or is a declaration DIE.
11267 (resolve_addr): Do not add the linkage name twice when
11268 generating a stub DIE for the DW_TAG_GNU_call_site target.
11270 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11272 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11273 Use machine_mode rather than int for arg1_mode.
11275 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
11278 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
11279 that the second argument of the built-in functions to unpack
11280 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
11281 switch statement instead a lot of if statements.
11282 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
11283 Allow 64-bit values to be in Altivec registers as well as
11284 traditional floating point registers.
11285 (pack<mode>, FMOVE128_VSX iterator): Likewise.
11287 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
11289 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
11290 MAX_REGS_PER_ADDRESS == 1.
11292 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
11294 * config/i386/i386.opt (flag_fentry): Do not init to -1.
11295 * config/i386/i386.c (ix86_option_override_internal): Simplify
11296 setting of opts->x_flag_entry.
11298 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11299 Jakub Jelinek <jakub@redhat.com>
11300 Richard Biener <rguenther@suse.de>
11302 PR tree-optimization/81503
11303 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
11304 folded constant fits in the target type; reorder tests for clarity.
11306 2017-08-29 Martin Liska <mliska@suse.cz>
11308 * passes.def: Include pass_lower_switch.
11309 * stmt.c (dump_case_nodes): Remove and move to
11310 tree-switch-conversion.
11311 (case_values_threshold): Likewise.
11312 (expand_switch_as_decision_tree_p): Likewise.
11313 (emit_case_decision_tree): Likewise.
11314 (expand_case): Likewise.
11315 (balance_case_nodes): Likewise.
11316 (node_has_low_bound): Likewise.
11317 (node_has_high_bound): Likewise.
11318 (node_is_bounded): Likewise.
11319 (emit_case_nodes): Likewise.
11320 (struct simple_case_node): New struct.
11321 (add_case_node): Remove.
11322 (emit_case_dispatch_table): Use vector instead of case_list.
11323 (reset_out_edges_aux): Remove.
11324 (compute_cases_per_edge): Likewise.
11325 (expand_case): Build list of simple_case_node.
11326 (expand_sjlj_dispatch_table): Use it.
11327 * tree-switch-conversion.c (struct case_node): Moved from
11328 stmt.c and adjusted.
11329 (emit_case_nodes): Likewise.
11330 (node_has_low_bound): Likewise.
11331 (node_has_high_bound): Likewise.
11332 (node_is_bounded): Likewise.
11333 (case_values_threshold): Likewise.
11334 (reset_out_edges_aux): Likewise.
11335 (compute_cases_per_edge): Likewise.
11336 (add_case_node): Likewise.
11337 (dump_case_nodes): Likewise.
11338 (balance_case_nodes): Likewise.
11339 (expand_switch_as_decision_tree_p): Likewise.
11340 (emit_jump): Likewise.
11341 (emit_case_decision_tree): Likewise.
11342 (try_switch_expansion): Likewise.
11343 (do_jump_if_equal): Likewise.
11344 (emit_cmp_and_jump_insns): Likewise.
11345 (fix_phi_operands_for_edge): New function.
11346 (record_phi_operand_mapping): Likewise.
11347 (class pass_lower_switch): New pass.
11348 (pass_lower_switch::execute): New function.
11349 (make_pass_lower_switch): Likewise.
11350 (conditional_probability):
11351 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
11352 * tree-pass.h: Add make_pass_lower_switch.
11354 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11357 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
11360 2017-08-29 Richard Biener <rguenther@suse.de>
11362 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
11363 we do not add a DW_AT_inline attribute twice.
11364 (gen_subprogram_die): Remove code setting DW_AT_inline on
11365 DECL_ABSTRACT_P nodes.
11367 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
11369 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
11370 calls to internal functions.
11371 (gimplify_modify_expr): Likewise.
11372 * tree-call-cdce.c (use_internal_fn): Likewise.
11373 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
11374 (convert_to_divmod): Set the nothrow flag.
11375 * tree-if-conv.c (predicate_mem_writes): Likewise.
11376 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
11377 (vectorizable_call): Likewise.
11378 (vectorizable_store): Likewise.
11379 (vectorizable_load): Likewise.
11380 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
11381 (vect_recog_mask_conversion_pattern): Likewise.
11383 2017-08-29 Martin Liska <mliska@suse.cz>
11386 * gcc.c (driver_handle_option): Add new argument.
11387 * opts-common.c (handle_option): Pass
11388 target_option_override_hook.
11389 * opts-global.c (lang_handle_option): Add new option.
11390 (set_default_handlers): Add new argument.
11391 (decode_options): Likewise.
11392 * opts.c (target_handle_option): Likewise.
11393 (common_handle_option): Call target_option_override_hook.
11394 * opts.h (struct cl_option_handler_func): Add hook for
11395 target option override.
11396 (struct cl_option_handlers): Likewise.
11397 (set_default_handlers): Add new argument.
11398 (decode_options): Likewise.
11399 (common_handle_option): Likewise.
11400 (target_handle_option): Likewise.
11401 * toplev.c (toplev::main): Pass targetm.target_option.override
11404 2017-08-29 Richard Biener <rguenther@suse.de>
11405 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
11407 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
11408 life to the active subtree.
11410 2017-08-28 Jeff Law <law@redhat.com>
11412 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
11413 derive_equivalences.
11414 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
11415 Code moved into....
11416 (edge_info::derive_equivalences): New private member function
11418 * tree-ssa-dom.c (class edge_info): Changed from a struct
11419 to a class. Add ctor/dtor, methods and data members.
11420 (edge_info::edge_info): Renamed from allocate_edge_info.
11421 Initialize additional members.
11422 (edge_info::~edge_info): New.
11423 (free_dom_edge_info): Delete the edge info.
11424 (record_edge_info): Use new class & associated member functions.
11425 Tighten forms for testing for edge equivalences.
11426 (record_temporary_equivalences): Iterate over the simple
11427 equivalences rather than assuming there's only one per edge.
11428 (cprop_into_successor_phis): Iterate over the simple
11429 equivalences rather than assuming there's only one per edge.
11430 (optimize_stmt): Use operand_equal_p rather than pointer
11431 equality for mini-DSE code.
11433 2017-08-28 Nathan Sidwell <nathan@acm.org>
11435 * gcc.c (execute): Fold SIGPIPE handling into switch
11436 statement. Adjust internal error message.
11438 2017-08-28 Richard Biener <rguenther@suse.de>
11441 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
11442 Do nothing for removed DIEs.
11444 2017-08-28 Richard Biener <rguenther@suse.de>
11446 PR tree-optimization/81977
11447 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
11450 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
11453 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
11454 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
11455 using targetm.gen_mem_thread_fence.
11457 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
11460 * config/i386/i386.md (*<btsc><mode>): Change operand 2
11461 predicate to register_operand. Reorder operands.
11462 (*btr<mode>): Ditto.
11463 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
11464 (*btr<mode>_mask): Ditto.
11466 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
11468 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
11469 * config/rs6000/xmmintrin.h: New file.
11470 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
11472 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11475 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
11476 parameter and_insn and return it.
11477 (recombine_lvx_pattern): Insert a copy to ensure availability of
11478 the base register of the copied masking operation at the point of
11479 the instruction replacement.
11480 (recombine_stvx_pattern): Likewise.
11482 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
11484 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
11485 undocumented switches.
11486 (-mpower9-dform-vector): Likewise.
11487 (-mpower9-dform): Likewise.
11488 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
11489 comments to delete references to -mpower9-dform* switches.
11490 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
11491 Delete reference to -mpower9-dform* switches, test for
11492 -mpower9-vector instead.
11493 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
11494 (OTHER_P9_VECTOR_MASKS): Likewise.
11495 (POWERPC_MASKS): Likewise.
11496 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
11497 tests against -mpower9-dform* to -mpower9-vector. Delete code
11498 that checked for -mpower9-dform* consistancy with other options.
11499 Add test for -mpower9-misc to enable other power9 switches.
11500 (rs6000_init_hard_regno_mode_ok): Likewise.
11501 (rs6000_option_override_internal): Likewise.
11502 (rs6000_emit_prologue): Likewise.
11503 (rs6000_emit_epilogue): Likewise.
11504 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
11505 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
11506 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
11508 (emit_fusion_p9_store): Likewise.
11509 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
11510 resetting these macros if the assembler does not support ISA 3.0
11512 (TARGET_P9_DFORM_VECTOR): Likewise.
11513 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
11514 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
11516 2017-08-25 Alan Modra <amodra@gmail.com>
11518 PR rtl-optimization/81747
11519 * cse.c (cse_extended_basic_block): Don't attempt to record
11520 equivalences for degenerate conditional branches that branch
11521 to their fall-through.
11523 2017-08-24 Martin Sebor <msebor@redhat.com>
11525 PR middle-end/81908
11526 * gimple-fold.c (size_must_be_zero_p): New function.
11527 (gimple_fold_builtin_memory_op): Call it.
11529 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
11531 * config/rs6000/mm_malloc.h: New file.
11533 2017-08-24 Bin Cheng <bin.cheng@arm.com>
11535 PR tree-optimization/81913
11536 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
11537 analysis when either IVs in condition can wrap.
11539 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
11541 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
11542 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
11544 2017-08-24 Richard Biener <rguenther@suse.de>
11547 * targhooks.c (default_target_can_inline_p): Properly
11548 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
11549 is present and always compare.
11550 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
11551 infer -mfpmath=sse from TARGET_SSE_P.
11552 (ix86_can_inline_p): Properly use target_option_default_node when
11553 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
11555 2017-08-24 Richard Biener <rguenther@suse.de>
11558 * dwarf2out.c (output_die): Handle flag_generate_offload like
11560 (output_comp_unit): Likewise.
11561 (gen_array_type_die): Likewise.
11562 (dwarf2out_early_finish): Likewise.
11563 (note_variable_value_in_expr): Likewise.
11564 (dwarf2out_finish): Likewise. Adjust assert.
11565 * cgraphunit.c (symbol_table::compile): Move setting of
11566 flag_generate_offload earlier ...
11567 (symbol_table::finalize_compilation_unit): ... here, before
11568 early debug finalization.
11570 2017-08-24 Richard Biener <rguenther@suse.de>
11572 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
11573 and ipa-fnsummary.h.
11574 (ix86_can_inline_p): When ix86_fpmath flags do not match
11575 check whether the callee uses FP math at all.
11577 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
11579 PR middle-end/81931
11580 * tree-ssanames.c (get_nonzero_bits): Use element_precision
11581 instead of TYPE_PRECISION.
11583 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
11584 Alan Hayward <alan.hayward@arm.com>
11585 David Sherwood <david.sherwood@arm.com>
11587 * combine.c (make_extraction): Use subreg_offset_from_lsb.
11589 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
11591 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
11593 (ix86_frame::stack_realign_allocate): New field.
11594 (struct machine_frame_state): Modify comments.
11595 (machine_frame_state::sp_realigned_fp_end): New field.
11596 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
11597 layout calculation.
11598 (sp_valid_at): Add assertion to assure no attempt to access invalid
11599 offset of a realigned stack.
11600 (fp_valid_at): Likewise.
11601 (choose_baseaddr): Modify comments.
11602 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
11603 ix86_expand_prologue.
11604 (ix86_expand_prologue): Modify stack realignment and allocation.
11605 (ix86_expand_epilogue): Modify comments.
11606 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
11607 avx2_runtime, avx512f, and avx512f_runtime.
11609 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
11611 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
11612 (mstackrealign): Do not init to -1.
11613 * config/i386/i386.c (ix86_option_override_internal):
11614 Check opts_set, not opts when setting default value of
11615 opts->x_ix86_force_align_arg_pointer.
11617 2017-08-23 Richard Biener <rguenther@suse.de>
11619 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
11620 lang_hooks.decl_printable_name.
11621 * print-rtl-function.c (print_rtx_function): Likewise.
11622 * tree-pretty-print.c (dump_function_header): Likewise.
11624 2017-08-23 Richard Biener <rguenther@suse.de>
11627 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
11628 -g0 at compile-time.
11630 2017-08-23 Tamar Christina <tamar.christina@arm.com>
11632 PR middle-end/19706
11633 * doc/sourcebuild.texi (Other hardware attributes):
11636 2017-08-23 Tamar Christina <tamar.christina@arm.com>
11638 PR middle-end/19706
11639 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
11640 Fix single-use check.
11642 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
11644 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
11646 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
11648 * doc/install.texi: Modify to add more details on running selected
11651 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
11653 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
11654 is combined with -mabi=ms.
11655 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
11658 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11660 PR tree-optimization/81488
11661 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
11662 and cached_basis fields.
11663 (MAX_SPREAD): New constant.
11664 (alloc_cand_and_find_basis): Initialize new fields.
11665 (clear_visited): New function.
11666 (create_phi_basis_1): Rename from create_phi_basis, set visited
11667 and cached_basis fields.
11668 (create_phi_basis): New wrapper function.
11669 (phi_add_costs_1): Rename from phi_add_costs, add spread
11670 parameter, set visited field, short-circuit when limits reached.
11671 (phi_add_costs): New wrapper function.
11672 (record_phi_increments_1): Rename from record_phi_increments, set
11674 (record_phi_increments): New wrapper function.
11675 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
11676 (phi_incr_cost): New wrapper function.
11677 (all_phi_incrs_profitable_1): Rename from
11678 all_phi_incrs_profitable, set visited field.
11679 (all_phi_incrs_profitable): New wrapper function.
11681 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
11682 Alan Hayward <alan.hayward@arm.com>
11683 David Sherwood <david.sherwood@arm.com>
11685 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
11686 that takes the outer and inner modes.
11687 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
11688 comparison as the canonical test for a paradoxical subreg.
11689 * combine.c (simplify_set): Use paradoxical_subreg_p.
11690 (make_extraction): Likewise.
11691 (force_to_mode): Likewise.
11692 (rtx_equal_for_field_assignment_p): Likewise.
11693 (gen_lowpart_for_combine): Likewise.
11694 (simplify_comparison): Likewise.
11695 * cse.c (equiv_constant): Likewise.
11696 * expmed.c (store_bit_field_1): Likewise.
11697 * final.c (alter_subreg): Likewise.
11698 * fwprop.c (propagate_rtx): Likewise.
11699 (forward_propagate_subreg): Likewise.
11700 * ira-conflicts.c (ira_build_conflicts): Likewise.
11701 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
11702 * lra-constraints.c (curr_insn_transform): Likewise.
11703 (split_reg): Likewise.
11704 * lra-eliminations.c (move_plus_up): Likewise.
11705 (lra_eliminate_regs_1): Likewise.
11706 * recog.c (general_operand): Likewise.
11707 * ree.c (combine_reaching_defs): Likewise.
11708 * reload.c (push_reload): Likewise.
11709 (find_reloads): Likewise.
11710 * reload1.c (elimination_effects): Likewise.
11711 (compute_reload_subreg_offset): Likewise.
11712 (choose_reload_regs): Likewise.
11713 * rtlanal.c (subreg_lsb_1): Likewise.
11714 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11715 (simplify_subreg): Likewise.
11716 * var-tracking.c (track_loc_p): Likewise.
11717 * emit-rtl.c (byte_lowpart_offset): Likewise.
11718 (paradoxical_subreg_p): Delete out-of-line definition.
11720 2017-08-22 Jeff Law <law@redhat.com>
11722 PR tree-optimization/81741
11723 PR tree-optimization/71947
11724 * tree-ssa-dom.c: Include tree-inline.h.
11725 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
11726 equivalences if one is more expensive to compute than the other.
11727 * tree-ssa-scopedtables.h (class const_or_copies): Make
11728 record_const_or_copy_raw method private.
11729 (class avail_exprs_stack): New method simplify_binary_operation.
11730 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
11731 avail_exprs_stack::simplify_binary_operation as needed.
11732 (avail_exprs_stack::simplify_binary_operation): New function.
11734 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
11736 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
11737 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
11738 (DOT_SYMBOLS): Likewise.
11739 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
11740 (RELOCATABLE_NEEDS_FIXUP): Likewise.
11741 (RS6000_ABI_NAME): Likewise.
11742 (TARGET_CMODEL): Likewise.
11743 (TOC_SECTION_ASM_OP): Likewise.
11744 (SET_CMODEL): New macro.
11745 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
11747 2017-08-22 Richard Biener <rguenther@suse.de>
11749 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
11750 to allow for free-lang-data replacements similar to verify_type_variant.
11752 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
11753 Alan Hayward <alan.hayward@arm.com>
11754 David Sherwood <david.sherwood@arm.com>
11756 * config/aarch64/aarch64.md (casesi): Use DImode rather than
11757 VOIDmode for the LABEL_REF.
11759 2017-08-22 Richard Biener <rguenther@suse.de>
11761 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
11763 2017-08-22 Richard Biener <rguenther@suse.de>
11765 * common.opt (feliminate-dwarf2-dups): Ignore.
11766 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
11767 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
11768 same_die_p_wrap, compute_section_prefix,
11769 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
11770 (comdat_symbol_id, comdat_symbol_number): Likewise.
11771 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
11773 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
11774 (output_die): Mark unreachable path unreachable.
11775 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
11776 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
11777 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
11778 (dwarf2out_early_finish): Likewise.
11780 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
11782 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
11784 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
11787 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
11788 not VAR_P. Filter attribute warnings with OPT_Wattributes.
11789 (avr_attribute_table) <io, io_low, address>: Initialize
11790 .decl_required with true.
11792 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
11794 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
11795 undocumented debugging options.
11796 (-mvsx-scalar-double): Likewise.
11797 (-mallow-df-permute): Likewise.
11798 (-mvectorize-builtins): Likewise.
11799 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
11800 (rs6000_builtin_vectorized_function): Likewise.
11801 (rs6000_builtin_md_vectorized_function): Likewise.
11802 (rs6000_opt_vars): Likewise.
11804 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
11807 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
11808 (*btrq_imm): Rename from *btrq.
11809 (*btcq_imm): Rename from *btcq.
11810 (btsc): New code attribute.
11811 (*<btsc><mode>): New insn pattern.
11812 (*btr<mode>): Ditto.
11813 (*<btsc><mode>_mask): New insn_and_split pattern.
11814 (*btr<mode>_mask): Ditto.
11816 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11817 Alan Hayward <alan.hayward@arm.com>
11818 David Sherwood <david.sherwood@arm.com>
11820 * function.c (pad_below): Simplify padding calculation.
11822 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11823 Alan Hayward <alan.hayward@arm.com>
11824 David Sherwood <david.sherwood@arm.com>
11826 * target.def (function_prologue): Remove frame size argument.
11827 (function_epilogue): Likewise.
11828 * doc/tm.texi: Regenerate.
11829 * final.c (final_start_function): Update call to function_prologue.
11830 (final_end_function): Update call to function_epilogue.
11831 (default_function_pro_epilogue): Remove frame size argument.
11832 * output.h (default_function_pro_epilogue): Likewise.
11833 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
11834 (arm_output_function_prologue): Likewise.
11835 * config/frv/frv.c (frv_function_prologue): Likewise.
11836 (frv_function_epilogue): Likewise.
11837 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
11838 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
11839 (ia64_output_function_epilogue): Likewise.
11840 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
11841 (m32r_output_function_epilogue): Likewise.
11842 * config/microblaze/microblaze.c (microblaze_function_prologue)
11843 (microblaze_function_epilogue): Likewise.
11844 * config/mips/mips.c (mips_output_function_prologue): Likewise.
11845 (mips_output_function_epilogue): Likewise.
11846 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
11847 (mmix_target_asm_function_epilogue): Likewise.
11848 * config/msp430/msp430.c (msp430_start_function): Likewise.
11849 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
11850 (nds32_asm_function_epilogue): Likewise.
11851 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
11852 * config/pa/pa.c (pa_output_function_prologue): Likewise.
11853 (pa_output_function_epilogue): Likewise.
11854 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
11855 (rs6000_output_function_epilogue): Likewise.
11856 * config/rl78/rl78.c (rl78_start_function): Likewise.
11857 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
11858 (rs6000_output_function_epilogue): Likewise.
11859 * config/rx/rx.c (rx_output_function_prologue): Likewise.
11860 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
11861 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
11862 (sparc_asm_function_epilogue): Likewise.
11864 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11866 * tree.h (type_has_mode_precision_p): New function.
11867 * convert.c (convert_to_integer_1): Use it.
11868 * expr.c (expand_expr_real_2): Likewise.
11869 (expand_expr_real_1): Likewise.
11870 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
11871 * match.pd: Likewise.
11872 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
11873 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
11874 * tree-tailcall.c (process_assignment): Likewise.
11875 * tree-vect-loop.c (vectorizable_reduction): Likewise.
11876 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
11877 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
11878 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11879 (vectorizable_assignment): Likewise.
11880 (vectorizable_shift): Likewise.
11881 (vectorizable_operation): Likewise.
11882 * tree-vrp.c (register_edge_assert_for_2): Likewise.
11884 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
11886 * match.pd: Add pow (C, x) simplification.
11888 2017-08-21 Richard Biener <rguenther@suse.de>
11890 PR tree-optimization/81900
11891 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
11892 for blocks with abnormal predecessors.
11893 (compute_antic): Do not set visited flag prematurely.
11895 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
11898 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
11900 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11902 * stor-layout.h (vector_type_mode): Move to...
11903 * tree.h (vector_type_mode): ...here.
11904 * stor-layout.c (vector_type_mode): Move to...
11905 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
11907 2017-08-21 Richard Biener <rguenther@suse.de>
11909 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
11910 register_external_die hooks.
11911 (debug_false_tree_charstarstar_uhwistar): Declare.
11912 (debug_nothing_tree_charstar_uhwi): Likewise.
11913 * debug.c (do_nothing_debug_hooks): Adjust.
11914 (debug_false_tree_charstarstar_uhwistar): New do nothing.
11915 (debug_nothing_tree_charstar_uhwi): Likewise.
11916 * dbxout.c (dbx_debug_hooks): Adjust.
11917 (xcoff_debug_hooks): Likewise.
11918 * sdbout.c (sdb_debug_hooks): Likewise.
11919 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
11920 * dwarf2out.c (macinfo_label_base): New global.
11921 (dwarf2out_register_external_die): New function for the
11922 register_external_die hook.
11923 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
11924 (dwarf2_debug_hooks): Use them.
11925 (dwarf2_lineno_debug_hooks): Adjust.
11926 (struct die_struct): Add with_offset flag.
11927 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
11928 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
11929 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
11930 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
11931 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
11932 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
11933 defining section names for the early LTO debug variants.
11934 (reset_indirect_string): New helper.
11935 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
11936 (print_dw_val): Add support for offsetted symbol references.
11937 (get_ultimate_context): Split out from is_cxx.
11938 (is_cxx): Use get_ultimate_context.
11939 (is_fortran): Add decl overload.
11940 (compute_comp_unit_symbol): Split out worker from
11941 compute_section_prefix.
11942 (compute_section_prefix): Call compute_comp_unit_symbol and
11943 set comdat_type_p here.
11944 (output_die): Skip DIE symbol output for the LTO added one.
11945 Handle DIE symbol references with offset.
11946 (output_comp_unit): Guard section name mangling properly.
11947 For LTO debug sections emit a symbol at the section beginning
11948 which we use to refer to its DIEs.
11949 (add_abstract_origin_attribute): For DIEs registered via
11950 dwarf2out_register_external_die directly refer to the early
11951 DIE rather than indirectly through the shadow one we created.
11952 Remove obsolete call to dwarf2out_abstract_function for
11953 non-function/block origins.
11954 (gen_array_type_die): When generating early LTO debug do
11955 not emit DW_AT_string_length.
11956 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
11957 late when in LTO. As suggested place a gcc_unreachable for
11958 the DECL_ABSTRACT_P case.
11959 (gen_subprogram_die): Avoid another specification DIE
11960 for early built declarations/definitions for the late LTO case.
11961 (gen_variable_die): Add type references for late duplicated VLA dies
11963 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
11964 we have the abstract instance already.
11965 (process_scope_var): Adjust decl DIE contexts in LTO which
11966 first puts them in limbo.
11967 (gen_decl_die): Do not generate type DIEs late apart from
11968 types for VLAs or for decls we do not yet have a DIE. Do not
11969 call dwarf2out_abstract_function late.
11970 (dwarf2out_early_global_decl): Make sure to create DIEs
11971 for abstract instances of a decl first.
11972 (dwarf2out_late_global_decl): Adjust comment.
11973 (output_macinfo_op): With multiple macro sections use
11974 macinfo_label_base to distinguish labels.
11975 (output_macinfo): Likewise. Update macinfo_label_base.
11976 Pass in the line info label.
11977 (note_variable_value_in_expr): When generating LTO resolve
11978 all variable values here by generating DIEs as needed.
11979 (init_sections_and_labels): Add early LTO debug flag parameter
11980 and generate different sections and names if set. Add generation
11981 counter for the labels so we can have multiple of them.
11982 (reset_dies): Helper to allow DIEs to be output multiple times.
11983 (dwarf2out_finish): When outputting DIEs to the fat part of an
11984 LTO object first reset DIEs.
11985 (dwarf2out_early_finish): Output early DIEs when generating LTO.
11986 (modified_type_die): Check for decl_ultimate_origin being self
11988 (gen_type_die_with_usage): Likewise.
11989 (gen_typedef_die): Allow decl_ultimate_origin being self.
11990 (set_decl_abstract_flags): Remove.
11991 (set_block_abstract_flags): Likewise.
11992 (dwarf2out_abstract_function): Treat the early generated DIEs
11993 as the abstract copy and only add DW_AT_inline and
11994 DW_AT_artificial here and call set_decl_origin_self.
11995 If the DIE has an abstract origin don't do anything.
11996 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
11997 if we have none yet (Go fails to build one, PR78628).
11998 (variably_modified_type_p): Prevent endless recursion for Ada
11999 cyclic pointer types.
12000 * lto-streamer-in.c: Include debug.h.
12001 (dref_queue): New global.
12002 (lto_read_tree_1): Stream in DIE references.
12003 (lto_input_tree): Register DIE references.
12004 (input_function): Stream DECL_DEBUG_ARGS.
12005 * lto-streamer-out.c: Include debug.h.
12006 (lto_write_tree_1): Output DIE references.
12007 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
12008 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
12009 (output_function): Stream DECL_DEBUG_ARGS.
12010 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
12011 Stream DECL_ABSTRACT_ORIGIN.
12012 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
12013 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
12014 DECL_CONTEXT for file-scope decls.
12015 * lto-streamer.h (struct dref_entry): Declare.
12016 (dref_queue): Likewise.
12017 * cfgexpand.c (pass_expand::execute): Do not call the
12018 outlining_inline_function hook here.
12019 * lto-wrapper.c (debug_obj): New global.
12020 (tool_cleanup): Unlink it if required.
12021 (debug_objcopy): New function.
12022 (run_gcc): Handle early debug sections in the IL files by
12023 extracting them to separate files, partially linkin them and
12024 feeding the result back as result to the linker.
12025 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
12026 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
12027 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
12028 sections into a separate segment.
12029 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
12031 (darwin_asm_dwarf_section): Likewise.
12032 (darwin_asm_output_dwarf_offset): Likewise.
12033 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
12035 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12036 Alan Hayward <alan.hayward@arm.com>
12037 David Sherwood <david.sherwood@arm.com>
12039 * read-md.h (md_reader::record_potential_iterator_use): Replace
12040 pointer argument with an rtx and an index.
12041 * read-rtl.c (iterator_group::apply_iterator): Likewise.
12042 (apply_mode_iterator): Likewise.
12043 (apply_code_iterator): Likewise.
12044 (apply_int_iterator): Likewise.
12045 (apply_subst_iterator): Likewise.
12046 (record_iterator_use): Likewise.
12047 (record_attribute_use): Likewise.
12048 (md_reader::record_potential_iterator_use): Likewise. Update calls
12049 to record_iterator_use and apply_iterator.
12050 (iterator_use): Replace ptr with x and index.
12051 (attribute_use): Likewise.
12052 (apply_attribute_uses): Update calls to apply_iterator.
12053 (apply_iterators): Likewise. Update initialization of iterator_use.
12054 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
12055 and record_potential_iterator_use.
12056 (rtx_reader::read_rtx_operand): Likewise.
12058 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12059 Alan Hayward <alan.hayward@arm.com>
12060 David Sherwood <david.sherwood@arm.com>
12062 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
12065 2017-08-21 Richard Biener <rguenther@suse.de>
12067 PR middle-end/81884
12068 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
12069 at struct end conservatively when comparing common bases.
12071 2017-08-21 Richard Biener <rguenther@suse.de>
12073 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
12074 (mem_ref_in_stmt): Remove.
12075 (determine_max_movement): Use ref index to get at the reference.
12076 (invariantness_dom_walker::before_dom_children): Deal with
12077 lim data already initialized.
12078 (gather_mem_refs_stmt): Initialize lim data and record ref index.
12080 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12082 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
12083 (TARGET_ISA_ROUND): Ditto.
12084 (TARGET_ROUND): Ditto.
12085 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
12086 * config/i386/i386.md: Ditto.
12087 * config/i386/sse.md: Ditto.
12088 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
12089 with OPTION_MASK_ISA_SSE4_1.
12091 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12094 * doc/extend.texi (x86 Built-in Functions): Correct the name of
12095 __builtin_ia32_lzcnt_u16.
12097 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
12100 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
12101 (rs6000_set_current_function): Rewrite function to use it.
12103 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
12106 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
12107 and TYPE_WARN_IF_NOT_ALIGN.
12108 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
12109 (handle_warn_if_not_align): New.
12110 (place_union_field): Call handle_warn_if_not_align.
12111 (place_field): Call handle_warn_if_not_align.
12112 Copy TYPE_WARN_IF_NOT_ALIGN.
12113 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
12114 (layout_type): Likewise.
12115 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
12117 (tree_decl_common): Add warn_if_not_align.
12118 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
12119 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
12120 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
12121 (DECL_WARN_IF_NOT_ALIGN): Likewise.
12122 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
12123 * doc/extend.texi: Document warn_if_not_aligned attribute.
12124 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
12126 2017-08-17 Martin Liska <mliska@suse.cz>
12129 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
12130 (get_data_dependence): Use it as pointer type.
12131 (distribute_loop): Likewise.
12133 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12135 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
12136 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
12137 (p8_vmrgow_<mode>_direct): New define_insn.
12138 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
12139 handle endianness for vmrgew and vmrgow permute patterns.
12141 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
12143 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
12144 * config/rs6000/rs6000-cpus.def: Remove comment.
12145 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
12146 (POWERPC_MASKS): Likewise.
12147 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
12148 use of TARGET_VSX_TIMODE.
12149 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
12150 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
12151 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
12152 (rs6000_option_override_internal): Remove dead code.
12153 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
12154 (rs6000_legitimize_reload_address): Likewise.
12155 (rs6000_legitimate_address_p): Likewise.
12156 (rs6000_opt_masks): Delete "vsx-timode".
12157 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
12158 from function comment.
12159 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
12160 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
12161 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
12163 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
12164 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
12165 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
12166 (VSX_TI): Likewise.
12168 (define_peephole2): Likewise.
12170 2017-08-17 Martin Sebor <msebor@redhat.com>
12173 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
12174 past the end of an array.
12175 (test_pp_format): Add test cases.
12177 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
12179 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
12180 missing ECF_NOTHROW flags.
12182 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
12185 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
12186 operands residing in integer registers.
12187 (*vsx_le_perm_load_<mode>): Likewise.
12188 (*vsx_le_perm_store_<mode>): Likewise.
12189 (define_peephole2): Add peepholes to optimize the above.
12191 2017-08-17 Marek Polacek <polacek@redhat.com>
12193 PR middle-end/81814
12194 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
12195 to mimic what shorten_compare did. Change the return type to bool.
12196 (fold_cond_expr_with_comparison): Update call to
12197 operand_equal_for_comparison_p.
12198 (fold_ternary_loc): Likewise.
12200 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
12202 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
12204 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
12206 2017-08-17 Richard Biener <rguenther@suse.de>
12208 * tree-ssa-structalias.c (solve_graph): When propagating
12209 to successors update the graphs succ edges and avoid duplicate work.
12211 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
12214 * config/i386/i386.c (ix86_option_override_internal): Save target
12215 specific options after ix86_stack_protector_guard_reg was changed.
12217 2017-08-17 Richard Biener <rguenther@suse.de>
12219 PR tree-optimization/81827
12220 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
12221 (new_var_info): Initialize it conservatively.
12222 (get_call_vi): Mark register vars.
12223 (new_scalar_tmp_constraint_exp): Likewise.
12224 (handle_rhs_call): Likewise.
12225 (handle_const_call): Likewise.
12226 (create_function_info_for): Likewise.
12227 (solve_constraints): Sort varinfos to separate register from
12228 non-register vars to pack points-to solution bitmaps during
12231 2017-08-17 Marek Polacek <polacek@redhat.com>
12233 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
12235 2017-08-17 Richard Biener <rguenther@suse.de>
12237 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
12238 to true when overflow is undefined and we saturated the result.
12240 2017-08-17 Alan Modra <amodra@gmail.com>
12243 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
12244 Don't use store multiple if only one reg needs saving.
12245 (interesting_frame_related_regno): New function.
12246 (rs6000_frame_related): Don't emit frame info for regs that
12248 (rs6000_emit_epilogue): Likewise.
12250 2017-08-16 Nathan Sidwell <nathan@acm.org>
12252 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
12253 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
12254 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
12255 * tree.c (free_lang_data_in_type): Use else-if chain. Always
12256 clear TYPE_LANG_1. Remove obsolete member-function stripping.
12257 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
12258 (verify_type): Adjust for TYPE_BINFO move.
12259 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
12260 process TYPE_BINFO directly.
12261 (hash_tree): Likewise.
12262 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
12264 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
12267 2017-08-16 David Malcolm <dmalcolm@redhat.com>
12269 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
12271 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
12274 * config/i386/i386.md (*anddi_1_btr): Change predicates of
12275 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
12276 Add ix86_binary_operator_ok to insn constraint.
12277 (*iordi_1_bts): Ditto.
12278 (*xordi_1_btc): Ditto.
12279 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
12280 Update corresponding peephole2 pattern.
12284 2017-08-16 Bin Cheng <bin.cheng@arm.com>
12286 PR tree-optimization/81832
12287 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
12288 copy loop header which has IFN_LOOP_DIST_ALIAS call.
12290 2017-08-16 Marek Polacek <polacek@redhat.com>
12293 * fold-const.c (fold_indirect_ref_1): Restore original behavior
12294 regarding size_zero_node.
12296 2017-08-16 Martin Liska <mliska@suse.cz>
12299 * config.gcc: Respect previously set extra_objs in case
12302 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
12304 PR tree-optimization/81835
12305 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
12306 the phi SSA_NAME. Check that the condition in a COND_EXPR does
12307 not depend on the phi.
12309 2017-08-16 Alan Modra <amodra@gmail.com>
12311 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
12314 2017-08-16 Alan Modra <amodra@gmail.com>
12316 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
12317 (save_reg_p): ..into this. Update all callers.
12318 (first_reg_to_save): Simplify.
12320 2017-08-16 Alan Modra <amodra@gmail.com>
12322 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
12325 2017-08-15 Joseph Myers <joseph@codesourcery.com>
12329 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
12330 constant count if that count is less than 32.
12332 2017-08-15 Nathan Sidwell <nathan@acm.org>
12334 * gcc.c (execute): Emit friendlier message if inferior is killed
12335 by an external cause.
12337 2017-08-15 Richard Biener <rguenther@suse.de>
12339 PR tree-optimization/81790
12340 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
12341 CONSTRUCTORs from simplifying and VN.
12343 2017-08-14 Martin Sebor <msebor@redhat.com>
12345 * builtin-attrs.def: Add comments.
12347 2017-08-14 Martin Sebor <msebor@redhat.com>
12350 * doc/extend.texi (attribute nonstring): Document new attribute.
12352 2017-08-14 Martin Sebor <msebor@redhat.com>
12355 * tree-diagnostic.c (default_tree_printer): Handle %G.
12356 * gimple-pretty-print.h (percent_G_format): Declare new function.
12357 * gimple-pretty-print.c (percent_G_format): Define.
12358 * tree-pretty-print.c (percent_K_format): Add argument.
12360 2017-08-14 Martin Sebor <msebor@redhat.com>
12362 PR translation/79998
12363 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
12364 Remove a stray space.
12366 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
12369 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
12370 (*iordi_1_bts): Ditto.
12371 (*xordi_1_btc): Ditto.
12373 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12376 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
12377 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12379 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
12380 quoted strings, and make more translator-friendly.
12381 (darwin_rs6000_override_options): Likewise.
12382 (rs6000_option_override_internal): Likewise.
12383 (rs6000_return_in_memory): Fix overlong line.
12384 (init_cmulative_args): Use quoted strings, and make more
12385 translator-friendly.
12386 (rs6000_pass_by_reference): Fix overlong line.
12387 (def_builtin): Use quoted strings.
12388 (altivec_expand_predicate_builtin): Use quoted strings, and make
12389 more translator-friendly.
12390 (htm_expand_builtin): Use quoted strings.
12391 (cpu_expand_builtin): Use quoted strings, and make more
12392 translator-friendly.
12393 (altivec_expand_builtin): Likewise.
12394 (paired_expand_predicate_builtin): Likewise.
12395 (rs6000_invalid_builtin): Likewise.
12396 (builtin_function_type): Use quoted strings.
12397 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
12398 more translator-friendly.
12399 (rs6000_trampoline_init): Likewise.
12400 (rs6000_handle_altivec_attribute): Likewise.
12401 (rs6000_inner_target_options): Use quoted strings.
12402 (rs6000_disable_incompatible_switches): Likewise.
12403 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
12404 strings, and make more translator-friendly.
12405 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
12407 2017-08-14 Bin Cheng <bin.cheng@arm.com>
12409 PR tree-optimization/81799
12410 * tree-loop-distribution.c (version_loop_by_alias_check): Force
12411 cond_expr to simple gimple operand.
12413 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
12415 PR middle-end/46932
12416 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
12418 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
12422 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
12423 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
12424 TARGET_GASISR_PROLOGUES.
12425 * config/avr/avr.c (avr_option_override): Same.
12426 (avr_pass_pre_proep::execute): Same.
12428 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
12431 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
12432 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
12434 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
12436 * config/i386/i386.md (*load_tp_<mode>): Redefine as
12437 define_insn_and_split. Split to a memory load from 0 in
12438 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
12439 using PTR mode iterator.
12440 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
12441 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
12442 (*add_tp_<mode>): Redefine as define_insn_and_split.
12443 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
12444 address space. Merge with *add_tp_x32 using PTR mode iterator.
12445 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
12446 Split to an add with a memory load from 0 in
12447 DEFAULT_TLS_SEG_REG address space.
12449 2017-08-12 Andrew Pinski <apinski@cavium.com>
12451 * config/aarch64/aarch64-option-extensions.def (rdma):
12452 Fix feature string to what Linux prints out in /proc/cpuinfo.
12454 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
12457 * dwarf2out.c (modified_type_die): For C typedef types that have
12458 an ultimate origin, process the ultimate origin instead of the
12460 (gen_typedef_die): Assert that input DECLs have no ultimate
12462 (gen_type_die_with_usage): For typedef variants that have an
12463 ultimate origin, just call gen_decl_die on the original DECL.
12464 (process_scope_var): Avoid creating DIEs for local typedefs and
12465 concrete static variables.
12467 2017-08-12 Alan Modra <amodra@gmail.com>
12471 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
12472 match gnu-user.h startfile.
12473 (ENDFILE_LINUX_SPEC): Similarly.
12475 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
12478 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
12480 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
12482 2017-08-11 Tamar Christina <tamar.christina@arm.com>
12483 * config/aarch64/aarch64.md (mov<mode>): Change.
12484 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
12485 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
12486 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
12488 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
12490 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
12491 for storage order barriers.
12493 2017-08-11 Martin Liska <mliska@suse.cz>
12496 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
12497 variables of void type.
12499 2017-08-11 Martin Liska <mliska@suse.cz>
12501 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
12502 TARGET_SUPPORTS_ALIASES.
12503 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
12504 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
12505 (optimize_weakref): Likewise.
12506 * symtab.c (symtab_node::noninterposable_alias): Likewise.
12507 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
12508 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
12510 2017-08-11 Martin Liska <mliska@suse.cz>
12513 * config/i386/i386.c (make_resolver_func): Do complete
12514 refactoring of the function.
12516 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
12519 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
12520 * config/i386/i386.c (ix86_stack_protect_guard): Use
12521 ix86_stack_protect_guard_symbol_str to generate varible declaration.
12522 * doc/invoke.texi (x86 Options): Document
12523 -mstack-protector-guard-symbol= option.
12525 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
12527 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
12528 * config/i386/i386.c (ix86_split_stack_guard): New function.
12529 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
12530 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
12531 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
12532 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
12533 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
12534 (split_stack_space_check): Call ix86_split_stack_guard.
12536 2017-08-10 Martin Sebor <msebor@redhat.com>
12538 * print-tree.c (print_node): Print location using the established
12540 Replace spaces with colons.
12541 (debug_raw, debug): Ditto.
12543 2017-08-10 Martin Sebor <msebor@redhat.com>
12546 * pretty-print.c (pp_format): Correct the handling of %s precision.
12548 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
12551 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
12553 (ix86_finalize_stack_frame_flags): This. Also clear
12554 frame_pointer_needed if -fno-omit-frame-pointer is used without
12556 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
12557 with ix86_finalize_stack_frame_flags.
12558 (ix86_expand_epilogue): Likewise.
12559 (ix86_expand_split_stack_prologue): Likewise.
12560 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
12562 2017-08-10 Martin Liska <mliska@suse.cz>
12565 * c-attribs.c (handle_target_attribute):
12566 Report warning for an empty string argument of target attribute.
12568 2017-08-09 Jakub Jelinek <jakub@redhat.com>
12571 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
12573 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
12574 or DECL_NONLOCAL labels.
12575 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
12576 or DECL_NONLOCAL labels here.
12578 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
12580 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
12581 to indicate when early gimple folding has been disabled.
12582 (rs6000_gimple_fold_builtin): Add debug content.
12583 (rs6000_invalid_builtin): Fix whitespace.
12584 (rs6000_expand_builtin): Fix whitespace.
12585 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
12587 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
12590 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
12591 SAVE_MULTIPLE if not all the registers that saves, should be saved.
12593 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
12595 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
12596 (qdf24xx): Likewise.
12597 * config/aarch64/aarch64.md: Include falkor.md.
12598 * config/aarch64/falkor.md: New.
12600 2017-08-09 Marek Polacek <polacek@redhat.com>
12603 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
12604 * diagnostic.c (emit_diagnostic): Add a comment.
12605 (emit_diagnostic_valist): New function.
12607 2017-08-09 Marek Polacek <polacek@redhat.com>
12610 * input.c (make_location): New overload.
12611 * input.h (make_location): Declare.
12613 2017-08-08 Alan Modra <amodra@gmail.com>
12614 H.J. Lu <hongjiu.lu@intel.com>
12617 * gcc.c (NO_PIE_SPEC): Delete.
12618 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
12620 (LINK_PIE_SPEC): ..to here.
12621 (LINK_COMMAND_SPEC): Support -no-pie.
12622 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
12623 chain of crtbegin*.o selection, update for PIE_SPEC changes and
12625 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
12626 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
12627 (ENDFILE_CRTEND_SPEC): Similarly.
12629 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
12632 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
12633 (mstack-protector-guard-offset=): Ditto.
12634 * config/i386/i386.c (ix86_option_override): Handle
12635 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
12637 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
12638 ix86_stack_protect_guard_offset variables.
12639 (TARGET_STACK_PROTECT_GUARD): Always define.
12640 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
12641 and -mstack-protector-guard-offset= options.
12643 2017-08-08 Bin Cheng <bin.cheng@arm.com>
12645 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
12646 boundary case for the last candidate.
12648 2017-08-08 Bin Cheng <bin.cheng@arm.com>
12650 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
12651 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
12653 2017-08-08 Tamar Christina <tamar.christina@arm.com>
12655 PR middle-end/19706
12656 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
12657 * config/aarch64/aarch64-builtins.c
12658 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
12659 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
12660 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
12662 2017-08-08 Tamar Christina <tamar.christina@arm.com>
12663 Andrew Pinski <pinskia@gmail.com>
12665 PR middle-end/19706
12666 * internal-fn.def (XORSIGN): New.
12667 * optabs.def (xorsign_optab): New.
12668 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
12669 (convert_expand_mult_copysign): New.
12670 (pass_optimize_widening_mul::execute): Call
12671 convert_expand_mult_copysign.
12673 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12675 PR tree-optimization/81354
12676 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
12677 Insert on edges rather than explicitly creating landing pads.
12678 (analyze_candidates_and_replace): Commit edge inserts.
12680 2017-08-08 Richard Biener <rguenther@suse.de>
12682 PR middle-end/81719
12683 * tree-ssa-loop-niter.c: Include tree-dfa.h.
12684 (expand_simple_operations): Also look through ADDR_EXPRs with
12685 MEM_REF bases treating them as POINTER_PLUS_EXPR.
12687 2017-08-08 Richard Biener <rguenther@suse.de>
12689 PR tree-optimization/81723
12690 * tree-vect-slp.c (struct bst_traits): New hash traits.
12691 (bst_fail): New global.
12692 (vect_build_slp_tree_2): New worker, split out from ...
12693 (vect_build_slp_tree): ... this now wrapping it with using
12694 bst_fail set to cache SLP tree build fails. Properly handle
12696 (vect_analyze_slp_instance): Allocate and free bst_fail.
12698 2017-08-08 Martin Liska <mliska@suse.cz>
12701 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
12702 LABEL_DECLs that can be from a different function.
12704 2017-08-08 Bin Cheng <bin.cheng@arm.com>
12706 PR tree-optimization/81744
12707 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
12708 loop's number of iterations.
12710 2017-08-08 Martin Liska <mliska@suse.cz>
12712 * asan.c: Include header files.
12713 * attribs.c (build_decl_attribute_variant): New function moved
12715 (build_type_attribute_qual_variant): Likewise.
12716 (cmp_attrib_identifiers): Likewise.
12717 (simple_cst_list_equal): Likewise.
12718 (omp_declare_simd_clauses_equal): Likewise.
12719 (attribute_value_equal): Likewise.
12720 (comp_type_attributes): Likewise.
12721 (build_type_attribute_variant): Likewise.
12722 (lookup_ident_attribute): Likewise.
12723 (remove_attribute): Likewise.
12724 (merge_attributes): Likewise.
12725 (merge_type_attributes): Likewise.
12726 (merge_decl_attributes): Likewise.
12727 (merge_dllimport_decl_attributes): Likewise.
12728 (handle_dll_attribute): Likewise.
12729 (attribute_list_equal): Likewise.
12730 (attribute_list_contained): Likewise.
12731 * attribs.h (lookup_attribute): New function moved from tree.[ch].
12732 (lookup_attribute_by_prefix): Likewise.
12733 * bb-reorder.c: Include header files.
12734 * builtins.c: Likewise.
12735 * calls.c: Likewise.
12736 * cfgexpand.c: Likewise.
12737 * cgraph.c: Likewise.
12738 * cgraphunit.c: Likewise.
12739 * convert.c: Likewise.
12740 * dwarf2out.c: Likewise.
12741 * final.c: Likewise.
12742 * fold-const.c: Likewise.
12743 * function.c: Likewise.
12744 * gimple-expr.c: Likewise.
12745 * gimple-fold.c: Likewise.
12746 * gimple-pretty-print.c: Likewise.
12747 * gimple.c: Likewise.
12748 * gimplify.c: Likewise.
12749 * hsa-common.c: Likewise.
12750 * hsa-gen.c: Likewise.
12751 * internal-fn.c: Likewise.
12752 * ipa-chkp.c: Likewise.
12753 * ipa-cp.c: Likewise.
12754 * ipa-devirt.c: Likewise.
12755 * ipa-fnsummary.c: Likewise.
12756 * ipa-inline.c: Likewise.
12757 * ipa-visibility.c: Likewise.
12759 * lto-cgraph.c: Likewise.
12760 * omp-expand.c: Likewise.
12761 * omp-general.c: Likewise.
12762 * omp-low.c: Likewise.
12763 * omp-offload.c: Likewise.
12764 * omp-simd-clone.c: Likewise.
12765 * opts-global.c: Likewise.
12766 * passes.c: Likewise.
12767 * predict.c: Likewise.
12768 * sancov.c: Likewise.
12769 * sanopt.c: Likewise.
12770 * symtab.c: Likewise.
12771 * toplev.c: Likewise.
12772 * trans-mem.c: Likewise.
12773 * tree-chkp.c: Likewise.
12774 * tree-eh.c: Likewise.
12775 * tree-into-ssa.c: Likewise.
12776 * tree-object-size.c: Likewise.
12777 * tree-parloops.c: Likewise.
12778 * tree-profile.c: Likewise.
12779 * tree-ssa-ccp.c: Likewise.
12780 * tree-ssa-live.c: Likewise.
12781 * tree-ssa-loop.c: Likewise.
12782 * tree-ssa-sccvn.c: Likewise.
12783 * tree-ssa-structalias.c: Likewise.
12784 * tree-ssa.c: Likewise.
12785 * tree-streamer-in.c: Likewise.
12786 * tree-vectorizer.c: Likewise.
12787 * tree-vrp.c: Likewise.
12788 * tsan.c: Likewise.
12789 * ubsan.c: Likewise.
12790 * varasm.c: Likewise.
12791 * varpool.c: Likewise.
12792 * tree.c: Remove functions moved to attribs.[ch].
12793 * tree.h: Likewise.
12794 * config/aarch64/aarch64.c: Add attrs.h header file.
12795 * config/alpha/alpha.c: Likewise.
12796 * config/arc/arc.c: Likewise.
12797 * config/arm/arm.c: Likewise.
12798 * config/avr/avr.c: Likewise.
12799 * config/bfin/bfin.c: Likewise.
12800 * config/c6x/c6x.c: Likewise.
12801 * config/cr16/cr16.c: Likewise.
12802 * config/cris/cris.c: Likewise.
12803 * config/darwin.c: Likewise.
12804 * config/epiphany/epiphany.c: Likewise.
12805 * config/fr30/fr30.c: Likewise.
12806 * config/frv/frv.c: Likewise.
12807 * config/ft32/ft32.c: Likewise.
12808 * config/h8300/h8300.c: Likewise.
12809 * config/i386/winnt.c: Likewise.
12810 * config/ia64/ia64.c: Likewise.
12811 * config/iq2000/iq2000.c: Likewise.
12812 * config/lm32/lm32.c: Likewise.
12813 * config/m32c/m32c.c: Likewise.
12814 * config/m32r/m32r.c: Likewise.
12815 * config/m68k/m68k.c: Likewise.
12816 * config/mcore/mcore.c: Likewise.
12817 * config/microblaze/microblaze.c: Likewise.
12818 * config/mips/mips.c: Likewise.
12819 * config/mmix/mmix.c: Likewise.
12820 * config/mn10300/mn10300.c: Likewise.
12821 * config/moxie/moxie.c: Likewise.
12822 * config/msp430/msp430.c: Likewise.
12823 * config/nds32/nds32-isr.c: Likewise.
12824 * config/nds32/nds32.c: Likewise.
12825 * config/nios2/nios2.c: Likewise.
12826 * config/nvptx/nvptx.c: Likewise.
12827 * config/pa/pa.c: Likewise.
12828 * config/pdp11/pdp11.c: Likewise.
12829 * config/powerpcspe/powerpcspe.c: Likewise.
12830 * config/riscv/riscv.c: Likewise.
12831 * config/rl78/rl78.c: Likewise.
12832 * config/rx/rx.c: Likewise.
12833 * config/s390/s390.c: Likewise.
12834 * config/sh/sh.c: Likewise.
12835 * config/sol2.c: Likewise.
12836 * config/sparc/sparc.c: Likewise.
12837 * config/spu/spu.c: Likewise.
12838 * config/stormy16/stormy16.c: Likewise.
12839 * config/tilegx/tilegx.c: Likewise.
12840 * config/tilepro/tilepro.c: Likewise.
12841 * config/v850/v850.c: Likewise.
12842 * config/vax/vax.c: Likewise.
12843 * config/visium/visium.c: Likewise.
12844 * config/xtensa/xtensa.c: Likewise.
12846 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
12849 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
12850 constraints since the -mupper-regs-* switches have been
12852 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
12853 into a vector from a double word element that was extracted from
12854 another vector, and eliminate extra XXPERMDI instructions.
12855 (vsx_concat_<mode>_2): Likewise.
12856 (vsx_concat_<mode>_3): Likewise.
12857 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
12858 concat to allow optimizing inserts from previous extracts.
12860 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
12862 * config/i386/i386.c (ix86_stack_protect_guard): Generate
12863 memory reference to a SSP offset in TLS address space.
12864 (ix86_print_operand) <case '@'>: Remove.
12865 (ix86_print_operand_punct_valid_p): Remove '@' code.
12866 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
12867 UNSPEC_SP_TLS_TEST.
12868 (stack_tls_protect_set_<mode>): Remove.
12869 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
12870 (stack_tls_protect_test_<mode>): Remove.
12871 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
12873 2017-08-07 Olivier Hainque <hainque@adacore.com>
12876 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
12878 2017-08-07 Douglas Rupp <rupp@adacore.com>
12880 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
12881 variable was referenced as multidir in command.
12883 2017-08-07 Jakub Jelinek <jakub@redhat.com>
12886 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
12889 2017-08-07 Martin Liska <mliska@suse.cz>
12891 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
12892 * config/rl78/rl78.c: Add include of attribs.h.
12893 * config/sh/sh.c: Likewise.
12894 * config/v850/v850.c: Likewise.
12896 2017-08-07 Tom de Vries <tom@codesourcery.com>
12898 PR middle-end/78266
12899 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
12901 2017-08-07 Martin Liska <mliska@suse.cz>
12903 * config/mips/mips.c: Include attribs.h.
12905 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
12908 * doc/invoke.texi: Document change in behvaior for -Ofast for
12911 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
12913 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
12915 (aarch64_pop_regs): Likewise.
12916 (aarch64_gen_load_pair): Likewise.
12917 (aarch64_save_callee_saves): Likewise.
12918 (aarch64_restore_callee_saves): Likewise.
12920 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
12922 * config/i386/i386.c: Revert the last change.
12924 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
12927 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
12929 (ix86_finalize_stack_frame_flags): This. Also clear
12930 frame_pointer_needed if -fno-omit-frame-pointer is used without
12932 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
12933 with ix86_finalize_stack_frame_flags.
12934 (ix86_expand_epilogue): Likewise.
12935 (ix86_expand_split_stack_prologue): Likewise.
12937 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
12940 * config/i386/i386.c (get_builtin_code_for_version): Set priority
12941 to P_AES for Westmere.
12943 2017-08-07 Jonathan Yong <10walls@gmail.com>
12945 * config/i386/mingw.opt (fset-stack-executable): Removed.
12946 * config/i386/cygming.opt (fset-stack-executable): Moved
12948 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
12950 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
12952 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
12954 2017-08-07 Marek Polacek <polacek@redhat.com>
12956 PR middle-end/81737
12957 * fold-const.c (fold_indirect_ref_1): Check type_domain.
12959 2017-08-07 Martin Liska <mliska@suse.cz>
12961 * attribs.h (canonicalize_attr_name): New function.
12962 (cmp_attribs): Move from c-format.c and adjusted.
12963 (is_attribute_p): Moved from tree.h.
12964 * tree-inline.c: Add new includes.
12965 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
12966 (private_is_attribute_p): Remove.
12967 (private_lookup_attribute): Likewise.
12968 (private_lookup_attribute_by_prefix): Simplify.
12969 (remove_attribute): Use is_attribute_p.
12970 * tree.h: Remove removed declarations.
12972 2017-08-07 Jakub Jelinek <jakub@redhat.com>
12974 PR middle-end/81698
12975 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
12976 instead of computing it in the function. Formatting fix.
12977 (expand_case): Don't rely on default_edge being the first edge,
12978 clear it if removing it, pass default_edge to
12979 emit_case_dispatch_table.
12980 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
12983 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
12985 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
12986 insn in the function, emit NOP after the insn.
12988 2017-08-06 Tom de Vries <tom@codesourcery.com>
12990 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
12993 2017-08-06 Tom de Vries <tom@codesourcery.com>
12995 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
12998 2017-08-04 Yury Gribov <tetra2005@gmail.com>
13000 PR tree-optimization/57371
13001 * match.pd: New pattern.
13003 2017-08-04 Marek Polacek <polacek@redhat.com>
13005 PR middle-end/81695
13006 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
13007 perform the computation in offset_int.
13009 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13011 PR tree-optimization/81136
13012 * tree-vectorizer.h: Include tree-hash-traits.h.
13013 (vec_base_alignments): New typedef.
13014 (vec_info): Add a base_alignments field.
13015 (vect_record_base_alignments): Declare.
13016 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
13018 (DR_IS_CONDITIONAL_IN_STMT): New macro.
13019 (create_data_ref): Add an is_conditional_in_stmt argument.
13020 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
13021 the is_conditional_in_stmt field.
13022 (data_ref_loc): Add an is_conditional_in_stmt field.
13023 (get_references_in_stmt): Set the is_conditional_in_stmt field.
13024 (find_data_references_in_stmt): Update call to create_data_ref.
13025 (graphite_find_data_references_in_stmt): Likewise.
13026 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
13027 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
13028 (vect_record_base_alignment): New function.
13029 (vect_record_base_alignments): Likewise.
13030 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
13031 for nested statements even if we fail to compute a misalignment.
13032 Use pooled base alignments for unconditional references.
13033 (vect_find_same_alignment_drs): Compare base addresses instead
13035 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
13036 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
13038 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13040 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
13041 Add an explicit name for the enum. Use auto_vec for slp_instances
13042 and grouped_stores.
13043 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
13045 (_bb_vec_info): Add a constructor and destructor.
13046 (vinfo_for_stmt): Return NULL for uids of -1 as well.
13047 (destroy_loop_vec_info): Delete.
13048 (vect_destroy_datarefs): Likewise.
13049 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
13050 (vec_info::vec_info): New function.
13051 (vec_info::~vec_info): Likewise.
13052 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
13053 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
13054 destroy_loop_vec_info.
13055 * tree-vect-loop.c (new_loop_vec_info): Replace with...
13056 (_loop_vec_info::_loop_vec_info): ...this.
13057 (destroy_loop_vec_info): Replace with...
13058 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
13059 the stmt_vec_infos. Leave handling of vec_info information to its
13060 destructor. Remove explicit vector releases.
13061 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
13062 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
13063 * tree-vect-slp.c (new_bb_vec_info): Replace with...
13064 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
13065 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
13066 (destroy_bb_vec_info): Replace with...
13067 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
13068 information to its destructor.
13069 (vect_slp_analyze_bb_1): Use new and delete instead of
13070 new_bb_vec_info and destroy_bb_vec_info.
13071 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
13074 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13076 * tree-data-ref.h (subscript): Add access_fn field.
13077 (data_dependence_relation): Add could_be_independent_p.
13078 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
13079 (same_access_functions): Move to tree-data-ref.c.
13080 * tree-data-ref.c (ref_contains_union_access_p): New function.
13081 (access_fn_component_p): Likewise.
13082 (access_fn_components_comparable_p): Likewise.
13083 (dr_analyze_indices): Add a reference to access_fn_component_p.
13084 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
13086 (constant_access_functions): Likewise.
13087 (add_other_self_distances): Likewise.
13088 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
13089 (initialize_data_dependence_relation): Use XCNEW and remove
13090 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
13091 of access functions that have the same type. Allow the
13092 subsequence to end with different bases in some circumstances.
13093 Record the chosen access functions in SUB_ACCESS_FN.
13094 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
13095 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
13096 (subscript_dependence_tester_1): Likewise dra and drb.
13097 (build_classic_dist_vector): Update calls accordingly.
13098 (subscript_dependence_tester): Likewise.
13099 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
13100 DDR_COULD_BE_INDEPENDENT_P.
13101 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
13102 comp_alias_ddrs instead of may_alias_ddrs.
13103 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
13105 (vect_analyze_data_ref_dependence): Use it if
13106 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
13107 distance vectors if that fails.
13108 (dependence_distance_ge_vf): New function.
13109 (vect_prune_runtime_alias_test_list): Use it. Don't clear
13110 LOOP_VINFO_MAY_ALIAS_DDRS.
13112 2017-08-04 Richard Biener <rguenther@suse.de>
13114 PR middle-end/81705
13115 * fold-const.c (fold_binary_loc): Properly restrict
13116 minus_var0 && minus_var1 case when associating undefined overflow
13119 2017-08-04 Tom de Vries <tom@codesourcery.com>
13121 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
13123 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13125 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13126 Don't start diagnostic messages with a capital letter.
13127 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13129 (rs6000_invalid_builtin): Likewise.
13130 (rs6000_trampoline_init): Likewise.
13132 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13135 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
13136 after setting changeable df flags.
13138 2017-08-03 Richard Biener <rguenther@suse.de>
13140 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
13141 up if the use is in USE - X.
13143 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
13145 * toplev.c (dumpfile.h): New include.
13146 (internal_error_reentered): New static function. Use it...
13147 (internal_error_function): ...here to handle reentered internal_error.
13149 2017-08-03 Richard Biener <rguenther@suse.de>
13151 PR middle-end/81148
13152 * fold-const.c (split_tree): Add minus_var and minus_con
13153 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
13154 here but always use minus_*.
13155 (associate_trees): Assert we never associate with MINUS_EXPR
13156 and NULL first operand. Do not recurse for PLUS_EXPR operands
13157 when associating as MINUS_EXPR either.
13158 (fold_binary_loc): Track minus_var and minus_con.
13160 2017-08-03 Tom de Vries <tom@codesourcery.com>
13163 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
13164 ACCEL_COMPILER, apply finish_options on
13165 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
13167 2017-08-03 Tom de Vries <tom@codesourcery.com>
13170 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
13171 function_entry_patch_area_size > 0.
13173 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13176 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
13177 instead of 10??LU, perform unit multiplication in wide_int,
13178 don't change alloc_object_size_limit if the limit is larger
13181 PR tree-optimization/81655
13182 PR tree-optimization/81588
13183 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
13184 the case when ranges[i].low and high are 1 for unsigned type with
13187 PR middle-end/81052
13188 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
13189 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
13191 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13193 * tree-vrp.h: Add include guard.
13195 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
13198 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
13199 (ud2): New insn pattern.
13200 * config/i386/i386.c (ix86_expand_epilogue):
13201 For naked functions, generate ud2 instead of trap insn.
13203 2017-08-02 Marek Polacek <polacek@redhat.com>
13206 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
13208 2017-08-02 Tom de Vries <tom@codesourcery.com>
13209 Cesar Philippidis <cesar@codesourcery.com>
13211 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
13212 Add missing edge probabilities.
13214 2017-08-02 Tamar Christina <tamar.christina@arm.com>
13216 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
13217 Correct endianness.
13219 2017-08-02 Jakub Jelinek <jakub@redhat.com>
13221 PR middle-end/79499
13222 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
13223 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
13224 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
13225 prologue_seq sequences - if any.
13227 2017-08-02 Richard Biener <rguenther@suse.de>
13229 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
13230 via vectors if supported, integer extracts via punning if supported
13231 or otherwise vector extracts.
13233 2017-08-02 Richard Biener <rguenther@suse.de>
13235 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
13237 (bitmap_insert_into_set): ... this.
13239 2017-08-02 Richard Biener <rguenther@suse.de>
13241 PR tree-optimization/81633
13243 2015-08-17 Alan Hayward <alan.hayward@arm.com>
13245 PR tree-optimization/71752
13246 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
13248 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
13250 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
13251 (machine_function::call_ms2sysv_pad_out): Remove field.
13252 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
13253 (ix86_compute_frame_layout): Likewise.
13255 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
13258 * config/i386/i386.c (ix86_set_func_type): Disallow naked
13259 attribute with interrupt attribute.
13261 2017-08-01 Andrew Pinski <apinski@cavium.com>
13263 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
13264 BIT_INSERT_EXPR's operand 1
13265 to see if the types precision matches.
13267 2017-08-01 Martin Liska <mliska@suse.cz>
13269 PR middle-end/70140
13270 * builtins.c (expand_builtin_memcpy_args): Remove.
13271 (expand_builtin_memcpy): Call newly added function
13272 expand_builtin_memory_copy_args.
13273 (expand_builtin_memcpy_with_bounds): Likewise.
13274 (expand_builtin_mempcpy): Remove last argument.
13275 (expand_builtin_mempcpy_with_bounds): Likewise.
13276 (expand_builtin_memory_copy_args): New function created from
13277 expand_builtin_mempcpy_args with small modifications.
13278 (expand_builtin_mempcpy_args): Remove.
13279 (expand_builtin_stpcpy): Remove unused argument.
13280 (expand_builtin): Likewise.
13281 (expand_builtin_with_bounds): Likewise.
13283 2017-08-01 Martin Liska <mliska@suse.cz>
13286 Make mempcpy more optimal (PR middle-end/70140).
13288 2017-08-01 Jakub Jelinek <jakub@redhat.com>
13291 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
13292 __builtin_vec_cmpne verify both arguments are compatible vectors
13293 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
13294 verify arg1_type is a pointer or array type. For __builtin_vec_st,
13295 move computation of aligned to after checking the argument types.
13299 * config/rs6000/vsx.md (vextract_fp_from_shorth,
13300 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
13303 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
13304 Doug Rupp <rupp@adacore.com>
13305 Olivier Hainque <hainque@adacore.com>
13307 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
13308 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
13310 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
13311 for TARGET_OS_CPP_BUILTIN.
13312 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
13313 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
13315 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
13316 passing required abi options to the assembler for EABI configurations.
13317 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
13318 of .text.hot and .text.unlikely sections for kernel modules when
13319 using ARM style exceptions.
13320 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
13321 options. Add EXTRA_CC1_SPEC.
13322 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
13324 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
13326 (ARM_TARGET2_DWARF_FORMAT): Define.
13327 * config/arm/t-vxworks: Adjust multilib control to removal of the
13328 Diab command line options.
13330 2017-08-01 Martin Liska <mliska@suse.cz>
13332 PR gcov-profile/81561
13333 * gcov.c (unblock): Make unblocking safe as we need to preserve
13334 index correspondence of blocks and block_lists.
13336 2017-08-01 Richard Biener <rguenther@suse.de>
13338 PR tree-optimization/81181
13339 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
13340 (compute_antic): ... end of iteration here.
13342 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
13344 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
13345 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
13346 (ftree-slp-vectorize): Likewise.
13347 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
13348 can no longer be set independent of flag_tree_loop_vectorize.
13349 * omp-general.c (emp_max_vf): Likewise.
13350 * opts.c (enable_fdo_optimizations): Remove references to
13351 flag_tree_vectorize, these are now implicit.
13352 (common_handle_option): Remove handling for OPT_ftree_vectorize,
13353 and leave it for the options machinery.
13355 2017-08-01 Martin Liska <mliska@suse.cz>
13357 PR middle-end/70140
13358 * builtins.c (expand_builtin_memcpy_args): Remove.
13359 (expand_builtin_memcpy): Call newly added function
13360 expand_builtin_memory_copy_args.
13361 (expand_builtin_memcpy_with_bounds): Likewise.
13362 (expand_builtin_mempcpy): Remove last argument.
13363 (expand_builtin_mempcpy_with_bounds): Likewise.
13364 (expand_builtin_memory_copy_args): New function created from
13365 expand_builtin_mempcpy_args with small modifications.
13366 (expand_builtin_mempcpy_args): Remove.
13367 (expand_builtin_stpcpy): Remove unused argument.
13368 (expand_builtin): Likewise.
13369 (expand_builtin_with_bounds): Likewise.
13371 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
13374 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
13375 print "ds:" only for immediates in generic address space.
13377 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
13380 * config/i386/i386.c (ix86_funciton_naked): New prototype.
13381 (ix86_function_ok_for_sibcall): Return false for naked functions.
13383 2017-08-01 Richard Biener <rguenther@suse.de>
13385 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
13386 (compute_antic): Seed worklist with exit block predecessors.
13387 * cfganal.c (dfs_find_deadend): For a cycle return the source
13388 of the edge closing it.
13390 2017-08-01 Tamar Christina <tamar.christina@arm.com>
13392 * config/aarch64/aarch64.c
13393 (aarch64_can_const_movi_rtx_p): Move 0 check.
13395 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13397 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
13398 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
13400 * match.pd: Ditto in address comparison pattern.
13402 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13404 PR tree-optimization/81627
13405 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
13406 closed ssa form for store-store chain.
13408 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13410 PR tree-optimization/81620
13411 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
13412 for store-store chain.
13414 2017-08-01 Jakub Jelinek <jakub@redhat.com>
13416 PR tree-optimization/81588
13417 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
13418 ranges[i].in_p, invert comparison code ccode. For >/>=,
13419 swap rhs1 and rhs2 and comparison code unconditionally,
13420 for </<= don't do that. Don't swap rhs1/rhs2 again if
13421 ranges[i].in_p, instead invert comparison code ccode if
13422 opcode or oe->rank is BIT_IOR_EXPR.
13425 * optabs.def (vec_extract_optab, vec_init_optab): Change from
13426 a direct optab to conversion optab.
13427 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
13428 with GET_MODE_INNER as last argument instead of optab_handler.
13429 * expmed.c (extract_bit_field_1): Likewise. Use vector from
13430 vector extraction if possible and optab is available.
13431 * expr.c (store_constructor): Use convert_optab_handler instead
13432 of optab_handler. Use vector initialization from smaller
13433 vectors if possible and optab is available.
13434 * tree-vect-stmts.c (vectorizable_load): Likewise.
13435 * doc/md.texi (vec_extract, vec_init): Document that the optabs
13436 now have two modes.
13437 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
13438 of vec_init from half-sized vectors with the same element mode.
13439 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
13440 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
13441 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
13442 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
13443 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
13444 after mode in gen_vec_extract* calls.
13445 (vec_extract<mode>): Renamed to ...
13446 (vec_extract<mode><ssescalarmodelower>): ... this.
13447 (vec_extract<mode><ssehalfvecmodelower>): New expander.
13448 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
13449 element mode after mode in gen_vec_init* calls.
13450 (VEC_INIT_HALF_MODE): New mode iterator.
13451 (vec_init<mode>): Renamed to ...
13452 (vec_init<mode><ssescalarmodelower>): ... this.
13453 (vec_init<mode><ssehalfvecmodelower>): New expander.
13454 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
13455 (vec_extractv2sfsf): ... this.
13456 (vec_initv2sf): Renamed to ...
13457 (vec_initv2sfsf): ... this.
13458 (vec_extractv2si): Renamed to ...
13459 (vec_extractv2sisi): ... this.
13460 (vec_initv2si): Renamed to ...
13461 (vec_initv2sisi): ... this.
13462 (vec_extractv4hi): Renamed to ...
13463 (vec_extractv4hihi): ... this.
13464 (vec_initv4hi): Renamed to ...
13465 (vec_initv4hihi): ... this.
13466 (vec_extractv8qi): Renamed to ...
13467 (vec_extractv8qiqi): ... this.
13468 (vec_initv8qi): Renamed to ...
13469 (vec_initv8qiqi): ... this.
13470 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
13471 (vec_init<mode>): Renamed to ...
13472 (vec_init<mode><VEC_base_l>): ... this.
13473 (vec_extract<mode>): Renamed to ...
13474 (vec_extract<mode><VEC_base_l>): ... this.
13475 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
13476 (vec_initv2sfsf): ... this.
13477 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
13478 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
13479 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
13480 element mode after mode in gen_vec_init* calls.
13481 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
13482 (vec_init<mode><Vel>): ... this.
13483 (vec_extract<mode>): Renamed to ...
13484 (vec_extract<mode><Vel>): ... this.
13485 * config/aarch64/iterators.md (Vel): New mode attribute.
13486 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
13487 Add element mode after mode in gen_vec_extract* calls.
13488 * config/s390/vector.md (non_vec_l): New mode attribute.
13489 (vec_extract<mode>): Renamed to ...
13490 (vec_extract<mode><non_vec_l>): ... this.
13491 (vec_init<mode>): Renamed to ...
13492 (vec_init<mode><non_vec_l>): ... this.
13493 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
13494 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
13496 * config/arm/iterators.md (V_elem_l): New mode attribute.
13497 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
13498 (vec_extract<mode><V_elem_l>): ... this.
13499 (vec_extractv2di): Renamed to ...
13500 (vec_extractv2didi): ... this.
13501 (vec_init<mode>): Renamed to ...
13502 (vec_init<mode><V_elem_l>): ... this.
13503 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
13504 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
13505 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
13506 Add element mode after gen_vec_extract* calls.
13507 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
13508 (vec_init<mode><unitmode>): ... this.
13509 (vec_extract<mode>): Renamed to ...
13510 (vec_extract<mode><unitmode>): ... this.
13511 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
13512 (vec_init<mode><unitmode>): ... this.
13513 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
13514 (vec_initv2sfsf): ... this.
13515 (vec_extractv2sf): Renamed to ...
13516 (vec_extractv2sfsf): ... this.
13517 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
13518 Add element mode after gen_vec_extract* calls.
13519 * config/mips/mips.md (unitmode): New mode iterator.
13520 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
13521 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
13522 * config/spu/spu.md (inner_l): New mode attribute.
13523 (vec_init<mode>): Renamed to ...
13524 (vec_init<mode><inner_l>): ... this.
13525 (vec_extract<mode>): Renamed to ...
13526 (vec_extract<mode><inner_l>): ... this.
13527 * config/sparc/sparc.md (veltmode): New mode iterator.
13528 (vec_init<VMALL:mode>): Renamed to ...
13529 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
13530 * config/ia64/vect.md (vec_initv2si): Renamed to ...
13531 (vec_initv2sisi): ... this.
13532 (vec_initv2sf): Renamed to ...
13533 (vec_initv2sfsf): ... this.
13534 (vec_extractv2sf): Renamed to ...
13535 (vec_extractv2sfsf): ... this.
13536 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
13537 (vec_init<mode>): Renamed to ...
13538 (vec_init<mode><VEC_base_l>): ... this.
13539 (vec_extract<mode>): Renamed to ...
13540 (vec_extract<mode><VEC_base_l>): ... this.
13541 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
13542 (vec_initv2sfsf): ... this.
13543 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
13544 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
13545 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
13546 gen_vec_init* calls.
13548 2017-08-01 Richard Biener <rguenther@suse.de>
13550 PR tree-optimization/81297
13551 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
13552 TREE_OVERFLOW from INTEGER_CSTs.
13554 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
13556 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
13558 2017-07-31 Carl Love <cel@us.ibm.com>
13560 * config/rs6000/rs6000-c: Add support for built-in functions
13561 vector signed char vec_xl_be (signed long long, signed char *);
13562 vector unsigned char vec_xl_be (signed long long, unsigned char *);
13563 vector signed int vec_xl_be (signed long long, signed int *);
13564 vector unsigned int vec_xl_be (signed long long, unsigned int *);
13565 vector signed long long vec_xl_be (signed long long, signed long long *);
13566 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
13567 vector signed short vec_xl_be (signed long long, signed short *);
13568 vector unsigned short vec_xl_be (signed long long, unsigned short *);
13569 vector double vec_xl_be (signed long long, double *);
13570 vector float vec_xl_be (signed long long, float *);
13571 * config/rs6000/altivec.h (vec_xl_be): Add #define.
13572 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
13573 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
13575 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
13576 (altivec_expand_builtin): Add switch statement to call
13577 altivec_expand_xl_be for each builtin.
13578 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
13579 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
13580 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
13581 __builtin_vsx_le_be_v16qi.
13582 * doc/extend.texi: Update the built-in documentation file for the
13583 new built-in functions.
13585 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
13588 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
13590 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
13592 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13594 * config.gcc: Add z14.
13595 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
13596 CPU model numbers for z13s and z14.
13597 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
13599 * config/s390/s390-opts.h (enum processor_type): Rename
13600 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
13601 * config/s390/s390.c (processor_table): Add field for CPU name to
13602 be passed to Binutils.
13603 (s390_asm_output_machine_for_arch): Use the new field in
13604 processor_table for Binutils.
13605 (s390_expand_builtin): Replace arch12 with z14.
13606 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
13607 (s390_get_sched_attrmask): Likewise.
13608 (s390_get_unit_mask): Likewise.
13609 * config/s390/s390.opt: Add z14 to processor_type enum.
13611 2017-07-31 Martin Jambor <mjambor@suse.cz>
13614 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
13615 regardless of optimization level.
13617 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
13618 Martin Liska <mliska@suse.cz>
13620 * predict.def: Remove old comment and adjust probability.
13621 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
13622 PREDICT statements.
13624 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
13627 * config/i386/i386.c (ix86_function_naked): New function.
13628 (ix86_can_use_return_insn_p): Return false for naked functions.
13629 (ix86_expand_prologue): Skip prologue for naked functions.
13630 (ix86_expand_epilogue): Skip epilogue for naked functions
13631 and emit trap instruction.
13632 (ix86_warn_func_return): New function.
13633 (ix86_attribute_table): Add "naked" attribute specification.
13634 (TARGET_WARN_FUNC_RETURN): Define.
13635 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
13637 2017-07-31 Martin Liska <mliska@suse.cz>
13639 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
13640 (dump_gimple_bb_header): Always dump BB info.
13641 (pp_cfg_jump): Do not append info about BB when dumping a jump.
13643 2017-07-31 Martin Liska <mliska@suse.cz>
13646 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
13647 also with current_function_decl non-null equality.
13649 2017-07-31 Jakub Jelinek <jakub@redhat.com>
13652 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
13653 change type to the element type, instead add eltype variable and
13654 use it where we are interested in the element type.
13656 PR tree-optimization/81603
13657 * ipa-polymorphic-call.c
13658 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
13659 offset arithmetic in offset_int, bail out if the resulting bit offset
13660 doesn't fit into shwi.
13662 2017-07-31 Martin Liska <mliska@suse.cz>
13664 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
13665 (gimplify_save_expr): Fix comment.
13667 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
13670 * config/i386/i386.c (ix86_function_arg): Update arguments for
13672 (ix86_compute_frame_layout): Set the initial stack offset to
13673 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
13674 INCOMING_FRAME_SP_OFFSET.
13675 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
13676 stack before exception handler returns.
13677 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
13678 the 'ERROR_CODE' for exception handler.
13680 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
13682 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
13683 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
13684 (ASM_OUTPUT_REG_POP): Ditto.
13685 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
13686 instead of asm_fprintf to output pure string.
13688 2017-07-29 Jakub Jelinek <jakub@redhat.com>
13690 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
13691 to imported_module_or_decl hook.
13692 (debug_nothing_tree_tree_tree_bool): Remove.
13693 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
13694 * debug.c (do_nothing_debug_hooks): Use
13695 debug_nothing_tree_tree_tree_bool_bool instead of
13696 debug_nothing_tree_tree_tree_bool.
13697 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
13698 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
13699 * sdbout.c (sdb_debug_hooks): Likewise.
13700 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
13701 (gen_namespace_die): Add DW_AT_export_symbols attribute if
13703 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
13704 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
13705 attribute, don't add anything.
13707 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13709 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
13710 (fold_build2_stat_loc): Likewise.
13711 (fold_build3_stat_loc): Likewise.
13712 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
13713 (fold_build1_loc): Remove macro.
13714 (fold_build2_loc): Likewise.
13715 (fold_build3_loc): Likewise.
13717 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13719 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
13720 (gimple_build_debug_bind_source_stat): Likewise.
13721 * gimple.h (gimple_build_debug_bind): Remove macro.
13722 (gimple_build_debug_bind_source): Likewise.
13724 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13726 * bitmap.c (bitmap_alloc): Adjust.
13727 (bitmap_gc_alloc): Likewise.
13728 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
13730 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13732 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
13733 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
13734 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
13735 (bitmap_gc_alloc_stat): Likewise.
13736 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
13738 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13740 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
13741 * rtl.h (shallow_copy_rtx): Remove macro.
13743 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13745 * emit-rtl.c (gen_raw_REG): Adjust.
13746 * gengenrtl.c (gendef): Likewise.
13747 * rtl.c (rtx_alloc_stat): Remove _stat from name.
13748 * rtl.h (rtx_alloc): Remove macro.
13750 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13752 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
13753 (build_tree_list_stat): Likewise.
13754 * tree.h (build_tree_list): Remove macro.
13755 (build_tree_list_vec): Likewise.
13757 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13759 * tree.c (make_vector_stat): Remove _stat from name.
13760 (build_vector_stat): Likewise.
13761 * tree.h (make_vector_stat): Remove macro.
13762 (build_vector_stat): Likewise.
13764 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13766 * tree.h (build_var_debug_value): Remove prototype.
13768 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13770 * tree.c (tree_cons_stat): Remove _stat from name.
13771 * tree.h (tree_cons): Remove macro.
13773 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13775 * tree.c (build_vl_exp_stat): Remove _stat from name.
13776 * tree.h (build_vl_exp): Remove macro.
13778 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13780 * tree.c (build_decl_stat): Remove _stat from name.
13781 * tree.h (build_decl): Remove macro.
13783 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13785 * gimple.c (gimple_build_with_ops_stat): Adjust.
13786 (gimple_alloc_stat): Remove _stat from name.
13787 * gimple.h (gimple_alloc): Remove macro.
13789 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13791 * tree.c (make_tree_vec_stat): Remove _stat from name.
13792 (grow_tree_vec_stat): Likewise.
13793 * tree.h (make_tree_vec_stat): Adjust prototype.
13794 (grow_tree_vec_stat): Likewise.
13795 (make_tree_vec): Remove macro.
13796 (grow_tree_vec): Likewise.
13798 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13800 * fold-const.c (fold_build1_stat_loc): Adjust.
13801 (fold_build2_stat_loc): Likewise.
13802 (fold_build3_stat_loc): Likewise.
13803 * tree.c (build0_stat): Remove _stat from name.
13804 (build1_stat): Likewise.
13805 (build2_stat): Likewise.
13806 (build3_stat): Likewise.
13807 (build4_stat): Likewise.
13808 (build5_stat): Likewise.
13809 * tree.h (build1_loc): Remove macro, and rename _stat function
13811 (build2_loc): Likewise.
13812 (build3_loc): Likewise.
13813 (build4_loc): Likewise.
13814 (build5_loc): Likewise.
13816 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13818 * tree.c (make_int_cst_stat): Remove _stat from name.
13819 * tree.h (make_int_cst_stat): Adjust prototype.
13820 (make_int_cst): Remove macro.
13822 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13824 * tree.c (make_tre_binfo_stat): Remove _stat from name.
13825 * tree.h (make_tree_binfo_stat): Adjust prototype.
13826 (make_tree_binfo): Remove.
13828 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13830 * tree.c (copy_node_stat): Rename to copy_node.
13831 (build_distinct_type_copy): Adjust.
13832 * tree.h (copy_node_stat): Adjust prototype.
13833 (copy_node): Remove macro.
13835 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13837 * tree.c (make_node_stat): rename to make_node.
13838 (build_tree_list_stat): Adjust.
13839 (build0_stat): Likewise.
13840 (build2_stat): Likewise.
13841 (build3_stat): Likewise.
13842 (build4_stat): Likewise.
13843 (build5_stat): Likewise.
13844 (build_decl_stat): Likewise.
13845 * tree.h (make_node_stat): Adjust prototype.
13846 (make_node): remove macro.
13848 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
13850 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
13851 (PPC_FEATURE2_SCV): Likewise.
13852 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
13854 2017-07-28 Tamar Christina <tamar.christina@arm.com>
13856 * config/aarch64/aarch64.c
13857 (aarch64_internal_mov_immediate): Add new special pattern.
13858 * config/aarch64/aarch64.md (*movdi_aarch64):
13859 Add reg/32bit const mov case.
13861 2017-07-28 Tamar Christina <tamar.christina@arm.com>
13862 Richard Sandiford <richard.sandiford@linaro.org>
13864 * config/aarch64/aarch64.md (mov<mode>): Generalize.
13865 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
13866 Add integer and movi cases.
13867 (movi-split-hf-df-sf split, fp16): New.
13868 (enabled): Added TARGET_FP_F16INST.
13869 * config/aarch64/iterators.md (GPF_HF): New.
13870 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
13872 2017-07-28 Tamar Christina <tamar.christina@arm.com>
13874 * config/aarch64/aarch64.c
13875 (aarch64_simd_container_mode): Add prototype.
13876 (aarch64_expand_mov_immediate): Add HI support.
13877 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
13878 (aarch64_can_const_movi_rtx_p): New.
13879 (aarch64_preferred_reload_class):
13880 Remove restrictions of using FP registers for certain SIMD operations.
13881 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
13882 (aarch64_valid_floating_const): Add integer move validation.
13883 (aarch64_simd_imm_scalar_p): Remove.
13884 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
13885 (aarch64_legitimate_constant_p): Expand list of supported cases.
13886 * config/aarch64/aarch64-protos.h
13887 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
13888 (aarch64_reinterpret_float_as_int): New.
13889 (aarch64_simd_imm_scalar_p): Remove.
13890 * config/aarch64/constraints.md (Uvi): New.
13891 (Dd): Split into Ds and new Dd.
13892 * config/aarch64/aarch64.md (*movsi_aarch64):
13894 (*movdi_aarch64): Add SIMD mov case.
13896 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13898 * tree-predcom.c: (struct chain): Handle store-store chain in which
13899 stores for elimination only store loop invariant values.
13900 (execute_pred_commoning_chain): Ditto.
13901 (prepare_initializers_chain_store_elim): Ditto.
13902 (prepare_finalizers): Ditto.
13903 (is_inv_store_elimination_chain): New function.
13904 (initialize_root_vars_store_elim_1): New function.
13906 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13908 * tree-predcom.c: Revise general description of the pass.
13909 (enum chain_type): New enum type for store elimination.
13910 (struct chain): New field supporting store elimination.
13911 (struct component): Ditto.
13912 (dump_chain): Dump store-stores chain.
13913 (release_chain): Release resources.
13914 (split_data_refs_to_components): Compute and create component
13915 contains only stores for elimination.
13916 (get_chain_last_ref_at): New function.
13917 (make_invariant_chain): Initialization.
13918 (make_rooted_chain): Specify chain type in parameter and record it.
13919 (add_looparound_copies): Skip for store-stores chain.
13920 (determine_roots_comp): Compute type of chain and pass it to
13922 (initialize_root_vars_store_elim_2): New function.
13923 (finalize_eliminated_stores): New function.
13924 (remove_stmt): Handle store for elimination.
13925 (execute_pred_commoning_chain): Execute predictive commoning on
13926 store-store chains.
13927 (determine_unroll_factor): Skip unroll for store-stores chain.
13928 (prepare_initializers_chain_store_elim): New function.
13929 (prepare_initializers_chain): Hanlde store-store chain.
13930 (prepare_finalizers_chain, prepare_finalizers): New function.
13931 (tree_predictive_commoning_loop): Return integer value indicating
13932 if loop is unrolled or lcssa form is corrupted.
13933 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
13935 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13937 * tree-predcom.c (initialize_root): Delete.
13938 (execute_pred_commoning_chain): Initialize root vars and replace
13939 reference of non-combined chain directly, rather than call above
13942 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13944 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
13945 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
13947 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13949 * tree-predcom.c (struct chain): New field init_seq.
13950 (release_chain): Release init_seq.
13951 (prepare_initializers_chain): Record intialization stmts in above
13953 (insert_init_seqs): New function.
13954 (tree_predictive_commoning_loop): Call insert_init_seqs.
13956 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13958 * tree-predcom.c (determine_roots_comp): Skip trivial components.
13960 2017-07-28 Richard Biener <rguenther@suse.de>
13962 * match.pd: Remove superfluous :c.
13963 * genmatch.c (simplify::id): Add member.
13964 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
13966 (current_id): New global.
13967 (dt_node::parent): Move from ...
13968 (dt_operand::parent): ... here. Add for_id member.
13969 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
13970 (decision_tree::find_node): Relax order requirement when
13971 merging DT_TRUE nodes to ones inbetween the current simplify
13972 and the one we try to merge with. Add diagnostic whenever
13973 we need to enforce pattern order by not merging.
13974 (decision_tree::insert): Set current_id.
13975 (decision_tree::print_node): Dump parent node and for_id.
13976 (parser::last_id): Add member.
13977 (parser::push_simplify): Assign unique id.
13978 (parser::parser): Initialize last_id.
13980 2017-07-28 Martin Liska <mliska@suse.cz>
13983 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
13984 gimple_build_debug_bind.
13986 2017-07-28 Richard Biener <rguenther@suse.de>
13988 PR tree-optimization/81502
13989 * match.pd: Add pattern combining BIT_INSERT_EXPR with
13991 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
13993 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
13994 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
13995 for BIT_FIELD_REF args.
13996 * fold-const.c (make_bit_field_ref): Likewise.
13997 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
13999 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14002 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
14003 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
14004 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
14005 Or it into SANITIZER_UNDEFINED.
14006 * ubsan.c: Include gimple-fold.h and varasm.h.
14007 (ubsan_expand_ptr_ifn): New function.
14008 (instrument_pointer_overflow): New function.
14009 (maybe_instrument_pointer_overflow): New function.
14010 (instrument_object_size): Formatting fix.
14011 (pass_ubsan::execute): Call instrument_pointer_overflow
14012 and maybe_instrument_pointer_overflow.
14013 * internal-fn.c (expand_UBSAN_PTR): New function.
14014 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
14015 * sanitizer.def (__ubsan_handle_pointer_overflow,
14016 __ubsan_handle_pointer_overflow_abort): New builtins.
14017 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
14018 * internal-fn.def (UBSAN_PTR): New internal function.
14019 * opts.c (sanitizer_opts): Add pointer-overflow.
14020 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
14021 * fold-const.c (build_range_check): Compute pointer range check in
14022 integral type if pointer arithmetics would be needed. Formatting
14025 2017-07-28 Martin Liska <mliska@suse.cz>
14028 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
14029 parameters that are of a variable-length.
14031 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
14033 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
14035 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
14036 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
14037 (CRT_CALL_STATIC_FUNCTION): Likewise.
14038 (ASM_DEFAULT_SPEC): New define.
14039 (ASM_SPEC32): Likewise.
14040 (ASM_SPEC64): Likewise.
14041 (ASM_SPEC_COMMON): Likewise.
14042 (ASM_SPEC): Likewise.
14043 (INVALID_64BIT): Likewise.
14044 (LINK_OS_DEFAULT_SPEC): Likewise.
14045 (LINK_OS_SPEC32): Likewise.
14046 (LINK_OS_SPEC64): Likewise.
14047 (POWERPC_LINUX): Likewise.
14048 (PTRDIFF_TYPE): Likewise.
14049 (RESTORE_FP_PREFIX): Likewise.
14050 (RESTORE_FP_SUFFIX): Likewise.
14051 (SAVE_FP_PREFIX): Likewise.
14052 (SAVE_FP_SUFFIX): Likewise.
14053 (SIZE_TYPE): Likewise.
14054 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
14055 (TARGET_64BIT): Likewise.
14056 (TARGET_64BIT): Likewise.
14057 (TARGET_AIX): Likewise.
14058 (WCHAR_TYPE_SIZE): Likewise.
14059 (WCHAR_TYPE): Undefine.
14060 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
14061 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
14062 (CPP_OS_RTEMS_SPEC): Delete.
14063 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
14064 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
14066 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
14068 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14070 PR tree-optimization/81578
14071 * tree-parloops.c (build_new_reduction): Bail out if
14072 reduction_code isn't one of the standard OpenMP reductions.
14073 Move the details printing after that decision.
14075 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14077 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
14078 related to reload_in_progress.
14079 (splat_input_operand): Likewise.
14080 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
14082 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
14084 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
14085 (TARGET_INSTANTIATE_DECLS): Likewise.
14086 (legitimate_indexed_address_p): Delete reload_in_progress code.
14087 (rs6000_debug_legitimate_address_p): Likewise.
14088 (rs6000_eliminate_indexed_memrefs): Likewise.
14089 (rs6000_emit_le_vsx_store): Likewise.
14090 (rs6000_emit_move_si_sf_subreg): Likewise.
14091 (rs6000_emit_move): Likewise.
14092 (register_to_reg_type): Likewise.
14093 (rs6000_pre_atomic_barrier): Likewise.
14094 (rs6000_machopic_legitimize_pic_address): Likewise.
14095 (rs6000_allocate_stack_temp): Likewise.
14096 (rs6000_address_for_fpconvert): Likewise.
14097 (rs6000_address_for_altivec): Likewise.
14098 (rs6000_secondary_memory_needed_rtx): Delete function.
14099 (rs6000_check_sdmode): Likewise.
14100 (rs6000_alloc_sdmode_stack_slot): Likewise.
14101 (rs6000_instantiate_decls): Likewise.
14102 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
14103 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
14104 Delete reload_in_progress.
14105 (*vec_reload_and_plus_<mptrsize>): Likewise.
14106 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
14107 (vsx_div_v2di): Likewise.
14108 (vsx_udiv_v2di): Likewise.
14110 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14112 * config/rs6000/rs6000.opt (mlra): Replace with stub.
14113 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
14114 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
14115 (rs6000_debug_reg_global): Delete print of LRA status.
14116 (rs6000_option_override_internal): Delete dead LRA related code.
14117 (rs6000_lra_p): Delete function.
14118 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
14120 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
14122 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
14123 * config/riscv/rtems.h: New file.
14125 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14126 Sudakshina Das <sudi.das@arm.com>
14128 * config/aarch64/aarch64.md
14129 (define_split for and<mode>3nr_compare): Move
14130 non aarch64_logical_operand to a register.
14131 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
14132 register immediate operand to a register.
14133 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
14135 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14137 PR middle-end/81564
14138 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
14140 2017-07-27 Richard Biener <rguenther@suse.de>
14142 PR tree-optimization/81573
14143 PR tree-optimization/81494
14144 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
14145 multi defuse cycle case.
14147 2017-07-27 Richard Biener <rguenther@suse.de>
14149 PR tree-optimization/81571
14150 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
14153 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
14155 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
14156 earlier and only if MASK_FPU is set. Adjust formatting.
14158 2017-07-27 Martin Liska <mliska@suse.cz>
14160 * opt-functions.awk: Add validation of value of Init.
14161 * optc-gen.awk: Pass new argument.
14163 2017-07-27 Martin Liska <mliska@suse.cz>
14165 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
14166 Fix wrong condition.
14168 2017-07-27 Martin Liska <mliska@suse.cz>
14170 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
14171 BBs and edges seen by autoFDO.
14173 2017-07-27 Richard Biener <rguenther@suse.de>
14175 PR tree-optimization/81502
14176 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
14177 with incompatible but same sized type.
14178 (execute_update_addresses_taken): Likewise.
14180 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
14182 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
14183 flag_tree_loop_vectorize rather than flag_tree_vectorize.
14185 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14188 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
14189 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
14190 Change s_operand to memory_operand.
14192 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
14194 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
14195 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
14196 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
14197 Emit instructions rather than returning an expression. Handle TFmode
14198 and KFmode by casting to TImode.
14199 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
14200 (rs6000_emit_le_vsx_store): Likewise.
14201 * config/rs6000/vsx.md (VSX_TI): New iterator.
14202 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
14203 (*vsx_le_undo_permute_<mode>): Likewise.
14204 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
14205 emit the split sequence.
14206 (*vsx_le_perm_store_<mode>): Likewise.
14208 2017-07-27 Jakub Jelinek <jakub@redhat.com>
14210 PR tree-optimization/81555
14211 PR tree-optimization/81556
14212 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
14213 if true, force CHANGED for the recursive invocation.
14214 (reassociate_bb): Remember original length of ops array, pass
14215 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
14217 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
14218 attributes for noipa attribute. For naked attribute use
14219 lookup_attribute first before lookup_attribute_spec.
14220 * final.c (rest_of_handle_final): Disable IPA RA for functions with
14222 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
14223 for functions with noipa attribute.
14224 (cgraph_externally_visible_p): Return true for functions with noipa
14226 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
14227 for functions with noipa attribute.
14228 * doc/extend.texi: Document noipa function attribute.
14229 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
14230 also for functions with noipa attribute.
14231 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
14233 2017-07-26 Andrew Pinski <apinski@cavium.com>
14235 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
14236 vec_unalign_load_cost and vec_unalign_store_cost.
14238 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14240 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
14241 -mvsx-small-integer option.
14242 (ISA_3_0_MASKS_IEEE): Likewise.
14243 (OTHER_VSX_VECTOR_MASKS): Likewise.
14244 (POWERPC_MASKS): Likewise.
14245 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
14246 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
14247 code, only testing for DImode being allowed in non-VSX floating
14249 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
14250 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
14252 (rs6000_option_override_internal): Delete -mvsx-small-integer.
14253 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
14254 TARGET_P8_VECTOR test.
14255 (rs6000_secondary_reload_simple_move): Likewise.
14256 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
14257 since TARGET_P9_VECTOR was already tested.
14258 (rs6000_opt_masks): Remove -mvsx-small-integer.
14259 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
14260 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14262 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
14263 test for TARGET_VEXTRACTUB was used, and that uses
14265 (p9 extract splitter): Likewise.
14266 (vsx_extract_<mode>_di_p9): Likewise.
14267 (vsx_extract_<mode>_store_p9): Likewise.
14268 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
14269 for TARGET_P9_VECTOR was used. Delete code that is now dead with
14270 the elimination of TARGET_VSX_SMALL_INTEGER.
14271 (vsx_extract_<mode>_p8): Likewise.
14272 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
14273 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
14274 (vsx_set_<mode>_p9): Likewise.
14275 (vsx_set_v4sf_p9): Likewise.
14276 (vsx_set_v4sf_p9_zero): Likewise.
14277 (vsx_insert_extract_v4sf_p9): Likewise.
14278 (vsx_insert_extract_v4sf_p9_2): Likewise.
14279 * config/rs6000/rs6000.md (sign extend splitter): Change
14280 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
14281 (floatsi<mode>2_lfiwax_mem): Likewise.
14282 (floatunssi<mode>2_lfiwzx_mem): Likewise.
14283 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
14284 since a test for TARGET_P9_VECTOR was used.
14285 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14286 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
14287 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14288 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
14289 TARGET_P8_VECTOR test.
14290 (fix_trunc<mode>si2_stfiwx): Likewise.
14291 (fix_trunc<mode>si2_internal): Likewise.
14292 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
14293 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14295 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
14296 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
14297 TARGET_P8_VECTOR test.
14298 (fixuns_trunc<mode>si2_stfiwx): Likewise.
14299 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
14300 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14302 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
14303 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
14304 since a test for TARGET_P9_VECTOR was used.
14305 (splitter for loading small constants): Likewise.
14307 2017-07-26 Andrew Pinski <apinski@cavium.com>
14309 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
14312 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
14315 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
14316 (fp_valid_at): Likewise.
14318 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
14320 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
14321 (qdf24xx_addrcost_table): Likewise.
14322 (cortexa57_tunings): Update to use generic_branch_cost.
14323 (cortexa72_tunings): Likewise.
14324 (cortexa73_tunings): Likewise.
14325 (qdf24xx_tunings): Likewise.
14327 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
14329 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
14330 (thunderx2t99_branch_cost): Likewise.
14331 (cortexa35_tunings): Update to use generic_branch_cost.
14332 (cortexa53_tunings): Likewise.
14333 (cortexa57_tunings): Likewise.
14334 (cortexa72_tunings): Likewise.
14335 (cortexa73_tunings): Likewise.
14336 (thunderx2t99_tunings): Likewise.
14338 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
14340 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
14341 (sparc_option_override): Honour MASK_FSMULD.
14342 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
14343 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
14344 * config/sparc/sparc.opt (mfsmuld): New option.
14345 * doc/invoke.texi (mfsmuld): Document option.
14347 2017-07-26 Marek Polacek <polacek@redhat.com>
14349 PR middle-end/70992
14350 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
14352 2017-07-26 Richard Biener <rguenther@suse.de>
14354 * gimple-match-head.c (do_valueize): Return OP if valueize
14356 (get_def): New helper to get at the def stmt of a SSA name
14357 if valueize allows.
14358 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
14359 do_valueize to get at the def stmt.
14360 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
14362 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
14364 PR middle-end/46932
14365 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
14367 2017-07-26 Martin Liska <mliska@suse.cz>
14370 * function.c (expand_function_start): Make expansion of
14371 nonlocal_goto_save_area after parm_birth_insn.
14373 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
14375 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
14376 from all CPU target flags enable members.
14378 2017-07-26 Richard Biener <rguenther@suse.de>
14380 * genmatch.c (dt_simplify::gen): Make iterator vars const.
14381 (decision_tree::gen): Make 'type' const.
14382 (write_predicate): Likewise.
14384 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
14386 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
14387 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
14388 (rs6000_option_override_internal): Likewise.
14389 (rs6000_expand_vector_set): Likewise.
14390 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
14391 (TARGET_UPPER_REGS_SF): Likewise.
14392 (TARGET_UPPER_REGS_DI): Likewise.
14393 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
14394 (TARGET_DIRECT_MOVE_64BIT): Likewise.
14395 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
14396 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14397 (Splitters for DI constants in Altivec registers): Likewise.
14398 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
14399 (vsx_set_v4sf_p9): Likewise.
14400 (vsx_set_v4sf_p9_zero): Likewise.
14401 (vsx_insert_extract_v4sf_p9): Likewise.
14402 (vsx_insert_extract_v4sf_p9_2): Likewise.
14404 2017-07-25 Carl Love <cel@us.ibm.com>
14406 * doc/extend.texi: Update the built-in documentation file for the
14407 existing built-in functions
14408 vector signed char vec_cnttz (vector signed char);
14409 vector unsigned char vec_cnttz (vector unsigned char);
14410 vector signed short vec_cnttz (vector signed short);
14411 vector unsigned short vec_cnttz (vector unsigned short);
14412 vector signed int vec_cnttz (vector signed int);
14413 vector unsigned int vec_cnttz (vector unsigned int);
14414 vector signed long long vec_cnttz (vector signed long long);
14415 vector unsigned long long vec_cnttz (vector unsigned long long);
14417 2017-07-25 Andrew Pinski <apinski@cavium.com>
14419 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
14420 accesses where the use is for the first operand of a BIT_INSERT.
14422 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
14425 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
14426 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
14428 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
14430 * config/i386/gstabs.h: Delete.
14431 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
14433 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
14435 * config/i386/i386.c (ix86_decompose_address): Do not check for
14436 register RTX when looking at index_reg or base_reg.
14437 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
14439 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
14441 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
14442 to update EH info here.
14444 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
14446 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
14448 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
14450 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
14452 2017-07-25 Torsten Duwe <duwe@suse.de>
14454 * common.opt: Introduce -fpatchable-function-entry
14455 command line option, and its variables function_entry_patch_area_size
14456 and function_entry_patch_area_start.
14457 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
14458 including a two-value parser.
14459 * target.def (print_patchable_function_entry): New target hook.
14460 * targhooks.h (default_print_patchable_function_entry): New function.
14461 * targhooks.c (default_print_patchable_function_entry): Likewise.
14462 * toplev.c (process_options): Switch off IPA-RA if
14463 patchable function entries are being generated.
14464 * varasm.c (assemble_start_function): Look at the
14465 patchable-function-entry command line switch and current
14466 function attributes and maybe generate NOP instructions by
14467 calling the print_patchable_function_entry hook.
14468 * doc/extend.texi: Document patchable_function_entry attribute.
14469 * doc/invoke.texi: Document -fpatchable_function_entry
14470 command line option.
14471 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
14473 * doc/tm.texi: Re-generate.
14475 2017-07-25 Jakub Jelinek <jakub@redhat.com>
14478 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
14479 TARGET_AVX512DQ rather than TARGET_AVX512BW.
14481 2017-07-25 Tamar Christina <tamar.christina@arm.com>
14483 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
14485 2017-07-25 Richard Biener <rguenther@suse.de>
14487 PR tree-optimization/81455
14488 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
14489 not walk in cycles when looking for guards.
14491 2017-07-25 Richard Biener <rguenther@suse.de>
14493 PR tree-optimization/81529
14494 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
14495 when optimizing backedge uses.
14497 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
14499 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
14501 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
14502 to dl_section_ref. On AIX, append an expression to subtract
14503 the size of the section length to dl_section_ref.
14505 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
14507 * configure.ac: If any of the config.* scripts fail, exit 1.
14508 * configure: Regenerate.
14510 2017-07-25 Richard Biener <rguenther@suse.de>
14512 PR middle-end/81546
14513 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
14514 of immediate uses, be more verbose on errors.
14516 2017-07-25 Richard Biener <rguenther@suse.de>
14518 PR tree-optimization/81510
14519 * tree-vect-loop.c (vect_is_simple_reduction): When the
14520 reduction stmt is not inside the loop bail out.
14522 2017-07-25 Richard Biener <rguenther@suse.de>
14524 PR tree-optimization/81303
14525 * tree-vect-loop-manip.c (vect_loop_versioning): Build
14526 profitability check against LOOP_VINFO_NITERSM1.
14528 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
14530 * domwalk.c (cmp_bb_postorder): Simplify.
14531 (sort_bbs_postorder): New function. Use it...
14532 (dom_walker::walk): ...here to optimize common cases.
14534 2017-07-25 Martin Liska <mliska@suse.cz>
14537 * ipa-visibility.c (function_and_variable_visibility): Make the
14538 redirection just on target that supports aliasing.
14539 Fix GNU coding style.
14541 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
14544 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
14546 * config/arm/rtems.h: Likewise.
14547 * config/bfin/rtems.h: Likewise.
14548 * config/i386/rtemself.h: Likewise.
14549 * config/lm32/rtems.h: Likewise.
14550 * config/m32c/rtems.h: Likewise.
14551 * config/m68k/rtemself.h: Likewise.
14552 * config/microblaze/rtems.h: Likewise.
14553 * config/mips/rtems.h: Likewise.
14554 * config/moxie/rtems.h: Likewise.
14555 * config/nios2/rtems.h: Likewise.
14556 * config/powerpcspe/rtems.h: Likewise.
14557 * config/rs6000/rtems.h: Likewise.
14558 * config/rtems.h: Likewise.
14559 * config/sh/rtems.h: Likewise.
14560 * config/sh/rtemself.h: Likewise.
14561 * config/sparc/rtemself.h: Likewise.
14563 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
14566 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
14567 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
14568 * tree-ssa-structalias.c (alias_get_name): Same.
14570 2017-07-25 Bin Cheng <bin.cheng@arm.com>
14573 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
14574 instructions if no du chain is found.
14576 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
14578 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
14580 2017-07-25 Richard Biener <rguenther@suse.de>
14582 PR middle-end/81505
14583 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
14586 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
14588 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
14589 upper-regs options.
14590 (ISA_2_7_MASKS_SERVER): Likewise.
14591 (ISA_3_0_MASKS_IEEE): Likewise.
14592 (OTHER_P8_VECTOR_MASKS): Likewise.
14593 (OTHER_VSX_VECTOR_MASKS): Likewise.
14594 (POWERPC_MASKS): Likewise.
14595 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
14596 duplicate list of options.
14597 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
14598 explicit -mupper-regs options.
14599 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
14600 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
14601 alias for -mupper-regs-df.
14602 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
14603 (rs6000_init_hard_regno_mode_ok): Likewise.
14604 (rs6000_option_override_internal): Likewise.
14605 (rs6000_opt_masks): Likewise.
14606 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
14607 options in terms of whether -mvsx or -mpower8-vector was used.
14608 (TARGET_UPPER_REGS_DI): Likewise.
14609 (TARGET_UPPER_REGS_SF): Likewise.
14610 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
14611 -mupper-regs-* options.
14613 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
14615 * passes.c (emergency_dump_function): Print some empty lines and a
14616 header before the RTL dump.
14618 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
14620 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
14622 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
14625 * config/aarch64/aarch64.c (aarch64_classify_symbol):
14626 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
14628 2017-07-24 Carl Love <cel@us.ibm.com>
14630 * config/rs6000/rs6000-c.c: Add support for built-in functions
14631 vector float vec_extract_fp32_from_shorth (vector unsigned short);
14632 vector float vec_extract_fp32_from_shortl (vector unsigned short);
14633 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
14634 vec_extract_fp_from_shortl): Add defines for the two builtins.
14635 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
14636 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
14638 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
14639 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
14640 * doc/extend.texi: Update the built-in documentation file for the
14641 new built-in function.
14643 2017-07-24 Jakub Jelinek <jakub@redhat.com>
14646 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
14648 * doc/generic.texi: Likewise.
14649 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
14650 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
14652 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
14654 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
14655 (aarch64_mls_elt_merge<mode>): Likewise.
14657 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
14659 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
14660 having __cxa_atexit.
14662 2017-07-23 Michael Collison <michael.collison@arm.com>
14664 * config/arm/arm.c (arm_option_override): Deprecate
14665 use of -mstructure-size-boundary.
14666 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
14667 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
14669 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14672 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
14673 Reduce cost estimate for direct moves.
14675 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
14678 * config/i386/i386.c (ix86_option_override_internal): Disable
14679 BMI, BMI2 and TBM instructions for -m16.
14681 2017-07-21 Carl Love <cel@us.ibm.com>
14683 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14684 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14685 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14686 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14687 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14688 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
14689 VMULOSW): New enum "unspec" values.
14690 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
14691 altivec_vmulosw): New patterns.
14692 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14693 VMULOSW): Add definitions.
14695 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
14697 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
14698 (qdf24xx): Likewise.
14699 * config/aarch64/aarch64-options-extensions.def (rdma); New.
14700 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
14701 (AARCH64_FL_V8_1): Renumber.
14702 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
14703 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
14704 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
14705 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
14706 rdma to feature modifiers list.
14708 2017-07-21 Yury Gribov <tetra2005@gmail.com>
14710 PR middle-end/56727
14711 * ipa-visibility (function_and_variable_visibility): Convert
14712 recursive PLT call to direct call if appropriate.
14714 2017-07-21 Andrew Pinski <apinski@cavium.com>
14716 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
14717 operand 1 to see if the types precision matches.
14718 * fold-const.c (operand_equal_p): Likewise.
14720 2017-07-21 Richard Biener <rguenther@suse.de>
14722 PR tree-optimization/81303
14723 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
14724 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
14725 (vect_peeling_hash_get_lowest_cost): Adjust.
14726 (vect_enhance_data_refs_alignment): Likewise. Use
14727 vect_get_peeling_costs_all_drs to compute the penalty for no
14728 peeling to match up costs.
14730 2017-07-21 Richard Biener <rguenther@suse.de>
14732 PR tree-optimization/81500
14733 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
14734 we didn't identify a reduction path.
14736 2017-07-21 Tom de Vries <tom@codesourcery.com>
14737 Cesar Philippidis <cesar@codesourcery.com>
14739 PR gcov-profile/81442
14740 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
14743 2017-07-21 Tom de Vries <tom@codesourcery.com>
14746 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
14748 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
14749 nvptx_override_options_after_change.
14751 2017-07-21 Ulrich Drepper <drepper@redhat.com>
14753 * dwarf2out.c (output_file_names): Avoid double testing for
14754 dwarf_version >= 5.
14756 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
14758 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
14760 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
14762 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
14764 (try_crossjump_to_edge): Do not punt on partitioned functions.
14766 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
14768 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
14769 Put all BBs reachable only via paths crossing cold region to cold
14771 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
14773 2016-07-21 Richard Biener <rguenther@suse.de>
14775 PR tree-optimization/81303
14776 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
14777 into account prologue and epilogue iterations when raising
14778 min_profitable_iters to sth at least covering one vector iteration.
14780 2017-07-21 Tamar Christina <tamar.christina@arm.com>
14782 * config/arm/arm.c (arm_test_cpu_arch_dat):
14785 2017-07-20 Nathan Sidwell <nathan@acm.org>
14787 Remove TYPE_METHODS.
14788 * tree.h (TYPE_METHODS): Delete.
14789 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
14790 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
14791 (dbxout_type_methods): Scan TYPE_FIELDS.
14792 (dbxout_type): Don't check TYPE_METHODS here.
14793 * function.c (use_register_for_decl): Always ignore register for
14794 class types when not optimizing.
14795 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
14796 * tree.c (free_lang_data_in_type): Stitch out member functions and
14797 templates from TYPE_FIELDS.
14798 (build_distinct_type_copy, verify_type_variant,
14799 verify_type): Member fns are on TYPE_FIELDS.
14800 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
14801 * tree-pretty-print.c (dump_generic_node): Likewise.
14803 2017-07-20 Jakub Jelinek <jakub@redhat.com>
14806 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
14807 V2TImode and V4TImode.
14808 (ix86_expand_vector_extract): Likewise.
14809 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
14810 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
14811 (ssescalarmode): Handle V4TImode and V2TImode.
14812 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
14813 (*vec_extractv2ti, *vec_extractv4ti): New insns.
14814 (VEXTRACTI128_MODE): New mode iterator.
14815 (splitter for *vec_extractv?ti first element): New.
14816 (VEC_INIT_MODE): New mode iterator.
14817 (vec_init<mode>): Consolidate 3 expanders into one using
14818 VEC_INIT_MODE mode iterator.
14820 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
14822 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
14823 non_spilled_static_chain_regno_p.
14825 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
14827 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
14829 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
14831 * bb-reorder.c (connect_traces): Allow copying of blocks within
14834 2017-07-20 Richard Biener <rguenther@suse.de>
14836 * gimple.h (gimple_phi_result): Add gphi * overload.
14837 (gimple_phi_result_ptr): Likewise.
14838 (gimple_phi_arg): Likewise. Adjust index assert to only
14839 allow actual argument accesses rather than all slots available
14841 (gimple_phi_arg_def): Add gphi * overload.
14842 * tree-phinodes.c (make_phi_node): Initialize only actual
14844 (resize_phi_node): Clear memory not covered by old node,
14845 do not initialize excess argument slots.
14846 (reserve_phi_args_for_new_edge): Initialize new argument slot
14849 2017-07-20 Bin Cheng <bin.cheng@arm.com>
14851 PR tree-optimization/81388
14853 2016-07-21 Bin Cheng <bin.cheng@arm.com>
14855 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
14856 by removing computation of may_be_zero.
14858 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
14859 Tom de Vries <tom@codesourcery.com>
14861 PR middle-end/81030
14862 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
14863 when gimple level profile disagrees with what RTL expander did.
14865 2017-07-20 Richard Biener <rguenther@suse.de>
14867 PR tree-optimization/61171
14868 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
14869 (vect_analyze_stmt): Add slp instance parameter.
14870 (vectorizable_reduction): Likewise.
14871 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
14872 (vect_is_simple_reduction): Deal with chains not detected
14873 as SLP reduction chain, specifically not properly associated
14874 chains containing a mix of plus/minus.
14875 (get_reduction_op): Remove.
14876 (get_initial_defs_for_reduction): Simplify, pass in whether
14877 this is a reduction chain, pass in the SLP node for the PHIs.
14878 (vect_create_epilog_for_reduction): Get the SLP instance as
14880 (vectorizable_reduction): Get the SLP instance as arg.
14881 During analysis remember the SLP node with the PHIs in the
14882 instance. Simplify getting at the vectorized reduction PHIs.
14883 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
14884 through SLP instance.
14885 (vect_slp_analyze_operations): Likewise.
14886 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
14887 (vect_transform_stmt): Likewise.
14889 2017-07-20 Tom de Vries <tom@codesourcery.com>
14891 PR tree-optimization/81489
14892 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
14893 read of phi arg location to before loop that modifies phi.
14895 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14897 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
14900 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
14902 PR middle-end/81331
14903 * except.c (execute): Fix ordering issue.
14905 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
14907 PR rtl-optimization/81423
14908 * combine.c (make_compound_operation_int): Don't try to optimize
14909 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
14911 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
14913 PR rtl-optimization/81423
14914 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
14915 with a constant that is -1 in the truncated to mode.
14917 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
14919 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
14920 (determine_unlikely_bbs): ... here.
14921 * predict.h (propagate_unlikely_bbs_forward): Declare.
14922 * cfgexpand.c (pass_expand::execute): Use it.
14923 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
14925 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
14926 propagate_unlikely_bbs_forward.
14928 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
14930 PR middle-end/81331
14931 * except.c (maybe_add_nop_after_section_switch): New function.
14934 2017-07-19 Tom de Vries <tom@codesourcery.com>
14936 * gimple.h (gimple_phi_set_arg): Make assert more strict.
14938 2017-07-19 Tom de Vries <tom@codesourcery.com>
14940 * gimple.h (gimple_phi_arg): Make assert more strict.
14942 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
14944 * config.gcc (powerpc*-*-*): Add mmintrin.h.
14945 * config/rs6000/mmintrin.h: New file.
14946 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
14948 2017-07-19 Jakub Jelinek <jakub@redhat.com>
14950 PR tree-optimization/81346
14951 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
14953 2017-07-19 Tom de Vries <tom@codesourcery.com>
14955 * config/nvptx/nvptx.md (VECIM): Add V2DI.
14957 2017-07-19 Tom de Vries <tom@codesourcery.com>
14959 * config/nvptx/nvptx-modes.def: Add V2DImode.
14960 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
14961 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
14962 (nvptx_output_mov_insn): Handle lack of mov.b128.
14963 (nvptx_print_operand): Handle 'H' and 'L' codes.
14964 (nvptx_vector_mode_supported): Allow V2DImode.
14965 (nvptx_preferred_simd_mode): New function.
14966 (nvptx_data_alignment): New function.
14967 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
14968 nvptx_preferred_simd_mode.
14969 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
14971 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
14973 2017-07-19 Tom de Vries <tom@codesourcery.com>
14975 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
14976 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
14977 (nvptx_vector_mode_supported): New function. Allow V2SImode.
14978 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
14979 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
14980 (mov<VECIM>_insn): New define_insn.
14981 (define_expand "mov<VECIM>): New define_expand.
14983 2017-07-19 Tom de Vries <tom@codesourcery.com>
14985 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
14987 2017-07-19 Jakub Jelinek <jakub@redhat.com>
14989 PR tree-optimization/81346
14990 * fold-const.h (fold_div_compare, range_check_type): Declare.
14991 * fold-const.c (range_check_type): New function.
14992 (build_range_check): Use range_check_type.
14993 (fold_div_compare): No longer static, rewritten into
14994 a match.pd helper function.
14995 (fold_comparison): Don't call fold_div_compare here.
14996 * match.pd (X / C1 op C2): New optimization using fold_div_compare
14997 as helper function.
14999 2017-07-19 Nathan Sidwell <nathan@acm.org>
15001 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
15002 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
15003 * tree.c (find_decls_types_r, verify_type): Use
15004 TYPE_{MIN,MAX}_VALUE_RAW.
15005 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
15006 (hash_tree): Likewise.
15007 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
15009 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
15012 2017-07-18 Tom de Vries <tom@codesourcery.com>
15014 PR middle-end/81464
15015 * omp-expand.c (expand_omp_for_static_chunk): Handle
15016 equal-argument loop exit phi.
15018 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
15021 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
15022 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
15023 operand 2 predicate.
15024 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
15025 operand 2 predicate.
15026 (ror,rol -> rorx splitters): Use const_int_operand as
15027 operand 2 predicate.
15029 2017-06-18 Richard Biener <rguenther@suse.de>
15031 PR tree-optimization/81410
15032 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
15033 the gap in the ! slp_perm SLP case after each group.
15035 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15037 PR middle-end/81463
15038 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
15041 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15043 PR middle-end/81462
15044 * predict.c (set_even_probabilities): Cleanup; do not affect
15045 probabilities that are already known.
15046 (combine_predictions_for_bb): Call even when count is set.
15048 2017-07-18 Nathan Sidwell <nathan@acm.org>
15050 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
15053 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15056 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
15057 optimization for loop niter analysis.
15059 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
15062 * config/avr/avr.c (avr_optimize_casesi): Don't use
15063 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
15065 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
15067 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
15068 body_cost_vec from _vect_peel_extended_info.
15069 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
15070 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
15073 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15075 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
15077 2017-07-18 Richard Biener <rguenther@suse.de>
15079 PR tree-optimization/80620
15080 PR tree-optimization/81403
15081 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
15082 info when re-using a VN table entry.
15084 2017-07-18 Richard Biener <rguenther@suse.de>
15086 PR tree-optimization/81418
15087 * tree-vect-loop.c (vectorizable_reduction): Properly compute
15088 vectype_in. Verify that with lane-reducing reduction operations
15089 we have a single def-use cycle.
15091 2017-07-17 Carl Love <cel@us.ibm.com>
15093 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
15095 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15096 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15097 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15098 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15099 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15100 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15101 VMULOSW): New enum "unspec" values.
15102 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
15103 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
15104 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15105 altivec_vmulosw): New patterns.
15106 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15107 VMULOSW): Add definitions.
15108 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
15110 * config/alpha/alpha.c: Include predict.h.
15112 2017-07-17 Yury Gribov <tetra2005@gmail.com>
15114 * tree-vrp.c (compare_assert_loc): Fix comparison function
15115 to return predictable results.
15117 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15119 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
15121 (subdi3): Likewise.
15122 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
15123 * doc/invoke.texi (mexpand-adddi): Update text.
15125 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15127 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
15128 that also clobbers the CC register. The old expand code is moved
15130 (*arc_clzsi2): ... here.
15131 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
15132 the CC register. The old expand code is moved to ...
15133 (arc_ctzsi2): ... here.
15135 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15137 * config/arc/arc.opt (mindexed-loads): Use initial value
15138 TARGET_INDEXED_LOADS_DEFAULT.
15139 (mauto-modify-reg): Use initial value
15140 TARGET_AUTO_MODIFY_REG_DEFAULT.
15141 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15142 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15143 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15144 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15146 2017-07-17 Martin Liska <mliska@suse.cz>
15149 * opts.c (finish_options): Do not allow -fgnu-tm
15150 w/ -fsanitize={kernel-,}address. Say sorry.
15152 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15155 * tree-loop-distribution.c (classify_partition): Only assert on
15156 numer of iterations.
15157 (merge_dep_scc_partitions): Delete prameter. Update function call.
15158 (distribute_loop): Remove code handling loop with unknown niters.
15159 (pass_loop_distribution::execute): Skip loop with unknown niters.
15161 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15164 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
15165 function sort_partitions_by_post_order.
15167 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15169 PR tree-optimization/81374
15170 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
15171 the max index of basic blocks, rather than number of basic blocks.
15173 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15175 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
15177 (arc_legitimate_pic_operand_p): Likewise.
15178 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
15180 (arc_needs_pcl_p): Likewise.
15181 (arc_legitimate_pc_offset_p): Likewise.
15182 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
15183 function is also used in constrains.md.
15184 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
15185 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
15186 PLUS. Only return true/false in known cases, otherwise assert.
15187 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
15188 is already called in arc_legitimate_constant_p.
15189 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
15191 (LEGITIMATE_PIC_OPERAND_P): Use
15192 arc_raw_symbolic_reference_mentioned_p function.
15193 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
15198 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15199 Andrew Burgess <andrew.burgess@embecosm.com>
15201 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
15202 (arc_return_address_register): New function.
15203 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
15204 (arc_handle_fndecl_attribute): Add naked attribute.
15205 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15206 (TARGET_WARN_FUNC_RETURN): Likewise.
15207 (arc_allocate_stack_slots_for_args): New function.
15208 (arc_warn_func_return): Likewise.
15209 (machine_function): Change type fn_type.
15210 (arc_compute_function_type): Consider new naked function type,
15211 change function return type.
15212 (arc_must_save_register): Adapt to handle new
15213 arc_compute_function_type's return type.
15214 (arc_expand_prologue): Likewise.
15215 (arc_expand_epilogue): Likewise.
15216 (arc_return_address_regs): Delete.
15217 (arc_return_address_register): New function.
15218 (arc_epilogue_uses): Use above function.
15219 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
15220 (arc_function_type): Change encoding, add naked type.
15221 (ARC_INTERRUPT_P): Change to handle the new encoding.
15222 (ARC_FAST_INTERRUPT_P): Likewise.
15223 (ARC_NORMAL_P): Define.
15224 (ARC_NAKED_P): Likewise.
15225 (arc_compute_function_type): Delete prototype.
15226 * config/arc/arc.md (in_ret_delay_slot): Use
15227 arc_return_address_register function.
15228 (simple_return): Likewise.
15229 (p_return_i): Likewise.
15231 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15233 PR tree-optimization/81428
15234 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
15235 can't be built for those types.
15237 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
15239 Remove stuff dead since r239246.
15241 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
15242 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
15243 (avr_inform_devices): Remove dead stuff.
15245 2017-07-17 Tamar Christina <tamar.christina@arm.com>
15247 * config/arm/arm_neon.h: Fix softp typo.
15249 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15251 PR tree-optimization/81365
15252 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
15253 aggregate moves onto bb predecessor edges, make sure there are no
15254 loads that could alias the lhs in between the start of bb and the
15257 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
15260 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
15261 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
15262 [LSHIFTRT, outer_code = TRUNCATE]: Same.
15264 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15266 PR tree-optimization/81396
15267 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
15268 (init_symbolic_number): Initialize it to 1.
15269 (perform_symbolic_merge): Add n_ops from both operands into the new
15271 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
15272 without base_addr as useless if they need more than one operation.
15273 (bswap_replace): Handle !bswap case for NULL base_addr.
15275 2017-07-17 Tom de Vries <tom@codesourcery.com>
15278 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
15281 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
15283 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
15284 conditional builtin define __FIX_LEON3FT_B2BST.
15286 2017-07-17 Daniel Cederman <cederman@gaisler.com>
15288 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
15289 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
15292 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
15294 PR rtl-optimization/81424
15295 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
15296 to remove potential trapping from operands if -fnon-call-exceptions.
15298 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15300 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
15301 profile_proability for scalling.
15302 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
15304 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15306 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
15308 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15310 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
15311 fixpoint arithmetics.
15313 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15315 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
15316 fixpoint arithmetics.
15318 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15320 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
15321 fixpoint arithmetics.
15323 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15325 * profile-count.h (profile_probability::from_reg_br_prob_note,
15326 profile_probability::to_reg_br_prob_note): New functions.
15327 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
15328 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
15329 * predict.c (probability_reliable_p): Update.
15330 (edge_probability_reliable_p): Update.
15331 (br_prob_note_reliable_p): Update.
15332 (invert_br_probabilities): Update.
15333 (add_reg_br_prob_note): New function.
15334 (combine_predictions_for_insn): Update.
15335 * asan.c (asan_clear_shadow): Update.
15336 * cfgbuild.c (compute_outgoing_frequencies): Update.
15337 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
15338 (update_br_prob_note): Update.
15339 (rtl_verify_edges): Update.
15340 (purge_dead_edges): Update.
15341 (fixup_reorder_chain): Update.
15342 * emit-rtl.c (try_split): Update.
15343 * ifcvt.c (cond_exec_process_insns): Update.
15344 (cond_exec_process_if_block): Update.
15345 (dead_or_predicable): Update.
15346 * internal-fn.c (expand_addsub_overflow): Update.
15347 (expand_neg_overflow): Update.
15348 (expand_mul_overflow): Update.
15349 * loop-doloop.c (doloop_modify): Update.
15350 * loop-unroll.c (compare_and_jump_seq): Update.
15351 * optabs.c (emit_cmp_and_jump_insn_1): Update.
15352 * predict.h: Update.
15353 * reorg.c (mostly_true_jump): Update.
15355 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
15356 * config/alpha/alpha.c (emit_unlikely_jump): Update.
15357 * config/arc/arc.c: (emit_unlikely_jump): Update.
15358 * config/arm/arm.c: (emit_unlikely_jump): Update.
15359 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
15360 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
15361 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
15362 (ix86_print_operand): Update.
15363 (ix86_split_fp_branch): Update.
15364 (predict_jump): Update.
15365 * config/ia64/ia64.c (ia64_print_operand): Update.
15366 * config/mmix/mmix.c (mmix_print_operand): Update.
15367 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
15368 (rs6000_expand_split_stack_prologue): Update.
15369 * config/rs6000/rs6000.c: Update.
15370 * config/s390/s390.c (s390_expand_vec_strlen): Update.
15371 (s390_expand_vec_movstr): Update.
15372 (s390_expand_cs_tdsi): Update.
15373 (s390_expand_split_stack_prologue): Update.
15374 * config/sh/sh.c (sh_print_operand): Update.
15375 (expand_cbranchsi4): Update.
15376 (expand_cbranchdi4): Update.
15377 * config/sparc/sparc.c (output_v9branch): Update.
15378 * config/spu/spu.c (get_branch_target): Update.
15379 (ea_load_store_inline): Update.
15380 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
15381 * config/tilepro/tilepro.c: Update.
15383 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
15385 * gimplify.c (mostly_copy_tree_r): Revert latest change.
15386 (gimplify_save_expr): Likewise.
15388 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15390 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
15392 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15394 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
15396 * timevar.def (TV_IPA_FNSUMMARY): Define.
15398 2017-07-16 Daniel Cederman <cederman@gaisler.com>
15400 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
15401 to back store errata sensitive sequence from being generated.
15402 (sqrtdf2_fix): Likewise.
15404 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15406 * tree-ssa-threadupdate.c (compute_path_counts,
15407 update_joiner_offpath_counts): Use profile_probability.
15409 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
15412 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15414 * config/arm/arm-c.c (arm_cpu_builtins): Define
15415 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
15417 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
15419 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15420 array entries to represent __ieee128 versions of the
15421 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
15422 scalar_extract_sig, and scalar_insert_exp built-in functions.
15423 (altivec_resolve_overloaded_builtin): Add special case handling
15424 for the __builtin_scalar_insert_exp function, as represented by
15425 the P9V_BUILTIN_VEC_VSIEDP constant.
15426 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
15427 exponent support for __ieee128 argument.
15428 (VSESQP): Add scalar extract signature support for __ieee128
15430 (VSTDCNQP): Add scalar test negative support for __ieee128
15432 (VSIEQP): Add scalar insert exponent support for __int128 argument
15433 with __ieee128 result.
15434 (VSIEQPF): Add scalar insert exponent support for __ieee128
15435 argument with __ieee128 result.
15436 (VSTDCQP): Add scalar test data class support for __ieee128
15438 (VSTDCNQP): Add overload support for scalar test negative with
15439 __ieee128 argument.
15440 (VSTDCQP): Add overload support for scalar test data class
15441 __ieee128 argument.
15442 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
15443 UNSPEC_VSX_SXSIGDP.
15444 (UNSPEC_VSX_SIEXPQP): New constant.
15445 (xsxexpqp): New insn for VSX scalar extract exponent quad
15447 (xsxsigqp): New insn for VSX scalar extract significand quad
15449 (xsiexpqpf): New insn for VSX scalar insert exponent quad
15450 precision with floating point argument.
15451 (xststdcqp): New expand for VSX scalar test data class quad
15453 (xststdcnegqp): New expand for VSX scalar test negative quad
15455 (xststdcqp): New insn to match expansions for VSX scalar test data
15456 class quad precision and VSX scalar test negative quad precision.
15457 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
15458 special case operand checking to enforce that second operand of
15459 VSX scalar test data class with quad precision argument is a 7-bit
15461 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
15462 prototypes and descriptions of __ieee128 versions of
15463 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
15464 scalar_test_data_class, and scalar_test_neg built-in functions.
15466 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15468 PR tree-optimization/81162
15469 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
15470 replace a negate with an add.
15472 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
15474 * doc/invoke.texi (arm/-mcpu): Document +crypto.
15476 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15478 * config/arm/arm-c.c (arm_cpu_builtins): Define
15479 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
15481 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15483 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
15484 (armv8-r): Set ARM Cortex-R52 as default CPU.
15485 * config/arm/arm-tables.opt: Regenerate.
15486 * config/arm/arm-tune.md: Regenerate.
15487 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
15489 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
15490 extension for -mcpu=cortex-r52.
15492 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15494 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
15495 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
15496 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
15497 (fp-armv8): Define it as FP_ARMv8 only.
15498 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
15499 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
15501 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
15502 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
15503 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
15504 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
15505 than TARGET_FPU_ARMV8.
15506 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
15507 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
15508 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
15510 * config/arm/neon.md (neon_vrint): Likewise.
15511 (neon_vcvt): Likewise.
15512 (neon_<fmaxmin_op><mode>): Likewise.
15513 (<fmaxmin><mode>3): Likewise.
15514 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
15515 * config/arm/predicates.md (arm_cond_move_operator): Check against
15516 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
15518 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
15520 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
15521 to top of function.
15523 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15525 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
15526 loop in comment with memset.
15528 2017-07-14 Martin Liska <mliska@suse.cz>
15530 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
15531 * dwarf2out.c (is_java): Remove the function.
15532 (output_pubname): Remove usage of the function.
15533 (lower_bound_default): Remove usage of DW_LANG_Java.
15534 (gen_compile_unit_die): Likewise.
15535 * gcc.c: Remove compiler defaults for .java and .zip files.
15536 * gimple-expr.c (remove_suffix): Change as there's no longer
15537 extension than 4-letter one.
15538 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
15539 (gimplify_save_expr): Likewise.
15540 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
15541 as it's possible even for other languages than Java.
15542 * langhooks.h (struct lang_hooks): Remove Java from a comment.
15543 * lto-opts.c (lto_write_options): Remove reference to Java.
15544 * opts.c (strip_off_ending): Update file extension handling.
15545 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
15546 * tree-eh.c (lower_resx): Likewise.
15547 * tree.c (free_lang_data_in_type): Remove dead code.
15548 (find_decls_types_r): Likewise.
15549 (build_common_builtin_nodes): Remove Java from a comment.
15550 (verify_type): Remove dead code.
15551 * varasm.c (assemble_external): Remove Java from a comment.
15553 2017-07-14 Martin Liska <mliska@suse.cz>
15555 * opts.c (finish_options): Add quotes.
15556 (common_handle_option): Likewise.
15558 2017-07-14 Martin Liska <mliska@suse.cz>
15560 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
15561 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
15562 Remove N_SO_PASCAL.
15563 * dwarf2out.c (lower_bound_default): Do not handle
15565 (gen_compile_unit_die): Likewise.
15566 * gcc.c: Remove default extension binding for GNU Pascal.
15567 * stmt.c: Remove Pascal language from a comment.
15568 * xcoffout.c: Likewise.
15570 2017-07-13 David Malcolm <dmalcolm@redhat.com>
15573 * diagnostic-show-locus.c (fixit_cmp): New function.
15574 (layout::layout): Sort m_fixit_hints.
15575 (column_range::column_range): Assert that the values are valid.
15576 (struct char_span): New struct.
15577 (correction::overwrite): New method.
15578 (struct source_line): New struct.
15579 (line_corrections::add_hint): Add assertions. Reimplement memcpy
15580 calls in terms of classes source_line and char_span, and
15581 correction::overwrite.
15582 (selftest::test_overlapped_fixit_printing_2): New function.
15583 (selftest::diagnostic_show_locus_c_tests): Call it.
15585 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
15587 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
15588 early if there is no lhs.
15590 2017-07-13 Martin Liska <mliska@suse.cz>
15592 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
15593 (gen_reference_type_die): Likewise.
15594 * stor-layout.c: Remove Pascal-related comment.
15596 2017-07-13 Martin Liska <mliska@suse.cz>
15598 * opts.c (finish_options): Add quotes to error messages.
15599 (parse_sanitizer_options): Likewise.
15601 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
15603 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
15605 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
15607 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
15609 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
15611 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
15613 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
15615 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
15618 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
15619 provides the hardware capability bits, define the macro
15620 __BUILTIN_CPU_SUPPORTS__.
15621 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
15622 if GLIBC does not provide the hardware capability bits. Add a
15623 gcc_unreachable call if the built-in cpu function is neither
15624 __builtin_cpu_is nor __builtin_cpu_supports.
15625 (rs6000_get_function_versions_dispatcher): Change the warning
15626 that an old GLIBC is used which does not export the capability
15627 bits to be an error.
15628 * doc/extend.texi (target_clones attribute): Document the
15629 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
15630 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
15631 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
15632 the macros defined by GCC if the newer GLIBC is available.
15634 2017-07-12 Jeff Law <law@redhat.com>
15636 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
15637 remaining includes slightly.
15638 * config/riscv/riscv-builtins.c: Include profile-count.h.
15640 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
15643 * config/avr/avr.c (avr_set_current_function): In diagnostic
15644 messages: Quote keywords and (parts of) identifiers.
15645 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
15648 2017-07-12 Carl Love <cel@us.ibm.com>
15650 * config/rs6000/rs6000-c.c: Add support for built-in functions
15651 vector bool char vec_revb (vector bool char);
15652 vector bool short vec_revb (vector short char);
15653 vector bool int vec_revb (vector bool int);
15654 vector bool long long vec_revb (vector bool long long);
15655 * doc/extend.texi: Update the built-in documentation file for the
15656 new built-in functions.
15658 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15660 * config/s390/s390.md: Remove movcc splitter.
15662 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15664 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
15665 load/store on condition.
15667 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
15670 * config/avr/avr.c (avr_encode_section_info)
15671 [progmem && !TREE_READONLY]: Error if progmem object needs
15674 2017-07-11 Michael Collison <michael.collison@arm.com>
15676 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
15679 2017-07-11 Carl Love <cel@us.ibm.com>
15681 * config/rs6000/rs6000-c.c: Add support for builtins
15682 vector unsigned int vec_parity_lsbb (vector signed int);
15683 vector unsigned int vec_parity_lsbb (vector unsigned int);
15684 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
15685 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
15686 vector unsigned long long vec_parity_lsbb (vector signed long long);
15687 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
15688 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
15689 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
15690 * doc/extend.texi: Update the built-in documentation file for the
15691 new built-in functions.
15693 2017-07-11 David Malcolm <dmalcolm@redhat.com>
15695 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
15696 (layout::m_primary_loc): New field.
15697 (layout::layout): Initialize new field. Move location filtering
15698 logic from here to...
15699 (layout::maybe_add_location_range): ...this new method. Add
15700 support for filtering to just the lines already specified by other
15702 (layout::will_show_line_p): New method.
15703 (gcc_rich_location::add_location_if_nearby): New method.
15704 (selftest::test_add_location_if_nearby): New test function.
15705 (selftest::diagnostic_show_locus_c_tests): Call it.
15706 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
15709 2017-07-11 Tom de Vries <tom@codesourcery.com>
15711 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
15712 (bb_first_real_insn): New function.
15713 (nvptx_single): Add extra initialization of broadcasted condition
15716 2017-07-11 Nathan Sidwell <nathan@acm.org>
15718 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
15720 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
15722 * doc/extend.texi (AVR Function Attributes): Remove weblink to
15723 Binutils doc as TEXI will mess them up.
15724 * doc/invoke.texi (AVR Options): Same here.
15726 2017-07-11 Daniel Cederman <cederman@gaisler.com>
15728 * config/sparc/sparc.opt (mfix-ut700): New option.
15729 (mfix-gr712rc): Likewise.
15730 (sparc_fix_b2bst): New variable.
15731 * doc/invoke.texi (SPARC options): Document them.
15732 (ARM options): Fix warnings.
15733 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
15734 instructions to prevent sequences that can trigger the store-store
15735 errata for certain LEON3FT processors.
15736 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
15737 (sparc_option_override): Set sparc_fix_b2bst appropriately.
15738 * config/sparc/sparc.md (fix_b2bst): New attribute.
15739 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
15741 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
15744 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
15746 (*rsqrtsf2_sse): Ditto.
15748 (div<mode>3): Macroize insn from divdf3 and divsf3
15749 using MODEF mode iterator.
15751 2017-07-10 Martin Sebor <msebor@redhat.com>
15753 PR tree-optimization/80397
15754 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
15755 instead of testing for equality to INTEGER_TYPE.
15757 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
15759 * config.gcc: Remove uclibc from arc target spec.
15761 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
15763 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
15765 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15768 * lto-wrapper.c (remove_option): New function.
15769 (merge_and_complain): Merge PIC/PIE options more realistically.
15771 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
15773 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
15777 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
15778 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
15779 * config.in: Regenerate.
15780 * configure: Regenerate.
15781 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
15782 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
15783 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
15784 (TARGET_GASISR_PROLOGUES): ...target mask.
15785 * common/config/avr/avr-common.c
15786 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
15787 Set -mgas-isr-prologues.
15788 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
15789 INSERT_PASS_BEFORE for it.
15790 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
15791 * config/avr/avr.c (avr_option_override)
15792 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
15793 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
15794 (avr_attribute_table) <no_gccisr>: Add new function attribute.
15795 (avr_set_current_function) <is_no_gccisr>: Init machine field.
15796 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
15798 (make_avr_pass_pre_proep): New function.
15799 (emit_push_sfr) <treg>: Add argument to function and use it
15800 instead of TMP_REG.
15801 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
15802 and set machine->gasisr.yes.
15803 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
15804 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
15805 __gcc_isr.n_pushed to .L__stack_usage.
15806 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
15807 (avr_asm_final_postscan_insn): ...this new static function.
15808 * config/avr/avr.h (machine_function)
15809 <is_no_gccisr, use_L__stack_usage>: New fields.
15810 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
15811 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
15812 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
15813 (gasisr, *gasisr): New expander and insn.
15814 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
15815 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
15816 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
15818 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
15820 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
15823 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
15825 Move jump-tables out of .text again.
15828 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
15829 (ASM_OUTPUT_ADDR_VEC): New function.
15830 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
15831 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
15832 INSN_ADDRESSes as asm comment.
15833 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
15834 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
15835 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
15836 * config/avr/avr.md (*tablejump): Adjust comment.
15837 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
15838 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
15840 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
15841 (avr_output_addr_vec): New proto.
15842 (avr_log_t) <insn_addresses>: New field.
15844 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
15847 * config/i386/i386.c (ix86_function_arg_advance): Set
15848 outgoing_args_on_stack to true if there are outgoing arguments
15850 (ix86_function_arg): Likewise.
15851 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
15852 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
15853 * config/i386/i386.h (machine_function): Add
15854 outgoing_args_on_stack.
15856 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
15858 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
15859 supporting pthreds.
15860 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
15862 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
15864 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
15865 (REAL_H): Remove $(MACHMODE_H).
15866 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
15868 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
15869 $(MACHMODE_H) and double-int.h.
15870 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
15872 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
15875 2017-07-07 Andrew Pinski <apinski@cavium.com>
15877 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
15878 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
15880 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15882 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
15883 Add warning if GCC was not configured to link against a GLIBC that
15884 exports the hardware capability bits.
15885 (make_resolver_func): Make resolver function private and not a
15886 COMDAT function. Create the name with clone_function_name instead
15887 of make_unique_name.
15890 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
15891 correct operand in doing the split.
15893 2017-07-07 Carl Love <cel@us.ibm.com>
15895 * config/rs6000/rs6000-c: Add support for built-in function
15896 vector unsigned short vec_pack_to_short_fp32 (vector float,
15898 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
15899 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
15900 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
15901 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
15902 (convert_4f32_8i16): Add define_expand.
15903 * doc/extend.texi: Update the built-in documentation file for the
15904 new built-in function.
15906 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15908 * config/sparc/m8.md: New file.
15909 * config/sparc/sparc.md: Include m8.md.
15911 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15913 * config/sparc/sparc.opt: New option -mvis4b.
15914 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
15915 (sparc_option_override): Handle VIS4B.
15916 (enum sparc_builtins): Define
15917 SPARC_BUILTIN_DICTUNPACK{8,16,32},
15918 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
15919 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
15920 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
15921 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
15922 (check_constant_argument): New function.
15923 (sparc_vis_init_builtins): Define builtins
15924 __builtin_vis_dictunpack{8,16,32},
15925 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
15926 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
15927 __builtin_vis_fpcmpde{8,16,32}shl and
15928 __builtin_vis_fpcmpur{8,16,32}shl.
15929 (sparc_expand_builtin): Check that the constant operands to
15930 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
15931 constant and in range.
15932 * config/sparc/sparc-c.c (sparc_target_macros): Handle
15934 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
15935 (SPARC_IMM5_P): Likewise.
15936 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
15937 (enabled): Handle vis4b.
15938 (UNSPEC_DICTUNPACK): New unspec.
15939 (UNSPEC_FPCMPSHL): Likewise.
15940 (UNSPEC_FPUCMPSHL): Likewise.
15941 (UNSPEC_FPCMPDESHL): Likewise.
15942 (UNSPEC_FPCMPURSHL): Likewise.
15943 (cpu_feature): New CPU feature `vis4b'.
15944 (dictunpack{8,16,32}): New insns.
15945 (FPCSMODE): New mode iterator.
15946 (fpcscond): New code iterator.
15947 (fpcsucond): Likewise.
15948 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
15949 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
15950 (fpcmpde{8,16,32}{si,di}shl): Likewise.
15951 (fpcmpur{8,16,32}{si,di}shl): Likewise.
15952 * config/sparc/constraints.md: Define constraints `q' for unsigned
15953 2-bit integer constants and `t' for unsigned 5-bit integer
15955 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
15957 (imm5_operand_dictunpack16): Likewise.
15958 (imm5_operand_dictunpack32): Likewise.
15959 (imm2_operand): Likewise.
15960 * doc/invoke.texi (SPARC Options): Document -mvis4b.
15961 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
15962 ditunpack* and fpcmp*shl builtins.
15964 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15966 * config.gcc: Handle m8 in --with-{cpu,tune} options.
15967 * config.in: Add HAVE_AS_SPARC6 define.
15968 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
15970 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
15972 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
15973 (CPP_CPU_SPEC): Handle m8.
15974 (ASM_CPU_SPEC): Likewise.
15975 * config/sparc/sparc-opts.h (enum processor_type): Add
15977 * config/sparc/sparc.c (m8_costs): New struct.
15978 (sparc_option_override): Handle TARGET_CPU_m8.
15979 (sparc32_initialize_trampoline): Likewise.
15980 (sparc64_initialize_trampoline): Likewise.
15981 (sparc_issue_rate): Likewise.
15982 (sparc_register_move_cost): Likewise.
15983 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
15984 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
15985 (ASM_CPU64_DEFAULT_SPEC): Likewise.
15986 (CPP_CPU_SPEC): Handle M8.
15987 (ASM_CPU_SPEC): Likewise.
15988 (AS_M8_FLAG): Define.
15989 * config/sparc/sparc.md: Add m8 to the cpu attribute.
15990 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
15991 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
15993 * configure: Regenerate.
15994 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
15997 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15999 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
16001 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
16002 ("*movdi_insn_sp32"): Do not set v3pipe.
16003 ("*movsi_insn"): Likewise.
16004 ("*movdi_insn_sp64"): Likewise.
16005 ("*movsf_insn"): Likewise.
16006 ("*movdf_insn_sp32"): Likewise.
16007 ("*movdf_insn_sp64"): Likewise.
16008 ("*zero_extendsidi2_insn_sp64"): Likewise.
16009 ("*sign_extendsidi2_insn"): Likewise.
16010 ("*mov<VM32:mode>_insn"): Likewise.
16011 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16012 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16013 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16014 ("<vlop:code><VL:mode>3"): Likewise.
16015 ("*not_<vlop:code><VL:mode>3"): Likewise.
16016 ("*nand<VL:mode>_vis"): Likewise.
16017 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
16018 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
16019 ("one_cmpl<VL:mode>2"): Likewise.
16020 ("faligndata<VM64:mode>_vis"): Likewise.
16021 ("alignaddrsi_vis"): Likewise.
16022 ("alignaddrdi_vis"): Likweise.
16023 ("alignaddrlsi_vis"): Likewise.
16024 ("alignaddrldi_vis"): Likewise.
16025 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16026 ("bmaskdi_vis"): Likewise.
16027 ("bmasksi_vis"): Likewise.
16028 ("bshuffle<VM64:mode>_vis"): Likewise.
16029 ("cmask8<P:mode>_vis"): Likewise.
16030 ("cmask16<P:mode>_vis"): Likewise.
16031 ("cmask32<P:mode>_vis"): Likewise.
16032 ("pdistn<P:mode>_vis"): Likewise.
16033 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16035 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16037 * config/sparc/sparc.md ("subtype"): New insn attribute.
16038 ("*wrgsr_sp64"): Set insn subtype.
16039 ("*rdgsr_sp64"): Likewise.
16040 ("alignaddrsi_vis"): Likewise.
16041 ("alignaddrdi_vis"): Likewise.
16042 ("alignaddrlsi_vis"): Likewise.
16043 ("alignaddrldi_vis"): Likewise.
16044 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16045 ("fexpand_vis"): Likewise.
16046 ("fpmerge_vis"): Likewise.
16047 ("faligndata<VM64:mode>_vis"): Likewise.
16048 ("bshuffle<VM64:mode>_vis"): Likewise.
16049 ("cmask8<P:mode>_vis"): Likewise.
16050 ("cmask16<P:mode>_vis"): Likewise.
16051 ("cmask32<P:mode>_vis"): Likewise.
16052 ("fchksm16_vis"): Likewise.
16053 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
16054 ("fmean16_vis"): Likewise.
16055 ("fp<plusminus_insn>64_vis"): Likewise.
16056 ("<plusminus_insn>v8qi3"): Likewise.
16057 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16058 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
16059 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
16060 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
16061 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
16062 ("*movqi_insn"): Likewise.
16063 ("*movhi_insn"): Likewise.
16064 ("*movsi_insn"): Likewise.
16065 ("movsi_pic_gotdata_op"): Likewise.
16066 ("*movdi_insn_sp32"): Likewise.
16067 ("*movdi_insn_sp64"): Likewise.
16068 ("movdi_pic_gotdata_op"): Likewise.
16069 ("*movsf_insn"): Likewise.
16070 ("*movdf_insn_sp32"): Likewise.
16071 ("*movdf_insn_sp64"): Likewise.
16072 ("*zero_extendhisi2_insn"): Likewise.
16073 ("*zero_extendqihi2_insn"): Likewise.
16074 ("*zero_extendqisi2_insn"): Likewise.
16075 ("*zero_extendqidi2_insn"): Likewise.
16076 ("*zero_extendhidi2_insn"): Likewise.
16077 ("*zero_extendsidi2_insn_sp64"): Likewise.
16078 ("ldfsr"): Likewise.
16079 ("prefetch_64"): Likewise.
16080 ("prefetch_32"): Likewise.
16081 ("tie_ld32"): Likewise.
16082 ("tie_ld64"): Likewise.
16083 ("*tldo_ldub_sp32"): Likewise.
16084 ("*tldo_ldub1_sp32"): Likewise.
16085 ("*tldo_ldub2_sp32"): Likewise.
16086 ("*tldo_ldub_sp64"): Likewise.
16087 ("*tldo_ldub1_sp64"): Likewise.
16088 ("*tldo_ldub2_sp64"): Likewise.
16089 ("*tldo_ldub3_sp64"): Likewise.
16090 ("*tldo_lduh_sp32"): Likewise.
16091 ("*tldo_lduh1_sp32"): Likewise.
16092 ("*tldo_lduh_sp64"): Likewise.
16093 ("*tldo_lduh1_sp64"): Likewise.
16094 ("*tldo_lduh2_sp64"): Likewise.
16095 ("*tldo_lduw_sp32"): Likewise.
16096 ("*tldo_lduw_sp64"): Likewise.
16097 ("*tldo_lduw1_sp64"): Likewise.
16098 ("*tldo_ldx_sp64"): Likewise.
16099 ("*mov<VM32:mode>_insn"): Likewise.
16100 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16101 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16103 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16105 * config/sparc/sparc.md ("type"): New insn type viscmp.
16106 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
16108 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
16109 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
16110 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16111 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
16113 ("n7_vis_logical_11cycle"): Likewise.
16114 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
16115 * config/sparc/niagara2.md ("niag3_vis": Likewise.
16116 * config/sparc/niagara.md ("niag_vis"): Likewise.
16117 * config/sparc/ultra3.md ("us3_fga"): Likewise.
16118 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
16120 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16122 * config/sparc/sparc.md: New instruction type `bmask'.
16123 (bmaskdi_vis): Use the `bmask' type.
16124 (bmasksi_vis): Likewise.
16125 * config/sparc/ultra3.md (us3_array): Likewise.
16126 * config/sparc/niagara7.md (n7_array): Likewise.
16127 * config/sparc/niagara4.md (n4_array): Likewise.
16128 * config/sparc/niagara2.md (niag2_vis): Likewise.
16129 (niag3_vis): Likewise.
16130 * config/sparc/niagara.md (niag_vis): Likewise.
16132 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16134 * ipa-comdats.c: Remove optimize check from gate.
16135 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
16136 for functions not optimized.
16137 (ipa_fn_summary_read): Skip optimize check.
16138 (ipa_fn_summary_write): Likewise.
16139 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
16141 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
16143 (can_inline_edge_p): Check flag_pcc_struct_return for match.
16144 (check_callers): Give up on caller which is not optimized.
16145 (inline_small_functions): Likewise.
16146 (ipa_inline): Do not give up when not optimizing.
16147 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
16148 away unoptimizes cdtors.
16149 (whole_program_function_and_variable_visibility): Do
16150 ipa_discover_readonly_nonaddressable_vars in LTO mode.
16151 * ipa.c (process_references): Do not check optimize.
16152 (symbol_table::remove_unreachable_nodes): Update optimize check.
16153 (set_writeonly_bit): Update optimize check.
16154 (pass_ipa_cdtor_merge::gate): Do not check optimize.
16155 (pass_ipa_single_use::gate): Remove.
16157 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16159 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
16160 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
16161 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
16162 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
16163 permute_load, permute_store, adjust_extract, adjust_splat,
16164 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
16165 replace_swap_with_copy, dump_swap_insn_table,
16166 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
16167 recombine_lvx_pattern, recombine_stvx_pattern,
16168 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
16169 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
16170 to file rs6000-p8swap.c.
16171 * config/rs6000/rs6000-p8swap.c: New file.
16172 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
16173 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
16174 and rs6000*-*-* targets.
16176 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16178 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
16180 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16182 * lto-wrapper.c (merge_and_complain): Do not merge
16183 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
16184 fsigned_zeros, ftrapping_math, fwrapv.
16185 (append_compiler_options): Do not track these options.
16186 (append_linker_options): Likewie
16188 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16190 * cgraphunit.c (cgraph_node::finalize_function): When
16191 !flag_toplevel_reorde set no_reorder flag.
16192 (varpool_node::finalize_decl): Likewise.
16193 (symbol_table::compile): Drop no toplevel reorder path.
16195 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16197 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
16198 edges; zero probability is not better than uninitialized.
16200 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
16202 * asan.h (asan_sanitize_allocas_p): Declare.
16203 * asan.c (asan_sanitize_allocas_p): New function.
16204 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
16205 (handle_builtin_alloca): Likewise.
16206 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
16207 if !asan_sanitize_allocas_p.
16208 * params.def (asan-instrument-allocas): Add new option.
16209 * params.h (ASAN_PROTECT_ALLOCAS): Define.
16210 * opts.c (common_handle_option): Disable allocas sanitization for
16213 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
16215 * asan.c: Include gimple-fold.h.
16216 (get_last_alloca_addr): New function.
16217 (handle_builtin_stackrestore): Likewise.
16218 (handle_builtin_alloca): Likewise.
16219 (asan_emit_allocas_unpoison): Likewise.
16220 (get_mem_refs_of_builtin_call): Add new parameter, remove const
16221 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
16222 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
16223 (instrument_builtin_call): Pass gimple iterator to
16224 get_mem_refs_of_builtin_call.
16225 (last_alloca_addr): New global.
16226 * asan.h (asan_emit_allocas_unpoison): Declare.
16227 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
16228 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
16229 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
16230 if function calls alloca.
16231 * gimple-fold.c (replace_call_with_value): Remove static keyword.
16232 * gimple-fold.h (replace_call_with_value): Declare.
16233 * internal-fn.c: Include asan.h.
16234 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
16235 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
16237 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16239 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
16240 (C_SELFTEST_FLAGS): New.
16241 (CPP_SELFTEST_FLAGS): New.
16242 (SELFTEST_DEPS): New, from deps of s-selftest.
16243 (C_SELFTEST_DEPS): New, from deps of s-selftest.
16244 (CPP_SELFTEST_DEPS): New.
16245 (selftest): Add dependency on s-selftest-c++.
16246 (s-selftest): Rename to...
16247 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
16248 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
16249 than SELFTEST_FLAGS.
16250 (selftest-gdb): Rename to...
16251 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
16253 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
16254 (selftest-valgrind): Rename to...
16255 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
16257 (selftest-valgrind): Reintroduce as an alias for
16258 selftest-c-valgrind.
16259 (s-selftest-c++): New.
16260 (selftest-c++-gdb): New.
16261 (selftest-c++-valgrind): New.
16263 2017-07-06 Olivier Hainque <hainque@adacore.com>
16265 * gcc.c (process_command): When deciding if undefined variables
16266 should be ignored when processing specs, accept "gcc -v" as well.
16268 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16270 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
16271 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
16273 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
16275 * config/arm/arm-cpus.in (armv8-r): Add new entry.
16276 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
16277 * config/arm/arm-tables.opt: Regenerate.
16278 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
16280 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
16282 2017-07-06 Carl Love <cel@us.ibm.com>
16284 * ChangeLog: Clean up from mid air collision
16286 2017-07-06 Carl Love <cel@us.ibm.com>
16288 * config/rs6000/rs6000-c.c: Add support for built-in functions
16289 vector signed int vec_subc (vector signed int, vector signed int);
16290 vector signed __int128 vec_subc (vector signed __int128,
16291 vector signed __int128);
16292 vector unsigned __int128 vec_subc (vector unsigned __int128,
16293 vector unsigned __int128);
16294 vector signed int vec_sube (vector signed int, vector signed int,
16295 vector signed int);
16296 vector unsigned int vec_sube (vector unsigned int,
16297 vector unsigned int,
16298 vector unsigned int);
16299 vector signed __int128 vec_sube (vector signed __int128,
16300 vector signed __int128,
16301 vector signed__int128);
16302 vector unsigned __int128 vec_sube (vector unsigned __int128,
16303 vector unsigned __int128,
16304 vector unsigned __int128);
16305 vector signed int vec_subec (vector signed int, vector signed int,
16306 vector signed int);
16307 vector unsigned int vec_subec (vector unsigned int,
16308 vector unsigned int,
16309 vector unsigned int);
16310 vector signed __int128 vec_subec (vector signed __int128,
16311 vector signed __int128,
16312 vector signed__int128);
16313 vector unsigned __int128 vec_subec (vector unsigned __int128,
16314 vector unsigned __int128,
16315 vector unsigned __int128);
16316 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
16317 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
16318 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
16319 BU_ALTIVEC_OVERLOAD_X definitions.
16320 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
16321 * doc/extend.texi: Update the built-in documentation file for the new
16322 built-in functions.
16324 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16327 * diagnostic-show-locus.c (layout::layout): Use start and finish
16328 spelling location for the start and finish of each range.
16329 * genmatch.c (linemap_client_expand_location_to_spelling_point):
16330 Add unused aspect param.
16331 * input.c (expand_location_1): Add "aspect" param, and use it
16332 to access the correct part of the location.
16333 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
16335 (expand_location_to_spelling_point): Likewise.
16336 (linemap_client_expand_location_to_spelling_point): Add "aspect"
16337 param, and pass it to expand_location_1.
16339 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
16341 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
16342 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
16343 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
16344 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
16345 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
16346 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
16347 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
16348 _mm_maskz_getmant_ss): New intrinsics.
16349 (__builtin_ia32_getexpss128_mask): Changed to ...
16350 __builtin_ia32_getexpss128_round ... this.
16351 (__builtin_ia32_getexpsd128_mask): Changed to ...
16352 __builtin_ia32_getexpsd128_round ... this.
16353 * config/i386/i386-builtin-types.def
16354 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
16355 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
16356 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
16357 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
16358 __builtin_ia32_getmantss_mask_round): New builtins.
16359 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
16360 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
16361 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
16362 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
16363 * config/i386/sse.md
16364 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
16365 avx512f_sgetexp<mode><mask_scalar_name>
16366 <round_saeonly_scalar_name> ... this.
16367 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
16368 %0, %1, %2<round_saeonly_op3>}): Changed to ...
16369 vgetexp<ssescalarmodesuffix>
16370 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16371 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
16372 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
16373 avx512f_vgetmant<mode><mask_scalar_name>
16374 <round_saeonly_scalar_name> ... this.
16375 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
16376 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
16377 vgetmant<ssescalarmodesuffix>
16378 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
16379 %0<mask_scalar_operand4>, %1, %2
16380 <round_saeonly_scalar_mask_op4>, %3} ... this.
16381 * config/i386/subst.md (mask_scalar_operand4,
16382 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
16383 round_saeonly_scalar_nimm_predicate): New subst attributes.
16385 2017-07-06 Julia Koval <julia.koval@intel.com>
16387 * config/i386/i386.c (ix86_erase_embedded_rounding):
16388 Remove code for old rounding pattern.
16390 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
16392 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
16394 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
16396 * doc/sourcebuild.texi (Test Directives, Variants of
16397 dg-require-support): Add documentation for dg-require-stack-check.
16399 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
16401 * config/i386/subst.md (mask_scalar, round_scalar,
16402 round_saeonly_scalar): New meta-templates.
16403 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
16404 round_scalar_mask_operand3, round_scalar_mask_op3,
16405 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
16406 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
16407 round_saeonly_scalar_constraint,
16408 round_saeonly_scalar_prefix): New subst attribute.
16409 * config/i386/sse.md
16410 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
16411 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
16412 <round_scalar_name> ... this.
16413 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
16414 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
16415 <round_scalar_name> ... this.
16416 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
16417 <sse>_vm<code><mode>3<mask_scalar_name>
16418 <round_saeonly_scalar_name> ... this.
16419 (v<plusminus_mnemonic><ssescalarmodesuffix>
16420 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
16421 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
16422 v<plusminus_mnemonic><ssescalarmodesuffix>
16423 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16424 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
16425 (v<multdiv_mnemonic><ssescalarmodesuffix>
16426 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
16427 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
16428 v<multdiv_mnemonic><ssescalarmodesuffix>
16429 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16430 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
16431 (v<maxmin_float><ssescalarmodesuffix>
16432 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
16433 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
16434 v<maxmin_float><ssescalarmodesuffix>
16435 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16436 %0<mask_scalar_operand3>, %1, %<iptr>2
16437 <round_saeonly_scalar_mask_op3>} ... this.
16439 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
16441 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
16442 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
16444 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
16445 Alan Hayward <alan.hayward@arm.com>
16446 David Sherwood <david.sherwood@arm.com>
16448 * combine.c (simplify_if_then_else): Remove "enum" before
16450 * compare-elim.c (can_eliminate_compare): Likewise.
16451 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
16453 (aarch64_lookup_simd_builtin_type): Likewise.
16454 (aarch64_simd_builtin_type): Likewise.
16455 (aarch64_init_simd_builtin_types): Likewise.
16456 (aarch64_simd_expand_args): Likewise.
16457 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
16459 (aarch64_reverse_mask): Likewise.
16460 (aarch64_simd_emit_reg_reg_move): Likewise.
16461 (aarch64_gen_adjusted_ldpstp): Likewise.
16462 (aarch64_ccmp_mode_to_code): Likewise.
16463 (aarch64_operands_ok_for_ldpstp): Likewise.
16464 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16465 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
16467 (aarch64_min_divisions_for_recip_mul): Likewise.
16468 (aarch64_reassociation_width): Likewise.
16469 (aarch64_get_condition_code_1): Likewise.
16470 (aarch64_simd_emit_reg_reg_move): Likewise.
16471 (aarch64_simd_attr_length_rglist): Likewise.
16472 (aarch64_reverse_mask): Likewise.
16473 (aarch64_operands_ok_for_ldpstp): Likewise.
16474 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16475 (aarch64_gen_adjusted_ldpstp): Likewise.
16476 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
16478 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
16479 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
16480 (arm_lookup_simd_builtin_type): Likewise.
16481 (arm_simd_builtin_type): Likewise.
16482 (arm_init_simd_builtin_types): Likewise.
16483 (arm_expand_builtin_args): Likewise.
16484 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
16485 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
16486 (ft32_setup_incoming_varargs): Likewise.
16487 (ft32_function_arg): Likewise.
16488 (ft32_function_arg_advance): Likewise.
16489 (ft32_pass_by_reference): Likewise.
16490 (ft32_arg_partial_bytes): Likewise.
16491 (ft32_valid_pointer_mode): Likewise.
16492 (ft32_addr_space_pointer_mode): Likewise.
16493 (ft32_addr_space_legitimate_address_p): Likewise.
16494 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
16496 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
16497 (ix86_emit_outlined_ms2sysv_restore): Likewise.
16498 (iamcu_alignment): Likewise.
16499 (canonicalize_vector_int_perm): Likewise.
16500 (ix86_noce_conversion_profitable_p): Likewise.
16501 (ix86_mpx_bound_mode): Likewise.
16502 (ix86_operands_ok_for_move_multiple): Likewise.
16503 * config/microblaze/microblaze-protos.h
16504 (microblaze_expand_conditional_branch_reg): Likewise.
16505 * config/microblaze/microblaze.c
16506 (microblaze_expand_conditional_branch_reg): Likewise.
16507 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
16509 (rs6000_reassociation_width): Likewise.
16510 (rs6000_invalid_binary_op): Likewise.
16511 (fusion_p9_p): Likewise.
16512 (emit_fusion_p9_load): Likewise.
16513 (emit_fusion_p9_store): Likewise.
16514 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
16516 (riscv_hard_regno_mode_ok_p): Likewise.
16517 (riscv_address_insns): Likewise.
16518 (riscv_split_symbol): Likewise.
16519 (riscv_legitimize_move): Likewise.
16520 (riscv_function_value): Likewise.
16521 (riscv_hard_regno_nregs): Likewise.
16522 (riscv_expand_builtin): Likewise.
16523 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
16524 (riscv_build_integer): Likewise.
16525 (riscv_split_integer): Likewise.
16526 (riscv_legitimate_constant_p): Likewise.
16527 (riscv_cannot_force_const_mem): Likewise.
16528 (riscv_regno_mode_ok_for_base_p): Likewise.
16529 (riscv_valid_base_register_p): Likewise.
16530 (riscv_valid_offset_p): Likewise.
16531 (riscv_valid_lo_sum_p): Likewise.
16532 (riscv_classify_address): Likewise.
16533 (riscv_legitimate_address_p): Likewise.
16534 (riscv_address_insns): Likewise.
16535 (riscv_load_store_insns): Likewise.
16536 (riscv_force_binary): Likewise.
16537 (riscv_split_symbol): Likewise.
16538 (riscv_force_address): Likewise.
16539 (riscv_legitimize_address): Likewise.
16540 (riscv_move_integer): Likewise.
16541 (riscv_legitimize_const_move): Likewise.
16542 (riscv_legitimize_move): Likewise.
16543 (riscv_address_cost): Likewise.
16544 (riscv_subword): Likewise.
16545 (riscv_output_move): Likewise.
16546 (riscv_canonicalize_int_order_test): Likewise.
16547 (riscv_emit_int_order_test): Likewise.
16548 (riscv_function_arg_boundary): Likewise.
16549 (riscv_pass_mode_in_fpr_p): Likewise.
16550 (riscv_pass_fpr_single): Likewise.
16551 (riscv_pass_fpr_pair): Likewise.
16552 (riscv_get_arg_info): Likewise.
16553 (riscv_function_arg): Likewise.
16554 (riscv_function_arg_advance): Likewise.
16555 (riscv_arg_partial_bytes): Likewise.
16556 (riscv_function_value): Likewise.
16557 (riscv_pass_by_reference): Likewise.
16558 (riscv_setup_incoming_varargs): Likewise.
16559 (riscv_print_operand): Likewise.
16560 (riscv_elf_select_rtx_section): Likewise.
16561 (riscv_save_restore_reg): Likewise.
16562 (riscv_for_each_saved_reg): Likewise.
16563 (riscv_register_move_cost): Likewise.
16564 (riscv_hard_regno_mode_ok_p): Likewise.
16565 (riscv_hard_regno_nregs): Likewise.
16566 (riscv_class_max_nregs): Likewise.
16567 (riscv_memory_move_cost): Likewise.
16568 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
16569 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
16570 (rl78_addr_space_address_mode): Likewise.
16571 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16573 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
16574 (rs6000_reassociation_width): Likewise.
16575 (rs6000_invalid_binary_op): Likewise.
16576 (fusion_p9_p): Likewise.
16577 (emit_fusion_p9_load): Likewise.
16578 (emit_fusion_p9_store): Likewise.
16579 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
16580 (ok_for_simple_move_operands): Likewise.
16581 (ok_for_simple_move_strict_operands): Likewise.
16582 (ok_for_simple_arith_logic_operands): Likewise.
16583 (visium_legitimize_reload_address): Likewise.
16584 (visium_select_cc_mode): Likewise.
16585 (output_cbranch): Likewise.
16586 (visium_split_double_move): Likewise.
16587 (visium_expand_copysign): Likewise.
16588 (visium_expand_int_cstore): Likewise.
16589 (visium_expand_fp_cstore): Likewise.
16590 * config/visium/visium.c (visium_pass_by_reference): Likewise.
16591 (visium_function_arg): Likewise.
16592 (visium_function_arg_advance): Likewise.
16593 (visium_libcall_value): Likewise.
16594 (visium_setup_incoming_varargs): Likewise.
16595 (visium_legitimate_constant_p): Likewise.
16596 (visium_legitimate_address_p): Likewise.
16597 (visium_legitimize_address): Likewise.
16598 (visium_secondary_reload): Likewise.
16599 (visium_register_move_cost): Likewise.
16600 (visium_memory_move_cost): Likewise.
16601 (prepare_move_operands): Likewise.
16602 (ok_for_simple_move_operands): Likewise.
16603 (ok_for_simple_move_strict_operands): Likewise.
16604 (ok_for_simple_arith_logic_operands): Likewise.
16605 (visium_function_value_1): Likewise.
16606 (rtx_ok_for_offset_p): Likewise.
16607 (visium_legitimize_reload_address): Likewise.
16608 (visium_split_double_move): Likewise.
16609 (visium_expand_copysign): Likewise.
16610 (visium_expand_int_cstore): Likewise.
16611 (visium_expand_fp_cstore): Likewise.
16612 (visium_split_cstore): Likewise.
16613 (visium_select_cc_mode): Likewise.
16614 (visium_split_cbranch): Likewise.
16615 (output_cbranch): Likewise.
16616 (visium_print_operand_address): Likewise.
16617 * expmed.c (flip_storage_order): Likewise.
16618 * expmed.h (emit_cstore): Likewise.
16619 (flip_storage_order): Likewise.
16620 * genrecog.c (validate_pattern): Likewise.
16621 * hsa-gen.c (gen_hsa_addr): Likewise.
16622 * internal-fn.c (expand_arith_overflow): Likewise.
16623 * ira-color.c (allocno_copy_cost_saving): Likewise.
16624 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16625 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
16626 (process_invariant_for_inheritance): Likewise.
16627 * lra-eliminations.c (move_plus_up): Likewise.
16628 * omp-low.c (lower_oacc_reductions): Likewise.
16629 * simplify-rtx.c (simplify_subreg): Likewise.
16630 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
16631 (TARGET_CHKP_BOUND_MODE): Likewise..
16632 * targhooks.c (default_chkp_bound_mode): Likewise.
16633 (default_setup_incoming_vararg_bounds): Likewise.
16634 * targhooks.h (default_chkp_bound_mode): Likewise.
16635 (default_setup_incoming_vararg_bounds): Likewise.
16636 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
16637 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
16638 (have_whole_vector_shift): Likewise.
16639 * tree-vect-stmts.c (vectorizable_load): Likewise.
16640 * doc/tm.texi: Regenerate.
16642 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
16644 Graceful degrade if Binutils PR21472 is not available.
16647 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
16648 .rodata in flash test fails.
16649 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
16650 * confgure: Regenerate.
16651 * config.in: Regenerate.
16652 * config/avr/avr.c (avr_asm_named_section)
16653 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
16654 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
16655 (avr_asm_init_sections): Same.
16657 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16659 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
16660 (fma<VH:mode>4_intrinsic): Likewise.
16661 (*fmsub<VCVTF:mode>4): Likewise.
16662 (*fmsub<VH:mode>4_intrinsic): Likewise.
16664 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
16667 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
16668 Don't depend on "optimize > 0".
16669 (out_movhi_r_mr, out_movqi_mr_r): Same.
16670 (out_movhi_mr_r, out_movqi_r_mr): Same.
16671 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
16672 io_address_operand on "optimize > 0".
16674 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16676 * tree-loop-distribution.c: Add general explanantion on the pass.
16677 (generate_loops_for_partition): Mark distributed loop.
16678 (pg_add_dependence_edges): New parameter. Handle alias data
16679 dependence specially and record it in the parameter if asked.
16680 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
16681 (init_partition_graph_vertices, add_partition_graph_edge): New.
16682 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
16683 (free_partition_graph_vdata, build_partition_graph): New.
16684 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
16685 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
16686 (data_ref_segment_size, latch_dominated_by_data_ref): New.
16687 (compute_alias_check_pairs, version_loop_by_alias_check): New.
16688 (version_for_distribution_p, finalize_partitions): New.
16689 (distribute_loop): Handle alias data dependence specially. Factor
16690 out loop fusion code as functions and call these functions.
16692 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16694 * tree-loop-distribution.c (classify_partition): New parameter and
16695 better handle reduction statement.
16696 (rdg_build_partitions): Revise comment.
16697 (distribute_loop): Compute statements in all partitions and pass it
16698 to classify_partition.
16700 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16702 * tree-loop-distribution.c (enum partition_type): New.
16703 (struct partition): New field type.
16704 (partition_merge_into): Add parameter. Update partition type.
16705 (data_dep_in_cycle_p, update_type_for_merge): New functions.
16706 (build_rdg_partition_for_vertex): Compute partition type.
16707 (rdg_build_partitions): Dump partition type.
16708 (distribute_loop): Update calls to partition_merge_into.
16710 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16712 * tree-loop-distribution.c (struct ddr_hasher): New.
16713 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
16715 (classify_partition): Call get_data_dependence.
16716 (pg_add_dependence_edges): Ditto.
16717 (distribute_loop): Release data dependence hash table.
16719 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16721 * tree-loop-distribution.c (ref_base_address): Delete.
16722 (similar_memory_accesses): Rename ...
16723 (share_memory_accesses): ... to this. Check if partitions access
16724 the same memory reference.
16725 (distribute_loop): Call share_memory_accesses.
16727 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16729 * tree-loop-distribution.c (struct partition): New field recording
16730 its data reference.
16731 (partition_alloc, partition_free): Init and release data refs.
16732 (partition_merge_into): Merge data refs.
16733 (build_rdg_partition_for_vertex): Collect data refs for partition.
16734 (pg_add_dependence_edges): Change parameters from vector to bitmap.
16736 (distribute_loop): Remve data refs from vertice data of partition
16739 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16741 * tree-loop-distribution.c (params.h): Include header file.
16742 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
16743 (datarefs_vec): New global var.
16744 (create_rdg_vertices): Use datarefs_vec directly.
16745 (free_rdg): Don't free data references.
16746 (build_rdg): Update use. Don't free data references.
16747 (distribute_loop): Compute global variable for data references.
16748 Bail out if there are too many data references.
16750 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16752 * tree-loop-distribution.c (loop_nest): New global var.
16753 (build_rdg): Use loop directly, rather than loop nest.
16754 (pg_add_dependence_edges): Remove loop nest parameter. Use global
16756 (distribute_loop): Compute global variable loop nest. Update use.
16758 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16760 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
16761 (partition_merge_into): New parameter. Dump reason for fusion.
16762 (distribute_loop): Update use of partition_merge_into.
16764 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16766 * tree-loop-distribution.c (bb_top_order_index): New.
16767 (bb_top_order_index_size, bb_top_order_cmp): New.
16768 (stmts_from_loop): Use topological order.
16769 (pass_loop_distribution::execute): Compute and release topological
16770 order for basic blocks.
16772 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16774 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
16777 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16779 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
16780 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
16781 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
16782 * internal-fn.def (LOOP_DIST_ALIAS): New.
16783 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
16784 (fold_loop_internal_call): ... this.
16785 (vect_loop_dist_alias_call): New function.
16786 (set_uid_loop_bbs): Call fold_loop_internal_call.
16787 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
16790 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
16793 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
16794 Require dead FLAGS_REG at the beginning of a peephole.
16796 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
16799 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
16800 arguments in the call to __builtin_ia32_sbb_u32.
16801 (_subborrow_u64): Swap _X and _Y arguments in the call to
16802 __builtin_ia32_sbb_u64.
16804 2017-07-04 Jakub Jelinek <jakub@redhat.com>
16807 * tree-vrp.c (compare_assert_loc): Turn into a function template
16808 with stable template parameter. Only test if a->e is NULL,
16809 !a->e == !b->e has been verified already. Use e == NULL or
16810 e != NULL instead of e or ! e tests. If stable is true, don't use
16811 iterative_hash_expr, on the other side allow a or b or both NULL
16812 and sort the NULLs last.
16813 (process_assert_insertions): Sort using compare_assert_loc<false>
16814 instead of compare_assert_loc, later sort using
16815 compare_assert_loc<true> before calling process_assert_insertions_for
16816 in a loop. Use break instead of continue once seen NULL pointer.
16818 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
16820 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
16821 Cortex-R7 and Cortex-R8 processors.
16823 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
16825 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
16826 uninitialized while src is not.
16828 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
16830 * common/config/arm/arm-common.c: Adjust include path for
16832 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
16833 (arm-cpu.h): Create in build directory. Adjust dependency rules.
16834 (arm-cpu-data.h): Likewise.
16835 (arm-cpu-cdata.h): Likewise.
16836 * config/arm/arm-cpu.h: Delete.
16837 * config/arm/arm-cpu-cdata.h: Delete.
16838 * config/arm/arm-cpu-data.h: Delete.
16840 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
16842 * config/arm/arm-cpus.in (cortex-a55): New.
16843 (cortex-a75): Likewise.
16844 (cortex-a75.cortex-a55): Likewise.
16845 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
16847 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
16848 * config/arm/arm-cpu-cdata.h: Regenerate.
16849 * config/arm/arm-cpu-data.h: Regenerate.
16850 * config/arm/arm-cpu.h: Regenerate.
16851 * config/arm/arm-tables.opt: Regenerate.
16852 * config/arm/arm-tune.md: Regenerate.
16854 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
16856 * haifa-sched.c (sched_create_recovery_edges): Update profile.
16858 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
16860 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
16863 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
16865 PR tree-optimization/81292
16866 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
16867 full_string_p, also call adjust_related_strinfos if the adjustment
16868 is simple, otherwise invalidate related strinfos.
16870 2017-07-04 Martin Liska <mliska@suse.cz>
16873 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
16874 newly created variable as DECL_IGNORED_P.
16876 2017-07-04 Martin Liska <mliska@suse.cz>
16879 * ipa-inline.c (inline_small_functions):
16880 Use xstrdup_for_dump.
16882 2017-07-04 Tom de Vries <tom@codesourcery.com>
16884 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
16886 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
16889 * config/darwin.c (darwin_function_switched_text_sections):
16892 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
16894 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
16896 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
16898 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
16900 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16902 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
16903 min_profitable_iters, and th as inclusive lower bounds.
16904 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
16905 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
16906 for min_profitable_iters and min_profitable_estimate.
16907 (vect_transform_loop): Treat th as an inclusive lower bound.
16908 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
16910 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
16913 * config/darwin.c (darwin_function_switched_text_sections):
16914 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
16915 in two pieces, and suppress the use of buf.
16917 2017-07-03 Nathan Sidwell <nathan@acm.org>
16919 * hash-table.h (hash_table_mod1): Fix indentation.
16921 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16923 PR middle-end/81290
16924 * predict.c (force_edge_cold): Be more careful about propagation
16926 * profile-count.h (profile_probability::guessed,
16927 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
16929 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
16931 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
16933 * doc/invoke.texi (rcpc architecture extension): Document it.
16935 2017-07-03 Richard Biener <rguenther@suse.de>
16937 PR tree-optimization/60510
16938 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
16939 the scalar reduction PHI and use it.
16940 (vectorizable_reduction): Properly guard the single_defuse_cycle
16941 path for non-SLP reduction chains where we cannot use it.
16942 Rework reduc_def/index and vector type deduction. Rework
16943 vector operand gathering during reduction op code-gen.
16944 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
16945 chains dissolve the chain and leave it to non-SLP reduction
16948 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16950 * tree-data-ref.h (dr_alignment): Declare.
16951 * tree-data-ref.c (dr_alignment): New function.
16952 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
16953 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
16955 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
16957 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16959 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
16960 and base_misalignment fields.
16961 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
16962 * tree-data-ref.c: Include builtins.h.
16963 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
16964 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
16965 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
16966 * tree-vect-data-refs.c: Include tree-cfg.h.
16967 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
16968 fields instead of calculating an alignment here.
16969 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
16970 innermost_loop_behavior fields.
16972 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16974 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
16976 (DR_STEP_ALIGNMENT): New macro.
16977 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
16978 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
16979 (create_data_ref): Print it.
16980 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
16981 to tell whether the step preserves vector (mis)alignment.
16982 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
16983 Move the check for an integer step and generalise to all INTEGER_CST.
16984 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
16985 Print the outer step alignment.
16987 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16989 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
16990 with offset_alignment.
16991 (DR_ALIGNED_TO): Delete.
16992 (DR_OFFSET_ALIGNMENT): New macro.
16993 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
16994 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
16995 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
16996 (create_data_ref): Likewise.
16997 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
16998 (vect_analyze_data_refs): Likewise.
16999 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
17000 creating dummy innermost behavior.
17002 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17004 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
17005 with a "innermost_loop_behavior *" and refeence tree.
17006 * tree-data-ref.c (dr_analyze_innermost): Likewise.
17007 (create_data_ref): Update call accordingly.
17008 * tree-predcom.c (find_looparound_phi): Likewise.
17010 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17012 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
17013 fields with dr_wrt_vec_loop.
17014 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
17015 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
17016 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
17017 (vect_dr_behavior): New function.
17018 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17019 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
17020 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
17021 track whether the step preserves the misalignment.
17022 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17023 Use vect_dr_behavior.
17024 (vect_setup_realignment): Update call accordingly.
17025 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
17026 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
17027 call to vect_create_addr_base_for_vector_ref.
17028 (vect_create_cond_for_align_checks): Likewise.
17029 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
17030 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
17031 (vect_recog_mask_conversion_pattern): Likewise.
17032 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
17033 (new_stmt_vec_info): Remove redundant zeroing.
17035 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17037 * common/config/arm/arm-common.c (arm_be8_option): New function.
17038 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
17039 (ISA_ARMv6): Add isa_bit_be8.
17040 * config/arm/arm.h (arm_be8_option): Add prototype.
17041 (BE8_SPEC_FUNCTION): New define.
17042 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
17043 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
17044 (mlittle-endian): Similarly.
17045 (mbe8, mbe32): New options.
17046 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
17047 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
17049 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17051 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
17053 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17055 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
17056 (cleanup_tree_cfg_bb): Use it.
17057 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
17059 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
17061 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17064 * loop-doloop.c (add_test): Update profile.
17066 2017-07-03 Martin Liska <mliska@suse.cz>
17069 * sanopt.c (rewrite_usage_of_param): New function.
17070 (sanitize_rewrite_addressable_params): Likewise.
17071 (pass_sanopt::execute): Call rewrite_usage_of_param.
17073 2017-07-03 Richard Biener <rguenther@suse.de>
17075 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
17076 back to using VIEW_CONVERT_EXPR.
17078 2017-07-03 Martin Liska <mliska@suse.cz>
17081 * doc/extend.texi: Document when a resolver function is
17082 generated for target_clones.
17084 2017-07-03 Martin Liska <mliska@suse.cz>
17086 * asan.c (asan_emit_stack_protection): Unpoison just red zones
17087 and shadow memory of auto variables which are subject of
17088 use-after-scope sanitization.
17089 (asan_expand_mark_ifn): Add do set only when is_poison.
17091 2016-07-03 Richard Biener <rguenther@suse.de>
17093 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
17095 (vect_force_simple_reduction): Record reduction def -> phi mapping.
17096 (vectorizable_reduction): Perform reduction PHI creation when
17097 visiting a reduction PHI and adjust and simplify code generation
17098 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
17099 (vect_transform_loop): Visit reduction PHIs.
17100 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
17101 defs into the SLP tree.
17102 (vect_build_slp_tree): Reduction defs terminate the recursion.
17103 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
17105 (vect_get_vec_defs_for_stmt_copy): Export.
17106 (vect_get_vec_defs): Likewise.
17107 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
17109 (vect_get_vec_defs_for_stmt_copy): Declare.
17110 (vect_get_vec_defs): Likewise.
17112 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17114 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
17115 parameter with a "loop" parameter and use it instead of the
17116 loop containing DR_STMT. Don't check simple_iv when doing
17117 BB analysis. Describe the two analysis modes in the comment.
17119 2017-07-03 Tom de Vries <tom@codesourcery.com>
17121 PR tree-optimization/69468
17122 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
17123 (find_same_succ_bb): Handle ignore_edge_flags.
17125 2017-07-03 Tom de Vries <tom@codesourcery.com>
17127 PR tree-optimization/81192
17128 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
17130 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
17132 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
17134 2017-07-03 Tom de Vries <tom@codesourcery.com>
17136 PR tree-optimization/81192
17137 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
17138 BB_SAME_SUCC (bb) == NULL.
17140 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17142 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
17145 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17147 * dumpfile.c: Include profile-count.h
17148 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
17150 (insert_cond_bb): Update profile.
17151 * tree-cfg.h (insert_cond_bb): Update prototype.
17152 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
17153 * tree-dump.c: Do not include tree-cfg.
17155 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17157 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
17159 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17161 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
17164 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17166 * tree-complex.c (expand_complex_div_wide): update profile.
17168 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17169 Alan Hayward <alan.hayward@arm.com>
17170 David Sherwood <david.sherwood@arm.com>
17172 * Makefile.in (MACHMODE_H): Remove insn-modes.h
17173 (CORETYPES_H): New define.
17174 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
17175 (insn-modes-inline.h, s-modes-inline-h): New rules.
17176 (generated_files): Add insn-modes-inline.h.
17177 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
17178 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
17179 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
17180 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
17181 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
17182 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
17183 (build/gencodes.o, build/genconditions.o): Likewise.
17184 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
17185 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
17186 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
17187 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
17188 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
17189 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
17190 * coretypes.h: Include everything up to real.h for generators.
17191 Include insn-modes.h first. Include wide-int-print.h after
17192 wide-int.h. Include insn-modes-inline.h and then machmode.h.
17193 * machmode.h: Don't include insn-modes.h here.
17194 * function-tests.c: Remove includes of signop.h, machmode.h,
17195 double-int.h and wide-int.h.
17197 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
17199 * optc-save-gen.awk: Likewise.
17200 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
17201 * godump.c: Remove include of wide-int-print.h.
17202 * pretty-print.h: Likewise.
17203 * wide-int-print.cc: Likewise.
17204 * wide-int.cc: Likewise.
17205 * hash-map-tests.c: Remove include of signop.h.
17206 * hash-set-tests.c: Likewise.
17207 * rtl-tests.c: Likewise.
17208 * mkconfig.sh: Remove include of machmode.h.
17209 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
17211 (emit_insn_modes_inline_h): ...this new function. Emit the code
17212 into an insn-modes-inline.h header file, adding appropriate
17213 include guards and end comments.
17214 (emit_insn_modes_c_header): Remove include of machmode.h.
17215 (emit_min_insn_modes_c_header): Include coretypes.h rather than
17217 (main): Handle -i flag and call emit_insn_modes_inline_h when
17220 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17222 * tree-ssa-strlen.c (strinfo): Rename the length field to
17223 nonzero_chars. Add a full_string_p field.
17224 (compare_nonzero_chars, zero_length_string_p): New functions.
17225 (get_addr_stridx): Add an offset_out parameter.
17226 Use compare_nonzero_chars.
17227 (get_stridx): Update accordingly. Use compare_nonzero_chars.
17228 (new_strinfo): Update after above changes to strinfo.
17229 (set_endptr_and_length): Set full_string_p.
17230 (get_string_length): Update after above changes to strinfo.
17231 (unshare_strinfo): Update call to new_strinfo.
17232 (maybe_invalidate): Likewise.
17233 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
17234 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
17235 as a uhwi instead of an shwi. Update after above changes to
17236 strinfo and new_strinfo.
17237 (zero_length_string): Assert that chainsi contains full strings.
17238 Use zero_length_string_p. Update call to new_strinfo.
17239 (adjust_related_strinfos): Update after above changes to strinfo.
17240 Copy full_string_p from origsi.
17241 (adjust_last_stmt): Use zero_length_string_p.
17242 (handle_builtin_strlen): Update after above changes to strinfo and
17243 new_strinfo. Install the lhs as the string length if the previous
17244 entry didn't describe a full string.
17245 (handle_builtin_strchr): Update after above changes to strinfo
17247 (handle_builtin_strcpy): Likewise.
17248 (handle_builtin_strcat): Likewise.
17249 (handle_builtin_malloc): Likewise.
17250 (handle_pointer_plus): Likewise.
17251 (handle_builtin_memcpy): Likewise. Track nonzero characters
17252 that aren't necessarily followed by a nul terminator.
17253 (handle_char_store): Likewise.
17255 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17257 PR tree-optimization/80769
17258 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
17259 for malloc and calloc. Document the new invariant that all related
17260 strinfos have delayed lengths or none do.
17261 (verify_related_strinfos): Move earlier in file.
17262 (set_endptr_and_length): New function, split out from...
17263 (get_string_length): ...here. Also set the lengths of related
17265 (zero_length_string): Assert that chainsi has known (rather than
17267 (adjust_related_strinfos): Likewise.
17269 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17271 PR tree-optimization/81136
17272 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
17273 assert that two references with the same misalignment have the same
17274 compile-time misalignment if those compile-time misalignments
17277 2017-07-01 Andi Kleen <ak@linux.intel.com>
17279 * print-tree.c (print_node): Print all attributes.
17281 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17283 * cfg.c (scale_bbs_frequencies): New function.
17284 * cfg.h (scale_bbs_frequencies): Declare it.
17285 * cfgloopanal.c (single_likely_exit): Cleanup.
17286 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
17288 (scale_loop_profile): Likewise.
17289 (loop_version): Likewise.
17290 (create_empty_loop_on_edge): Update.
17291 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
17292 scale_loop_frequencies, scale_loop_profile, loopify,
17293 loop_version): Update prototypes.
17294 * modulo-sched.c (sms_schedule): Update.
17295 * predict.c (unlikely_executed_edge_p): Also check probability.
17296 (probably_never_executed_edge_p): Fix typo.
17297 * tree-if-conv.c (version_loop_for_if_conversion): Update.
17298 * tree-parloops.c (gen_parallel_loop): Update.
17299 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
17300 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
17301 * tree-ssa-loop-split.c (split_loop): Update.
17302 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
17303 * tree-vect-loop-manip.c (vect_do_peeling): Update.
17304 (vect_loop_versioning): Update.
17305 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
17307 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17309 * trans-mem.c (split_bb_make_tm_edge): Update profile.
17311 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17313 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
17314 to keep profile consistent.
17316 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17318 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
17319 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
17320 * profile-count.h (max_safe_multiplier): Make unsigned.
17321 (profile_count::guessed_zero): New.
17323 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17325 * bb-reorder.c (fix_up_crossing_landing_pad,
17326 fix_crossing_conditional_branches): Use make_single_succ_edge
17327 to keep profile consistent.
17329 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17331 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
17334 2017-07-01 Jakub Jelinek <jakub@redhat.com>
17337 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
17338 the right scopes, make sure cond_jump isn't preserved between multiple
17339 iterations. Search for fallthru edge whenever there are 3+ edges and
17340 use find_fallthru_edge for it.
17342 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17344 Patch by Alexander Monakov <amonakov@ispras.ru>
17345 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
17346 probabilities consistently.
17348 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17350 * pa.c (pa_expand_compare_and_swap_loop): Update call of
17351 emit_cmp_and_jump_insns.
17353 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17356 * tree-inline.c (expand_call_inline): Combine profile statuses.
17358 2017-06-30 Andrew Pinski <apinski@cavium.com>
17360 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
17361 fold_stmt returned true.
17363 2017-06-30 Nathan Sidwell <nathan@acm.org>
17365 * ggc.h (empty_string): Delete.
17366 * cfgexpand.c (expand_asm_stmt): Use plain "".
17367 * optabs.c (expand_asm_memory_barrier): Likewise.
17368 * stringpool.c (empty_string): Delete.
17369 (digit_vector, digit_string): Delete.
17370 (ggc_alloc_string): Use plain "", don't optimize single digit
17371 strings. Use ggc_alloc_atomic.
17373 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
17375 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
17376 comparison set and one other set, use the cost of the non-comparison
17379 2017-06-30 Nathan Sidwell <nathan@acm.org>
17381 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
17384 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
17386 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
17387 loops. Remove now unneeded calls to gimple_switch_set_label() that
17388 just set removed labels to NULL_TREE.
17390 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
17392 * tree-ssanames.c (set_range_info_raw): Abstract from ...
17393 (set_range_info): ...here. Only call set_range_info_raw if domain
17395 (set_nonzero_bits): Call set_range_info_raw.
17396 * tree-ssanames.h (set_range_info_raw): New.
17398 2017-06-30 Jakub Jelinek <jakub@redhat.com>
17401 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
17402 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
17403 of nonimmediate_operand and <store_mask_constraint> instead of m
17404 for the input operand. For V8FI iterator, always split if input
17405 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
17406 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
17407 <store_mask_predicate> instead of register_operand and
17408 <store_mask_constraint> instead of v for the input operand. Make
17409 sure both operands aren't MEMs for if not <mask_applied>.
17411 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
17413 * lto-wrapper.c (copy_file) Close both file descriptors before
17416 2017-06-30 Martin Liska <mliska@suse.cz>
17419 * multiple_target.c (create_dispatcher_calls): Make ifunc
17420 also for function that don't have calls or are not referenced.
17422 2017-06-30 Richard Biener <rguenther@suse.de>
17424 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
17425 analyze the first scalar stmt. Move vector type computation
17426 for the BB case here from ...
17427 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
17428 live operation processing in the SLP case properly.
17430 2017-06-30 Richard Biener <rguenther@suse.de>
17432 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
17434 2017-06-30 Martin Liska <mliska@suse.cz>
17437 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
17438 before BUILT_IN_UNWIND_RESUME when ASAN is used.
17440 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
17442 * doc/invoke.texi (AArch64): Add missing options and remove redundant
17445 2017-06-30 Richard Biener <rguenther@suse.de>
17447 PR tree-optimization/81249
17448 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
17449 condition reduction result to original scalar type.
17451 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17453 * profile-count.h (enum profile_quality): Fix typos and whitespace
17456 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17458 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
17459 type for branch probabilities.
17461 2017-06-29 Julian Brown <julian@codesourcery.com>
17462 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17464 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
17465 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
17466 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
17467 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
17469 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17471 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
17472 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
17473 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
17474 CC usage from generic code to here.
17475 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
17476 CC usage into the target macros.
17478 2017-06-29 Maya Rashish <coypu@sdf.org>
17480 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
17483 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17485 * arm/arm-builtins.c: Include profile-count.h
17486 * except.c (sjlj_emit_function_enter): Use
17487 profile_probability::unlikely.
17489 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17491 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
17492 and tocrel_offset be pointer args rather than implicitly using
17494 (legitimate_constant_pool_address_p, rs6000_emit_move,
17495 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
17496 tocrel_offset and use in toc_relative_expr_p call.
17497 (print_operand, print_operand_address): Use static tocrel_base_oac
17498 and tocrel_offset_oac.
17499 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
17502 2017-06-29 Maya Rashish <coypu@sdf.org>
17504 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
17506 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
17508 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
17509 objects, take into account only the alignment of 'op0' and 'mode1' if
17512 2017-06-29 Steve Ellcey <sellcey@cavium.com>
17514 * ccmp.c (ccmp_tree_comparison_p): New function.
17515 (ccmp_candidate_p): Update to use above function.
17516 (get_compare_parts): New function.
17517 (expand_ccmp_next): Update to use new functions.
17518 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
17520 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
17521 take mode as argument.
17522 * ccmp.h (expand_ccmp_expr): Add mode as argument.
17523 * expr.c (expand_expr_real_1): Pass mode as argument.
17525 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
17527 * combine.c (combine_instructions): Print insns to dump_file, together
17530 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17532 * asan.c (asan_emit_stack_protection): Update.
17533 (create_cond_insert_point): Update.
17534 * auto-profile.c (afdo_propagate_circuit): Update.
17535 * basic-block.h (struct edge_def): Turn probability to
17536 profile_probability.
17537 (EDGE_FREQUENCY): Update.
17538 * bb-reorder.c (find_traces_1_round): Update.
17539 (better_edge_p): Update.
17540 (sanitize_hot_paths): Update.
17541 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
17542 (make_single_succ_edge): Update.
17543 (check_bb_profile): Update.
17544 (dump_edge_info): Update.
17545 (update_bb_profile_for_threading): Update.
17546 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
17547 probabilitycount to 0.
17548 * cfgbuild.c (compute_outgoing_frequencies): Update.
17549 * cfgcleanup.c (try_forward_edges): Update.
17550 (outgoing_edges_match): Update.
17551 (try_crossjump_to_edge): Update.
17552 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
17553 (expand_gimple_tailcall): Update.
17554 (construct_init_block): Use make_single_succ_edge.
17555 (construct_exit_block): Use make_single_succ_edge.
17556 * cfghooks.c (verify_flow_info): Update.
17557 (redirect_edge_succ_nodup): Update.
17558 (split_edge): Update.
17559 (account_profile_record): Update.
17560 * cfgloopanal.c (single_likely_exit): Update.
17561 * cfgloopmanip.c (scale_loop_profile): Update.
17562 (set_zero_probability): Remove.
17563 (duplicate_loop_to_header_edge): Update.
17564 * cfgloopmanip.h (loop_version): Update prototype.
17565 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
17566 (force_nonfallthru_and_redirect): Update.
17567 (update_br_prob_note): Update.
17568 (rtl_verify_edges): Update.
17569 (purge_dead_edges): Update.
17570 (rtl_lv_add_condition_to_bb): Update.
17571 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
17572 * cgraphunit.c (init_lowered_empty_function): Update.
17573 (cgraph_node::expand_thunk): Update.
17574 * cilk-common.c: Include profile-count.h
17575 * dojump.c (inv): Remove.
17576 (jumpifnot): Update.
17577 (jumpifnot_1): Update.
17578 (do_jump_1): Update.
17580 (do_jump_by_parts_greater_rtx): Update.
17581 (do_compare_rtx_and_jump): Update.
17582 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
17583 do_jump_1. do_compare_rtx_and_jump): Update prototype.
17584 * dwarf2cfi.c: Include profile-count.h
17585 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
17586 (sjlj_emit_dispatch_table): Likewise.
17587 * explow.c: Include profile-count.h
17588 * expmed.c (emit_store_flag_force): Update.
17589 (do_cmp_and_jump): Update.
17590 * expr.c (compare_by_pieces_d::generate): Update.
17591 (compare_by_pieces_d::finish_mode): Update.
17592 (emit_block_move_via_loop): Update.
17593 (store_expr_with_bounds): Update.
17594 (store_constructor): Update.
17595 (expand_expr_real_2): Update.
17596 (expand_expr_real_1): Update.
17597 * expr.h (try_casesi, try_tablejump): Update prototypes.
17598 * gimple-pretty-print.c (dump_probability): Update.
17599 (dump_profile): New.
17600 (dump_gimple_label): Update.
17601 (dump_gimple_bb_header): Update.
17602 * graph.c (draw_cfg_node_succ_edges): Update.
17603 * hsa-gen.c (convert_switch_statements): Update.
17604 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
17605 (find_if_case_1): Update.
17606 (find_if_case_2): Update.
17607 * internal-fn.c (expand_arith_overflow_result_store): Update.
17608 (expand_addsub_overflow): Update.
17609 (expand_neg_overflow): Update.
17610 (expand_mul_overflow): Update.
17611 (expand_vector_ubsan_overflow): Update.
17612 * ipa-cp.c (good_cloning_opportunity_p): Update.
17613 * ipa-split.c (split_function): Use make_single_succ_edge.
17614 * ipa-utils.c (ipa_merge_profiles): Update.
17615 * loop-doloop.c (add_test): Update.
17616 (doloop_modify): Update.
17617 * loop-unroll.c (compare_and_jump_seq): Update.
17618 (unroll_loop_runtime_iterations): Update.
17619 * lra-constraints.c (lra_inheritance): Update.
17620 * lto-streamer-in.c (input_cfg): Update.
17621 * lto-streamer-out.c (output_cfg): Update.
17622 * mcf.c (adjust_cfg_counts): Update.
17623 * modulo-sched.c (sms_schedule): Update.
17624 * omp-expand.c (expand_omp_for_init_counts): Update.
17625 (extract_omp_for_update_vars): Update.
17626 (expand_omp_ordered_sink): Update.
17627 (expand_omp_for_ordered_loops): Update.
17628 (expand_omp_for_generic): Update.
17629 (expand_omp_for_static_nochunk): Update.
17630 (expand_omp_for_static_chunk): Update.
17631 (expand_cilk_for): Update.
17632 (expand_omp_simd): Update.
17633 (expand_omp_taskloop_for_outer): Update.
17634 (expand_omp_taskloop_for_inner): Update.
17635 * omp-simd-clone.c (simd_clone_adjust): Update.
17636 * optabs.c (expand_doubleword_shift): Update.
17637 (expand_abs): Update.
17638 (emit_cmp_and_jump_insn_1): Update.
17639 (expand_compare_and_swap_loop): Update.
17640 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
17641 * predict.c (predictable_edge_p): Update.
17642 (edge_probability_reliable_p): Update.
17643 (set_even_probabilities): Update.
17644 (combine_predictions_for_insn): Update.
17645 (combine_predictions_for_bb): Update.
17646 (propagate_freq): Update.
17647 (estimate_bb_frequencies): Update.
17648 (force_edge_cold): Update.
17649 * profile-count.c (profile_count::dump): Add missing space into dump.
17650 (profile_count::debug): Add newline.
17651 (profile_count::differs_from_p): Explicitly convert to unsigned.
17652 (profile_count::stream_in): Update.
17653 (profile_probability::dump): New member function.
17654 (profile_probability::debug): New member function.
17655 (profile_probability::differs_from_p): New member function.
17656 (profile_probability::differs_lot_from_p): New member function.
17657 (profile_probability::stream_in): New member function.
17658 (profile_probability::stream_out): New member function.
17659 * profile-count.h (profile_count_quality): Rename to ...
17660 (profile_quality): ... this one.
17661 (profile_probability): New.
17662 (profile_count): Update.
17663 * profile.c (compute_branch_probabilities): Update.
17664 * recog.c (peep2_attempt): Update.
17665 * sched-ebb.c (schedule_ebbs): Update.
17666 * sched-rgn.c (find_single_block_region): Update.
17667 (compute_dom_prob_ps): Update.
17668 (schedule_region): Update.
17669 * sel-sched-ir.c (compute_succs_info): Update.
17670 * stmt.c (struct case_node): Update.
17671 (do_jump_if_equal): Update.
17672 (get_outgoing_edge_probs): Update.
17673 (conditional_probability): Update.
17674 (emit_case_dispatch_table): Update.
17675 (expand_case): Update.
17676 (expand_sjlj_dispatch_table): Update.
17677 (emit_case_nodes): Update.
17678 * targhooks.c: Update.
17679 * tracer.c (better_p): Update.
17680 (find_best_successor): Update.
17681 * trans-mem.c (expand_transaction): Update.
17682 * tree-call-cdce.c: Update.
17683 * tree-cfg.c (gimple_split_edge): Upate.
17684 (move_sese_region_to_fn): Upate.
17685 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
17686 * tree-eh.c (lower_resx): Upate.
17687 (cleanup_empty_eh_move_lp): Upate.
17688 * tree-if-conv.c (version_loop_for_if_conversion): Update.
17689 * tree-inline.c (copy_edges_for_bb): Update.
17690 (copy_cfg_body): Update.
17691 * tree-parloops.c (gen_parallel_loop): Update.
17692 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
17693 (gimple_gen_time_profiler): Update.
17694 * tree-ssa-dce.c (remove_dead_stmt): Update.
17695 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
17696 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
17697 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
17698 (unloop_loops): Update.
17699 (try_peel_loop): Update.
17700 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
17701 * tree-ssa-loop-split.c (connect_loops): Update.
17702 (split_loop): Update.
17703 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
17704 (hoist_guard): Update.
17705 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
17706 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
17707 (value_replacement): Update.
17708 * tree-ssa-reassoc.c (branch_fixup): Update.
17709 * tree-ssa-tail-merge.c (replace_block_by): Update.
17710 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
17711 (create_edge_and_update_destination_phis): Update.
17712 (compute_path_counts): Update.
17713 (recompute_probabilities): Update.
17714 (update_joiner_offpath_counts): Update.
17715 (freqs_to_counts_path): Update.
17716 (duplicate_thread_path): Update.
17717 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
17718 (struct switch_conv_info): Update.
17719 (gen_inbound_check): Update.
17720 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
17721 (vect_do_peeling): Update.
17722 (vect_loop_versioning): Update.
17723 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
17724 (optimize_mask_stores): Update.
17725 * ubsan.c (ubsan_expand_null_ifn): Update.
17726 * value-prof.c (gimple_divmod_fixed_value): Update.
17727 (gimple_divmod_fixed_value_transform): Update.
17728 (gimple_mod_pow2): Update.
17729 (gimple_mod_pow2_value_transform): Update.
17730 (gimple_mod_subtract): Update.
17731 (gimple_mod_subtract_transform): Update.
17732 (gimple_ic): Update.
17733 (gimple_stringop_fixed_value): Update.
17734 (gimple_stringops_transform): Update.
17735 * value-prof.h: Update.
17737 2017-06-29 Carl Love <cel@us.ibm.com>
17739 * config/rs6000/rs6000-c.c: Add support for built-in functions
17740 vector signed int vec_signed (vector float);
17741 vector signed long long vec_signed (vector double);
17742 vector signed int vec_signed2 (vector double, vector double);
17743 vector signed int vec_signede (vector double);
17744 vector signed int vec_signedo (vector double);
17745 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
17746 instruction generator.
17747 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
17748 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
17749 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
17751 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
17752 vunsignede_v2df): Add define_expands.
17753 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
17754 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
17755 VEC_UNSIGNEDO): Add definitions.
17756 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
17757 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
17758 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
17759 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
17760 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
17761 * config/rs6000/altivec.h (vec_signed, vec_signed2,
17762 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
17763 vec_unsignede, vec_unsignedo): Add builtin defines.
17764 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
17766 * doc/extend.texi: Update the built-in documentation file for the
17767 new built-in functions.
17769 2017-06-29 Richard Biener <rguenther@suse.de>
17771 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
17772 reduction chains to LOOP_VINFO_REDUCTIONS.
17773 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
17774 SLP reductions after processing reduction chains.
17776 2017-06-29 Nathan Sidwell <nathan@acm.org>
17778 * builtins.c (fold_builtin_FUNCTION): Use
17779 lang_hooks.decl_printable_name.
17781 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
17783 PR middle-end/81194
17784 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
17785 with only one label.
17786 * stmt.c (expand_case): Assert NCASES is greater than one.
17788 2017-06-29 Richard Biener <rguenther@suse.de>
17790 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
17792 (group_case_labels): Likewise.
17793 (find_taken_edge): Push sanity checking on val to workers...
17794 (find_taken_edge_cond_expr): ... here
17795 (find_taken_edge_switch_expr): ... and here, handle cases
17796 with just a default label.
17797 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
17798 (group_case_labels): Likewise.
17799 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
17800 group_case_labels does anything cleanup the CFG again.
17802 2017-06-29 Bin Cheng <bin.cheng@arm.com>
17804 PR tree-optimization/81196
17805 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
17806 exit condition comparing two IVs.
17808 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
17810 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
17811 profile to the dummy entry at the end of the list of architectures.
17812 * config/arm/arm-cpu-cdata.h: Regenerated.
17814 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17815 Michael Collison <michael.collison@arm.com>
17818 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
17820 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
17821 (*aarch64_reg_<mode>3_minus_mask): New pattern.
17822 (*aarch64_<optab>_reg_di3_mask2): New pattern.
17823 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
17824 of shift when the shift amount is masked with constant equal to
17825 the size of the mode.
17826 * config/aarch64/predicates.md (subreg_lowpart_operator): New
17829 2017-06-29 Martin Liska <mliska@suse.cz>
17831 * config/i386/i386.opt: Change range from [1,5] to [0,5].
17833 2017-06-29 Yury Gribov <tetra2005@gmail.com>
17836 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
17839 (edge_growth_cache_entry::edge_growth_cache_entry): New
17841 (reset_edge_growth_cache): Update to use constructor.
17843 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17845 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
17846 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
17847 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
17849 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
17851 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
17852 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
17854 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
17856 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
17857 (*-linux-uclibc*): Add t-uclibc tmake_file.
17858 * config/t-musl: New.
17859 * config/t-uclibc: New.
17861 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
17863 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
17865 (gen_comm_data): Emit architectural setting of arch_prof.
17866 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
17868 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
17869 (armv8-m.base, armv8-m.main): Likewise.
17870 * arm-protos.h (arm_build_target): Add profile field.
17871 (arch_option): Likewise.
17872 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
17874 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
17875 arm_active_target.profile.
17877 2017-06-28 Richard Biener <rguenther@suse.de>
17879 PR middle-end/81227
17880 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
17881 TYPE_OVERFLOW_WRAPS.
17882 * match.pd (negate_expr_p): Likewise.
17883 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
17884 fold_build2, not fold_binary.
17886 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
17888 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
17889 Convert memory address to Pmode.
17890 (aarch64_print_operand): Assert MEM operands are always Pmode.
17892 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
17895 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
17896 Remove redundant if.
17897 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
17898 * config/arm/aarch-common-protos.h
17899 (aarch_forward_to_shift_is_not_shifted_re): Remove.
17900 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
17902 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
17905 * multiple_target.c (create_dispatcher_calls): Set the default
17906 clone to be static, not public.
17908 2017-06-28 Richard Biener <rguenther@suse.de>
17910 * tree-vect-loop.c (vectorizable_reduction): Move special
17911 cond reduction IV var creation ...
17912 (vect_create_epilog_for_reduction): ... here. Remove induction_index
17913 parameter. Use STMT_VINFO_VECTYPE.
17914 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
17917 2017-06-28 Martin Liska <mliska@suse.cz>
17920 * ipa-visibility.c (non_local_p): Handle visibility.
17922 2017-06-28 Martin Liska <mliska@suse.cz>
17925 * common.opt: Add IntegerRange to various options.
17926 * opt-functions.awk (integer_range_info): New function.
17927 * optc-gen.awk: Add integer_range_info to cl_options struct.
17928 * opts-common.c (decode_cmdline_option): Handle
17929 CL_ERR_INT_RANGE_ARG.
17930 (cmdline_handle_error): Likewise.
17931 * opts.c (print_filtered_help): Show valid interval in
17932 when --help is provided.
17933 * opts.h (struct cl_option): Add range_min and range_max fields.
17934 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
17936 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
17938 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
17939 (x * C EQ/NE y * C): New transformation.
17941 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
17943 * genmultilib (combination_space): Accept '+' in option names.
17945 2017-06-28 Martin Liska <mliska@suse.cz>
17948 * asan.c (instrument_derefs): Bail out inner references
17949 that are hard register variables.
17951 2017-06-28 Jakub Jelinek <jakub@redhat.com>
17954 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
17955 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
17957 2017-06-28 Richard Biener <rguenther@suse.de>
17959 * tree-vectorizer.h (vect_get_vec_defs): Remove.
17960 (vect_get_slp_defs): Adjust.
17961 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
17963 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
17965 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
17966 get_initial_defs_for_reduction instead of vect_get_vec_defs.
17967 (vectorizable_reduction): Adjust.
17968 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
17970 (vect_get_slp_defs): Likewise.
17971 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
17972 (vectorizable_bswap): Adjust.
17973 (vectorizable_call): Likewise.
17974 (vectorizable_conversion): Likewise.
17975 (vectorizable_assignment): Likewise.
17976 (vectorizable_shift): Likewise.
17977 (vectorizable_operation): Likewise.
17978 (vectorizable_store): Likewise.
17979 (vectorizable_condition): Likewise.
17980 (vectorizable_comparison): Likewise.
17982 2017-06-28 Michael Collison <michael.collison@arm.com>
17985 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
17987 (arm_gen_movmemqi): Removed unused variable 'i'.
17988 Convert 'for' loop into 'while' loop.
17989 (arm_expand_prologue): Remove last unnecessary set of insn.
17990 (thumb_pop): Remove unused variable 'pushed_words'.
17991 (thumb_exit): Remove last unnecessary set of regs_to_pop.
17993 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17995 * config/s390/predicates.md: Use s390_rel_address_ok_p.
17996 * config/s390/s390-protos.h: Add prototype of
17997 s390_rel_address_ok_p.
17998 * config/s390/s390.c (s390_got_symbol): New function.
17999 (s390_rel_address_ok_p): New function.
18000 (legitimize_pic_address): Use s390_rel_address_ok_p.
18001 (s390_load_got): Use s390_got_symbol.
18002 (s390_option_override): Issue error if
18003 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
18004 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
18006 * config/s390/s390.opt: New option mpic-data-is-text-relative.
18008 2017-06-27 Andrew Pinski <apinski@cavium.com>
18010 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
18011 (X * copysign (1.0, X)): New pattern.
18012 (X * copysign (1.0, -X)): New pattern.
18013 (copysign (-1.0, CST)): New pattern.
18015 2017-06-27 Joseph Myers <joseph@codesourcery.com>
18017 * genmultilib (combination_space): Remove variable.
18018 Validate reuse rules against regular expression for any sequence
18019 of multilib options in any order.
18021 2017-06-27 Michael Collison <michael.collison@arm.com>
18023 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
18024 call aarch64_split_simd_combine.
18025 * (aarch64_combine_internal<mode>): Delete pattern.
18026 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
18027 Allow register and subreg operands.
18029 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18031 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
18032 specific need, just fallback on defaults.
18033 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
18035 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18036 Olivier Hainque <hainque@adacore.com>
18038 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
18040 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
18041 targets. Pick a default if no particular attempt applied.
18042 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
18045 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18047 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
18048 (x86_64-wrs-vxworks7): Likewise.
18050 2017-06-27 Marek Polacek <polacek@redhat.com>
18053 * ubsan.c (instrument_null): Check get_base_address's result for null.
18055 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18057 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
18059 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18061 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
18062 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
18063 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
18064 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
18065 New function types.
18066 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
18067 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
18068 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
18069 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
18070 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
18071 BUILT_IN_FEUPDATEENV): New builtins.
18072 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
18073 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
18074 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
18075 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
18077 (builtin_structptr_types): Adjust size.
18078 * tree.c (builtin_structptr_types): Add four entries.
18080 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18081 Olivier Hainque <hainque@adacore.com>
18083 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
18084 (TLS_SYM): New local macro, forcing reference to __tls__ on
18085 link command lines for VxWorks 7 RTPs, triggering initialization
18087 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
18088 OS features TLS support, true for RTPs on VxWorks 7.
18089 * config/vxworks.c (vxworks_override_options): Setup emutls
18092 2017-06-27 Jakub Jelinek <jakub@redhat.com>
18094 * predict.c (test_prediction_value_range): Use -1U instead of -1
18095 to avoid narrowing conversion warning.
18096 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
18097 to avoid narrowing conversion warning.
18098 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
18100 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
18102 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18104 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
18105 64bit configurations.
18106 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
18107 (SIZE_TYPE): Likewise.
18108 * config/vxworks.c (vxworks_emutls_var_fields): Use
18109 long_unsigned_type_node instead of unsigned_type_node as the offset
18110 field type, which is "pointer" mode in emutls.c.
18112 2017-06-27 Jakub Jelinek <jakub@redhat.com>
18115 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
18117 PR middle-end/81207
18118 * gimple-fold.c (replace_call_with_call_and_fold): Handle
18119 gimple_vuse copying separately from gimple_vdef copying.
18121 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18123 * value-prof.c (free_hist): Remove call to memset and the enclosing if
18126 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18127 Olivier Hainque <hainque@adacore.com>
18129 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
18130 for all vxworks7 targets.
18131 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
18132 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
18133 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
18134 variations for VX6/VX7 and 32/64bits later on in ...
18135 (VXWORKS_LIB_SPEC): Leverage new macros.
18136 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
18137 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
18139 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18141 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
18142 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
18144 2017-06-26 Carl Love <cel@us.ibm.com>
18146 * config/rs6000/rs6000-c.c: Add support for built-in functions
18147 vector bool char vec_reve (vector bool char);
18148 vector signed char vec_reve (vector signed char);
18149 vector unsigned char vec_reve (vector unsigned char);
18150 vector bool int vec_reve (vector bool int);
18151 vector signed int vec_reve (vector signed int);
18152 vector unsigned int vec_reve (vector unsigned int);
18153 vector bool long long vec_reve (vector bool long long);
18154 vector signed long long vec_reve (vector signed long long);
18155 vector unsigned long long vec_reve (vector unsigned long long);
18156 vector bool short vec_reve (vector bool short);
18157 vector signed short vec_reve (vector signed short);
18158 vector double vec_reve (vector double);
18159 vector float vec_reve (vector float);
18160 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
18161 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
18162 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
18163 (altivec_vreve): New pattern.
18164 * config/rs6000/altivec.h (vec_reve): New define.
18165 * doc/extend.texi (vec_rev): Update the built-in documentation file
18166 for the new built-in functions.
18168 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18170 PR tree-optimization/71815
18171 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
18173 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
18175 (slsr_process_phi): Likewise.
18176 (replace_uncond_cands_and_profitable_phis): Don't replace a
18177 multiply candidate with a stride of 1 (copy or cast).
18178 (phi_incr_cost): Call uses_consumed_by_stmt rather than
18180 (lowest_cost_path): Likewise.
18181 (total_savings): Likewise.
18183 2017-06-26 Richard Biener <rguenther@suse.de>
18186 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
18187 Use def_builtin_pure for all gather builtins.
18189 2017-06-26 Richard Biener <rguenther@suse.de>
18191 PR tree-optimization/81203
18192 * tree-tailcall.c (find_tail_calls): Do not move stmts into
18193 non-dominating BBs.
18195 2017-06-26 Marek Polacek <polacek@redhat.com>
18198 * doc/invoke.texi: Document -Wmultistatement-macros.
18200 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
18202 * doc/sourcebuild.texi (ARM-specific attributes): Document new
18203 arm_neon_ok_no_float_abi effective target.
18205 2017-06-26 Richard Biener <rguenther@suse.de>
18207 PR tree-optimization/80928
18208 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
18209 (copy_bbs): Set BB_DUPLICATED flag early.
18210 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
18212 (execute_on_shrinking_pred): Likewise.
18213 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
18214 BB_DUPLICATED blocks.
18215 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
18216 iterate over all PHIs considering removal of *gsi.
18218 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
18220 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
18223 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18225 * config/rs6000/rs6000-string.c: (expand_block_clear,
18226 do_load_for_compare, select_block_compare_mode,
18227 compute_current_alignment, expand_block_compare,
18228 expand_strncmp_align_check, expand_strn_compare,
18229 expand_block_move, rs6000_output_load_multiple)
18230 Move functions related to string/block move/compare
18231 to a separate file.
18232 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
18233 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
18234 for this function which is now used in two files.
18235 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
18236 * config.gcc: Add rs6000-string.o to extra_objs for
18237 targets powerpc*-*-* and rs6000*-*-*.
18239 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
18242 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
18243 32-bit, since indexed is not valid for DImode.
18244 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
18245 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
18246 (define_peephole2 for Altivec d-form load): Add 32-bit support.
18247 (define_peephole2 for Altivec d-form store): Likewise.
18250 * multiple_target.c (create_dispatcher_calls): Only create the
18251 dispatcher call if the function is the default clone of a
18252 versioned function.
18254 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
18256 PR middle-end/80902
18257 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
18258 a call, force the call to not be a tail call.
18260 2017-06-23 Jeff Law <law@redhat.com>
18262 * doc/contrib.texi: Add entry for Steven Pemberton's work on
18265 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
18267 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
18268 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
18269 handling for early expansion of vector shifts (sl,sr,sra,rl).
18270 (builtin_function_type): Add vector shift right instructions
18271 to the unsigned argument list.
18273 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
18275 rtl-optimizatoin/79286
18276 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
18277 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
18278 trap. PIC register plus a const unspec without offset can never trap.
18280 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
18282 * tree.h (builtin_structptr_type): New type.
18283 (builtin_structptr_types): Declare new array.
18284 * tree.c (builtin_structptr_types): New array.
18285 (free_lang_data, build_common_tree_nodes): Use it.
18287 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
18290 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
18293 2017-06-22 Matt Turner <mattst88@gmail.com>
18295 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
18296 Lake models to skylake case. Assume skylake for unknown
18297 models with clflushopt.
18299 2017-06-22 Jeff Law <law@redhat.com>
18301 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
18302 frame sizes that do not satisfy aarch64_uimm12_shift.
18304 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
18306 * profile-count.h (apply_probability,
18307 apply_scale, probability_in): Fix checks for zero.
18309 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18311 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
18312 * doc/cppdiropts.texi (-I @var{dir}): Document it.
18314 2016-06-22 Richard Biener <rguenther@suse.de>
18316 * tree-vect-loop.c (vect_model_reduction_cost): Handle
18317 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
18318 REDUC_MAX_EXPR support.
18319 (vectorizable_reduction): Likewise.
18320 (vect_create_epilog_for_reduction): Likewise.
18322 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
18324 * match.pd (A / (1 << B) -> A >> B): New.
18325 * generic-match-head.c: Include optabs-tree.h.
18326 * gimple-match-head.c: Likewise.
18327 * optabs-tree.h (target_supports_op_p): New.
18328 * optabs-tree.c (target_supports_op_p): New.
18330 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18332 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
18333 $gcc_cv_ld --help output.
18334 (gcc_cv_ld_demangle): Likewise.
18335 (gcc_cv_ld_eh_frame_hdr): Likewise.
18336 (gcc_cv_ld_pie): Likewise.
18337 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
18338 (gcc_cv_ld_buildid): Likewise.
18339 (gcc_cv_ld_sysroot): Likewise.
18340 (ld_bndplt_support): Likewise.
18341 (ld_pushpopstate_support): Likewise.
18342 * configure: Regenerate.
18343 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
18345 2017-06-21 Jakub Jelinek <jakub@redhat.com>
18348 * config/i386/sse.md (round<mode>2): Renumber match_dup and
18349 operands indexes to avoid gap between operands and match_dups.
18351 2017-06-21 Andrew Pinski <apinski@cavium.com>
18353 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
18354 Increment Arith_shift and Arith_shift_reg by 1.
18355 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
18357 * config/aarch64/aarch64.c (thunderx_tunings): Enable
18358 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
18359 (aarch64_strip_extend): Add new argument and test for it.
18360 (aarch64_cheap_mult_shift_p): New function.
18361 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
18362 add a cost if it is true.
18363 Update calls to aarch64_strip_extend.
18364 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
18366 2017-06-21 Andrew Pinski <apinski@cavium.com>
18368 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
18370 (thunderxt88): Likewise.
18371 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
18372 (thunderx_prefetch_tune): New variable.
18373 (thunderx2t99_prefetch_tune): Update for the correct values.
18374 (thunderxt88_tunings): New variable.
18375 (thunderx_tunings): Use thunderx_prefetch_tune instead of
18376 generic_prefetch_tune.
18377 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
18379 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18381 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
18382 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
18383 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
18384 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
18385 (aarch64_atomic_cas<mode>, GPI): Likewise.
18387 2017-06-21 Martin Liska <mliska@suse.cz>
18389 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
18390 statements on cold and hot labels.
18391 * predict.c (tree_estimate_probability_bb): Remove the
18392 prediction from this place.
18394 2017-06-21 Martin Liska <mliska@suse.cz>
18396 PR tree-optimization/79489
18397 * gimplify.c (maybe_add_early_return_predict_stmt): New
18399 (gimplify_return_expr): Call the function.
18400 * predict.c (tree_estimate_probability_bb): Remove handling
18402 * predict.def: Update comment about early return predictor.
18403 * gimple-predict.h (is_gimple_predict): New function.
18404 * predict.def: Change default value of early return to 66.
18405 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
18407 * passes.def: Put pass_strip_predict_hints to the beginning of
18410 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
18412 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
18413 FUNCTION_DECL declarations.
18414 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
18416 (dwaf2out_decl): Likewise.
18417 * godump.c (go_early_global_decl): Skip call to the real debug hook
18418 for FUNCTION_DECL declarations.
18419 * passes.c (rest_of_decl_compilation): Skip call to the
18420 early_global_decl debug hook for FUNCTION_DECL declarations, unless
18421 -fdump-go-spec is passed.
18423 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
18425 * config/i386/i386.c (struct builtin_isa): New field pure_p.
18426 Reorder for compactness.
18427 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
18428 (def_builtin_pure, def_builtin_pure2): New functions.
18429 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
18431 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
18433 * match.pd (nop_convert): New predicate.
18434 ((A +- CST1) +- CST2): Allow some NOP conversions.
18436 2017-06-21 Jakub Jelinek <jakub@redhat.com>
18439 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
18440 with ctors/dtors if GOVD_SHARED is set.
18442 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
18444 * config/aarch64/aarch64.md (movti_aarch64):
18445 Emit mov rather than orr.
18446 (movtf_aarch64): Likewise.
18447 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
18448 Emit mov rather than orr.
18450 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
18452 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
18453 Swap alternatives, make integer dup more expensive.
18455 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
18457 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
18458 Return true for non-tls symbols.
18460 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
18462 * config/aarch64/aarch64-cores.def (cortex-a55): New.
18463 (cortex-a75): Likewise.
18464 (cortex-a75.cortex-a55): Likewise.
18465 * config/aarch64/aarch64-tune.md: Regenerate.
18466 * doc/invoke.texi (-mtune): Document new values for -mtune.
18468 2017-06-21 Tom de Vries <tom@codesourcery.com>
18470 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
18471 stack_size feature.
18472 (Effective-Target Keywords, Other attributes): Suggest using
18473 dg-add-options stack_size feature to get stack limit in stack_size
18474 effective target documentation.
18476 2017-06-21 Julian Brown <julian@codesourcery.com>
18477 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18479 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
18480 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
18481 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
18483 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
18484 attribute type list for neon_multiply.
18485 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
18486 attribute type list for neon_multiply.
18487 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
18488 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
18489 attribute type list for neon_multiply.
18490 * config/arm/types.md (crypto_pmull): Add.
18491 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
18492 attribute type list.
18494 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
18496 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
18499 2017-06-20 Jakub Jelinek <jakub@redhat.com>
18501 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
18502 to make sure not to dereference a NULL cost_classes_ptr pointer.
18504 2017-06-20 Carl Love <cel@us.ibm.com>
18506 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18507 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
18508 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
18509 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
18510 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
18511 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
18512 VMULOSW): New enum "unspec" values.
18513 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
18514 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
18515 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
18516 altivec_vmulosw): New patterns.
18517 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
18518 VMULOSW): Add definitions.
18520 2017-06-20 Julia Koval <julia.koval@intel.com>
18522 * config/i386/i386.c: Fix rounding expand for new pattern.
18523 * config/i386/subst.md: Fix pattern (parallel -> unspec).
18525 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
18527 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
18528 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
18530 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
18532 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
18535 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
18537 * config/aarch64/aarch64-cores.def: Rearrange to sort by
18538 architecture, then by implementer ID.
18539 * config/aarch64/aarch64-tune.md: Regenerate.
18541 2017-06-20 Richard Biener <rguenther@suse.de>
18543 PR middle-end/81097
18544 * fold-const.c (split_tree): Fold to type before negating.
18546 2017-06-20 David Malcolm <dmalcolm@redhat.com>
18548 * diagnostic-show-locus.c
18549 (selftest::test_fixit_deletion_affecting_newline): New function.
18550 (selftest::diagnostic_show_locus_c_tests): Call it.
18552 2017-06-20 Andreas Schwab <schwab@suse.de>
18555 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
18558 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18560 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
18561 __ARM_FEATURE_COPROC according to support.
18563 2017-06-20 Jakub Jelinek <jakub@redhat.com>
18565 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
18566 Rewritten to avoid overflow for > 32-bit pointers.
18569 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
18570 by removing enum keyword.
18571 (ubsan_type_descriptor): Likewise. Formatting fix.
18574 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
18575 splitter): Require TARGET_SSE2 in the condition.
18577 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
18580 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
18581 for doing vector set of SFmode on ISA 3.0.
18582 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
18583 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
18585 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
18586 SFmode value into a V4SF variable that was extracted from another
18587 V4SF variable without converting the element to double precision
18588 and back to single precision vector format.
18589 (vsx_insert_extract_v4sf_p9_2): Likewise.
18591 2017-06-19 Jakub Jelinek <jakub@redhat.com>
18593 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
18594 in UWHI to avoid undefined overflow.
18597 * ubsan.h (enum ubsan_encode_value_phase): New.
18598 (ubsan_encode_value): Change second argument to
18599 enum ubsan_encode_value_phase with default value of
18600 UBSAN_ENCODE_VALUE_GENERIC.
18601 * ubsan.c (ubsan_encode_value): Change second argument to
18602 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
18603 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
18604 create_tmp_var_raw instead of create_tmp_var and use a
18606 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
18607 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
18608 ubsan_encode_value callers.
18611 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
18612 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
18613 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
18615 2017-06-19 Richard Biener <rguenther@suse.de>
18617 PR middle-end/81118
18618 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
18619 estimates if we changed anything.
18621 2017-06-19 Richard Biener <rguenther@suse.de>
18623 PR tree-optimization/80887
18624 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
18625 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
18626 simplified lookups, then reset mprts_hook.
18627 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
18629 (try_to_simplify): Likewise.
18631 2017-06-19 Martin Liska <mliska@suse.cz>
18634 * gimplify.c (gimplify_switch_expr):
18635 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
18637 2017-06-19 Martin Liska <mliska@suse.cz>
18639 * doc/install.texi: Document that PGO runs in 4 stages.
18641 2017-06-19 Martin Liska <mliska@suse.cz>
18644 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
18645 to dispatcher function name.
18646 * multiple_target.c (replace_function_decl): New function.
18647 (create_dispatcher_calls): Redirect both edges and references.
18649 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
18651 * profile-count.c (profile_count::dump): Dump quality.
18652 (profile_count::differs_from_p): Update for unsigned val.
18653 * profile-count.h (profile_count_quality): New enum.
18654 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
18656 2017-06-19 Richard Biener <rguenther@suse.de>
18658 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
18659 struct function as arg.
18660 (estimate_numbers_of_iterations): Export overload with loop arg.
18661 (free_numbers_of_iterations_estimates_loop): Use an overload of
18662 free_numbers_of_iterations_estimates instead.
18663 * tree-cfg.c (remove_bb): Adjust.
18664 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
18665 * tree-parloops.c (gen_parallel_loop): Likewise.
18666 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
18668 (tree_unroll_loops_completely): Likewise.
18669 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
18670 Use an overload instead and export.
18671 (estimated_loop_iterations): Adjust.
18672 (max_loop_iterations): Likewise.
18673 (likely_max_loop_iterations): Likewise.
18674 (estimate_numbers_of_iterations): Take struct function as arg
18676 (loop_exits_before_overflow): Adjust.
18677 (free_numbers_of_iterations_estimates_loop): Use an overload.
18678 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
18679 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
18681 2017-06-19 Richard Biener <rguenther@suse.de>
18684 * ipa-prop.c (find_constructor_constant_at_offset): Handle
18685 RANGE_EXPR conservatively.
18687 2017-06-16 Carl Love <cel@us.ibm.com>
18689 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18690 definitions for vec_float, vec_float2, vec_floato,
18691 vec_floate built-ins.
18692 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
18693 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
18695 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
18696 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
18697 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
18698 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
18699 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
18700 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
18701 vec_floato): Add builtin defines.
18702 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
18703 Update the built-in documentation file for the new built-in
18706 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18708 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
18709 (mthumb): Mark as the negative of -marm.
18711 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18713 * doc/invoke.texi (ARM Options, -mcpu): Document supported
18715 (ARM Options, -mtune): Document that this accepts the same
18716 extension options as -mcpu.
18717 (ARM Options, -mfpu): Document addition of -mfpu=auto.
18719 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18721 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
18722 permitted extensions.
18724 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18726 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
18727 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
18728 (armv8-m.main): Add option +nodsp.
18729 * config/arm/arm-cpu-cdata.h: Regenerated.
18731 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18733 * config/arm/t-fuchsia: New file.
18734 * config.gcc (arm*-*-fuchsia*): Use it.
18736 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18738 * config/arm/t-symbian: Rewrite for new option infrastructure.
18740 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18742 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
18743 (MULTILIB_REQUIRED): Likewise.
18745 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18747 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
18748 (MULTILIB_RESUE): Likewise.
18749 (MULTILIB_MATCHES): Likewise.
18750 (MULTLIB_REQUIRED): Likewise.
18752 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18754 * config/arm/t-rtems: Rewrite for new option framework.
18756 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18758 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
18759 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
18760 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
18761 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
18762 * config/arm/t-multilib: ... here.
18763 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
18764 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
18765 armv7-a and armv8*-a when A-profile libraries have not been built.
18766 * config/arm/t-rmprofile: Rewrite.
18768 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18770 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
18771 with a backslash. Remove the backslash after substituting unescaped
18773 * doc/fragments.texi (MULTILIB_REUSE): Document it.
18775 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18777 * config.gcc: (arm*-*-*): When building a-profile libraries, force
18778 the driver to pass through the default setting of -mfloat-abi.
18779 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
18781 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
18782 (all_feat_combs): New rule.
18783 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
18785 * config/arm/t-aprofile: Rewrite.
18787 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18789 * config/arm/arm.h (FPUTYPE_AUTO): Define.
18790 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
18791 fpu is not specified by the user/command-line.
18792 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
18793 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
18794 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
18795 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
18796 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
18797 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
18799 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18801 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
18802 * config/arm/t-arm-elf: Rewritten.
18804 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18806 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
18807 have some floating-point instructions.
18808 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
18809 (TARGET_MAYBE_HARD_FLOAT): New macro.
18810 * config/arm/arm-builtins.c (arm_init_builtins): Use
18811 TARGET_MAYBE_HARD_FLOAT.
18812 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
18814 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18816 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
18817 (configargs.h): Include it.
18818 (arm_print_hint_for_fpu_option): New function.
18819 (arm_parse_fpu_option): New function.
18820 (candidate_extension): New class.
18821 (arm_canon_for_multilib): New function.
18822 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
18823 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
18824 (ARCH_CANONICAL_SPECS): New macro.
18825 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
18827 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18829 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
18830 are set after handling multilib fragments. Set target_cpu_default2
18833 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18835 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
18837 (arm*-*-*): Set target_cpu_default2 to a quoted string.
18838 * config/arm/parsecpu.awk (check_cpu): Validate any extension
18840 (check_arch): Likewise.
18841 * config/arm/arm.c (arm_configure_build_target): Handle
18842 TARGET_CPU_DEFAULT being a string constant. Scan any feature
18843 options in the default.
18845 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18847 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
18848 when an option is an alias of another.
18849 * config/arm/parsecpu.awk (optalias): New parser token.
18850 (gen_comm_data): Mark non-alias options as such. Emit entries
18851 for extension aliases.
18852 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
18853 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
18854 (armv6kz, armv6zk, armv6t2): Likewise.
18855 (armv7): Make vfpv3-d16 an alias.
18856 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
18858 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
18859 Sort in canonical order.
18860 (armv8-a): Sort in canonical order.
18861 (armv8.1-a, armv8.2-a): Likewise.
18862 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
18864 (cortex-a9): Sort in canonical order.
18865 * config/arm/arm.c (selftests.h): Include it.
18866 (arm_test_cpu_arch_data): New function.
18867 (arm_run_self_tests): New function.
18868 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
18869 (targetm): Move declaration to the end of the file.
18870 * arm-cpu-cdata.h: Regenerated.
18872 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18874 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
18875 call to target_mode_check describing the type of option passed.
18876 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
18877 (arm_target_thumb_only): Use arm_parse_arch_option_name or
18878 arm_parse_cpu_option_name to match parameters against list of
18880 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
18881 arm_arch_core_flags data structure.
18882 * config/arm/arm-cpu_cdata.h: Regenerated.
18884 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18886 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
18888 (arm_print_hint_for_cpu_option): Likewise.
18889 (arm_print_hint_for_arch_option): Likewise.
18890 (arm_parse_cpu_option_name): Likewise.
18891 (arm_parse_arch_option_name): Likewise.
18892 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
18893 of entries in the all_fpus list.
18894 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
18895 (arm_parse_cpu_option_name): Declare.
18896 (arm_parse_arch_option_name): Declare.
18897 (arm_parse_option_features): Declare.
18898 (arm_intialize_isa): Declare.
18899 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
18901 (gen_comm_data): ... here. Make definitions non-static.
18902 * config/arm/arm-cpu-data.h: Regenerated.
18903 * config/arm/arm-cpu-cdata.h: Regenerated.
18905 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18907 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
18908 (cpu_arch_extension): New structure.
18909 (cpu_arch_option, arch_option, cpu_option): New structures.
18910 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
18911 architecture types.
18912 (gen_data): Generate new format data tables.
18913 * config/arm/arm.c (cpu_tune): New structure.
18914 (cpu_option, processors): Delete.
18915 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
18916 (arm_print_hint_for_cpu_option): ... this and ...
18917 (arm_print_hint_for_arch_option): ... this.
18918 (arm_parse_arch_cpu_name): Delete. Replace with ...
18919 (arm_parse_cpu_option_name): ... this and ...
18920 (arm_parse_arch_option_name): ... this.
18921 (arm_unrecognized_feature): Change type of target parameter to
18923 (arm_parse_arch_cpu_features): Delete. Replace with ...
18924 (arm_parse_option_features): ... this.
18925 (arm_configure_build_target): Rework to use new configuration data
18927 (arm_print_tune_info): Rework for new configuration data tables.
18928 * config/arm/arm-cpu-data.h: Regenerated.
18929 * config/arm/arm-cpu.h: Regenerated.
18931 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18933 * Makefile.in (OBJS): Move sbitmap.o from here ...
18934 (OBJS-libcommon): ... to here.
18936 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18938 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
18939 (ISA_ALL_CRYPTO): New macro.
18940 (ISA_ALL_SIMD): New macro
18941 (ISA_ALL_FP): New macro.
18942 * config/arm/arm.c (fpu_bitlist): Update initializer.
18943 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
18945 (arm9e): Add fpu. Add option for nofp
18946 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
18947 (arm926ej-s, arm1026ej-s): Likewise.
18948 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
18949 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
18950 neon-fp16, neon-vfpv4, nofp and nosimd.
18951 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
18952 (cortex-a8): Add fpu. Add option for nofp.
18953 (cortex-a9): Add fpu. Add options for nosimd and nofp.
18954 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
18955 (cortex-r4f): Add fpu.
18956 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
18957 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
18959 (cortex-r8): Likewise.
18960 (cortex-m4): Add fpu. Add option for nofp.
18961 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
18962 (cortex-a17.cortex-a7): Likewise.
18963 (cortex-a32): Add fpu. Add options for crypto and nofp.
18964 (cortex-a35, cortex-a53): Likewise.
18965 (cortex-a57): Add fpu. Add option for crypto.
18966 (cortex-a72, cortex-a73): Likewise.
18967 (exynos-m1): Likewise.
18968 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
18969 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
18970 (cortex-m33): Add fpu. Add option for nofp.
18971 * config/arm/arm-cpu-cdata.h: Regenerated
18972 * config/arm/arm-cpu-data.h: Regenerated.
18974 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18976 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
18977 (armv5te, armv5tej): Likewise.
18978 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
18979 (armv7): Add options fp and vfpv3-d16.
18980 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
18981 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
18983 (armv7ve): Likewise.
18984 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
18985 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
18986 (armv8-a): Add nocrypto option.
18987 (armv8.1-a, armv8.2-a): Likewise.
18988 (armv8-m.main): add options fp, fp.dp and nofp.
18990 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18992 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
18994 (armv8-a+crc): Delete.
18995 (armv8.1-a): Add options simd, crypto and nofp.
18996 (armv8.2-a): Add options fp16, simd, crypto and nofp.
18997 (armv8.2-a+fp16): Delete.
18998 (armv8-m.main): Add option dsp.
18999 (armv8-m.main+dsp): Delete.
19000 (cortex-a8): Add fpu. Add nofp option.
19001 (cortex-a9): Add fpu. Add nofp and nosimd options.
19002 * config/arm/parsecpu.awk (gen_data): Generate option tables and
19003 link to main cpu and architecture data structures.
19004 (gen_comm_data): Only put isa attributes from the main architecture
19006 (option): New statement for architecture and CPU entries.
19007 * arm.c (struct cpu_option): New structure.
19008 (struct processors): Add entry for options.
19009 (arm_unrecognized_feature): New function.
19010 (arm_parse_arch_cpu_name): Ignore any characters after the first
19012 (arm_parse_arch_cpu_feature): New function.
19013 (arm_configure_build_target): Separate out any CPU and architecture
19014 features and parse separately. Don't error out if -mfpu=auto is
19015 used with only an architecture string.
19016 (arm_print_asm_arch_directives): New function.
19017 (arm_file_start): Call it.
19018 * config/arm/arm-cpu-cdata.h: Regenerated.
19019 * config/arm/arm-cpu-data.h: Likewise.
19020 * config/arm/arm-tables.opt: Likewise.
19022 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19024 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
19025 assembler when it is not -mfpu=auto.
19027 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19029 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
19030 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
19031 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
19032 (ASM_CPU_SPEC): Rewrite.
19033 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
19034 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
19035 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
19036 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
19037 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
19038 copied string is NUL-terminated. Also strip any characters prefixed
19040 (arm_rewrite_selected_arch): New function.
19041 (arm_rewrite_march): New function.
19043 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19045 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
19046 (x_arm_cpu_string, x_arm_tune_string): Likewise.
19047 (march, mcpu, mtune): Convert to string-based options.
19048 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
19049 (arm_parse_arch_cpu_name): New function.
19050 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
19051 identify selected architecture or CPU.
19052 (arm_option_save): New function.
19053 (TARGET_OPTION_SAVE): Redefine.
19054 (arm_option_restore): Restore string options.
19055 (arm_option_print): Print string options.
19057 2017-06-16 Martin Sebor <msebor@redhat.com>
19059 PR tree-optimization/80933
19060 PR tree-optimization/80934
19061 * builtins.c (fold_builtin_3): Do not handle bcmp here.
19062 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
19063 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
19064 (gimple_fold_builtin): Call them.
19066 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19068 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
19069 as unlikely; update profile.
19071 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19073 * predict.c (force_edge_cold): Handle declaring edges impossible
19076 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19078 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
19080 (try_unroll_loop_completely): Fix reporting.
19082 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19084 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
19086 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
19089 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
19090 if given a non-constant argument for an intrinsic which requires a
19093 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19095 * profile.c (compare_freqs): New function.
19096 (branch_prob): Sort edge list.
19097 (find_spanning_tree): Assume that the list is priority sorted.
19099 2017-06-16 Richard Biener <rguenther@suse.de>
19101 PR tree-optimization/81090
19102 * passes.def (pass_record_bounds): Remove.
19103 * tree-pass.h (make_pass_record_bounds): Likewise.
19104 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
19105 make_pass_record_bounds): Likewise.
19106 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
19107 not free niter estimates at the beginning but at the end.
19108 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
19110 2017-06-16 Richard Biener <rguenther@suse.de>
19112 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
19113 initializer to workaround ICE in host GCC 4.8.
19115 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19117 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
19119 (clone_inlined_nodes): Update.
19121 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19123 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
19124 prefetch settings, and enable prefetching by default at -O3.
19126 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19128 * config/aarch64/aarch64.c (aarch64_override_options_internal):
19129 Set flag_prefetch_loop_arrays according to tuning data.
19131 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19133 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
19134 New tune structure.
19135 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
19136 [Unrelated to main purpose of the patch] Place the pointer field last
19137 to enable type checking errors when tune structure are wrongly merged.
19138 * config/aarch64/aarch64.c (generic_prefetch_tune,)
19139 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
19140 (thunderx2t99_prefetch_tune): New tune constants.
19141 (tune_params *_tunings): Update all tunings (no functional change).
19142 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
19143 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
19144 from tunings structures.
19146 2017-06-16 Jakub Jelinek <jakub@redhat.com>
19149 * ubsan.c (instrument_null): Add T argument, use it instead
19150 of computing it based on IS_LHS.
19151 (instrument_object_size): Likewise.
19152 (pass_ubsan::execute): Adjust instrument_null and
19153 instrument_object_size callers to pass gimple_get_lhs or
19154 gimple_assign_rhs1 result to it. Use instrument_null instead of
19155 calling get_base_address and instrument_mem_ref. Handle
19156 aggregate call arguments for object-size sanitization.
19158 2017-06-16 Yury Gribov <tetra2005@gmail.com>
19160 PR tree-optimization/81089
19161 * tree-vrp.c (is_masked_range_test): Validate operands of
19164 2017-06-15 Martin Sebor <msebor@redhat.com>
19167 * dumpfile.c (dump_register): Avoid calling memset to initialize
19168 a class with a default ctor.
19169 * gcc.c (struct compiler): Remove const qualification.
19170 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
19171 * hash-table.h: Ditto.
19172 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
19174 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
19175 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
19177 * params.h (struct param_info): Make struct members non-const.
19178 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
19179 with default initialization.
19180 * vec.h (vec_copy_construct, vec_default_construct): New helper
19182 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
19183 with vec_copy_construct.
19184 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
19185 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
19186 * doc/invoke.texi (-Wclass-memaccess): Document.
19188 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19190 * emit-rtl.h (is_leaf): Update comment about local
19191 register allocator.
19193 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19196 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
19197 for a variable to have a section before checking if the section has a
19199 Set section to.persistent if persistent attribute is set.
19200 Warn if .persistent attribute is used on an automatic variable.
19202 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
19204 PR rtl-optimization/80474
19205 * reorg.c (update_block): Do not ignore instructions in a delay slot.
19207 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
19209 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
19212 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
19214 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
19215 (casesi): Emit bounds checking as RTL.
19216 (casesi_internal_mips16_<mode>): Remove bounds checking.
19218 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
19220 * config/xtensa/xtensa.c (xtensa_option_override): Append
19221 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
19222 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
19223 xtensa_doloop_hooks): Define unconditionally.
19224 (xtensa_reorg_loops): Only call reorg_loops in the presence of
19226 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
19227 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
19228 for it in xtensa_option_override.
19229 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
19230 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
19232 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
19234 * doc/cppopts.texi: Document '-' special value to -MF.
19236 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
19238 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
19239 (cortex_a53_fconst): Likewise.
19240 (cortex_a53_fpmul): Likewise.
19241 (cortex_a53_f_load_64): Likewise.
19242 (cortex_a53_f_load_many): Likewise.
19243 (cortex_a53_advsimd_alu): Likewise.
19244 (cortex_a53_advsimd_alu_q): Likewise.
19245 (cortex_a53_advsimd_mul): Likewise.
19246 (cortex_a53_advsimd_mul_q): Likewise.
19247 (fpmac bypass): Add new bypass for fpmac-fpmac case.
19248 Add missing fmul, r2f_cvt and fconst cases.
19250 2017-06-14 Richard Biener <rguenther@suse.de>
19252 PR middle-end/81088
19253 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
19255 (fold_binary_loc): When associating do not treat pre-existing
19256 TREE_OVERFLOW on literal constants as a reason to allow
19257 TREE_OVERFLOW on associated literal constants.
19259 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19261 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
19262 (MASK_FEATURES): New macro.
19263 * config/sparc/sparc.c (sparc_option_override): Remove the special
19264 handling of -mfpu and generalize it to all MASK_FEATURES switches.
19266 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19268 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
19269 a division of 0 if non-call exceptions are enabled.
19271 2017-06-14 Andrew Pinski <apinski@cavium.com>
19272 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19275 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
19276 Improve vector initialization code gen for only variable case.
19278 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19280 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
19282 2017-06-14 Richard Biener <rguenther@suse.de>
19284 PR tree-optimization/81083
19285 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
19288 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19290 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
19291 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
19292 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
19293 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
19294 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
19295 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
19297 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19299 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
19300 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
19302 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19304 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
19306 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19308 * config/rs6000/t-rtems: Don't handle SPE.
19310 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19312 * config/rs6000/t-linux: Don't handle SPE.
19314 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19316 * config/rs6000/eabispe.h: Delete file.
19318 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19320 * config/rs6000/t-spe: Delete file.
19322 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19324 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
19325 (rs6000_legitimate_offset_address_p): Return false for anything in
19326 V2SImode or V2SFmode.
19328 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19330 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
19331 except V2SF and V2SI. Rearrange the vector modes, and add comments.
19332 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
19334 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
19335 (rs6000_legitimate_offset_address_p): Ditto.
19336 (rs6000_emit_move): Ditto.
19337 (rs6000_init_builtins): Remove V4HI_type_node.
19339 2017-06-13 Martin Liska <mliska@suse.cz>
19342 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
19343 (gate_asan): Likewise.
19344 * asan.h (asan_no_sanitize_address_p): Remove the function.
19345 (sanitize_flags_p): New function.
19346 * builtins.def: Fix coding style.
19347 * common.opt: Use renamed enum value.
19348 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
19349 * doc/extend.texi: Document no_sanitize attribute.
19350 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
19351 to SANITIZE_UNDEFINED_NONDEFAULT.
19352 * gcc.c (sanitize_spec_function): Use the renamed enum value.
19353 * gimple-fold.c (optimize_atomic_compare_exchange_p):
19354 Use sanitize_flags_p.
19355 * gimplify.c (gimplify_function_tree): Likewise.
19356 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
19357 * opts.c (parse_no_sanitize_attribute): New function.
19358 (common_handle_option): Use renamed enum value.
19359 * opts.h (parse_no_sanitize_attribute): Declare.
19360 * tree.c (sanitize_flags_p): New function.
19361 * tree.h: Declared here.
19362 * tsan.c: Use sanitize_flags_p.
19363 * ubsan.c (ubsan_expand_null_ifn): Likewise.
19364 (instrument_mem_ref): Likewise.
19365 (instrument_bool_enum_load): Likewise.
19366 (do_ubsan_in_current_function): Remove the function.
19367 (pass_ubsan::execute): Use sanitize_flags_p.
19368 * ubsan.h: Remove do_ubsan_in_current_function
19369 * tree-cfg.c (print_no_sanitize_attr_value): New function.
19370 (dump_function_to_file): Use it here.
19372 2017-06-13 Martin Jambor <mjambor@suse.cz>
19374 PR tree-optimization/80803
19375 PR tree-optimization/81063
19376 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
19377 (propagate_subaccesses_across_link): Enqueue subtree whenever
19378 necessary instead of relying on the caller.
19380 2017-06-13 Martin Jambor <mjambor@suse.cz>
19382 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
19383 that have a first_link.
19384 (sort_and_splice_var_accesses): Do not check first_link before
19386 (subtree_mark_written_and_enqueue): Likewise.
19387 (propagate_all_subaccesses): Likewise and do not stop at first
19388 parent with a first_link.
19390 2017-06-13 Martin Jambor <mjambor@suse.cz>
19392 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
19395 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19397 * match.pd: New pattern.
19399 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19401 * tree-vrp.c (is_masked_range_test): New function.
19402 (register_edge_assert_for): Determine ranges for
19405 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19407 PR tree-optimization/67328
19408 * fold-const.c (maskable_range_p): New function.
19409 (build_range_check): Generate bittests if possible.
19411 2017-06-13 Martin Liska <mliska@suse.cz>
19413 * gimple-pretty-print.c (dump_probability): Add new argument.
19414 (dump_edge_probability): Dump both probability and count.
19415 (dump_gimple_label): Likewise.
19416 (dump_gimple_bb_header): Likewise.
19418 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
19421 * config/avr/avr-devices.c: Fix indentation.
19422 * config/avr/gen-avr-mmcu-specs.c: Dito.
19424 2017-06-13 Richard Biener <rguenther@suse.de>
19426 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
19427 instead get vector type from stmt_info.
19428 (vectorizable_reduction): Adjust. Remove dead code.
19430 2017-06-13 Richard Biener <rguenther@suse.de>
19432 PR middle-end/81065
19433 * fold-const.c (extract_muldiv_1): Remove bogus distribution
19434 case of C * (x * C2 + C3).
19435 (fold_addr_of_array_ref_difference): Properly fold index difference.
19437 2017-06-12 David S. Miller <davem@davemloft.net>
19440 * config/sparc/sparc.md (return expander): Emit frame blockage if
19441 function uses alloca.
19443 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
19445 * combine.c (make_field_assignment): Check len rather than the mode
19446 precision when calling force_to_mode.
19448 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
19450 Support multilibs and devices that see flash in RAM address range.
19453 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
19454 (avr_mcu_t) <flash_pm_offset>: New field.
19455 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
19456 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
19457 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
19458 (AVR_TINY_PM_OFFSET): Remove macro.
19459 * config/avr/avr.opt (-mshort-calls): New option.
19460 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
19461 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
19462 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
19463 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
19464 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
19465 instead of avr_arch->have_jmp_call.
19466 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
19467 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
19468 avr_arch->flash_pm_offset to define.
19469 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
19470 new field flash_pm_offset. Add entry for avrxmega3.
19471 (avr_texinfo): Add entry for avrxmega3.
19472 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
19473 attiny212, attiny214,
19474 attiny412, attiny414, attiny416, attiny417,
19475 attiny814, attiny816, attiny817,
19476 attiny1614, attiny1616, attiny1617,
19477 attiny3214, attiny3216, attiny3217.
19478 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
19479 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
19480 (avr_print_operand_address) [AVR_TINY]: Same.
19481 (avr_asm_init_sections) <readonly_data_section>: Only patch
19482 callback if avr_arch->flash_pm_offset = 0.
19483 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
19484 for rodata if avr_arch->flash_pm_offset != 0.
19485 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
19486 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
19487 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
19488 (m_options): Append opt_rcall.
19489 (m_dirnames): Append dir_rcall.
19490 * config/avr/t-multilib: Regenerate.
19492 * configure.ac [target=avr]: Check whether avrxmega3 default
19493 linker description file works as needed.
19494 * configure: Regenerate.
19495 * doc/avr-mmcu.texi: Regenerate.
19496 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
19497 <__AVR_ARCH__>: Document avrxmega3 and 103.
19498 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
19499 <__AVR_SHORT_CALLS__>: Document it.
19500 <__AVR_PM_BASE_ADDRESS__>: Document it.
19501 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
19502 (AVR Variable Attributes) <progmem>: Document this is
19503 not needed for avrxmega3.
19504 (AVR Named Address Spaces) <__flash>: Dito.
19506 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
19508 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
19510 2017-06-12 Doug Rupp <rupp@adacore.com>
19512 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
19513 Append vxworks-stdint.h to the tm_file list.
19514 * config/vxworks-stdint.h: New file.
19516 2017-06-12 Martin Liska <mliska@suse.cz>
19518 PR tree-optimization/81041
19519 * tree-profile.c (gimple_gen_ic_func_profiler):
19520 Create an extra BB in profile-generate
19521 (gimple_gen_time_profiler): Likewise.
19523 2017-06-12 Jakub Jelinek <jakub@redhat.com>
19525 PR tree-optimization/81003
19526 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
19527 (update_range_test): Use it instead of force_gimple_operand_gsi.
19529 2017-06-12 Richard Biener <rguenther@suse.de>
19531 PR tree-optimization/81053
19532 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
19533 with backedge value not defined in loop. Simplify def stmt
19536 2017-06-11 Tom de Vries <tom@codesourcery.com>
19539 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
19541 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
19542 nvptx_cannot_force_const_mem.
19544 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19546 * opts.c (finish_options): Move test for flag_split_stack after
19547 it has been initialized.
19549 2017-06-11 Jason Merrill <jason@redhat.com>
19551 * tree.h (id_equal): New.
19552 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
19553 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
19554 instead of strcmp of IDENTIFIER_POINTER.
19556 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19558 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
19559 (mark_all_inlined_calls_cdtor): Fix formating.
19560 (inline_transform): Rescale profile before inlining.
19562 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19564 * cgraph.h (cgraph_edge::clone): Update prototype.
19565 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
19566 (cgraph_node::create_clone): Update.
19567 (cgraph_node::create_version_clone): Update.
19568 * tree-inline.c (copy_bb): Update.
19569 (expand_call_inline): Update.
19571 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
19573 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
19574 factored out from ...
19575 (rs6000_emit_prologue): ... here.
19577 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
19579 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
19580 factored out from ...
19581 (rs6000_emit_prologue): ... here.
19583 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19585 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
19587 (handle_missing_profiles): Fix computation of tp_first_run.
19588 (counts_to_freqs): Do not touch freqs when count is 0.
19590 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19592 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
19595 2017-06-10 Tom de Vries <tom@codesourcery.com>
19597 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
19598 attributes): Document signal effective target.
19600 2017-06-10 Tom de Vries <tom@codesourcery.com>
19602 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
19603 Document effective target stack_size.
19605 2017-06-09 David Malcolm <dmalcolm@redhat.com>
19607 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
19608 to the edit_context if they can be auto-applied.
19610 2017-06-9 Ian Lance Taylor <iant@golang.org>
19612 * opts.c (finish_options): If -fsplit-stack, disable implicit
19613 -forder-blocks-and-partition.
19614 * doc/invoke.texi (Optimize Options): Document that when using
19615 -fsplit-stack -forder-blocks-and-partition is not implicitly
19618 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
19620 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
19621 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
19622 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
19623 * builtins.def (abort, trap, unreachable): Declare cold.
19624 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
19625 * tree-core.h (ECF_COLD): New.
19626 * tree.c (set_call_expr_flags): Handle ECF_COLD.
19627 (build_common_builtin_nodes): Mark unreachable and abort as cold.
19629 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
19631 * predict.c (unlikely_executed_stmt_p): Cleanup.
19633 2017-06-09 Richard Biener <rguenther@suse.de>
19635 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
19636 model if the ref is always written to.
19638 2017-06-09 Tamar Christina <tamar.christina@arm.com>
19640 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
19642 2017-06-09 Tamar Christina <tamar.christina@arm.com>
19644 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
19647 2017-06-09 Tom de Vries <tom@codesourcery.com>
19650 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
19651 "target cannot support label values" when encountering LABEL_REF.
19653 2017-06-09 Martin Liska <mliska@suse.cz>
19655 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
19656 (gimple_gen_ic_func_profiler): Emit direct comparison
19657 of __gcov_indirect_call_callee with NULL.
19658 (gimple_gen_time_profiler): Change probability from
19659 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
19661 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
19663 * profile.c (edge_gcov_counts): Turn to pointer.
19664 (compute_branch_probabilities, compute_branch_probabilities): Update.
19665 (branch_prob): Do not clear edge_gcov_count.
19666 * profile.h (edge_gcov_counts): Turn to pointer.
19667 (edge_gcov_count): Update.
19669 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
19671 * gimple.h (gimple_check_failed): Mark cold.
19673 2017-06-09 Richard Biener <rguenther@suse.de>
19675 PR tree-optimization/66623
19676 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
19677 refactor check_reduction into two parts, properly computing
19678 whether we have to check reduction validity for outer loop
19681 2017-06-09 Richard Biener <rguenther@suse.de>
19683 PR tree-optimization/79483
19684 * graphite-scop-detection.c (order): New global.
19685 (get_order): Compute bb to order mapping that satisfies code
19686 generation constraints.
19687 (cmp_pbbs): New helper.
19688 (build_scops): Start domwalk at entry block, sort generated
19691 2017-06-09 Richard Biener <rguenther@suse.de>
19693 PR middle-end/81007
19694 * ipa-polymorphic-call.c
19695 (ipa_polymorphic_call_context::restrict_to_inner_class):
19696 Skip FIELD_DECLs with error_mark_node type.
19697 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
19700 2017-06-09 Martin Liska <mliska@suse.cz>
19702 * predict.c (struct branch_predictor): New struct.
19703 (test_prediction_value_range): New test.
19704 (predict_c_tests): New function.
19705 * selftest-run-tests.c (selftest::run_tests): Run the function.
19706 * selftest.h: Declare new tests.
19708 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
19711 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
19712 gen_add3_insn did not fail.
19713 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
19714 r0, construct that number in a temporary reg and add that reg to r0.
19715 If asked to put the result in r0 as well, fail.
19717 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
19719 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
19720 for early expansion of vec_eqv.
19722 2017-06-08 Jakub Jelinek <jakub@redhat.com>
19724 PR middle-end/81005
19725 * ubsan.c (instrument_null): Avoid pointless code temporary.
19726 (pass_ubsan::execute): Instrument aggregate arguments of calls.
19728 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
19732 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
19735 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
19738 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19740 * predict.c (unlikely_executed_edge_p): Move ahead.
19741 (probably_never_executed_edge_p): Use it.
19743 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
19745 PR middle-end/79988
19746 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
19747 gimple_call_builtin_p call.
19749 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19751 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
19752 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
19753 rtl_check_failed_type2, rtl_check_failed_code1,
19754 rtl_check_failed_code2, rtl_check_failed_code_mode,
19755 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
19756 rtvec_check_failed_bounds, rtl_check_failed_flag,
19757 _fatal_insn_not_found, _fatal_insn): Likewise.
19758 * tree.h (tree_contains_struct_check_failed,
19759 tree_check_failed, tree_not_check_failed,
19760 tree_class_check_failed, tree_range_check_failed,
19761 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
19762 tree_vec_elt_check_failed, phi_node_elt_check_failed,
19763 tree_operand_check_failed, omp_clause_check_failed,
19764 omp_clause_operand_check_failed, omp_clause_range_check_failed):
19767 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19769 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
19770 flag_branch_probabilities.
19771 * ipa-inline.c (edge_badness): Likewise.
19772 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
19773 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
19774 * predict.c (maybe_hot_frequency_p): Likewise.
19775 (probably_never_executed): Likewise.
19776 * sched-ebb.c (schedule_ebbs): Likewise.
19777 * sched-rgn.c (find_single_block_region): Likewise.
19778 * tracer.c (tail_duplicate): Likewise.
19780 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19782 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
19783 longer requires x_flag_profile_use.
19785 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19787 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
19788 instead of flag_reorder_blocks_and_partition.
19789 * dbxout.c (dbxout_function_end): Likewise.
19790 * dwarf2out.c (gen_subprogram_die): Likewise.
19791 * haifa-sched.c (sched_create_recovery_edges): Likewise.
19792 * hw-doloop.c (reorg_loops): Likewise.
19793 * varasm.c (assemble_start_function,
19794 assemble_end_function): Likewise.
19795 (decide_function_section): Do not check for
19796 flag_reorder_blocks_and_partition.
19798 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
19800 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
19802 (chkp_get_hard_register_fake_addr_expr): Ditto.
19803 (chkp_build_addr_expr): Add check for hard reg case.
19804 (chkp_parse_array_and_component_ref): Ditto.
19805 (chkp_find_bounds_1): Ditto.
19806 (chkp_process_stmt): Don't generate bounds store for
19809 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19811 * predict.c (maybe_hot_bb_p): Do not check profile status.
19812 (maybe_hot_edge_p): Likewise.
19813 (probably_never_executed): Check for zero counts even if profile
19815 (unlikely_executed_edge_p): New function.
19816 (unlikely_executed_stmt_p): New function.
19817 (unlikely_executed_bb_p): New function.
19818 (set_even_probabilities): Use unlikely predicates.
19819 (combine_predictions_for_bb): Likewise.
19820 (predict_paths_for_bb): Likewise.
19821 (predict_paths_leading_to_edge): Likewise.
19822 (determine_unlikely_bbs): New function.
19823 (estimate_bb_frequencies): Use it.
19824 (compute_function_frequency): Use zero counts even if profile is
19826 * profile-count.h: Fix typo.
19828 2017-08-08 Julia Koval <julia.koval@intel.com>
19830 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
19831 _mm512_mask_cvtsepi16_storeu_epi8,
19832 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
19833 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
19834 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
19835 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
19836 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
19837 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
19838 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
19839 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
19840 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
19841 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
19842 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
19843 __builtin_ia32_pmovuswb256mem_mask,
19844 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
19845 __builtin_ia32_pmovwb512mem_mask): New builtins.
19847 2017-08-08 Julia Koval <julia.koval@intel.com>
19849 PR target/73350,80862
19850 * config/i386/subst.md (round): Fix round pattern.
19851 * config/i386/i386.c (ix86_erase_embedded_rounding):
19852 Fix erasing rounding for the fixed pattern.
19854 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19856 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
19858 2017-06-08 Martin Liska <mliska@suse.cz>
19860 PR gcov-profile/80911
19861 * gcov.c (block_info::block_info): New constructor.
19863 2017-06-07 Carl Love <cel@us.ibm.com>
19865 * config/rs6000/rs6000-c: The return type of the following
19866 built-in functions was implemented as int not long long. Fix sign
19867 of return value for the unsigned version of vec_mulo and vec_mule.
19868 vector unsigned long long vec_bperm (vector unsigned long long,
19869 vector unsigned char)
19870 vector signed long long vec_mule (vector signed int,
19872 vector unsigned long long vec_mule (vector unsigned int,
19873 vector unsigned int)
19874 vector signed long long vec_mulo (vector signed int,
19876 vector unsigned long long vec_mulo (vector unsigned int,
19877 vector unsigned int)
19878 * doc/extend.texi: Fix the documentation for the built-in
19881 2017-06-07 Carl Love <cel@us.ibm.com>
19884 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
19887 2017-06-07 Carl Love <cel@us.ibm.com>
19889 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
19890 support, Generate doublehv for signed int/float for BE case only.
19892 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
19894 * doc/invoke.texi (mcx16): Rewrite.
19896 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19898 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
19899 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
19900 *mov<mode>_softfloat, and an anonymous splitter): Use
19901 nonimmediate_operand instead of rs6000_nonimmediate_operand.
19903 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19905 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
19907 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
19908 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
19909 (rs6000_debug_reg_global): Adjust.
19910 (rs6000_init_hard_regno_mode_ok): Adjust.
19911 (rs6000_dbx_register_number): Adjust.
19912 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
19913 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
19914 Remove SPE_ACC and SPEFSCR.
19915 (REG_ALLOC_ORDER): Ditto.
19916 (FRAME_POINTER_REGNUM): Change to 111.
19917 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
19918 (REG_CLASS_NAMES): Ditto.
19919 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
19920 (REGISTER_NAMES): Ditto.
19921 (ADDITIONAL_REG_NAMES): Ditto.
19922 (rs6000_reg_names): Ditto.
19923 * config/rs6000/rs6000.md: Renumber some register number
19926 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19928 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
19930 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
19931 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
19933 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
19934 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
19935 Delete the SPE high registers.
19936 (REG_ALLOC_ORDER): Ditto.
19937 (enum reg_class): Remove SPE_HIGH_REGS.
19938 (REG_CLASS_NAMES): Ditto.
19939 (REG_CLASS_CONTENTS): Delete the SPE high registers.
19940 (REGISTER_NAMES): Ditto.
19941 (rs6000_reg_names): Ditto.
19942 * doc/tm.texi.in: Remove SPE as example.
19943 * doc/tm.texi: Regenerate.
19945 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19947 * config/rs6000/8540.md (ppc8540_brinc): Delete.
19948 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
19949 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
19950 * config/rs6000/rs6000.md (type): Remove "brinc".
19952 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19954 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
19955 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
19956 * config/rs6000/linuxspe.h: Delete file.
19957 * config/rs6000/rs6000.md: Don't include spe.md.
19958 * config/rs6000/spe.h: Delete file.
19959 * config/rs6000/spe.md: Delete file.
19960 * config/rs6000/t-rs6000: Remove spe.md.
19962 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19964 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
19965 (reg_or_none500mem_operand): Delete.
19966 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
19967 instead of reg_or_none500mem_operand.
19969 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19971 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
19972 handling of SPE flags.
19973 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
19975 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19977 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
19979 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
19980 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
19981 paired_divv2sf3): Similar.
19982 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
19983 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
19984 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
19986 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
19987 Rename the paired_* instruction patterns.
19988 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
19990 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
19991 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
19992 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
19993 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
19994 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
19995 PAIRED_VECTOR_MODE.
19996 (struct machine_function): Delete field spe_insn_chain_scanned_p.
19997 (spe_func_has_64bit_regs_p): Delete.
19998 (spe_expand_predicate_builtin): Delete.
19999 (spe_expand_evsel_builtin): Delete.
20000 (TARGET_DWARF_REGISTER_SPAN): Do not define.
20001 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
20002 (invalid_e500_subreg): Delete.
20003 (rs6000_legitimize_address): Always force_reg op2 as well, for
20004 paired single memory accesses.
20005 (rs6000_member_type_forces_blk): Delete.
20006 (rs6000_spe_function_arg): Delete.
20007 (rs6000_expand_unop_builtin): Delete SPE handling.
20008 (rs6000_expand_binop_builtin): Ditto.
20009 (spe_expand_stv_builtin): Delete.
20010 (bdesc_2arg_spe): Delete.
20011 (spe_expand_builtin): Delete.
20012 (spe_expand_predicate_builtin): Delete.
20013 (spe_expand_evsel_builtin): Delete.
20014 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
20015 (spe_init_builtins): Delete.
20016 (spe_func_has_64bit_regs_p): Delete.
20017 (savres_routine_name): Delete "info" parameter. Adjust callers.
20018 (rs6000_emit_stack_reset): Ditto.
20019 (rs6000_dwarf_register_span): Delete.
20020 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
20021 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
20022 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
20024 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
20026 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
20027 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
20028 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
20029 mulv2sf3, divv2sf3): Delete expanders.
20031 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20033 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
20035 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20037 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
20038 * config/rs6000/rs6000.c: Ditto.
20040 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20042 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
20043 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
20044 comparison_operator.
20046 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20048 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
20049 * config/rs6000/rs6000.opt: Ditto.
20050 * config/rs6000/t-rtems: Ditto.
20052 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20054 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
20055 TARGET_E500_SINGLE by 0, simplify.
20056 * config/rs6000/rs6000.c: Ditto.
20057 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
20058 (spe_build_register_parallel): Delete.
20059 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
20060 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
20061 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
20062 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
20063 (E500_CONVERT): Delete.
20064 * config/rs6000/spe.md: Remove many patterns and all define_constants.
20066 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20068 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
20069 * config/rs6000/dfp.md: Ditto.
20070 (negdd2, *negdd2_fpr): Merge.
20071 (absdd2, *absdd2_fpr): Merge.
20072 (negtd2, *negtd2_fpr): Merge.
20073 (abstd2, *abstd2_fpr): Merge.
20074 * config/rs6000/e500.h: Delete file.
20075 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
20076 TARGET_FPRS by 1 and simplify.
20077 * config/rs6000/rs6000-c.c: Ditto.
20078 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
20079 TARGET_DF_SPE by 0.
20080 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
20082 * config/rs6000/rs6000.md: Ditto.
20083 (floatdidf2, *floatdidf2_fpr): Merge.
20084 (move_from_CR_gt_bit): Delete.
20085 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
20086 (E500_CR_IOR_COMPARE): Delete.
20087 (All patterns that require !TARGET_FPRS): Delete.
20088 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
20090 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20092 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
20094 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20096 * graphds.c (add_edge): Intitialize edge's attached data.
20097 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
20098 pointer parameter. Call pointed function on each edge during
20099 graph traversing. Skip traversing the edge when the function
20101 (graphds_dfs, graphds_scc): Ditto.
20102 (for_each_edge): New parameter. Pass the new parameter to callback
20104 * graphds.h (skip_edge_callback): New function pointer type.
20105 (graphds_dfs, graphds_scc): New function pointer parameter.
20106 (graphds_edge_callback, for_each_edge): New parameter.
20108 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20110 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
20111 out code checking if runtime alias check is possible to below ...
20112 Call the new function.
20113 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
20114 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
20116 2017-06-07 Marek Polacek <polacek@redhat.com>
20119 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
20120 TYPE_OVERFLOW_WRAPS check.
20122 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20124 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
20125 if versioning is required.
20126 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
20127 peeling with the check for versioning.
20129 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20131 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
20132 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
20133 Set true to new parameter if new ssa variable is defined.
20134 (vect_gen_vector_loop_niters): Refactor. Set range information
20135 for the new vector loop bound variable.
20136 (vect_do_peeling): Ditto.
20138 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20140 * tree-affine.c (ssa.h): Include header file.
20141 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
20142 has wrapping overflow behavior.
20144 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20146 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
20148 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20150 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
20151 (tree_to_aff_combination): ... here.
20153 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20155 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
20156 reg_pressure model function.
20157 (ivopts_global_cost_for_size): Delete.
20158 (determine_set_costs, iv_ca_recount_cost): Call new model function
20159 ivopts_estimate_reg_pressure.
20161 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20163 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
20164 expensive than udiv. Remove floating point cases from mod.
20166 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20168 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
20169 Increase idiv cost.
20171 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20173 * config/aarch64/aarch64.md
20174 (copysignsf3): Fix mask generation.
20176 2017-06-07 Jakub Jelinek <jakub@redhat.com>
20178 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
20180 (class dump_manager): Add register_dumps method.
20181 * dumpfile.c: Include langhooks.h.
20182 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
20183 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
20184 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
20185 (dump_manager::dump_register): Start with 512 entries instead of 32.
20186 (dump_manager::register_dumps): New method.
20187 * toplev.c (general_init): Instead of invoking register_dumps
20188 langhook, invoke register_dumps method on the dump manager.
20189 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
20192 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
20194 * doc/md.texi: Clarify the restrictions on a define_insn condition.
20195 Say that # requires an associated define_split to exist, and that
20196 the define_split must be suitable for use after register allocation.
20198 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20200 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
20201 (compute_outgoing_frequencies): Also initialize zero counts.
20202 (find_many_sub_basic_blocks): Do not produce uninitialized profile
20203 around loops; preserve more of profile when nothing changes.
20205 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
20207 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
20209 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
20210 * config/arm/arm-cpu-cdata.h: Regenerate.
20211 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
20212 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
20213 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
20214 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
20215 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
20217 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
20219 * config/arm/t-rmprofile: Likewise.
20220 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
20222 2017-06-06 David S. Miller <davem@davemloft.net>
20225 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
20226 blockage if function uses alloca.
20228 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20230 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
20231 New "uid" fields to hold pretty-print IDs of group and ref.
20232 Memory references are now identified as <group_id>:<ref_id>
20233 instead of using [random] addresses.
20234 (dump_mem_details): Simplify, no functional change.
20235 (dump_mem_ref): Simplify and make output more concise.
20236 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
20237 (find_or_create_group): Initialize group uid.
20238 (record_ref): Initialize ref uid. Improve debug output.
20239 (prune_group_by_reuse, should_issue_prefetch_p,)
20240 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
20241 (mark_nontemporal_store, determine_loop_nest_reuse):
20242 Improve debug output.
20244 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20246 * dbgcnt.def (prefetch): New debug counter.
20247 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
20248 (schedule_prefetches): Stop issueing prefetches if debug counter
20251 2017-06-06 Tom de Vries <tom@codesourcery.com>
20253 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
20254 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
20256 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20258 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
20259 Use aarch64_reg_or_zero predicate for operand 4.
20260 (aarch64_compare_and_swap<mode> define_insn_and_split):
20261 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
20262 (aarch64_store_exclusive<mode>): Likewise for operand 2.
20264 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
20266 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
20267 (arm_compute_save_core_reg_mask): This.
20268 (thumb1_compute_save_reg_mask): Rename into ...
20269 (thumb1_compute_save_core_reg_mask): This.
20270 (arm_compute_save_reg0_reg12_mask): Adapt comment.
20271 (arm_compute_frame_layout): Likewise.
20273 2017-06-06 Richard Biener <rguenther@suse.de>
20275 PR tree-optimization/80974
20276 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
20277 keep or clear leaders SSA info.
20279 2017-06-06 Tom de Vries <tom@codesourcery.com>
20281 * config/nvptx/nvptx.c (split_mode_p): New function.
20282 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
20284 2017-06-06 Tom de Vries <tom@codesourcery.com>
20286 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
20288 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20291 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
20294 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20296 * shrink-wrap.c (handle_simple_exit): Update profile.
20297 (try_shrink_wrapping): Upate profile.
20299 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20301 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
20302 (tree_guess_outgoing_edge_probabilities): New.
20303 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
20304 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
20306 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20308 * ipa-split.c (split_function): Initialize return bb profile.
20310 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20312 * profile.c (compute_branch_probabilities): Also initialize
20313 EXIT_BLOCK profile.
20315 2017-06-06 Richard Biener <rguenther@suse.de>
20317 PR tree-optimization/80928
20318 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
20319 (vect_analyze_loop_operations): Properly guard analysis for
20321 (vect_transform_loop): Likewise.
20322 (vect_analyze_loop_2): Also reset SLP type on PHIs.
20323 (vect_model_induction_cost): Do not cost for pure SLP.
20324 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
20325 of induction in inner loop vectorization.
20326 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
20327 (vect_get_and_check_slp_defs): Handle vect_induction_def.
20328 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
20330 (vect_analyze_slp_cost_1): Cost induction.
20331 (vect_detect_hybrid_slp_stmts): Handle PHIs.
20332 (vect_get_slp_vect_defs): Likewise.
20333 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
20334 (vect_transform_stmt): Handle SLP reductions.
20335 * tree-vectorizer.h (vectorizable_induction): Adjust.
20337 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
20339 * config/rs6000/rs6000.c (make_resolver_func): Update
20340 init_lowered_empty_function call.
20342 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
20344 * doc/invoke.texi: Document the -fprofile-abs-path option.
20345 * common.opt (fprofile-abs-path): New option.
20346 * gcov-io.h (gcov_write_filename): Declare.
20347 * gcov-io.c (gcov_write_filename): New function.
20348 * coverage.c (coverage_begin_function): Use gcov_write_filename.
20349 * profile.c (output_location): Likewise.
20351 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20353 * shring-wrap.c: Revert accidental commit.
20355 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
20357 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
20359 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20361 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
20363 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
20364 profile in callgraph edge.
20365 * profile-count.h (apply_probability): If THIS is 0, then result is 0
20366 (apply_scale): Likewise.
20367 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
20368 Also scale profile when inlining function with zero profile.
20369 (initialize_cfun): Update exit block profile even when it is zero.
20370 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
20371 when profile is read.
20373 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
20375 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
20376 (CLONE_*): New constants to define the processors we can generate
20377 code for with the target_clone attribute.
20378 (rs6000_clone_map): New array to identify which clone processors
20379 the current program is running on.
20380 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
20381 target_clone attribute.
20382 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
20383 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
20384 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
20385 (cpu_expand_builtin): Add support for target_clone attribute.
20386 (rs6000_valid_attribute_p): Allow "default" attribute.
20387 (get_decl_name): New debug function to simplify printing the
20388 current function name in debugging statements.
20389 (rs6000_clone_priority): New functions to support the target_clone
20390 attribute, and be able to generate code to switch between ISA 2.05
20391 through ISA 3.0 (power6 through power9).
20392 (rs6000_compare_version_priority): Likewise.
20393 (rs6000_get_function_versions_dispatcher): Likewise.
20394 (make_resolver_func): Likewise.
20395 (add_condition_to_bb): Likewise.
20396 (dispatch_function_versions): Likewise.
20397 (rs6000_generate_version_dispatcher_body): Likewise.
20398 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
20399 (fusion_gpr_load_p): Fix a spacing issue.
20400 * doc/extend.texi (Common Function Attributes): Document that the
20401 PowerPC supports the target_clone attribute.
20403 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
20405 * config/arm/arm.h: explain F symbol found in description of ARM
20406 register allocation in its legend.
20408 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20410 * config/mips/frame-header-opt.c: Include profile-count.h.
20411 * config/riscv/riscv.c: Include profile-count.h
20413 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20415 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
20417 (sm_set_flag_if_changed): Add bbs field.
20418 (execute_sm_if_changed_flag_set): Pass BBS.
20419 (execute_sm): Update.
20421 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20423 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
20426 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20428 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
20429 (peephole2): New peephole2 to emit the above.
20430 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
20432 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20434 * config/aarch64/aarch64.c (define_peephole2 above
20435 *sub_<shift>_<mode>): New peephole.
20437 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
20439 * config/i386/i386.c (make_resolver_func): Update.
20440 * Makefile.in: Add profile-count.h and profile-count.o
20441 * auto-profile.c (afdo_indirect_call): Update to new API.
20442 (afdo_set_bb_count): Update.
20443 (afdo_propagate_edge): Update.
20444 (afdo_propagate_circuit): Update.
20445 (afdo_calculate_branch_prob): Update.
20446 (afdo_annotate_cfg): Update.
20447 * basic-block.h: Include profile-count.h
20448 (struct edge_def): Turn count to profile_count.
20449 (struct basic_block_def): Likewie.
20450 (REG_BR_PROB_BASE): Move to profile-count.h
20451 (RDIV): Move to profile-count.h
20452 * bb-reorder.c (max_entry_count): Turn to profile_count.
20453 (find_traces): Update.
20454 (rotate_loop):Update.
20455 (connect_traces):Update.
20456 (sanitize_hot_paths):Update.
20457 * bt-load.c (migrate_btr_defs): Update.
20458 * cfg.c (RDIV): Remove.
20459 (init_flow): Use alloc_block.
20460 (alloc_block): Uninitialize count.
20461 (unchecked_make_edge): Uninitialize count.
20462 (check_bb_profile): Update.
20463 (dump_edge_info): Update.
20464 (dump_bb_info): Update.
20465 (update_bb_profile_for_threading): Update.
20466 (scale_bbs_frequencies_int): Update.
20467 (scale_bbs_frequencies_gcov_type): Update.
20468 (scale_bbs_frequencies_profile_count): New.
20469 * cfg.h (update_bb_profile_for_threading): Update.
20470 (scale_bbs_frequencies_profile_count): Declare.
20471 * cfgbuild.c (compute_outgoing_frequencies): Update.
20472 (find_many_sub_basic_blocks): Update.
20473 * cfgcleanup.c (try_forward_edges): Update.
20474 (try_crossjump_to_edge): Update.
20475 * cfgexpand.c (expand_gimple_tailcall): Update.
20476 (construct_exit_block): Update.
20477 * cfghooks.c (verify_flow_info): Update.
20478 (dump_bb_for_graph): Update.
20479 (split_edge): Update.
20480 (make_forwarder_block): Update.
20481 (duplicate_block): Update.
20482 (account_profile_record): Update.
20483 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
20484 (get_estimated_loop_iterations): Update.
20485 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
20486 (single_likely_exit): Update.
20487 * cfgloopmanip.c (scale_loop_profile): Update.
20489 (set_zero_probability): Update.
20490 (lv_adjust_loop_entry_edge): Update.
20491 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
20492 (purge_dead_edges): Update.
20493 (rtl_account_profile_record): Update.
20494 * cgraph.c (cgraph_node::create): Uninitialize count.
20495 (symbol_table::create_edge): Uninitialize count.
20496 (cgraph_update_edges_for_call_stmt_node): Update.
20497 (cgraph_edge::dump_edge_flags): Update.
20498 (cgraph_node::dump): Update.
20499 (cgraph_edge::maybe_hot_p): Update.
20500 * cgraph.h: Include profile-count.h
20501 (create_clone), create_edge, create_indirect_edge): Update.
20502 (cgraph_node): Turn count to profile_count.
20503 (cgraph_edge0: Likewise.
20504 (make_speculative, clone): Update.
20505 (create_edge): Update.
20506 (init_lowered_empty_function): Update.
20507 * cgraphclones.c (cgraph_edge::clone): Update.
20508 (duplicate_thunk_for_node): Update.
20509 (cgraph_node::create_clone): Update.
20510 * cgraphunit.c (cgraph_node::analyze): Update.
20511 (cgraph_node::expand_thunk): Update.
20512 * final.c (dump_basic_block_info): Update.
20513 * gimple-streamer-in.c (input_bb): Update.
20514 * gimple-streamer-out.c (output_bb): Update.
20515 * graphite.c (print_global_statistics): Update.
20516 (print_graphite_scop_statistics): Update.
20517 * hsa-brig.c: Include basic-block.h.
20518 * hsa-dump.c: Include basic-block.h.
20519 * hsa-gen.c (T sum_slice): Update.
20520 (convert_switch_statements):Update.
20521 * hsa-regalloc.c: Include basic-block.h.
20522 * ipa-chkp.c (chkp_produce_thunks): Update.
20523 * ipa-cp.c (struct caller_statistics): Update.
20524 (init_caller_stats): Update.
20525 (gather_caller_stats): Update.
20526 (ipcp_cloning_candidate_p): Update.
20527 (good_cloning_opportunity_p): Update.
20528 (get_info_about_necessary_edges): Update.
20529 (dump_profile_updates): Update.
20530 (update_profiling_info): Update.
20531 (update_specialized_profile): Update.
20532 (perhaps_add_new_callers): Update.
20533 (decide_about_value): Update.
20534 (ipa_cp_c_finalize): Update.
20535 * ipa-devirt.c (struct odr_type_warn_count): Update.
20536 (struct decl_warn_count): Update.
20537 (struct final_warning_record): Update.
20538 (possible_polymorphic_call_targets): Update.
20539 (ipa_devirt): Update.
20540 * ipa-fnsummary.c (redirect_to_unreachable): Update.
20541 * ipa-icf.c (sem_function::merge): Update.
20542 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
20543 * ipa-inline.c (compute_uninlined_call_time): Update.
20544 (compute_inlined_call_time): Update.
20545 (want_inline_small_function_p): Update.
20546 (want_inline_self_recursive_call_p): Update.
20547 (edge_badness): Update.
20548 (lookup_recursive_calls): Update.
20549 (recursive_inlining): Update.
20550 (inline_small_functions): Update.
20551 (dump_overall_stats): Update.
20552 (dump_inline_stats): Update.
20553 * ipa-profile.c (ipa_profile_generate_summary): Update.
20554 (ipa_propagate_frequency): Update.
20555 (ipa_profile): Update.
20556 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
20557 * ipa-utils.c (ipa_merge_profiles): Update.
20558 * loop-doloop.c (doloop_modify): Update.
20559 * loop-unroll.c (report_unroll): Update.
20560 (unroll_loop_runtime_iterations): Update.
20561 * lto-cgraph.c (lto_output_edge): Update.
20562 (lto_output_node): Update.
20563 (input_node): Update.
20564 (input_edge): Update.
20565 (merge_profile_summaries): Update.
20566 * lto-streamer-in.c (input_cfg): Update.
20567 * lto-streamer-out.c (output_cfg): Update.
20568 * mcf.c (create_fixup_graph): Update.
20569 (adjust_cfg_counts): Update.
20570 (sum_edge_counts): Update.
20571 * modulo-sched.c (sms_schedule): Update.
20572 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
20573 * predict.c (maybe_hot_count_p): Update.
20574 (probably_never_executed): Update.
20575 (dump_prediction): Update.
20576 (combine_predictions_for_bb): Update.
20577 (propagate_freq): Update.
20578 (handle_missing_profiles): Update.
20579 (counts_to_freqs): Update.
20580 (rebuild_frequencies): Update.
20581 (force_edge_cold): Update.
20582 * predict.h: Include profile-count.h
20583 (maybe_hot_count_p, counts_to_freqs): UPdate.
20584 * print-rtl-function.c: Do not include cfg.h
20585 * print-rtl.c: Include basic-block.h
20586 * profile-count.c: New file.
20587 * profile-count.h: New file.
20588 * profile.c (is_edge_inconsistent): Update.
20589 (correct_negative_edge_counts): Update.
20590 (is_inconsistent): Update.
20591 (set_bb_counts): Update.
20592 (read_profile_edge_counts): Update.
20593 (compute_frequency_overlap): Update.
20594 (compute_branch_probabilities): Update; Initialize and deinitialize
20596 (branch_prob): Update.
20597 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
20598 (edge_gcov_count): New.
20599 (bb_gcov_count): New.
20600 * shrink-wrap.c (try_shrink_wrapping): Update.
20601 * tracer.c (better_p): Update.
20602 * trans-mem.c (expand_transaction): Update.
20603 (ipa_tm_insert_irr_call): Update.
20604 (ipa_tm_insert_gettmclone_call): Update.
20605 * tree-call-cdce.c: Update.
20606 * tree-cfg.c (gimple_duplicate_sese_region): Update.
20607 (gimple_duplicate_sese_tail): Update.
20608 (gimple_account_profile_record): Update.
20609 (execute_fixup_cfg): Update.
20610 * tree-inline.c (copy_bb): Update.
20611 (copy_edges_for_bb): Update.
20612 (initialize_cfun): Update.
20613 (freqs_to_counts): Update.
20614 (copy_cfg_body): Update.
20615 (expand_call_inline): Update.
20616 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
20617 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
20618 (try_unroll_loop_completely): Update.
20619 (try_peel_loop): Update.
20620 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
20621 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
20622 * tree-ssa-loop-split.c (connect_loops): Update.
20623 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
20624 * tree-ssa-reassoc.c (branch_fixup): Update.
20625 * tree-ssa-tail-merge.c (replace_block_by): Update.
20626 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
20627 (compute_path_counts): Update.
20628 (update_profile): Update.
20629 (recompute_probabilities): Update.
20630 (update_joiner_offpath_counts): Update.
20631 (estimated_freqs_path): Update.
20632 (freqs_to_counts_path): Update.
20633 (clear_counts_path): Update.
20634 (ssa_fix_duplicate_block_edges): Update.
20635 (duplicate_thread_path): Update.
20636 * tree-switch-conversion.c (case_bit_test_cmp): Update.
20637 (struct switch_conv_info): Update.
20638 * tree-tailcall.c (decrease_profile): Update.
20639 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
20640 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
20641 * value-prof.c (check_counter): Update.
20642 (gimple_divmod_fixed_value): Update.
20643 (gimple_mod_pow2): Update.
20644 (gimple_mod_subtract): Update.
20645 (gimple_ic_transform): Update.
20646 (gimple_stringop_fixed_value): Update.
20647 * value-prof.h (gimple_ic): Update.
20649 2017-06-02 Carl Love <cel@us.ibm.com>
20651 * config/rs6000/rs6000-c: Add support for built-in functions
20652 vector double vec_doublee (vector signed int);
20653 vector double vec_doublee (vector unsigned int);
20654 vector double vec_doublee (vector float);
20655 vector double vec_doubleh (vector signed int);
20656 vector double vec_doubleh (vector unsigned int);
20657 vector double vec_doubleh (vector float);
20658 vector double vec_doublel (vector signed int);
20659 vector double vec_doublel (vector unsigned int);
20660 vector double vec_doublel (vector float);
20661 vector double vec_doubleo (vector signed int);
20662 vector double vec_doubleo (vector unsigned int);
20663 vector double vec_doubleo (vector float);.
20664 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
20665 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
20667 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
20668 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
20669 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
20671 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
20672 vec_doublel, vec_doubleh.
20673 * doc/extend.texi: Update the built-in documentation file for the
20674 new built-in functions.
20676 2017-06-02 David Malcolm <dmalcolm@redhat.com>
20679 * ipa-inline-analysis.c (free_growth_caches): Set
20680 edge_removal_hook_holder to NULL after removing it.
20682 2017-06-02 Sudakshina Das <sudi.das@arm.com>
20684 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
20685 comparision with zero.
20687 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
20688 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
20689 for early expansion of vec_min and vec_max builtins.
20690 (builtin_function_type): Add min/max unsigned variants to those
20691 identified as having unsigned arguments.
20693 2017-06-02 Olivier Hainque <hainque@adacore.com>
20695 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
20697 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20699 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
20700 Use VALL_F16 iterator rather than VALL.
20702 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20704 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
20705 Emit CBNZ inside loop when doing a strong exchange and comparing
20706 against zero. Generate the CC flags after the loop.
20708 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
20710 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
20711 (dl_section_ref): New.
20712 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
20713 On AIX, append an expression to subtract the size of the
20714 section length to dl_section_ref.
20716 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
20718 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
20719 for early expansion of vector absolute builtins.
20721 2017-06-02 Richard Biener <rguenther@suse.de>
20723 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
20724 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
20726 2017-06-02 Richard Biener <rguenther@suse.de>
20728 PR tree-optimization/80948
20729 * tree-tailcall.c (find_tail_calls): Track stmts to move in
20730 stmt order as well.
20732 2017-06-02 Richard Biener <rguenther@suse.de>
20734 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
20736 * tree-vect-stmts.c (process_use): Do not mark backedge defs
20737 for inductions as relevant.
20739 2017-06-02 Richard Biener <rguenther@suse.de>
20741 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
20742 (vectorizable_induction): ... this. Remove dead code.
20744 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
20746 * builtins. (expand_builtin_alloca): Remove second parameter and
20747 infer its value from the first parameter instead.
20748 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
20750 2017-06-02 Jakub Jelinek <jakub@redhat.com>
20752 PR rtl-optimization/80903
20753 * loop-doloop.c (add_test): Unshare sequence.
20755 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
20757 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
20759 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
20761 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
20763 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
20764 xlogue_layout::get_instance, logue_layout::xlogue_layout,
20765 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
20766 (xlogue_layout::get_stub_rtx): Make static.
20767 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
20768 (xlogue_layout::compute_stub_managed_regs): Rename to...
20769 (xlogue_layout::count_stub_managed_regs): ...this.
20770 (xlogue_layout::is_stub_managed_reg): New function.
20771 (xlogue_layout::m_stub_names): Rename to...
20772 (xlogue_layout::s_stub_names): ...this, make static.
20773 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
20774 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
20775 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
20776 xlogue_layout::s_stub_names): Instantiate statics.
20777 (stub_managed_regs): Remove.
20778 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
20779 (disable_call_ms2sysv_xlogues): Rename to...
20780 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
20781 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
20783 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
20784 change after reload_completed.
20785 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
20787 (ix86_expand_prologue): Likewise.
20788 (ix86_expand_epilogue): Likewise.
20789 (ix86_expand_split_stack_prologue): Likewise.
20790 (ix86_compute_frame_layout): Remove frame parameter ...
20791 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
20792 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
20794 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
20795 (ix86_frame): Move from here ...
20796 * config/i386/i386.h (ix86_frame): ... to here.
20797 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
20798 complete ix86_frame data structure instead. Remove some_ld_name.
20800 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
20802 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
20803 symbols that hold a DECL_VALUE_EXPR.
20805 2017-06-01 Martin Jambor <mjambor@suse.cz>
20807 PR tree-optimization/80898
20808 * tree-sra.c (process_subtree_disqualification): Removed.
20809 (disqualify_candidate): Do not acll
20810 process_subtree_disqualification.
20811 (subtree_mark_written_and_enqueue): New function.
20812 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
20813 RHS has been disqualified and re-queue LHS if necessary. Apart
20814 from that, ignore disqualified RHS.
20816 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20818 * config/s390/s390.c (s390_emit_epilogue): Disable early return
20819 address fetch for z10 or later.
20821 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20823 * config/arc/arc.md (tst_movb): Add guard when splitting.
20825 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20827 * config/arc/arc.c (arc_can_eliminate): Test against
20828 arc_frame_pointer_needed.
20830 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20832 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
20833 to prevent store reordering.
20834 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
20835 (type): Add block type.
20836 (stack_tie): Define special instruction to be used in
20839 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20841 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
20842 constraint. It is not valid for the pattern.
20843 (noncommutative_binary_comparison): Likewise.
20845 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20847 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
20850 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20852 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
20853 be used by the reg-alloc.
20855 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20857 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
20858 reg-alloc when having mul64 or mul32x16 instructions.
20859 (mulsidi3): Likewise.
20860 (umulsidi3): Likewise.
20861 (mulsi32x16): New pattern.
20862 (mulsi64): Likewise.
20863 (mulsidi64): Likewise.
20864 (umulsidi64): Likewise.
20865 (MUL32x16_REG): Define.
20866 (mul64_600): Use MUL32x16_REG.
20867 (mac64_600): Likewise.
20868 (umul64_600): Likewise.
20869 (umac64_600): Likewise.
20871 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20873 * config/arc/arc.md (mulsi3_700): Make it commutative.
20875 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
20877 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
20879 (*sign_extendsidi2_insn): Likewise for movstosw.
20881 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
20883 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
20884 the type of the input discriminant value. Convert the
20885 discriminant value of signedness vary.
20887 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
20889 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
20890 Add to -Wall section.
20892 2017-06-01 Richard Biener <rguenther@suse.de>
20894 PR middle-end/66313
20895 * fold-const.c (fold_plusminus_mult_expr): If the factored
20896 factor may be zero use a wrapping type for the inner operation.
20897 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
20898 and handle moved defs.
20899 (process_assignment): Properly guard the unary op case. Return a
20900 tri-state indicating that moving the stmt before the call may allow
20901 to continue. Pass through to_move.
20902 (find_tail_calls): Handle moving unrelated defs before
20905 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
20908 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
20909 splitter result in the canonical way.
20911 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
20913 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
20914 also for 32bit target. Update insn attributes.
20915 (zero-extendsidi2 splitter): Allow all registers for operand 1.
20917 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
20919 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
20920 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
20921 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
20922 (_mm_maskz_min_ss): New intrinsics.
20924 2017-05-31 Martin Liska <mliska@suse.cz>
20926 * tree-vect-loop.c (vect_create_epilog_for_reduction):
20927 Change comment style to one we normally use.
20928 (vectorizable_reduction): Likewise.
20929 (vectorizable_induction): Likewise.
20930 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
20931 (vectorizable_call): Likewise.
20932 (vectorizable_simd_clone_call): Likewise.
20933 (vectorizable_conversion): Likewise.
20934 (vectorizable_assignment): Likewise.
20935 (vectorizable_shift): Likewise.
20936 (vectorizable_operation): Likewise.
20937 (vectorizable_store): Likewise.
20938 (vectorizable_load): Likewise.
20939 * tree-vectorizer.h: Likewise.
20941 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
20943 * passes.c (emergency_dump_function): New.
20944 * tree-pass.h (emergency_dump_function): Declare.
20945 * plugin.c (plugins_internal_error_function): Remove.
20946 * plugin.h (plugins_internal_error_function): Remove declaration.
20947 * toplev.c (internal_error_function): New static function. Use it...
20948 (general_init): ...here.
20950 2017-05-31 Graham Markall <graham.markall@embecosm.com>
20952 * config/arc/arc.c (arc_print_operand): Handle constant operands.
20953 (arc_rtx_costs): Add costs for new patterns.
20954 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
20955 * config/arc/predicates.md: Add _1_2_3_operand predicate.
20957 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
20959 * tree-ssa-strlen.c (get_next_strinfo): New function.
20960 (get_stridx_plus_constant): Use it.
20961 (zero_length_string): Likewise.
20962 (adjust_related_strinfos): Likewise.
20963 (adjust_last_stmt): Likewise.
20965 2017-05-31 Richard Biener <rguenther@suse.de>
20968 * config/i386/i386.c (ix86_expand_builtin): Remove assert
20969 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
20971 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
20973 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
20974 loop_vinfo argument and use of dependence distance vectors.
20975 Check instead whether the two references differ only in their
20976 initial value and assume that they have the same alignment if the
20977 difference is a multiple of the vector alignment.
20978 (vect_analyze_data_refs_alignment): Update call accordingly.
20980 2017-05-31 Martin Liska <mliska@suse.cz>
20983 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
20985 2017-05-31 Bin Cheng <bin.cheng@arm.com>
20987 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
20988 (create_intersect_range_checks): Move from ...
20989 * tree-data-ref.c (create_intersect_range_checks_index)
20990 (create_intersect_range_checks): ... to here.
20991 (create_runtime_alias_checks): New function factored from ...
20992 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
20993 here. Call above function.
20994 * tree-data-ref.h (create_runtime_alias_checks): New function.
20996 2017-05-31 Bin Cheng <bin.cheng@arm.com>
20998 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
20999 segment length for dr_b and compute it in wide_int.
21001 2017-05-31 Richard Biener <rguenther@suse.de>
21003 PR tree-optimization/80906
21004 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
21005 and pass through iv_map.
21006 (copy_bb_and_scalar_dependences): Adjust.
21007 (translate_pending_phi_nodes): Likewise.
21008 (copy_loop_close_phi_args): Handle code-generating IVs instead
21011 2017-05-30 David Malcolm <dmalcolm@redhat.com>
21013 * diagnostic-color.c (color_dict): Add "type-diff".
21014 (parse_gcc_colors): Update comment.
21015 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
21016 -fdiagnostics-show-template-tree and -fno-elide-type.
21017 (GCC_COLORS): Add type-diff to example.
21019 (-fdiagnostics-show-template-tree): New.
21020 (-fno-elide-type): New.
21021 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
21022 the pp_format_decoder callback. Call any m_format_postprocessor's
21024 (pretty_printer::pretty_printer): Initialize
21025 m_format_postprocessor.
21026 (pretty_printer::~pretty_printer): Delete any
21027 m_format_postprocessor.
21028 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
21029 (class format_postprocessor): New class.
21030 (struct pretty_printer::format_decoder): Document the new parameters.
21031 (struct pretty_printer::m_format_postprocessor): New field.
21032 * tree-diagnostic.c (default_tree_printer): Update for new
21033 bool and const char ** params.
21034 * tree-diagnostic.h (default_tree_printer): Likewise.
21036 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
21038 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
21039 (lwa_operand): Delete rs6000_gen_cell_microcode test.
21040 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21041 rs6000_gen_cell_microcode code.
21042 (rs6000_final_prescan_insn): Delete.
21043 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
21044 "warn-cell-microcode" entries.
21045 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
21046 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
21047 throughout. Change cc_reg_not_micro_cr0_operand to
21048 cc_reg_not_cr0_operand throughout.
21049 (*extendhi<mode>2_noload): Delete.
21050 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
21051 (mwarn-cell-microcode): Delete.
21052 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
21053 -mgen-cell-microcode and -mwarn-cell-microcode.
21055 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
21058 * config/i386/constraints.md (Yd): New constraint.
21060 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
21061 and (?Yd, r) alternatives. Update insn attributes.
21062 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
21063 and (?*Yd, r) alternatives. Update insn attributes.
21064 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
21066 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
21068 * gimplify.c (gimplify_modify_expr): Don't create a
21069 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
21072 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
21074 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
21076 2017-05-30 Richard Biener <rguenther@suse.de>
21078 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
21079 and reduc_def fields.
21080 (STMT_VINFO_REDUC_TYPE): New define.
21081 (STMT_VINFO_REDUC_DEF): Likewise.
21082 (vect_force_simple_reduction): Adjust prototype.
21083 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
21084 (vect_is_simple_reduction): Remove check_reduction argument.
21085 (vect_force_simple_reduction): Adjust and set
21086 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21087 (vectorizable_reduction): Do not re-do reduction analysis
21088 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21089 * tree-parloops.c (gather_scalar_reductions): Adjust.
21091 2017-05-30 Richard Biener <rguenther@suse.de>
21093 PR middle-end/80901
21094 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
21097 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
21099 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
21100 Introduce unknown_misalignment parameter and remove vf.
21101 (vect_peeling_hash_get_lowest_cost):
21102 Pass unknown_misalignment parameter.
21103 (vect_enhance_data_refs_alignment):
21104 Fix unsupportable data ref treatment.
21106 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21108 * tree-vect-data-refs.c (vect_get_data_access_cost):
21109 Workaround for SLP handling.
21110 (vect_enhance_data_refs_alignment):
21111 Compute costs for doing no peeling at all, compare to the best
21112 peeling costs so far and avoid peeling if cheaper.
21114 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21116 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
21117 Return peeling info and set costs to zero for unlimited cost
21119 (vect_enhance_data_refs_alignment): Also inspect all datarefs
21120 with unknown misalignment. Compute and costs for unknown
21121 misalignment, compare them to the costs for known misalignment
21122 and choose the cheapest for peeling.
21124 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21126 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
21127 (vect_get_peeling_costs_all_drs): Create function.
21128 (vect_peeling_hash_get_lowest_cost):
21129 Use vect_get_peeling_costs_all_drs.
21130 (vect_peeling_supportable): Create function.
21131 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
21133 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21135 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
21136 DR_HAS_NEGATIVE_STEP.
21137 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
21138 (vect_enhance_data_refs_alignment): Use.
21139 (vect_duplicate_ssa_name_ptr_info): Use.
21140 * tree-vectorizer.h (dr_misalignment): Use.
21141 (known_alignment_for_access_p): Use.
21143 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21146 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
21148 (has_section_name): New function.
21150 2017-05-30 Martin Liska <mliska@suse.cz>
21153 * auto-profile.c (get_function_decl_from_block): Fix
21156 2017-05-30 Richard Biener <rguenther@suse.de>
21158 PR middle-end/80876
21159 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
21161 2017-05-30 Martin Liska <mliska@suse.cz>
21163 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
21164 * dumpfile.h (struct dump_file_info): Remove ctors.
21166 2017-05-30 Martin Liska <mliska@suse.cz>
21168 * predict.def: Fix GNU coding style.
21170 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21172 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
21173 Mark 'to' argument with ATTRIBUTE_UNUSED.
21175 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21177 * config/xtensa/xtensa.c (xtensa_emit_call): Use
21178 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
21179 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
21182 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
21184 * doc/install.texi (Options specification): Restore entry of
21185 --enable-sjlj-exceptions.
21187 2017-05-27 Michael Eager <eager@eagercon.com>
21190 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
21192 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
21194 * config/microblaze/microblaze.h
21195 (FIXED_REGISTERS): Update in macro.
21196 (CALL_USED_REGISTERS): Update in macro.
21198 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
21200 * doc/install.texi: Add links to macOS binary distributions.
21202 2017-05-27 Jakub Jelinek <jakub@redhat.com>
21206 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21208 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
21210 2017-05-26 Martin Liska <mliska@suse.cz>
21212 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
21214 2017-05-26 Martin Liska <mliska@suse.cz>
21216 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
21217 always leading ';; '.
21218 (dump_bb_info): Likewise.
21219 (brief_dump_cfg): Likewise.
21220 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
21221 * dumpfile.c: Remove usage of TDF_VERBOSE.
21222 * dumpfile.h (enum dump_kind): Likewise.
21223 (dump_gimple_bb_header): Do not use TDF_COMMENT.
21224 * print-tree.c (debug_verbose): Remove.
21225 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
21226 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
21227 * tree-diagnostic.c (default_tree_printer): Replace
21228 TDF_DIAGNOSTIC with TDF_SLIM.
21230 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21232 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
21233 in parameter loop, rather than loop_vinfo.
21234 (create_intersect_range_checks): Ditto.
21235 (vect_create_cond_for_alias_checks): Update call to above functions.
21237 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21239 PR tree-optimization/80815
21240 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
21241 for merging runtime alias checks. Handle negative DR_STEPs.
21243 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21245 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
21247 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
21248 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
21249 out code pruning runtime alias checks.
21250 * tree-data-ref.c (prune_runtime_alias_test_list): New function
21251 factored out from above.
21252 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
21254 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
21256 (prune_runtime_alias_test_list): New decalaration.
21258 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21260 * tree-vect-data-refs.c (compare_tree): Rename and move ...
21261 * tree-data-ref.c (data_ref_compare_tree): ... to here.
21262 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
21263 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
21264 (operator==, comp_dr_with_seg_len_pair): Ditto.
21265 (vect_prune_runtime_alias_test_list): Ditto.
21267 2017-05-26 Martin Liska <mliska@suse.cz>
21270 * params.def: Bound partial-inlining-entry-probability param.
21272 2017-05-26 Marek Polacek <polacek@redhat.com>
21275 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
21278 2017-05-26 Richard Biener <rguenther@suse.de>
21280 PR tree-optimization/80842
21281 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
21284 2017-05-26 Richard Biener <rguenther@suse.de>
21286 PR tree-optimization/80844
21287 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
21289 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
21291 * doc/md.texi (Machine Constraints): Update x86 family
21292 machine constraints section to match 'config/i386/constraints.md'.
21294 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
21296 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
21298 2017-05-25 Nathan Sidwell <nathan@acm.org>
21300 * doc/invoke.texi (--enable-languages): Update documentation.
21302 2017-05-25 Martin Liska <mliska@suse.cz>
21304 * dumpfile.c: Add TDF_FOLDING.
21305 * dumpfile.h (enum dump_kind): Likewise.
21306 * genmatch.c (dt_simplify::gen_1): Use it.
21308 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21310 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
21312 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21314 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
21315 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
21317 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21319 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
21320 * match.pd (X == C): Rewrite it here.
21321 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
21322 with_certain_nonzero_bits2): New predicates.
21323 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
21325 2017-05-24 Nathan Sidwell <nathan@acm.org>
21327 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
21330 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
21333 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
21335 * config/powerpcspe: New port. Files are copied from the rs6000
21336 port, with "rs6000" in filenames replaced by "powerpcspe".
21338 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
21340 PR rtl-optimization/80754
21341 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
21343 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
21345 * config/sparc/sparc.md (length): Return the correct value for -mflat
21346 sibcalls to match output_sibcall.
21348 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
21352 * config/rs6000/rs6000.c (struct machine_function): Add new field
21354 (rs6000_get_separate_components): Init that field, use it.
21355 (rs6000_components_for_bb): Use the field.
21357 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
21359 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
21361 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
21363 PR middle-end/80823
21364 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
21366 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21369 * config/s390/s390.c (s390_check_qrst_address): Check incoming
21370 address against address_operand predicate.
21371 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
21373 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
21375 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
21376 parameters passed indirectly.
21378 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
21380 * config/i386/i386.md (*movdi_internal): Remove SSE4
21381 alternative 18 (?r, *v). Update insn attributes.
21382 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
21383 Update insn attributes.
21384 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
21385 Update insn attributes.
21386 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
21387 alternative 1 (r, v). Remove isa attribute.
21388 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
21389 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
21390 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
21392 2017-05-23 Tom de Vries <tom@codesourcery.com>
21394 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
21397 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21399 * cgraphunit.c (symbol_table::process_new_functions): Update.
21400 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
21401 (inline_generate_summary): Rename to ...
21402 (ipa_fn_summary_generate): ... this one.
21403 (inline_read_summary): Rename to ...
21404 (ipa_fn_summary_read): ... this one.
21405 (inline_write_summary): Rename to ...
21406 (ipa_fn_summary_write): ... this one.
21407 (inline_free_summary): Rename to ...
21408 (ipa_free_fn_summary): ... this one.
21409 (pass_data_local_fn_summary, pass_local_fn_summary,
21410 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
21411 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
21412 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
21413 make_pass_ipa_fn_summary): New.
21414 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
21415 inline_write_summary, inline_free_summary): Remove.
21416 (ipa_free_fn_summary) : New.
21417 * ipa-inline.c (ipa_inline): Update.
21418 (pass_ipa_inline): Do not generate summaries.
21419 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
21421 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
21422 and add pass_ipa_fn_summary.
21423 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
21425 (make_pass_inline_parameters): Remove.
21427 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
21429 * omp-low.c (struct omp_context): Remove "default_kind" member.
21432 * omp-offload.c (execute_oacc_device_lower): Remove the
21433 parallelism dimensions function attributes for unparallelized
21434 OpenACC kernels constructs.
21436 2017-05-23 Martin Liska <mliska@suse.cz>
21438 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
21440 (cgraph_edge::make_speculative): Likewise.
21441 (cgraph_edge::resolve_speculation): Likewise.
21442 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
21443 (cgraph_node::dump): Likewise.
21444 * cgraph.h: Likewise.
21445 * cgraphunit.c (analyze_functions): Likewise.
21446 (symbol_table::compile): Likewise.
21447 * ipa-cp.c (print_all_lattices): Likewise.
21448 (determine_versionability): Likewise.
21449 (initialize_node_lattices): Likewise.
21450 (ipcp_verify_propagated_values): Likewise.
21451 (estimate_local_effects): Likewise.
21452 (update_profiling_info): Likewise.
21453 (create_specialized_node): Likewise.
21454 (perhaps_add_new_callers): Likewise.
21455 (decide_about_value): Likewise.
21456 (decide_whether_version_node): Likewise.
21457 (identify_dead_nodes): Likewise.
21458 (ipcp_store_bits_results): Likewise.
21459 * ipa-devirt.c (dump_targets): Likewise.
21460 (ipa_devirt): Likewise.
21461 * ipa-icf.c (sem_item::dump): Likewise.
21462 (sem_function::equals): Likewise.
21463 (sem_variable::equals): Likewise.
21464 (sem_item_optimizer::read_section): Likewise.
21465 (sem_item_optimizer::execute): Likewise.
21466 (congruence_class::dump): Likewise.
21467 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
21468 (dump_inline_summary): Likewise.
21469 (estimate_node_size_and_time): Likewise.
21470 (inline_analyze_function): Likewise.
21471 * ipa-inline-transform.c (inline_call): Likewise.
21472 * ipa-inline.c (report_inline_failed_reason): Likewise.
21473 (want_early_inline_function_p): Likewise.
21474 (edge_badness): Likewise.
21475 (update_edge_key): Likewise.
21476 (inline_small_functions): Likewise.
21477 * ipa-profile.c (ipa_profile): Likewise.
21478 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
21479 (ipa_make_edge_direct_to_target): Likewise.
21480 (remove_described_reference): Likewise.
21481 (ipa_impossible_devirt_target): Likewise.
21482 (propagate_controlled_uses): Likewise.
21483 (ipa_print_node_params): Likewise.
21484 (ipcp_transform_function): Likewise.
21485 * ipa-pure-const.c (pure_const_read_summary): Likewise.
21486 (propagate_pure_const): Likewise.
21487 * ipa-reference.c (generate_summary): Likewise.
21488 (read_write_all_from_decl): Likewise.
21489 (propagate): Likewise.
21490 (ipa_reference_read_optimization_summary): Likewise.
21491 * ipa-utils.c (ipa_merge_profiles): Likewise.
21492 * ipa.c (walk_polymorphic_call_targets): Likewise.
21493 (symbol_table::remove_unreachable_nodes): Likewise.
21494 (ipa_single_use): Likewise.
21495 * passes.c (execute_todo): Likewise.
21496 * predict.c (drop_profile): Likewise.
21497 * symtab.c (symtab_node::get_dump_name): New function.
21498 (symtab_node::dump_name): Likewise.
21499 (symtab_node::dump_asm_name): Likewise.
21500 (symtab_node::dump_references): Likewise.
21501 (symtab_node::dump_referring): Likewise.
21502 (symtab_node::dump_base): Likewise.
21503 (symtab_node::debug_symtab): Likewise.
21504 * tree-sra.c (convert_callers_for_node): Likewise.
21505 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
21506 * value-prof.c (init_node_map): Likewise.
21508 2017-05-23 Martin Liska <mliska@suse.cz>
21510 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
21511 and symtab_node::debug_symtab to symbol_table::debug.
21512 * cgraphunit.c (analyze_functions): Use the renamed function.
21513 (symbol_table::compile): Likewise.
21514 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
21515 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
21516 * passes.c (execute_todo): Likewise.
21517 * symtab.c (symbol_table::dump): New function.
21518 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
21520 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21522 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
21523 that nonconst implies exec.
21525 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21527 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
21528 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
21529 (inline_edge_summary_vec): Turn into ...
21530 (ipa_call_summaries): ... this one.
21531 (redirect_to_unreachable, edge_set_predicate,
21532 evaluate_properties_for_edge, inline_summary_alloc,
21533 reset_ipa_call_summary, reset_inline_summary,
21534 inline_summary_t::duplicate): Update.
21535 (inline_edge_duplication_hook): Turn to ...
21536 (ipa_call_summary_t::duplicate): ... this one.
21537 (inline_edge_removal_hook): Turn to ...
21538 (ipa_call_summary_t::remove): ... this one.
21539 (dump_inline_edge_summary): Turn to ...
21540 (dump_ipa_call_summary): ... this one.
21541 (estimate_function_body_sizes): Update.
21542 (inline_update_callee_summaries): Update.
21543 (remap_edge_change_prob): Update.
21544 (remap_edge_summaries): Update.
21545 (inline_merge_summary): Update.
21546 (do_estimate_edge_time): Update.
21547 (inline_generate_summary): Update.
21548 (inline_read_section): Update.
21549 (inline_read_summary): Update.
21550 (inline_free_summary): Update.
21551 * ipa-inline.c (can_inline_edge_p): Update.
21552 (compute_inlined_call_time): Update.
21553 (want_inline_small_function_p): Update.
21554 (edge_badness): Update.
21555 (early_inliner): Update.
21556 * ipa-inline.h (inline_edge_summary): Turn to ...
21557 (ipa_call_summary): ... this one.
21558 (ipa_call_summary_t): New class.
21559 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
21560 (ipa_call_summaries): New.
21561 (inline_edge_summary): Remove.
21562 (estimate_edge_growth): Update.
21563 * ipa-profile.c (ipa_propagate_frequency_1): Update.
21564 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21565 * ipa-split.c (execute_split_functions): Update.
21566 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
21568 2017-05-23 Tom de Vries <tom@codesourcery.com>
21570 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
21571 attributes): Document rdrand effective target.
21573 2017-05-23 Tom de Vries <tom@codesourcery.com>
21575 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
21576 attributes): Sort alphabetically.
21578 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
21580 * config/avr/genmultilib.awk: Use gsub instead of gensub.
21582 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
21585 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
21586 V2DF/V2DI splat into two separate patterns, one that handles
21587 registers, and the other that only handles memory. Drop support
21588 for splatting from a GPR on ISA 2.07 and then splitting the
21589 splat into direct move and splat.
21590 (vsx_splat_<mode>_reg): Likewise.
21591 (vsx_splat_<mode>_mem): Likewise.
21593 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
21595 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
21597 2017-05-22 Jakub Jelinek <jakub@redhat.com>
21599 PR middle-end/80809
21600 * omp-low.c (finish_taskreg_remap): New function.
21601 (finish_taskreg_scan): If unit size of ctx->record_type
21602 is non-constant, unshare the size expression and replace
21603 decls in it with possible outer var refs.
21605 PR middle-end/80809
21606 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
21607 GOVD_SHARED rather than GOVD_PRIVATE with it.
21608 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
21609 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
21611 PR middle-end/80853
21612 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
21613 as last argument to build_outer_var_ref for pointer bases of array
21614 section reductions.
21616 2017-05-19 Martin Sebor <msebor@redhat.com>
21618 * print-tree.c (print_node): Print DECL_READ_P flag.
21620 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21622 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
21623 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
21624 * cgraph.c: Likewise.
21625 * cgraphunit.c: Likewise.
21626 * gengtype.c: Likewise.
21627 * ipa-cp.c: Likewise.
21628 * ipa-devirt.c: Likewise.
21629 * ipa-icf.c: Likewise.
21630 * ipa-predicate.c: Likewise.
21631 * ipa-profile.c: Likewise.
21632 * ipa-prop.c: Likewise.
21633 * ipa-split.c: Likewise.
21635 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
21636 edge_predicate_pool, dump_inline_hints,
21637 inline_summary::account_size_time, redirect_to_unreachable,
21638 edge_set_predicate, set_hint_predicate,
21639 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
21640 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
21641 inline_summary_t::remove, remap_hint_predicate_after_duplication,
21642 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
21643 ipa_call_summary_t::remove, initialize_growth_caches,
21644 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
21645 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
21646 mark_modified, unmodified_parm_1, unmodified_parm,
21647 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
21648 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
21649 compute_bb_predicates, will_be_nonconstant_expr_predicate,
21650 will_be_nonconstant_predicate, record_modified_bb_info,
21651 get_minimal_bb, record_modified, param_change_prob,
21652 phi_result_unknown_predicate, predicate_for_phi_result,
21653 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
21654 estimate_function_body_sizes, compute_inline_parameters,
21655 compute_inline_parameters_for_curren, pass_data_inline_parameters,
21656 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
21657 inline_update_callee_summaries, remap_edge_change_prob,
21658 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
21659 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
21660 inline_analyze_function, inline_summary_t::insert,
21661 inline_generate_summary, read_ipa_call_summary, inline_read_section,
21662 inline_read_summary, write_ipa_call_summary, inline_write_summary,
21663 inline_free_summary): Move to ipa-fnsummary.h
21664 (predicate_t): Remove.
21665 * ipa-fnsummary.c: New file.
21666 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
21667 (enum inline_hints_vals, inline_hints, agg_position_info,
21668 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
21669 inline_summaries, ipa_call_summary, ipa_call_summary_t,
21670 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
21671 dump_inline_summary, dump_inline_hints, inline_generate_summary,
21672 inline_read_summary, inline_write_summary, inline_free_summary,
21673 inline_analyze_function, initialize_inline_failed,
21674 inline_merge_summary, inline_update_overall_summary,
21675 compute_inline_parameters): Move to ipa-fnsummary.h
21676 * ipa-fnsummary.h: New file.
21677 * ipa-inline-transform.h: Include ipa-inline.h.
21678 * ipa-inline.c: LIkewise.
21680 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21682 * ipa-inline.c (edge_badness): Use inlined_time instead of
21683 inline_summaries->get.
21685 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21687 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
21689 2017-05-22 Nathan Sidwell <nathan@acm.org>
21691 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
21692 (fdump-lang): Document 'raw' option.
21693 * dumpfile.h (TDI_tu): Delete.
21694 * dumpfile.c (dump_files): Remove translation-unit.
21695 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
21697 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
21699 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
21700 command option from $(AWK) call.
21701 * config/avr/genmultilib.awk: Simplify and rewrite so that it
21702 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
21703 [FORMAT]: Remove handling of variable.
21704 * config/avr/t-multilib: Regenerate.
21706 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21708 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
21710 (dump_inline_summary): Do not print self_time.
21711 (estimate_function_body_sizes): Do not set self_time.
21712 (compute_inline_parameters): Likewise.
21713 (inline_read_section, inline_write_summary): Do not stream self_time.
21714 * ipa-inline.h (inline_summary): Drop self_time.
21716 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21718 * ipa-inline-analysis.c (account_size_time): Rename to ...
21719 (inline_summary::account_size_time): ... this one.
21720 (reset_ipa_call_summary): Turn to ...
21721 (ipa_call_summary::reset): ... this one.
21722 (reset_inline_summary): Turn to ...
21723 (inline_summary::reset): ... this one.
21724 (inline_summary_t::remove): Update.
21725 (inline_summary_t::duplicate): Update.
21726 (ipa_call_summary_t::remove): Update.
21727 (dump_inline_summary): Update.
21728 (estimate_function_body_sizes): Update.
21729 (compute_inline_parameters): Update.
21730 (estimate_node_size_and_time): Update.
21731 (inline_merge_summary): Update.
21732 (inline_update_overall_summary): Update.
21733 (inline_read_section): Update.
21734 (inline_write_summary): Update.
21735 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
21736 add account_size_time and reset member functions.
21737 (ipa_call_summary): Add reset function.
21738 * ipa-predicate.h (predicate::operator &): Constify.
21740 2017-05-22 Richard Biener <rguenther@suse.de>
21742 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
21744 2017-05-19 Jason Merrill <jason@redhat.com>
21746 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
21748 2017-05-19 Marek Polacek <polacek@redhat.com>
21751 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
21752 TYPE_OVERFLOW_WRAPS checks.
21754 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
21756 * tree-core.h (enum omp_clause_default_kind): Add
21757 "OMP_CLAUSE_DEFAULT_PRESENT".
21758 * tree-pretty-print.c (dump_omp_clause): Handle it.
21759 * gimplify.c (enum gimplify_omp_var_data): Add
21760 "GOVD_MAP_FORCE_PRESENT".
21761 (gimplify_adjust_omp_clauses_1): Map it to
21762 "GOMP_MAP_FORCE_PRESENT".
21763 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
21765 * gimplify.c (oacc_default_clause): Clarify.
21767 2017-05-19 Nathan Sidwell <nathan@acm.org>
21769 LANG_HOOK_REGISTER_DUMPS
21770 * toplev.c (general_init): Call register dump lang hook.
21771 * doc/invoke.texi: Document -fdump-lang option family.
21772 * dumpfile.c (dump_files): Remove class dump here.
21773 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
21774 * dumpfile.h (tree_dump_index): Remove TDI_class.
21775 * langhooks-def.h (lhd_register_dumps): Declare.
21776 (LANG_HOOKS_REGISTER_DUMPS): Define.
21777 (LANG_HOOKS_INITIALIZER): Add it.
21778 * langhooks.c (lhd_register_dumps): Define.
21779 * langhooks.h (struct lang_hooks): Add register_dumps.
21781 2017-05-19 Nathan Sidwell <nathan@acm.org>
21783 * context.h (context::set_passes): New.
21784 * context.c (context::context): Do not create pass manager.
21785 * toplev.c (general_init): Create pass manager here.
21787 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
21789 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
21790 use this splitter if two add or or instructions would also work for
21791 the constant we want to generate.
21793 2017-05-19 Richard Biener <rguenther@suse.de>
21796 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
21797 predicate evaluation.
21799 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
21801 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
21803 * ipa-inline.c (want_inline_small_function_p): Do not cast to
21806 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
21808 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
21809 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
21810 (inline_edge_summary_vec): Turn into ...
21811 (ipa_call_summaries): ... this one.
21812 (redirect_to_unreachable, edge_set_predicate,
21813 evaluate_properties_for_edge, inline_summary_alloc,
21814 reset_ipa_call_summary, reset_inline_summary,
21815 inline_summary_t::duplicate): Update.
21816 (inline_edge_duplication_hook): Turn to ...
21817 (ipa_call_summary_t::duplicate): ... this one.
21818 (inline_edge_removal_hook): Turn to ...
21819 (ipa_call_summary_t::remove): ... this one.
21820 (dump_inline_edge_summary): Turn to ...
21821 (dump_ipa_call_summary): ... this one.
21822 (estimate_function_body_sizes): Update.
21823 (inline_update_callee_summaries): Update.
21824 (remap_edge_change_prob): Update.
21825 (remap_edge_summaries): Update.
21826 (inline_merge_summary): Update.
21827 (do_estimate_edge_time): Update.
21828 (inline_generate_summary): Update.
21829 (inline_read_section): Update.
21830 (inline_read_summary): Update.
21831 (inline_free_summary): Update.
21832 * ipa-inline.c (can_inline_edge_p): Update.
21833 (compute_inlined_call_time): Update.
21834 (want_inline_small_function_p): Update.
21835 (edge_badness): Update.
21836 (early_inliner): Update.
21837 * ipa-inline.h (inline_edge_summary): Turn to ...
21838 (ipa_call_summary): ... this one.
21839 (ipa_call_summary_t): New class.
21840 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
21841 (ipa_call_summaries): New.
21842 (inline_edge_summary): Remove.
21843 (estimate_edge_growth): Update.
21844 * ipa-profile.c (ipa_propagate_frequency_1): Update.
21845 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21846 * ipa-split.c (execute_split_functions): Update.
21847 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
21849 2017-05-19 Richard Biener <rguenther@suse.de>
21851 PR middle-end/80764
21852 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
21854 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
21856 * config/rs6000/rs6000.c (struct machine_function): Add field
21857 fpr_is_wrapped_separately.
21858 (rs6000_get_separate_components): Use 64 components. Handle the
21859 new FPR components.
21860 (rs6000_components_for_bb): Handle the FPR components.
21861 (rs6000_emit_prologue_components): Handle the FPR components.
21862 (rs6000_emit_epilogue_components): Handle the FPR components.
21863 (rs6000_set_handled_components): Handle the FPR components.
21864 (rs6000_emit_prologue): Don't output prologue code for those FPRs
21865 that are already separately shrink-wrapped.
21866 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
21867 that are already separately shrink-wrapped.
21869 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
21872 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
21875 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
21876 (define_peephole2 for Altivec d-form load): Add peepholes to catch
21877 cases where the register allocator uses a move and an offsettable
21878 memory operation to/from a FPR register on ISA 2.06/2.07.
21879 (define_peephole2 for Altivec d-form store): Likewise.
21881 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
21884 * config/i386/mmx.md (*mov<mode>_internal): Enable
21885 alternatives 11, 12, 13 and 14 also for 32bit targets.
21886 Remove alternatives 15, 16, 17 and 18.
21887 * config/i386/sse.md (vec_concatv2di): Change
21888 alternative (!x, *y) to (x, ?!*Yn).
21890 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
21892 * dumpfile.h (enum dump_kind): Remove stray comma.
21894 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
21896 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
21897 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
21898 predicate::num_conditions
21899 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
21900 (CHANGED): turn into predicate::changed.
21901 (agg_position_info): Move to ipa-predicate.h
21902 (add_condition, predicate::add_clause, predicate::operator &=,
21903 predicate::or_with, predicate::evaluate, predicate::probability,
21904 dump_condition, dump_clause, predicate::dump,
21905 predicate::remap_after_duplication, predicate::remap_after_inlining,
21906 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
21907 (evaluate_conditions_for_known_args): Update.
21908 (set_cond_stmt_execution_predicate): Update.
21909 * ipa-inline.h: Include ipa-predicate.h
21910 (condition, inline_param_summary, conditions, agg_position_info,
21911 predicate): Move to ipa-predicate.h
21912 * ipa-predicate.c: New file.
21913 * ipa-predicate.h: New file.
21915 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
21917 * final.c (leaf_function_p): Check we are not in a sequence.
21919 2017-05-18 Martin Liska <mliska@suse.cz>
21921 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
21922 * dumpfile.c (dump_register): Use new enum dump_kind.
21923 (get_dump_file_name): Likewise.
21924 (dump_enable_all): Likewise.
21925 (dump_switch_p_1): Likewise.
21926 (enable_rtl_dump_file): Remove usage of TDF_RTL.
21927 * dumpfile.h (enum dump_kind): New enum type.
21928 (struct dump_file_info): Create constructor and
21929 format fields and comments.
21930 * passes.c (pass_manager::register_one_dump_file):
21932 * statistics.c (statistics_early_init): Likewise.
21933 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
21934 TDF_TREE with TDF_SLIM.
21935 (gather_memory_references_ref): Likewise.
21937 2017-05-18 Martin Liska <mliska@suse.cz>
21939 * vec.h (struct vnull): Use it.
21941 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
21943 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
21944 (true_predicate, false_predicate, true_predicate_p,
21945 false_predicate_p): Remove.
21946 (single_cond_predicate, not_inlined_predicate): Turn to member function
21948 (add_condition): Update.
21949 (add_clause): Turn to...
21950 (predicate::add_clause): ... this one; update; allow passing NULL
21952 (and_predicates): Turn to ...
21953 (predicate::operator &=): ... this one.
21954 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
21955 (or_predicates): Turn to ...
21956 (predicate::or_with): ... this one.
21957 (evaluate_predicate): Turn to ...
21958 (predicate::evaluate): ... this one.
21959 (predicate_probability): Turn to ...
21960 (predicate::probability): ... this one.
21961 (dump_condition): Update.
21962 (dump_predicate): Turn to ...
21963 (predicate::dump): ... this one.
21964 (account_size_time): Update.
21965 (edge_set_predicate): Update.
21966 (set_hint_predicate): UPdate.
21967 (evaluate_conditions_for_known_args): Update.
21968 (evaluate_properties_for_edge): Update.
21969 (remap_predicate_after_duplication): Turn to...
21970 (predicate::remap_after_duplication): ... this one.
21971 (remap_hint_predicate_after_duplication): Update.
21972 (inline_summary_t::duplicate): UPdate.
21973 (dump_inline_edge_summary): Update.
21974 (dump_inline_summary): Update.
21975 (set_cond_stmt_execution_predicate): Update.
21976 (set_switch_stmt_execution_predicate): Update.
21977 (compute_bb_predicates): Update.
21978 (will_be_nonconstant_expr_predicate): Update.
21979 (will_be_nonconstant_predicate): Update.
21980 (phi_result_unknown_predicate): Update.
21981 (predicate_for_phi_result): Update.
21982 (array_index_predicate): Update.
21983 (estimate_function_body_sizes): Update.
21984 (estimate_node_size_and_time): Update.
21985 (estimate_ipcp_clone_size_and_time): Update.
21986 (remap_predicate): Rename to ...
21987 (predicate::remap_after_inlining): ... this one.
21988 (remap_hint_predicate): Update.
21989 (inline_merge_summary): Update.
21990 (inline_update_overall_summary): Update.
21991 (estimate_size_after_inlining): Update.
21992 (read_predicate): Rename to ...
21993 (predicate::stream_in): ... this one.
21994 (read_inline_edge_summary): Update.
21995 (write_predicate): Rename to ...
21996 (predicate::stream_out): ... this one.
21997 (write_inline_edge_summary): Update.
21998 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
21999 (clause_t): Turn to uint32_t
22000 (predicate): Turn to class; implement constructor and operators
22002 (size_time_entry): Update.
22003 (inline_summary): Update.
22004 (inline_edge_summary): Update.
22006 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
22008 * fold-const.c (fold_binary_loc): Move transformation...
22009 * match.pd (C - X CMP X): ... here.
22011 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
22013 * config/sparc/sparc.c (sparc_option_override): Set function
22014 alignment for -mcpu=niagara7 to 64 to match the I$ line.
22015 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
22017 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
22019 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
22021 2017-05-18 Marek Polacek <polacek@redhat.com>
22024 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
22025 (pass_ubsan::execute): Call gimple_assign_single_p instead of
22026 gimple_assign_load_p.
22028 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
22030 PR middle-end/80692
22031 * real.c (do_compare): Give decimal_do_compare preference over
22032 comparing just the signs.
22034 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
22036 * doc/md.texi (Canonicalization of Instructions): Describe the
22037 canonical form of instructions that inherently set a condition
22040 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
22042 PR middle-end/80775
22043 * tree-cfg.c: Move deletion of unreachable case statements to after
22044 the merging of consecutive case labels.
22046 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
22048 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
22049 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
22050 restoring of callee-saved registers.
22052 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
22054 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
22055 * config/visium/visium.c (single_set_and_flags): Likewise.
22056 * config/visium/visium.md (Substitutions): Likewise.
22058 2017-05-17 Martin Liska <mliska@suse.cz>
22060 * cfg.c: Introduce dump_flags_t type and
22061 use it instead of int type.
22063 * cfghooks.c: Likewise.
22064 * cfghooks.h (struct cfg_hooks): Likewise.
22065 * cfgrtl.c: Likewise.
22066 * cfgrtl.h: Likewise.
22067 * cgraph.c (cgraph_node::get_body): Likewise.
22068 * coretypes.h: Likewise.
22069 * domwalk.c: Likewise.
22070 * domwalk.h: Likewise.
22071 * dumpfile.c (struct dump_option_value_info): Likewise.
22072 (dump_enable_all): Likewise.
22073 (dump_switch_p_1): Likewise.
22074 (opt_info_switch_p): Likewise.
22075 * dumpfile.h (enum tree_dump_index): Likewise.
22076 (struct dump_file_info): Likewise.
22077 * genemit.c: Likewise.
22078 * generic-match-head.c: Likewise.
22079 * gengtype.c (open_base_files): Likewise.
22080 * gimple-pretty-print.c: Likewise.
22081 * gimple-pretty-print.h: Likewise.
22082 * graph.c (print_graph_cfg): Likewise.
22083 * graphite-scop-detection.c (dot_all_sese): Likewise.
22084 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
22085 * loop-unroll.c (report_unroll): Likewise.
22086 * passes.c (pass_manager::register_one_dump_file): Likewise.
22087 * print-tree.c: Likewise.
22088 * statistics.c: Likewise.
22089 * tree-cfg.c: Likewise.
22090 * tree-cfg.h: Likewise.
22091 * tree-dfa.c: Likewise.
22092 * tree-dfa.h: Likewise.
22093 * tree-dump.c (dump_function): Likewise.
22094 * tree-dump.h (struct dump_info): Likewise.
22095 * tree-pretty-print.c: Likewise.
22096 * tree-pretty-print.h: Likewise.
22097 * tree-ssa-live.c: Likewise.
22098 * tree-ssa-live.h: Likewise.
22099 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
22100 * tree-vect-loop.c: Likewise.
22101 * tree-vect-slp.c: Likewise.
22103 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
22104 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22106 PR tree-optimization/80457
22107 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
22108 of all arguments to a statement as scalar_to_vec operations.
22109 (vectorizable_call): Adjust call to vect_model_simple_cost for
22111 (vectorizable_conversion): Likewise.
22112 (vectorizable_assignment): Likewise.
22113 (vectorizable_shift): Likewise.
22114 (vectorizable_operation): Likewise.
22115 (vectorizable_comparison): Likewise.
22116 (vect_is_simple_cond): Record the def types for operands.
22117 (vectorizable_condition): Likewise, call vect_model_simple_cost.
22118 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
22119 for statement argument count.
22121 2017-05-16 Carl Love <cel@us.ibm.com>
22123 * config/rs6000/rs6000-c: Add support for built-in functions
22124 vector unsigned long long vec_bperm (vector unsigned long long,
22125 vector unsigned char)
22126 vector signed long long vec_mule (vector signed int,
22128 vector unsigned long long vec_mule (vector unsigned int,
22129 vector unsigned int)
22130 vector signed long long vec_mulo (vector signed int,
22132 vector unsigned long long vec_mulo (vector unsigned int,
22133 vector unsigned int)
22134 vector signed char vec_sldw (vector signed char,
22135 vector signed char,
22137 vector unsigned char vec_sldw (vector unsigned char,
22138 vector unsigned char,
22140 vector signed short vec_sldw (vector signed short,
22141 vector signed short,
22143 vector unsigned short vec_sldw (vector unsigned short,
22144 vector unsigned short,
22146 vector signed int vec_sldw (vector signed int,
22149 vector unsigned int vec_sldw (vector unsigned int,
22150 vector unsigned int,
22152 vector signed long long vec_sldw (vector signed long long,
22153 vector signed long long,
22155 vector unsigned long long vec_sldw (vector unsigned long long,
22156 vector unsigned long long,
22158 * config/rs6000/rs6000-c: Add support for built-in functions
22159 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
22160 * config/rs6000/altivec.h: Add defintion for vec_sldw.
22161 * doc/extend.texi: Update the built-in documentation for the
22162 new built-in functions.
22164 2017-05-16 Marek Polacek <polacek@redhat.com>
22168 * tree.c (save_expr): Don't fold the expression.
22170 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
22172 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
22173 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
22174 and (?*y,m). Update insn attributes.
22176 2017-05-16 Martin Liska <mliska@suse.cz>
22178 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
22179 flags argument of print_gimple_stmt, print_gimple_expr,
22180 print_generic_stmt and print_generic_expr.
22181 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
22182 * coretypes.h: Likewise.
22183 * except.c (dump_eh_tree): Likewise.
22184 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
22185 * gimple-pretty-print.h: Likewise.
22186 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
22187 (backprop::push_to_worklist): Likewise.
22188 (backprop::pop_from_worklist): Likewise.
22189 (backprop::process_use): Likewise.
22190 (backprop::intersect_uses): Likewise.
22191 (note_replacement): Likewise.
22192 * gimple-ssa-store-merging.c
22193 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
22194 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
22195 (pass_store_merging::execute): Likewise.
22196 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
22197 (ssa_base_cand_dump_callback): Likewise.
22198 (dump_incr_vec): Likewise.
22199 (replace_refs): Likewise.
22200 (replace_mult_candidate): Likewise.
22201 (create_add_on_incoming_edge): Likewise.
22202 (create_phi_basis): Likewise.
22203 (insert_initializers): Likewise.
22204 (all_phi_incrs_profitable): Likewise.
22205 (introduce_cast_before_cand): Likewise.
22206 (replace_one_candidate): Likewise.
22207 * gimplify.c (gimplify_expr): Likewise.
22208 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
22209 (set_rename): Likewise.
22210 (rename_uses): Likewise.
22211 (copy_loop_phi_nodes): Likewise.
22212 (add_close_phis_to_merge_points): Likewise.
22213 (copy_loop_close_phi_args): Likewise.
22214 (copy_cond_phi_args): Likewise.
22215 (graphite_copy_stmts_from_block): Likewise.
22216 (translate_pending_phi_nodes): Likewise.
22217 * graphite-poly.c (print_pdr): Likewise.
22218 (dump_gbb_cases): Likewise.
22219 (dump_gbb_conditions): Likewise.
22220 (print_scop_params): Likewise.
22221 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
22222 (build_cross_bb_scalars_use): Likewise.
22223 (gather_bbs::before_dom_children): Likewise.
22224 * hsa-dump.c (dump_hsa_immed): Likewise.
22225 * ipa-cp.c (print_ipcp_constant_value): Likewise.
22226 (get_replacement_map): Likewise.
22227 * ipa-inline-analysis.c (dump_condition): Likewise.
22228 (estimate_function_body_sizes): Likewise.
22229 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
22230 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
22231 * ipa-prop.c (ipa_dump_param): Likewise.
22232 (ipa_print_node_jump_functions_for_edge): Likewise.
22233 (ipa_modify_call_arguments): Likewise.
22234 (ipa_modify_expr): Likewise.
22235 (ipa_dump_param_adjustments): Likewise.
22236 (ipa_dump_agg_replacement_values): Likewise.
22237 (ipcp_modif_dom_walker::before_dom_children): Likewise.
22238 * ipa-pure-const.c (check_stmt): Likewise.
22239 (pass_nothrow::execute): Likewise.
22240 * ipa-split.c (execute_split_functions): Likewise.
22241 * omp-offload.c (dump_oacc_loop_part): Likewise.
22242 (dump_oacc_loop): Likewise.
22243 * trans-mem.c (tm_log_emit): Likewise.
22244 (tm_memopt_accumulate_memops): Likewise.
22245 (dump_tm_memopt_set): Likewise.
22246 (dump_tm_memopt_transform): Likewise.
22247 * tree-cfg.c (gimple_verify_flow_info): Likewise.
22248 (print_loop): Likewise.
22249 * tree-chkp-opt.c (chkp_print_addr): Likewise.
22250 (chkp_gather_checks_info): Likewise.
22251 (chkp_get_check_result): Likewise.
22252 (chkp_remove_check_if_pass): Likewise.
22253 (chkp_use_outer_bounds_if_possible): Likewise.
22254 (chkp_reduce_bounds_lifetime): Likewise.
22255 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
22256 (chkp_mark_completed_bounds): Likewise.
22257 (chkp_register_incomplete_bounds): Likewise.
22258 (chkp_mark_invalid_bounds): Likewise.
22259 (chkp_maybe_copy_and_register_bounds): Likewise.
22260 (chkp_build_returned_bound): Likewise.
22261 (chkp_get_bound_for_parm): Likewise.
22262 (chkp_build_bndldx): Likewise.
22263 (chkp_get_bounds_by_definition): Likewise.
22264 (chkp_generate_extern_var_bounds): Likewise.
22265 (chkp_get_bounds_for_decl_addr): Likewise.
22266 * tree-chrec.c (chrec_apply): Likewise.
22267 * tree-data-ref.c (dump_data_reference): Likewise.
22268 (dump_subscript): Likewise.
22269 (dump_data_dependence_relation): Likewise.
22270 (analyze_overlapping_iterations): Likewise.
22271 * tree-inline.c (expand_call_inline): Likewise.
22272 (tree_function_versioning): Likewise.
22273 * tree-into-ssa.c (dump_defs_stack): Likewise.
22274 (dump_currdefs): Likewise.
22275 (dump_names_replaced_by): Likewise.
22276 (dump_update_ssa): Likewise.
22277 (update_ssa): Likewise.
22278 * tree-object-size.c (pass_object_sizes::execute): Likewise.
22279 * tree-parloops.c (build_new_reduction): Likewise.
22280 (try_create_reduction_list): Likewise.
22281 (ref_conflicts_with_region): Likewise.
22282 (oacc_entry_exit_ok_1): Likewise.
22283 (oacc_entry_exit_single_gang): Likewise.
22284 * tree-pretty-print.h: Likewise.
22285 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
22286 (get_scalar_evolution): Likewise.
22287 (add_to_evolution): Likewise.
22288 (get_loop_exit_condition): Likewise.
22289 (analyze_evolution_in_loop): Likewise.
22290 (analyze_initial_condition): Likewise.
22291 (analyze_scalar_evolution): Likewise.
22292 (instantiate_scev): Likewise.
22293 (number_of_latch_executions): Likewise.
22294 (gather_chrec_stats): Likewise.
22295 (final_value_replacement_loop): Likewise.
22296 (scev_const_prop): Likewise.
22297 * tree-sra.c (dump_access): Likewise.
22298 (disqualify_candidate): Likewise.
22299 (create_access): Likewise.
22300 (reject): Likewise.
22301 (maybe_add_sra_candidate): Likewise.
22302 (create_access_replacement): Likewise.
22303 (analyze_access_subtree): Likewise.
22304 (analyze_all_variable_accesses): Likewise.
22305 (sra_modify_assign): Likewise.
22306 (initialize_constant_pool_replacements): Likewise.
22307 (find_param_candidates): Likewise.
22308 (decide_one_param_reduction): Likewise.
22309 (replace_removed_params_ssa_names): Likewise.
22310 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
22311 * tree-ssa-copy.c (dump_copy_of): Likewise.
22312 (copy_prop_visit_cond_stmt): Likewise.
22313 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
22314 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
22315 (record_equivalences_from_stmt): Likewise.
22316 * tree-ssa-dse.c (compute_trims): Likewise.
22317 (delete_dead_call): Likewise.
22318 (delete_dead_assignment): Likewise.
22319 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
22320 (forward_propagate_into_cond): Likewise.
22321 (pass_forwprop::execute): Likewise.
22322 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
22323 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
22325 (move_computations_worker): Likewise.
22326 (execute_sm): Likewise.
22327 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
22328 (remove_exits_and_undefined_stmts): Likewise.
22329 (remove_redundant_iv_tests): Likewise.
22330 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
22331 (adjust_iv_update_pos): Likewise.
22332 * tree-ssa-math-opts.c (bswap_replace): Likewise.
22333 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
22334 (value_replacement): Likewise.
22335 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
22336 * tree-ssa-pre.c (print_pre_expr): Likewise.
22337 (get_representative_for): Likewise.
22338 (create_expression_by_pieces): Likewise.
22339 (insert_into_preds_of_block): Likewise.
22340 (eliminate_insert): Likewise.
22341 (eliminate_dom_walker::before_dom_children): Likewise.
22342 (eliminate): Likewise.
22343 (remove_dead_inserted_code): Likewise.
22344 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
22345 * tree-ssa-reassoc.c (get_rank): Likewise.
22346 (eliminate_duplicate_pair): Likewise.
22347 (eliminate_plus_minus_pair): Likewise.
22348 (eliminate_not_pairs): Likewise.
22349 (undistribute_ops_list): Likewise.
22350 (eliminate_redundant_comparison): Likewise.
22351 (update_range_test): Likewise.
22352 (optimize_range_tests_var_bound): Likewise.
22353 (optimize_vec_cond_expr): Likewise.
22354 (rewrite_expr_tree): Likewise.
22355 (rewrite_expr_tree_parallel): Likewise.
22356 (linearize_expr): Likewise.
22357 (break_up_subtract): Likewise.
22358 (linearize_expr_tree): Likewise.
22359 (attempt_builtin_powi): Likewise.
22360 (attempt_builtin_copysign): Likewise.
22361 (transform_stmt_to_copy): Likewise.
22362 (transform_stmt_to_multiply): Likewise.
22363 (dump_ops_vector): Likewise.
22364 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
22365 (print_scc): Likewise.
22366 (set_ssa_val_to): Likewise.
22367 (visit_reference_op_store): Likewise.
22368 (visit_use): Likewise.
22369 (sccvn_dom_walker::before_dom_children): Likewise.
22370 (run_scc_vn): Likewise.
22371 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
22373 (expr_hash_elt::print): Likewise.
22374 (const_and_copies::pop_to_marker): Likewise.
22375 (const_and_copies::record_const_or_copy_raw): Likewise.
22376 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
22377 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
22378 (dump_predicates): Likewise.
22379 (find_uninit_use): Likewise.
22380 (warn_uninitialized_phi): Likewise.
22381 (pass_late_warn_uninitialized::execute): Likewise.
22382 * tree-ssa.c (verify_vssa): Likewise.
22383 (verify_ssa): Likewise.
22384 (maybe_optimize_var): Likewise.
22385 * tree-vrp.c (dump_value_range): Likewise.
22386 (dump_all_value_ranges): Likewise.
22387 (dump_asserts_for): Likewise.
22388 (register_edge_assert_for_2): Likewise.
22389 (vrp_visit_cond_stmt): Likewise.
22390 (vrp_visit_switch_stmt): Likewise.
22391 (vrp_visit_stmt): Likewise.
22392 (vrp_visit_phi_node): Likewise.
22393 (simplify_cond_using_ranges_1): Likewise.
22394 (fold_predicate_in): Likewise.
22395 (evrp_dom_walker::before_dom_children): Likewise.
22396 (evrp_dom_walker::push_value_range): Likewise.
22397 (evrp_dom_walker::pop_value_range): Likewise.
22398 (execute_early_vrp): Likewise.
22400 2017-05-16 Richard Biener <rguenther@suse.de>
22402 * dwarf2out.c (loc_list_from_tree_1): Do not create
22403 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
22405 2017-05-16 Richard Biener <rguenther@suse.de>
22407 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
22409 (note_variable_value_in_expr): If we resolved the decl ref
22410 do not push to the stack.
22412 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
22414 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
22415 operations in fast-math mode.
22416 (vaddq_f16): Likewise.
22417 (vmul_f16): Likewise.
22418 (vmulq_f16): Likewise.
22419 (vsub_f16): Likewise.
22420 (vsubq_f16): Likewise.
22421 * config/arm/neon.md (add<mode>3): New.
22423 (fma:<VH:mode>3): New. Also remove outdated comment.
22426 2017-05-16 Martin Liska <mliska@suse.cz>
22430 * ipa-devirt.c (warn_types_mismatch): Fix typo.
22431 (odr_types_equivalent_p): Likewise.
22433 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
22435 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
22437 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
22440 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
22441 non-interunit SSE move alternatives with '?'.
22442 (zero-extendsidi peephole2): New peephole to skip intermediate
22443 general register in SSE zero-extend sequence.
22445 2017-05-15 Jeff Law <law@redhat.com>
22447 * reorg.c (relax_delay_slots): Create a new variable to hold
22448 the temporary target rather than clobbering TARGET_LABEL.
22450 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
22451 missing argument to extract_bit_field call.
22452 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
22454 2017-05-15 Martin Liska <mliska@suse.cz>
22457 * gcc.c (process_command): Do not allow empty argument of -o option.
22459 2017-05-15 Renlin Li <renlin.li@arm.com>
22461 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
22462 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
22463 * config/aarch64/constraints.md (Usf): Add long call check.
22464 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
22465 (call_value): Likewise.
22466 (sibcall): Likewise.
22467 (sibcall_value): Likewise.
22469 (call_value_insn): New.
22470 (sibcall_insn): Update rtx pattern.
22471 (sibcall_value_insn): Likewise.
22472 (call_internal): Remove.
22473 (call_value_internal): Likewise.
22474 (sibcall_internal): Likewise.
22475 (sibcall_value_internal): Likewise.
22476 (call_reg): Likewise.
22477 (call_symbol): Likewise.
22478 (call_value_reg): Likewise.
22479 (call_value_symbol): Likewise.
22481 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
22484 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
22486 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
22488 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
22489 compatible with CCGOCmode and with CCZmode.
22491 2017-05-14 Martin Sebor <msebor@redhat.com>
22493 PR middle-end/77671
22494 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
22495 (gimple_fold_builtin_snprintf): Same.
22496 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
22497 (gimple_fold_builtin_snprintf): Same.
22498 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
22499 of character types.
22500 (is_call_safe): New function.
22501 (try_substitute_return_value): Call it.
22502 (try_simplify_call): New function.
22503 (pass_sprintf_length::handle_gimple_call): Call it.
22505 2017-05-14 Martin Sebor <msebor@redhat.com>
22507 PR middle-end/80669
22508 * builtins.c (expand_builtin_stpncpy): Simplify.
22510 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
22512 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
22513 * config/i386/i386.h
22514 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
22515 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
22516 (struct machine_function): Add new members call_ms2sysv,
22517 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
22518 (struct machine_frame_state): New fields sp_realigned and
22519 sp_realigned_offset.
22520 * config/i386/i386.c
22521 (enum xlogue_stub): New enum.
22522 (enum xlogue_stub_sets): New enum.
22523 (class xlogue_layout): New class.
22524 (struct ix86_frame): New fields stack_realign_allocate_offset,
22525 stack_realign_offset and outlined_save_offset. Modify comments to
22526 detail stack layout when using out-of-line stubs.
22527 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
22528 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
22529 -mcall-ms2sysv-xlogues.
22530 (stub_managed_regs): New static variable.
22531 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
22532 registers managed by out-of-line stub.
22533 (disable_call_ms2sysv_xlogues): New function.
22534 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
22535 m->call_ms2sysv when appropriate and compute frame layout for
22537 (sp_valid_at, fp_valid_at): New inline functions.
22538 (choose_basereg): New function.
22539 (choose_baseaddr): Add align parameter, use choose_basereg and modify
22541 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
22542 Use align parameter of choose_baseaddr to generated aligned SSE movs
22544 (pro_epilogue_adjust_stack): Modify to track
22545 machine_frame_state::sp_realigned.
22546 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
22547 (ix86_nsaved_sseregs): Likewise.
22548 (ix86_emit_save_regs): Likewise.
22549 (ix86_emit_save_regs_using_mov): Likewise.
22550 (ix86_emit_save_sse_regs_using_mov): Likewise.
22551 (get_scratch_register_on_entry): Likewise.
22552 (gen_frame_set): New function.
22553 (gen_frame_load): Likewise.
22554 (gen_frame_store): Likewise.
22555 (emit_outlined_ms2sysv_save): Likewise.
22556 (emit_outlined_ms2sysv_restore): Likewise.
22557 (ix86_expand_prologue): Modify stack re-alignment code and call
22558 emit_outlined_ms2sysv_save when appropriate.
22559 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
22560 parameter rtx_insn *insn, which allows the function to be used to only
22561 generate the notes.
22562 (ix86_expand_epilogue): Modify validity checks of frame and stack
22563 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
22564 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
22565 * config/i386/predicates.md
22566 (save_multiple): New predicate.
22567 (restore_multiple): Likewise.
22568 * config/i386/sse.md
22569 (save_multiple<mode>): New pattern.
22570 (save_multiple_realign<mode>): Likewise.
22571 (restore_multiple<mode>): Likewise.
22572 (restore_multiple_and_return<mode>): Likewise.
22573 (restore_multiple_leave_return<mode>): Likewise.
22574 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
22576 2017-05-14 Julia Koval <julia.koval@intel.com>
22578 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
22579 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
22580 (__builtin_ia32_xsetbv): New builtins.
22581 * config/i386/i386.c (ix86_expand_special_args_builtin):
22583 (ix86_expand_builtin): Special expand for new intrinsics.
22584 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
22585 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
22586 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
22588 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22590 * cfganal.c (inverted_post_order_compute): Change argument type
22592 * cfganal.h (inverted_post_order_compute): Adjust prototype.
22593 * df-core.c (rest_of_handle_df_initialize): Adjust.
22594 (rest_of_handle_df_finish): Likewise.
22595 (df_analyze_1): Likewise.
22596 (df_analyze): Likewise.
22597 (loop_inverted_post_order_compute): Change argument to be a vec *.
22598 (df_analyze_loop): Adjust.
22599 (df_get_n_blocks): Likewise.
22600 (df_get_postorder): Likewise.
22601 * df.h (struct df_d): Change field to be a vec.
22602 * lcm.c (compute_laterin): Adjust.
22603 (compute_available): Likewise.
22604 * lra-lives.c (lra_create_live_ranges_1): Likewise.
22605 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
22606 * tree-ssa-pre.c (compute_antic): Likewise.
22608 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22610 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
22611 (depth_first_search::depth_first_search): Change structure init
22612 function to this constructor.
22613 (depth_first_search::add_bb): Rename function to this member.
22614 (depth_first_search::execute): Likewise.
22615 (flow_dfs_compute_reverse_finish): Adjust.
22617 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22619 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
22620 (longest_simple_path): Likewise.
22621 * shrink-wrap.c (spread_components): Likewise.
22622 (disqualify_problematic_components): Likewise.
22623 (emit_common_heads_for_components): Likewise.
22624 (emit_common_tails_for_components): Likewise.
22625 (insert_prologue_epilogue_for_components): Likewise.
22627 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22629 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
22632 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22634 * df-core.c (df_set_blocks): Start using auto_bitmap.
22635 (df_compact_blocks): Likewise.
22636 * df-problems.c (df_rd_confluence_n): Likewise.
22637 * df-scan.c (df_insn_rescan_all): Likewise.
22638 (df_process_deferred_rescans): Likewise.
22639 (df_update_entry_block_defs): Likewise.
22640 (df_update_exit_block_uses): Likewise.
22641 (df_entry_block_bitmap_verify): Likewise.
22642 (df_exit_block_bitmap_verify): Likewise.
22643 (df_scan_verify): Likewise.
22644 * lra-constraints.c (lra_constraints): Likewise.
22645 (undo_optional_reloads): Likewise.
22646 (lra_undo_inheritance): Likewise.
22647 * lra-remat.c (calculate_gen_cands): Likewise.
22648 (do_remat): Likewise.
22649 * lra-spills.c (assign_spill_hard_regs): Likewise.
22650 (spill_pseudos): Likewise.
22651 * tree-ssa-pre.c (bitmap_set_and): Likewise.
22652 (bitmap_set_subtract_values): Likewise.
22654 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22656 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
22657 management with auto_bitmap.
22658 (fix_inter_tick): Likewise.
22659 (fix_recovery_deps): Likewise.
22660 * ira.c (add_store_equivs): Likewise.
22661 (find_moveable_pseudos): Likewise.
22662 (split_live_ranges_for_shrink_wrap): Likewise.
22663 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
22664 (rtx_reuse_manager::seen_def_p): Likewise.
22665 (rtx_reuse_manager::set_seen_def): Likewise.
22666 * print-rtl.h (class rtx_reuse_manager): Likewise.
22668 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22670 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
22672 (migrate_btr_def): Likewise.
22673 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
22674 * df-core.c (loop_post_order_compute): Likewise.
22675 (loop_inverted_post_order_compute): Likewise.
22676 * hsa-common.h: Likewise.
22677 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
22678 * init-regs.c (initialize_uninitialized_regs): Likewise.
22679 * ipa-inline.c (resolve_noninline_speculation): Likewise.
22680 (inline_small_functions): Likewise.
22681 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
22682 * ira.c (combine_and_move_insns): Likewise.
22683 (build_insn_chain): Likewise.
22684 * loop-invariant.c (find_invariants): Likewise.
22685 * lower-subreg.c (propagate_pseudo_copies): Likewise.
22686 * predict.c (tree_predict_by_opcode): Likewise.
22687 (predict_paths_leading_to): Likewise.
22688 (predict_paths_leading_to_edge): Likewise.
22689 (estimate_loops_at_level): Likewise.
22690 (estimate_loops): Likewise.
22691 * shrink-wrap.c (try_shrink_wrapping): Likewise.
22692 (spread_components): Likewise.
22693 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
22694 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
22695 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
22696 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
22697 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
22698 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
22699 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
22700 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
22701 (mark_threaded_blocks): Likewise.
22702 (thread_through_all_blocks): Likewise.
22703 * tree-ssa.c (verify_ssa): Likewise.
22704 (execute_update_addresses_taken): Likewise.
22705 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
22707 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22709 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
22711 (post_order_compute): Likewise.
22712 (inverted_post_order_compute): Likewise.
22713 (pre_and_rev_post_order_compute_fn): Likewise.
22715 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22717 * genrecog.c (int_set::int_set): Explicitly construct our
22718 auto_vec base class.
22719 * vec.h (auto_vec::auto_vec): New constructor.
22721 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22723 * bitmap.h (class auto_bitmap): New constructor taking
22724 bitmap_obstack * argument.
22726 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22728 * bitmap.h (class auto_bitmap): Change type of m_bits to
22729 bitmap_head, and adjust ctor / dtor and member operators.
22731 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
22733 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
22734 when returned register mode doesn't match original mode.
22736 2017-05-12 Jeff Law <law@redhat.com>
22737 Jakub Jelinek <jakub@redhat.com>
22739 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
22740 we look for cc setter after the compare-elim changes.
22741 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
22742 within the vector to match what compare-elim now expects.
22743 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
22744 (xorsi3_flags, one_cmplsi2_flags): Likewise.
22746 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
22747 after the compare-elim changes.
22748 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
22749 the vector to match what compare-elim now expects.
22750 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
22751 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
22752 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
22753 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
22754 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
22756 * config/visium/visium.c (single_set_and_flags): Fix where
22757 we look for cc setter after the compare-elim changes.
22758 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
22759 with the vector to match what compare-elim now expects.
22760 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
22761 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
22762 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
22763 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
22764 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
22765 (neg<mode>2_insn_set_overflow): Likewise.
22767 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
22769 PR middle-end/79794
22770 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
22771 maybe_expand_insn call, set ops[0].target. If still set after call,
22772 set alt_rtl. Add extra arg to recursive calls.
22773 (extract_bit_field): Add alt_rtl argument. Pass to
22775 * expmed.h (extract_bit_field): Fix prototype.
22776 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
22777 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
22778 to extract_bit_field_calls.
22779 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
22780 Pass alt_rtl to extract_bit_field calls.
22781 * calls.c (store_unaligned_arguments_into_psuedos)
22782 load_register_parameters): Pass extra NULL to extract_bit_field calls.
22783 * optabs.c (maybe_legitimize_operand): Clear op->target when call
22785 * optabs.h (struct expand_operand): Add target bitfield.
22787 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
22789 * compare-elim.c (try_eliminate_compare): Canonicalize
22790 operation with embedded compare to
22791 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
22792 (set (reg) (operation)].
22794 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
22796 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
22799 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
22800 cost of adding a carry flag for ADC instruction.
22801 [case MINUS]: Ignore the cost of subtracting a carry flag
22802 for SBB instruction.
22804 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
22806 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
22808 * config/rs6000/bmiintrin.h: New file.
22809 * config/rs6000/bmi2intrin.h: New file.
22810 * config/rs6000/x86intrin.h: New file.
22812 2017-05-12 Jeff Law <law@redhat.com>
22814 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
22817 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
22819 PR middle-end/80707
22820 * tree-cfg.c: Remove cfg edges of unreachable case statements.
22822 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
22824 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
22825 early expansion of vector divide builtins.
22826 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
22827 builtins identified as having unsigned arguments.
22829 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
22831 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
22832 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
22833 expansion of vector logical operations (and, andc, or, xor,
22836 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
22838 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
22839 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
22841 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
22843 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
22844 early GIMPLE expansion of vector multiplies.
22846 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22848 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
22849 TARGET_HAVE_MOVT conditional.
22850 (movt splitter): Likewise.
22852 2017-05-12 Richard Biener <rguenther@suse.de>
22854 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
22855 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
22856 Fold all stmts not inplace.
22858 2017-05-12 Richard Biener <rguenther@suse.de>
22860 PR tree-optimization/80713
22861 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
22862 inserted_exprs bit for not removed stmts.
22864 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
22866 PR middle-end/69921
22867 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
22868 parallelized" attribute for parallelized OpenACC kernels.
22869 * omp-offload.c (execute_oacc_device_lower): Use it.
22871 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
22872 Set "oacc kernels" attribute.
22873 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
22874 parameter. Adjust all users.
22875 (oacc_fn_attrib_kernels_p): Remove function.
22876 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
22877 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
22878 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
22879 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
22880 assert "oacc kernels" attribute is set.
22882 2017-05-11 Carl Love <cel@us.ibm.com>
22884 * config/rs6000/rs6000-c: Add support for built-in functions
22885 vector unsigned char vec_popcnt (vector signed char)
22886 vector unsigned char vec_popcnt (vector unsigned char)
22887 vector unsigned short vec_popcnt (vector signed short)
22888 vector unsigned short vec_popcnt (vector unsigned short)
22889 vector unsigned int vec_popcnt (vector signed int)
22890 vector unsigned int vec_popcnt (vector unsigned int)
22891 vector unsigned long long vec_popcnt (vector signed long long)
22892 vector unsigned long long vec_popcnt (vector unsigned long long)
22893 vector signed long long vec_slo (vector signed long long,
22894 vector signed char)
22895 vector signed long long vec_slo (vector signed long long,
22896 vector unsigned char)
22897 vector unsigned long long vec_slo (vector unsigned long long,
22898 vector signed char)
22899 vector unsigned long long vec_slo (vector unsigned long long,
22900 vector unsigned char)
22901 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
22902 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
22903 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
22904 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
22905 * doc/extend.texi: Update the built-in documentation file for the
22906 new built-in functions.
22908 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
22910 * attribs.h (sorted_attr_string): Move machine independent
22911 functions for target clone support from the i386 port to common
22912 code. Rename ix86_function_versions to common_function_versions.
22913 Rename make_name to make_unique_name.
22914 (common_function_versions): Likewise.
22915 (make_unique_name): Likewise.
22916 (make_dispatcher_decl): Likewise.
22917 (is_function_default_version): Likewise.
22918 * attribs.c (attr_strcmp): Likewise.
22919 (sorted_attr_string): Likewise.
22920 (common_function_versions): Likewise.
22921 (make_unique_name): Likewise.
22922 (make_dispatcher_decl): Likewise.
22923 (is_function_default_version): Likewise.
22924 * config/i386/i386.c (attr_strcmp): Likewise.
22925 (sorted_attr_string): Likewise.
22926 (ix86_function_versions): Likewise.
22927 (make_name): Likewise.
22928 (make_dispatcher_decl): Likewise.
22929 (is_function_default_version): Likewise.
22930 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
22932 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22935 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
22936 Account for direct move costs for vec_construct of integer
22939 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
22942 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
22943 (UNSPEC_STX_ATOMIC): Ditto.
22944 (loaddi_via_sse): New insn.
22945 (storedi_via_sse): Ditto.
22946 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
22947 Update corresponding peephole2 patterns.
22948 (atomic_storedi_fpu): Ditto.
22950 2017-05-11 Julia Koval <julia.koval@intel.com>
22952 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
22953 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
22955 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
22956 (__builtin_ia32_rsqrt14ss_mask): New builtins.
22957 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
22959 2017-05-11 Nathan Sidwell <nathan@acm.org>
22961 * graphite-poly.c: Include dumpfile.h.
22963 * dumpfle.h (dump_function): Declare here ...
22964 * tree-dump.h (dump_function): ... not here.
22965 * dumpfile.c: #include tree-cfg.h.
22966 (dump_function): Move here from ...
22967 * tree-dump.c (dump_function): ... here.
22968 * gimplify.c: #include splay-tree.h, not tree-dump.h.
22969 * graphite-poly.c: Don't include tree-dump.h.
22970 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
22971 * print-tree.c: Likewise.
22972 * stor-layout.c: Likewise.
22973 * tree-nested.c: Likewise.
22975 * dumpfile.c (dump_start): Use TDF_FLAGS.
22976 (dump_enable_all): Fix TDF_KIND check thinko.
22978 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
22980 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22981 array entries to represent two legal parameterizations of the
22982 overloaded __builtin_cmpb function, as represented by the
22983 P6_OV_BUILTIN_CMPB constant.
22984 (altivec_resolve_overloaded_builtin): Add special case handling
22985 for the __builtin_cmpb function, as represented by the
22986 P6_OV_BUILTIN_CMPB constant.
22987 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
22988 (BU_P6_64BIT_2): New macro.
22989 (BU_P6_OVERLOAD_2): New macro
22990 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
22991 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
22992 (CMPB): Add overload support to represent both 32-bit and 64-bit
22993 compare-bytes function.
22994 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
22995 support for TARGET_CMPB.
22996 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
22997 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
22998 documentation of the __builtin_cmpb overloaded built-in function.
23000 2017-05-11 Richard Biener <rguenther@suse.de>
23002 PR tree-optimization/80705
23003 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
23004 bases are not vectorizable.
23006 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23008 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
23009 when counting register pressure.
23011 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23013 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
23014 (iv_ca_more_deps): Renamed to ...
23015 (iv_ca_compare_deps): ... this.
23016 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
23018 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23020 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
23022 (determine_group_iv_costs): ... here.
23023 (find_inv_vars_cb): Record inv var if it's not recorded before.
23025 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23027 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
23028 (get_shiftadd_cost): Ditto.
23030 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23032 * tree-ssa-address.c: Include header file.
23033 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
23035 (add_to_parts): Refactor.
23036 (addr_to_parts): New parameter. Update use of move_hint_to_base.
23037 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
23040 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23042 PR tree-optimization/53090
23043 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
23045 (extract_cond_operands): Detect condition with IV on both sides
23046 and return COMP_IV_EXPR_2.
23047 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
23048 (rewrite_use_compare): Simplify by removing call to function
23049 extract_cond_operands.
23051 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23053 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
23054 (extract_cond_operands): Detect condition comparing against non-
23055 invariant bound and return appropriate enum value.
23056 (find_interesting_uses_cond): Update use of extract_cond_operands.
23057 Handle its return value accordingly.
23058 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
23060 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23062 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
23063 nonlinear iv_use computation in loop invariant sensitive way.
23065 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23067 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
23068 (find_iv_candidates): Call relate_compare_use_with_all_cands.
23070 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23072 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
23073 (dump_cand): Support iv_cand.inv_exprs.
23074 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
23076 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
23079 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23081 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
23083 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
23084 as local function. Include necessary header files.
23085 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
23087 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23089 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
23091 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23093 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
23094 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
23095 RSHIFT_EXPR and BIT_NOT_EXPR.
23097 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23099 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
23100 (adjust_setup_cost): New parameter supporting round up adjustment.
23101 (struct address_cost_data): Delete.
23102 (force_expr_to_var_cost): Don't bound cost with spill_cost.
23103 (split_address_cost, ptr_difference_cost): Delete.
23104 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
23105 (struct ainc_cost_data): New struct.
23106 (get_address_cost_ainc): New function.
23107 (get_address_cost, get_computation_cost): Reimplement.
23108 (determine_group_iv_cost_address): Record inv_expr for all uses of
23110 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
23111 (iv_ca_has_deps): Reimplemented to ...
23112 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
23114 (iv_ca_extend): Call iv_ca_more_deps.
23116 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23118 * tree-ssa-address.c (struct mem_address): Move to header file.
23119 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
23120 * tree-ssa-address.h (struct mem_address): Move from C file.
23121 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
23123 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23125 * tree-affine.h (aff_combination_type): New interface.
23126 (aff_combination_zero_p): Remove static.
23127 (aff_combination_const_p): New interface.
23128 (aff_combination_singleton_var_p): New interfaces.
23130 2017-05-11 Richard Biener <rguenther@suse.de>
23132 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23133 Skip unreachable blocks and destinations.
23134 (eliminate): Move stmt removal and fixup ...
23135 (fini_eliminate): ... here. Skip inserted exprs.
23136 (pass_pre::execute): Move fini_pre after fini_eliminate.
23137 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
23138 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
23139 PRE to get rid of dead code that has invalid SSA form and
23140 split critical edges again.
23142 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23144 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
23146 2017-05-11 Richard Biener <rguenther@suse.de>
23148 * passes.c (execute_function_todo): Verify loops if they are
23149 said to be up-to-date.
23150 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
23151 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
23153 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
23156 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
23157 handle calling assemble_external ourself.
23160 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
23161 modes with zero size. Enhance comment.
23163 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23165 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
23166 built-ins for vec_xl and vec_xst with short and char pointer
23169 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
23171 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
23172 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
23173 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
23174 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
23175 (_mm_maskz_min_round_ss): New intrinsics.
23176 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
23177 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
23178 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
23179 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
23180 (__builtin_ia32_minss_mask_round): New builtins.
23181 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23182 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23183 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
23185 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
23186 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
23188 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
23191 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
23193 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
23194 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
23195 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
23196 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
23197 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
23198 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
23199 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
23200 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23201 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
23202 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
23203 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
23204 (__builtin_ia32_mulss_mask_round): New builtins.
23205 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23206 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23207 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
23209 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
23210 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
23212 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
23215 2017-05-10 Julia Koval <julia.koval@intel.com>
23217 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
23218 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
23219 (_mm256_setr_m128i): New intrinsics.
23221 2017-05-10 Julia Koval <julia.koval@intel.com>
23223 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
23224 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
23225 (_mm_maskz_rcp14_ss): New intrinsics.
23226 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
23227 (__builtin_ia32_rcp14ss_mask): New builtins.
23228 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
23230 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
23232 PR tree-optimization/51513
23233 * tree-cfg.c (gimple_seq_unreachable_p): New function.
23234 (assert_unreachable_fallthru_edge_p): Use it.
23235 (group_case_labels_stmt): Likewise.
23236 * tree-cfg.h: Prototype it.
23237 * stmt.c: Include cfghooks.h and tree-cfg.h.
23238 (emit_case_dispatch_table) <gap_label>: New local variable.
23239 Use it to fill dispatch table gaps.
23240 Test for default_label before updating probabilities.
23241 (expand_case) <default_label>: Remove unneeded initialization.
23242 Test for unreachable default case statement and remove its edge.
23243 Set default_label accordingly.
23244 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
23246 2017-05-10 Carl Love <cel@us.ibm.com>
23248 * config/rs6000/rs6000-c: Add support for built-in functions
23249 vector signed char vec_neg (vector signed char)
23250 vector signed short int vec_neg (vector short int)
23251 vector signed int vec_neg (vector signed int)
23252 vector signed long long vec_neg (vector signed long long)
23253 vector float vec_neg (vector float)
23254 vector double vec_neg (vector double)
23255 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
23257 * config/rs6000/altivec.h: Add define for vec_neg
23258 * doc/extend.texi: Update the built-in documentation for the
23259 new built-in functions.
23261 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23263 PR tree-optimization/77644
23264 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
23266 2017-05-10 Nathan Sidwell <nathan@acm.org>
23268 * dumpfile.h (TDI_lang_all): New.
23269 (TDF_KIND): New. Renumber others
23270 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
23272 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
23274 (get_dump_file_name): Adjust suffix generation.
23275 (dump_enable_all): Use TDF_KIND.
23276 * doc/invoke.texi (-fdump-lang-all): Document.
23278 * dumpfile.h: Tabify.
23280 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
23283 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
23284 Move member access before delete.
23286 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
23288 * tree-inline.c (expand_call_inline): Split block at stmt
23291 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
23294 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
23295 are now unused after splitting mov{sf,sd}_hardfloat.
23296 (f32_lr2): Likewise.
23297 (f32_lm): Likewise.
23298 (f32_lm2): Likewise.
23299 (f32_li): Likewise.
23300 (f32_li2): Likewise.
23301 (f32_lv): Likewise.
23302 (f32_sr): Likewise.
23303 (f32_sr2): Likewise.
23304 (f32_sm): Likewise.
23305 (f32_sm2): Likewise.
23306 (f32_si): Likewise.
23307 (f32_si2): Likewise.
23308 (f32_sv): Likewise.
23309 (f32_dm): Likewise.
23310 (f32_vsx): Likewise.
23311 (f32_av): Likewise.
23312 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
23313 For movsf, order stores so the VSX stores occur before the GPR
23314 store which encourages the register allocator to use a traditional
23315 FPR instead of a GPR. For movsd, order the stores so that the GPR
23316 store comes before the VSX stores to allow the power6 to work.
23317 This is due to the power6 not having a 32-bit integer store
23318 instruction from a FPR.
23319 (movsf_hardfloat): Likewise.
23320 (movsd_hardfloat): Likewise.
23322 2017-05-09 Martin Sebor <msebor@redhat.com>
23324 PR translation/80280
23325 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
23328 PR translation/80280
23329 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
23330 data member added in r247778.
23331 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
23333 2017-05-09 Nathan Sidwell <nathan@acm.org>
23335 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
23337 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
23340 2017-05-09 Marek Polacek <polacek@redhat.com>
23342 * doc/invoke.texi: Fix typo.
23344 2017-05-09 Richard Biener <rguenther@suse.de>
23346 * tree-vrp.c (vrp_val_is_max): Adjust comment.
23347 (vrp_val_is_min): Likewise.
23348 (set_value_range_to_value): Likewise.
23349 (set_value_range_to_nonnegative): Likewise.
23350 (gimple_assign_nonzero_p): Likewise.
23351 (gimple_stmt_nonzero_p): Likewise.
23352 (vrp_int_const_binop): Likewise. Remove unreachable case.
23353 (adjust_range_with_scev): Adjust comments.
23354 (compare_range_with_value): Likewise.
23355 (extract_range_from_phi_node): Likewise.
23356 (test_for_singularity): Likewise.
23358 2017-05-09 Richard Biener <rguenther@suse.de>
23360 * tree-vrp.c (get_single_symbol): Add assert that we don't
23361 get overflowed constants as invariant part.
23362 (compare_values_warnv): Add comment before the TREE_NO_WARNING
23363 checks. Use wi::cmp instead of recursing for integer constants.
23364 (compare_values): Just ignore whether we assumed undefined
23365 overflow instead of failing the compare.
23366 (extract_range_for_var_from_comparison_expr): Add comment before the
23367 TREE_NO_WARNING sets.
23368 (test_for_singularity): Likewise.
23369 (extract_range_from_comparison): Do not disable optimization
23370 when we assumed undefined overflow.
23371 (extract_range_basic): Remove init of unused var.
23373 2017-05-09 Richard Biener <rguenther@suse.de>
23375 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
23376 (extract_range_from_multiplicative_op_1): Adjust.
23377 (extract_range_from_binary_expr_1): Use int_const_binop.
23379 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
23382 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
23383 rs6000_store_data_bypass_p in seven define_bypass directives and
23384 in several comments.
23385 * config/rs6000/rs6000-protos.h: Add prototype for
23386 rs6000_store_data_bypass_p function.
23387 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
23388 function implements slightly different (rs6000-specific) semantics
23389 than store_data_bypass_p, returning false rather than aborting
23390 with assertion error when arguments do not satisfy the
23391 requirements of store data bypass.
23392 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
23393 rs6000_store_data_bypass_p.
23395 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
23397 * config/xtensa/xtensa-protos.h
23398 (xtensa_initial_elimination_offset): New declaration.
23399 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
23400 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
23401 macro definition, add case for FRAME_POINTER_REGNUM when
23402 FRAME_GROWS_DOWNWARD.
23403 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
23404 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
23405 xtensa_initial_elimination_offset.
23407 2017-05-08 Nathan Sidwell <nathan@acm.org>
23409 * doc/invoke.texi: Alphabetize -fdump options.
23411 2017-05-08 Martin Sebor <msebor@redhat.com>
23413 PR translation/80280
23414 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
23416 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
23418 * target.def (compute_frame_layout): New optional target hook.
23419 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
23420 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
23421 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
23423 * reload1.c (verify_initial_elim_offsets): Likewise.
23424 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
23425 (use_simple_return_p): Call arm_compute_frame_layout if needed.
23426 (arm_get_frame_offsets): Split up into this ...
23427 (arm_compute_frame_layout): ... and this function.
23429 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
23431 * config/aarch64/constraints.md (Usa): New constraint.
23432 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
23434 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
23436 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
23437 with_multilib_list after it has been checked.
23439 2017-05-08 Richard Biener <rguenther@suse.de>
23441 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
23442 (bitmap_set_subtract_values): Likewise.
23444 2017-05-08 Richard Biener <rguenther@suse.de>
23446 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
23447 (gimple_assign_nonzero): ... this and remove strict_overflow_p
23449 (gimple_stmt_nonzero_warnv_p): Rename to ...
23450 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
23452 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
23453 (extract_range_basic): Adjust, do not disable propagation on
23454 strict overflow sensitive simplification.
23455 (vrp_visit_cond_stmt): Likewise.
23457 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
23459 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
23460 body size unconditionally.
23462 2017-05-07 Jeff Law <law@redhat.com>
23465 2017-05-06 Jeff Law <law@redhat.com>
23466 PR tree-optimization/78496
23467 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
23470 PR tree-optimization/78496
23471 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
23472 (simplify_stmt_using_ranges): Call it.
23473 (vrp_dom_walker::before_dom_children): Extract equivalences
23474 from an ASSERT_EXPR with an equality comparison against a
23477 2017-05-06 Jeff Law <law@redhat.com>
23479 PR tree-optimization/78496
23480 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
23483 PR tree-optimization/78496
23484 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
23485 (simplify_stmt_using_ranges): Call it.
23486 (vrp_dom_walker::before_dom_children): Extract equivalences
23487 from an ASSERT_EXPR with an equality comparison against a
23490 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
23492 * lra-constraints.c (lra_copy_reg_equiv): New function.
23493 (split_reg): Use it to copy equivalence information from the
23494 original register to the spill register.
23496 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
23498 PR rtl-optimization/75964
23499 * simplify-rtx.c (simplify_const_relational_operation): Remove
23500 invalid handling of comparisons of integer ABS.
23502 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
23504 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
23505 initialize to zero.
23506 (init_regs): Remove declaration.
23507 (function_arg_advance_32): Initialize error_p as boolean variable.
23509 2017-05-05 Nathan Sidwell <nathan@acm.org>
23511 * store-motion.c (remove_reachable_equiv_notes): Reformat long
23512 lines. Use for (;;).
23514 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23516 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
23517 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
23518 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
23519 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
23520 VF=2 that require versioning.
23522 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23524 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
23527 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23529 * diagnostic.h (diagnostic_override_option_index): Convert from
23530 macro to inline function.
23532 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23534 * diagnostic.c (last_module_changed_p): New function.
23535 (set_last_module): New function.
23536 (diagnostic_report_current_module): Convert macro usage to
23537 the above functions.
23538 * diagnostic.h (diagnostic_context::last_module): Strengthen
23539 from const line_map * to const line_map_ordinary *.
23540 (diagnostic_last_module_changed): Delete macro.
23541 (diagnostic_set_last_module): Delete macro.
23543 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23545 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
23546 with diagnostic_report_diagnostic.
23547 (diagnostic_n_impl_richloc): Likewise.
23548 * diagnostic.h (report_diagnostic): Delete macro.
23549 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
23550 with diagnostic_report_diagnostic.
23551 * substring-locations.c (format_warning_va): Likewise.
23553 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23555 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
23556 save/restor of format_spec. Move option-printing code to...
23557 (print_option_information): ...this new function, and
23558 reimplement by simply printing to the pretty_printer,
23559 rather than appending to the format string.
23561 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23563 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
23564 handling logic into...
23565 (update_effective_level_from_pragmas): ...this new function.
23567 2017-05-04 Andrew Waterman <andrew@sifive.com>
23569 * config/riscv/riscv.opt (mstrict-align): New option.
23570 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
23571 (SLOW_UNALIGNED_ACCESS): Define.
23572 (riscv_slow_unaligned_access): Declare.
23573 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
23575 (riscv_slow_unaligned_access): New variable.
23576 (rocket_tune_info): Set slow_unaligned_access to true.
23577 (optimize_size_tune_info): Set slow_unaligned_access to false.
23578 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
23579 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
23580 (riscv_option_override): Set riscv_slow_unaligned_access.
23581 * doc/invoke.texi: Add -mstrict-align to RISC-V.
23583 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
23585 * config/riscv/riscv.md: Unify indentation.
23587 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
23592 * config/rs6000/rs6000.md (u code attribute): Add FIX and
23594 (extendsi<mode>2): Add support for doing sign extension via
23595 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
23596 don't have ISA 3.0 instructions.
23597 (extendsi<mode>2 splitter): Likewise.
23598 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
23599 generate the normal insns since SImode can now go in vector
23600 registers. Disallow the special UNSPECs needed for previous
23601 machines to hide SImode being used. Add new insns
23602 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
23603 (fix_trunc<mode>si2_stfiwx): Likewise.
23604 (fix_trunc<mode>si2_internal): Likewise.
23605 (fixuns_trunc<mode>si2): Likewise.
23606 (fixuns_trunc<mode>si2_stfiwx): Likewise.
23607 (fctiw<u>z_<mode>_smallint): Likewise.
23608 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
23609 of floating point to 32-bit integer from doing a direct move to
23610 the GPR registers to do a store.
23611 (fctiwz_<mode>): Break long line.
23613 2017-05-05 Bin Cheng <bin.cheng@arm.com>
23615 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
23616 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
23617 (addr_list, addr_offset_valid_p): New.
23618 (split_address_groups): Check offset validity with above function.
23619 (gt-tree-ssa-loop-ivopts.h): Include header file.
23621 2017-05-05 Nathan Sidwell <nathan@acm.org>
23623 * config.gcc (arm*-*-*): Add missing 'fi'.
23625 2017-05-05 Steve Ellcey <sellcey@cavium.com>
23627 * doc/invoke.texi (-fopt-info): Explicitly say order of options
23628 included in -fopt-info does not matter.
23629 * doc/optinfo.texi (-fopt-info): Fix description of default
23630 behavour. Explicitly say order of options included in -fopt-info
23633 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
23635 * config.gcc: Allow combinations of aprofile and rmprofile values for
23636 --with-multilib-list.
23637 * config/arm/t-multilib: New file.
23638 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
23639 variables. Remove setting of ISA and floating-point ABI in
23640 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
23641 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
23642 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
23643 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
23645 * config/arm/t-rmprofile: Likewise except for the matches changes.
23646 * doc/install.texi (--with-multilib-list): Document the combination of
23647 aprofile and rmprofile values and warn about pitfalls in doing that.
23649 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
23651 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
23652 (movdi_aarch64): Likewise.
23654 2017-05-05 Jakub Jelinek <jakub@redhat.com>
23656 PR tree-optimization/80632
23657 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
23659 (build_arrays): Initialize it for virtual phis.
23660 (fix_phi_nodes): Use it for virtual phis.
23662 PR tree-optimization/80558
23663 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
23664 [x, y] op z into [x op, y op z] for op & or | if conditions
23667 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
23668 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23671 * config/arm/arm.md (use_literal_pool): Remove.
23672 (64-bit immediate split): No longer takes cost into consideration
23673 if arm_disable_literal_pool is enabled.
23674 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
23675 used when arm_disable_literal_pool is enabled.
23676 (arm_max_const_double_inline_cost): Remove use of
23677 arm_disable_literal_pool.
23678 (push_minipool_fix): Add assert.
23679 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
23680 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
23681 (no_literal_pool_sf_immediate): New.
23683 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23685 PR tree-optimization/80613
23686 * tree-ssa-dce.c (propagate_necessity): Remove cases for
23687 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
23689 2017-05-05 Richard Biener <rguenther@suse.de>
23691 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
23693 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
23695 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
23696 of this flag from insn conditions due to removal from r247495.
23698 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
23700 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
23702 (arm_early_store_addr_dep_ptr): Likewise.
23703 * config/arm/aarch-common-protos.h
23704 (arm_early_load_addr_dep_ptr): Add prototype.
23705 (arm_early_store_addr_dep_ptr): Likewise.
23706 * config/arm/cortex-a53.md: Add new bypasses.
23708 2017-05-05 Jakub Jelinek <jakub@redhat.com>
23710 * tree.c (next_type_uid): Change type to unsigned.
23711 (type_hash_canon): Decrement back next_type_uid if
23712 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
23713 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
23716 2017-05-04 Martin Sebor <msebor@redhat.com>
23718 * builtins.c: Fix a trivial typo in a comment.
23720 PR middle-end/79234
23721 * builtins.c (check_sizes): Adjust to handle reading past the end.
23722 Avoid printing excessive upper bound of ranges. Use %E to print
23723 tree nodes instead of converting them to %wu.
23724 (expand_builtin_memchr): New function.
23725 (compute_dest_size): Rename...
23726 (compute_objsize): ...to this.
23727 (expand_builtin_memcpy): Adjust.
23728 (expand_builtin_mempcpy): Adjust.
23729 (expand_builtin_strcat): Adjust.
23730 (expand_builtin_strcpy): Adjust.
23731 (check_strncat_sizes): Adjust.
23732 (expand_builtin_strncat): Adjust.
23733 (expand_builtin_strncpy): Adjust and simplify.
23734 (expand_builtin_memset): Adjust.
23735 (expand_builtin_bzero): Adjust.
23736 (expand_builtin_memcmp): Adjust.
23737 (expand_builtin): Handle memcmp.
23738 (maybe_emit_chk_warning): Check strncat just once.
23740 2017-05-04 Martin Sebor <msebor@redhat.com>
23742 PR preprocessor/79214
23743 PR middle-end/79222
23744 PR middle-end/79223
23745 * builtins.c (check_sizes): Add inlining context and issue
23746 warnings even when -Wno-system-headers is set.
23747 (check_strncat_sizes): Same.
23748 (expand_builtin_strncat): Same.
23749 (expand_builtin_memmove): New function.
23750 (expand_builtin_stpncpy): Same.
23751 (expand_builtin): Handle memmove and stpncpy.
23753 2017-05-04 Bin Cheng <bin.cheng@arm.com>
23755 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
23756 which is not used any more.
23758 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
23760 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
23762 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
23764 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
23765 (cortexa53_tunings): Likewise.
23766 (cortexa57_tunings): Likewise.
23767 (cortexa72_tunings): Likewise.
23768 (cortexa73_tunings): Likewise.
23770 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
23772 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
23773 Set loop alignment to 8.
23775 2017-05-04 Martin Sebor <msebor@redhat.com>
23777 PR translation/80280
23778 * builtins.c (expand_builtin_object_size): Add missing quoting to
23779 %D and like directives.
23780 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
23781 (hsa_type_for_tree_type): Same.
23782 (verify_function_arguments): Same.
23783 * symtab.c (symbol_table::change_decl_assembler_name): Same.
23784 * varasm.c (get_section): Same.
23787 2017-05-04 Martin Sebor <msebor@redhat.com>
23789 PR translation/80280
23790 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
23792 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
23794 * config/aarch64/aarch64.c (generic_addrcost_table):
23795 Change HI/TI mode setting.
23797 2017-05-04 Martin Jambor <mjambor@suse.cz>
23799 PR tree-optimization/80622
23800 * tree-sra.c (comes_initialized_p): New function.
23801 (build_accesses_from_assign): Only set write lazily when
23802 comes_initialized_p is false.
23803 (analyze_access_subtree): Use comes_initialized_p.
23804 (propagate_subaccesses_across_link): Assert !comes_initialized_p
23805 instead of testing for PARM_DECL.
23807 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23809 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
23810 constraint on operand 0 to allow more general addressing modes.
23811 Adjust output template.
23812 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
23814 * config/aarch64/aarch64-protos.h
23815 (aarch64_address_valid_for_prefetch_p): Declare prototype.
23816 * config/aarch64/constraints.md (Dp): New address constraint.
23817 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
23820 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
23822 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
23823 update use of estimate_ipcp_clone_size_and_time.
23824 (estimate_local_effects): Update use of
23825 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
23826 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
23827 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
23828 Return nonspecialized time.
23830 2017-05-04 Richard Biener <rguenther@suse.de>
23832 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
23833 for the last VUSE which def dominates the PHI. Directly call
23835 (get_continuation_for_phi_1): Remove.
23837 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
23839 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
23840 to explain the use of truncating division. Cap the number of
23841 iterations to the maximum given by nb_iterations_upper_bound,
23844 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
23846 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
23847 * configure: Regenerate.
23848 * config.in: Regenerate.
23849 * config/i386/driver-mingw32.c: new file.
23850 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
23851 * config.host: Link driver-mingw32.o on MinGW host.
23852 * doc/install.texi: Document new --enable-mingw-wildcard configure
23855 2017-05-04 Marek Polacek <polacek@redhat.com>
23857 PR tree-optimization/80612
23858 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
23860 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23861 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
23863 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
23864 (movt splitter): Likewise.
23865 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
23866 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
23867 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
23868 block for Thumb-1 with MOVT.
23869 (thumb2_legitimate_address_p): Move code block ...
23870 (can_avoid_literal_pool_for_label_p): ... into this new function.
23871 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
23873 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
23874 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
23875 "M-profile targets with the MOVT instruction".
23877 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23879 * config/arm/arm-builtins.c (arm_init_builtins): Rename
23880 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
23881 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
23883 2017-05-04 Martin Liska <mliska@suse.cz>
23885 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
23886 variable cond_code.
23888 2017-05-04 Richard Biener <rguenther@suse.de>
23890 * tree.c (array_at_struct_end_p): Handle arrays at struct
23891 end with flexarrays more conservatively. Refactor and treat
23892 arrays of arrays or aggregates more strict. Fix
23893 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
23894 * tree.c (array_at_struct_end_p): Adjust prototype.
23895 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
23896 * gimple-fold.c (get_range_strlen): Likewise.
23897 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
23899 2017-05-04 Richard Biener <rguenther@suse.de>
23901 PR tree-optimization/31130
23902 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
23904 (supports_overflow_infinity): Likewise.
23905 (is_negative_overflow_infinity): Likewise.
23906 (is_positive_overflow_infinity): Likewise.
23907 (is_overflow_infinity): Likewise.
23908 (stmt_overflow_infinity): Likewise.
23909 (overflow_infinity_range_p): Likewise.
23910 (usable_range_p): Remove as always returning true.
23911 (make_overflow_infinity): Remove.
23912 (negative_overflow_infinity): Likewise.
23913 (positive_overflow_infinity): Likewise.
23914 (avoid_overflow_infinity): Likewise.
23915 (set_value_range): Adjust accordingly.
23916 (set_value_range_to_nonnegative): Likewise, remove now unused
23917 overflow_infinity arg.
23918 (vrp_operand_equal_p): Adjust.
23919 (update_value_range): Likewise.
23920 (range_int_cst_singleton_p): Likewise.
23921 (operand_less_p): Likewise.
23922 (compare_values_warnv): Likewise.
23923 (extract_range_for_var_from_comparison_expr): Likewise.
23924 (vrp_int_const_binop): Likewise.
23925 (zero_nonzero_bits_from_vr): Likewise.
23926 (extract_range_from_multiplicative_op_1): Likewise.
23927 (extract_range_from_binary_expr_1): Likewise.
23928 (extract_range_from_unary_expr): Likewise.
23929 (extract_range_from_comparison): Likewise.
23930 (extract_range_basic): Likewise.
23931 (adjust_range_with_scev): Likewise.
23932 (compare_ranges): Likewise.
23933 (compare_range_with_value): Likewise.
23934 (dump_value_range): Likewise.
23935 (test_for_singularity): Likewise, remove strict_overflow_p parameter
23937 (simplify_cond_using_ranges): Adjust.
23939 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
23941 * brig-builtins.def: Added a builtin for class_f64.
23942 * builtin-types.def: Added a builtin type needed by class_f64.
23944 2017-05-03 Jason Merrill <jason@redhat.com>
23946 * timevar.def: Add TV_CONSTEXPR.
23948 2017-05-03 David Malcolm <dmalcolm@redhat.com>
23950 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
23952 2017-05-03 Martin Jambor <mjambor@suse.cz>
23954 * ipa-prop.c (ipa_update_after_lto_read): Removed.
23955 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
23956 * ipa-cp.c (ipcp_propagate_stage): Do not call
23957 ipa_update_after_lto_read.
23958 * ipa-inline.c (ipa_inline): Likewise.
23960 2017-05-03 Martin Jambor <mjambor@suse.cz>
23962 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
23963 tag. Added a default constructor and a destructor.
23964 (ipa_edge_args_sum_t): New class;
23965 (ipa_edge_args_sum): Declare.
23966 (ipa_edge_args_vector): Remove declaration.
23967 (IPA_EDGE_REF): Use ipa_edge_args_sum.
23968 (ipa_free_edge_args_substructures): Remove declaration.
23969 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
23970 (ipa_edge_args_info_available_for_edge_p): Likewise.
23971 * ipa-prop.c (ipa_edge_args_vector): Removed.
23972 (edge_removal_hook_holder): Likewise.
23973 (edge_duplication_hook_holder): Likewise.
23974 (ipa_edge_args_sum): New variable.
23975 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
23976 ipa_edge_args_vector.
23977 (ipa_free_edge_args_substructures): Likewise.
23978 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
23979 ipa_edge_args_vector.
23980 (ipa_edge_removal_hook): Turned into method
23981 ipa_edge_args_sum_t::remove.
23982 (ipa_edge_duplication_hook): Turned into method
23983 ipa_edge_args_sum_t::duplicate.
23984 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
23985 registering edge hooks.
23986 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
23987 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
23988 ipa_edge_args_sum instead of ipa_edge_args_vector.
23989 * ipa-profile.c (ipa_profile): Likewise.
23991 2017-05-03 Martin Jambor <mjambor@suse.cz>
23993 * symbol-summary.h (function_summary): New method exists.
23994 (function_summary::symtab_removal): Deallocate through release.
23995 (call_summary): New class.
23996 (gt_ggc_mx): New overload.
23997 (gt_pch_nx): Likewise.
23998 (gt_pch_nx): Likewise.
24000 2017-05-03 Jeff Law <law@redhat.com>
24002 PR tree-optimization/78496
24003 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
24004 from simplify_cond_using_ranges. Split off code to walk
24005 backwards through casts into ...
24006 (simplify_cond_using_ranges_2): New function.
24007 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
24008 (execute_vrp): After identifying jump threads, call
24009 simplify_cond_using_ranges_2.
24011 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
24014 * ipa-inline.h (inline_summary): Add ctor.
24015 (create_ggc): Do not use ggc_cleared_alloc.
24017 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
24018 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24020 * gcc.c (handle_braces): Support escaping in switch matching
24022 * doc/invoke.texi (Spec Files): Document it.
24023 Remove superfluous @code markup in items.
24025 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24027 * diagnostic-show-locus.c (struct column_range): New struct.
24028 (get_affected_columns): New function.
24029 (get_printed_columns): New function.
24030 (struct correction): New struct.
24031 (correction::ensure_capacity): New function.
24032 (correction::ensure_terminated): New function.
24033 (struct line_corrections): New struct.
24034 (line_corrections::~line_corrections): New dtor.
24035 (line_corrections::add_hint): New function.
24036 (layout::print_trailing_fixits): Reimplement in terms of the new
24038 (selftest::test_overlapped_fixit_printing): New function.
24039 (selftest::diagnostic_show_locus_c_tests): Call it.
24041 2017-05-03 Nathan Sidwell <nathan@acm.org>
24043 Canonicalize canonical type hashing
24044 * tree.h (type_hash_canon_hash): Declare.
24045 * tree.c (type_hash_list, attribute_hash_list): Move into
24046 type_hash_canon_hash.
24047 (build_type_attribute_qual_variant): Break out hash code calc into
24048 type_hash_canon_hash.
24049 (type_hash_canon_hash): New. Generic type hash computation.
24050 (build_range_type_1, build_array_type_1, build_function_type,
24051 build_method_type_directly, build_offset_type, build_complex_type,
24052 make_vector_type): Call it.
24054 2017-05-03 Richard Biener <rguenther@suse.de>
24056 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
24057 When all DRs have unknown misaligned do not always peel
24058 when there is a store but apply the same costing model as if
24059 there were only loads.
24061 2017-05-03 Richard Biener <rguenther@suse.de>
24064 PR tree-optimization/80492
24065 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
24066 compare_base_decls returning dont-know properly.
24068 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
24070 * config/arm/iterators.md (CCSI): New mode iterator.
24071 (arch): New mode attribute.
24072 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
24073 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
24074 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
24075 code iterator for success result mode.
24076 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
24077 the corresponding new insn generators.
24079 2017-05-03 Bin Cheng <bin.cheng@arm.com>
24082 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24083 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24085 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
24087 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
24088 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
24089 (DDR_A): Wrap DDR argument in brackets.
24090 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
24091 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
24092 (DDR_REVERSED_P): Likewise.
24094 2017-05-03 Jakub Jelinek <jakub@redhat.com>
24096 PR tree-optimization/79472
24097 * tree-switch-conversion.c (struct switch_conv_info): Add
24098 contiguous_range and default_case_nonstandard fields.
24099 (collect_switch_conv_info): Compute contiguous_range and
24100 default_case_nonstandard fields, don't clear final_bb if
24101 contiguous_range and only the default case doesn't have the required
24103 (check_all_empty_except_final): Set default_case_nonstandard instead
24104 of failing if contiguous_range and the default case doesn't have empty
24106 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
24107 and only the default case doesn't have the required constants. Skip
24109 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
24110 if default_case_nonstandard.
24111 (build_constructors): Build constant 1 just once. Assert that default
24112 values aren't inserted in between cases if contiguous_range. Skip
24114 (build_arrays): Skip virtual phis.
24115 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
24116 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
24117 Handle virtual phis.
24118 (gen_inbound_check): Handle default_case_nonstandard case.
24119 (process_switch): Adjust check_final_bb caller. Call
24120 gather_default_values with the first non-default case instead of
24121 default case if default_case_nonstandard.
24123 2017-05-02 Nathan Sidwell <nathan@acm.org>
24125 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
24126 check. Fix formatting.
24128 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
24130 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
24131 errors when comparing specialized and unspecialized times.
24133 2017-05-02 David Malcolm <dmalcolm@redhat.com>
24135 * diagnostic-show-locus.c
24136 (layout::should_print_annotation_line_p): Make private.
24137 (layout::print_annotation_line): Make private.
24138 (layout::annotation_line_showed_range_p): Make private.
24139 (layout::show_ruler): Make private.
24140 (layout::print_source_line): Make private. Pass in line and
24141 line_width, rather than calling location_get_source_line. Drop
24143 (layout::print_leading_fixits): New method.
24144 (layout::print_any_fixits): Rename to...
24145 (layout::print_trailing_fixits): ...this, and make private.
24146 Don't print newline fixits.
24147 (diagnostic_show_locus): Move logic for printing one row into...
24148 (layout::print_line): ...this new function. Move the
24149 location_get_source_line call and error-handling from
24150 print_source_line to here. Call print_leading_fixits, and rename
24151 print_any_fixits to print_trailing_fixits.
24152 (selftest::test_fixit_insert_containing_newline): Update now that
24153 newlines are partially supported.
24154 (selftest::test_fixit_insert_containing_newline_2): New test.
24155 (selftest::test_fixit_replace_containing_newline): Update comments.
24156 (selftest::diagnostic_show_locus_c_tests): Call the new test.
24157 * edit-context.c (class added_line): New class.
24158 (class edited_line): Describe newline handling in comment.
24159 (edited_line::actually_edited_p): New method.
24160 (edited_line::print_content): Delete redundant decl.
24161 (edited_line::m_predecessors): New field.
24162 (edited_file::print_content): Call edited_line::print_content.
24163 (edited_file::print_diff): Update to support newlines.
24164 (edited_file::print_diff_hunk): Likewise.
24165 (edited_file::print_run_of_changed_lines): New function.
24166 (edited_file::print_diff_line): Convert to...
24167 (print_diff_line): ...this.
24168 (edited_file::get_effective_line_count): New function.
24169 (edited_line::edited_line): Initialize new field m_predecessors.
24170 (edited_line::~edited_line): Clean up m_predecessors.
24171 (edited_line::apply_fixit): Handle newlines.
24172 (edited_line::get_effective_line_count): New function.
24173 (edited_line::print_content): New function.
24174 (edited_line::print_diff_lines): New function.
24175 (selftest::test_applying_fixits_insert_containing_newline): New
24177 (selftest::test_applying_fixits_replace_containing_newline): New
24179 (selftest::insert_line): New function.
24180 (selftest::test_applying_fixits_multiple_lines): Add example of
24182 (selftest::edit_context_c_tests): Call the new tests.
24184 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24186 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
24187 parameter cand. Update dump information.
24188 (get_computation_cost): Update uses.
24190 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24192 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
24193 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
24194 (get_computation_at, rewrite_use_address): Update use of
24195 get_computation_aff.
24197 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24199 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
24200 (get_computation): Delete.
24201 (get_computation_cost): Implement like get_computation_cost_at.
24202 Use get_computation_at.
24203 (get_computation_cost_at): Delete.
24204 (rewrite_use_nonlinear_expr): Use get_computation_at.
24205 (rewrite_use_compare, remove_unused_ivs): Ditto.
24207 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24209 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
24211 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24213 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
24214 (ivopts_global_cost_for_size): Rename parameter and update uses.
24215 (iv_ca_recount_cost): Update uses.
24216 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
24217 candidates seperately in n_invs and n_cands.
24218 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
24220 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24222 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
24223 (find_inv_vars_cb): New.
24224 (find_depends): Renamed to ...
24225 (find_inv_vars): ... this.
24226 (add_candidate_1, force_var_cost): Call find_inv_vars.
24227 (split_address_cost, determine_group_iv_cost_cond): Ditto.
24229 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24231 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
24232 inv_vars. Add inv_exprs.
24233 (struct iv_cand): Rename depends_on to inv_vars.
24234 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
24235 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
24236 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
24237 (dump_cand): Dump inv_vars.
24238 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
24239 (record_invariant, find_depends, add_candidate_1): Ditto.
24240 (set_group_iv_cost, force_var_cost): Ditto.
24241 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
24242 (get_computation_cost_at, get_computation_cost): Ditto.
24243 (determine_group_iv_cost_generic): Ditto.
24244 (determine_group_iv_cost_address): Ditto.
24245 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
24246 (determine_group_iv_costs): Ditto.
24247 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
24248 (iv_ca_set_remove_invariants): Renamed to ...
24249 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
24250 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
24251 (iv_ca_set_add_invariants): Renamed to ...
24252 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
24253 (iv_ca_set_cp): Use iv_ca_set_add_invs.
24254 (iv_ca_has_deps): Support inv_vars and inv_exprs.
24255 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
24256 (create_new_ivs): Remove useless dump.
24258 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24260 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
24262 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
24263 (iv_ca_set_no_cp, create_new_iv): Ditto.
24265 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24267 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24269 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
24271 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
24272 function tree_check2.
24274 2017-05-02 Martin Liska <mliska@suse.cz>
24276 * doc/gcov.texi: Add missing preposition.
24277 * gcov.c (function_info::function_info): Properly fill up
24278 all member variables.
24280 2017-05-02 Tamar Christina <tamar.christina@arm.com>
24282 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
24284 2017-05-02 Tamar Christina <tamar.christina@arm.com>
24286 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
24288 2017-05-02 Martin Liska <mliska@suse.cz>
24291 * lto-streamer-in.c (lto_read_tree_1): Remove
24292 LTO_STREAMER_DEBUG.
24293 * lto-streamer.c (struct tree_hash_entry): Likewise.
24294 (struct tree_entry_hasher): Likewise.
24295 (tree_entry_hasher::hash): Likewise.
24296 (tree_entry_hasher::equal): Likewise.
24297 (lto_streamer_init): Likewise.
24298 (lto_orig_address_map): Likewise.
24299 (lto_orig_address_get): Likewise.
24300 (lto_orig_address_remove): Likewise.
24301 * lto-streamer.h: Likewise.
24302 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
24303 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
24305 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
24307 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
24308 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
24309 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
24310 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
24311 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
24312 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
24313 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
24314 (mm_maskz_sub_ss): New intrinsics.
24315 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24316 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
24317 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
24318 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
24319 (__builtin_ia32_subss_mask_round): New builtins.
24320 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24321 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24322 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
24324 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
24325 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
24327 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
24330 2017-05-02 Martin Jambor <mjambor@suse.cz>
24332 PR tree-optimization/78687
24333 * tree-sra.c (access): New field parent.
24334 (process_subtree_disqualification): New function.
24335 (disqualify_candidate): Call it.
24336 (build_accesses_from_assign): Reset write flag if creating an
24338 (build_access_subtree): Fill in parent field and also prpagate
24339 down grp_write flag.
24340 (create_artificial_child_access): New parameter set_grp_write, set
24341 grp_write to its value.
24342 (propagate_subaccesses_across_link): Also propagate grp_write flag
24344 (propagate_all_subaccesses): Push the closest parent back to work
24345 queue if add_access_to_work_queue returned true.
24347 2017-05-02 Richard Biener <rguenther@suse.de>
24349 * common.opt (fstrict-overflow): Alias negative to fwrapv.
24350 * doc/invoke.texi (fstrict-overflow): Remove all traces of
24351 -fstrict-overflow documentation.
24352 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
24353 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
24354 flag_strict_overflow.
24355 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
24356 * lto-opts.c (lto_write_options): Do not stream it.
24357 * lto-wrapper.c (merge_and_complain): Do not handle it.
24358 * opts.c (default_options_table): Do not set -fstrict-overflow.
24359 (finish_options): Likewise do not clear it when sanitizing.
24360 * simplify-rtx.c (simplify_const_relational_operation): Do not
24361 test flag_strict_overflow.
24363 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
24365 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
24366 using enabled attribute.
24367 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
24368 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
24369 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
24370 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
24371 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
24372 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
24373 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
24374 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
24375 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
24376 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
24378 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
24380 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
24382 2017-05-02 Richard Biener <rguenther@suse.de>
24384 PR tree-optimization/80591
24386 2017-04-10 Richard Biener <rguenther@suse.de>
24388 * tree-ssa-structalias.c (find_func_aliases): Properly handle
24391 2017-05-02 Richard Biener <rguenther@suse.de>
24393 PR tree-optimization/80549
24394 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
24395 (cleanup_tree_cfg_noloop): Create forwarders to known loop
24396 headers if they do not have a preheader.
24398 2017-05-02 Martin Liska <mliska@suse.cz>
24401 * common.opt: Fix typo.
24402 * doc/invoke.texi: Likewise.
24404 2017-05-01 Jan Beulich <jbeulich@suse.com>
24406 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
24407 swapping, add (x,x,m,x,n) alternative.
24409 2017-05-01 Nathan Sidwell <nathan@acm.org>
24411 * calls.c (combine_pending_stack_adjustment_and_call): Remove
24412 unnecessary unadjusted_alignment check.
24414 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
24417 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
24419 * gimplify.c (gimplify_cilk_detach): New function.
24420 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
24421 * tree-core.h: Document EXPR_CILK_SPAWN.
24422 * tree.h (EXPR_CILK_SPAWN): Define.
24424 2017-05-01 David Malcolm <dmalcolm@redhat.com>
24426 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
24427 to use new fixit_hint representation, using the "replace" logic.
24428 (get_line_span_for_fixit_hint): Likewise.
24429 (layout::print_any_fixits): Likewise.
24430 (selftest::test_one_liner_many_fixits): Rename to...
24431 (selftest::test_one_liner_many_fixits_1): ...this, and update
24432 comment and expected output to reflect that the multiple fix-it
24433 hints are now consolidated into one insertion.
24434 (selftest::test_one_liner_many_fixits_2): New test.
24435 (selftest::test_diagnostic_show_locus_one_liner): Update for
24437 (selftest::test_fixit_consolidation): Update for fix-it API
24439 * diagnostic.c (print_parseable_fixits): Likewise.
24440 * edit-context.c (edited_line::m_line_events): Convert from
24441 auto_vec <line_event *> to auto_vec <line_event>.
24442 (class line_event): Convert from abstract base class to a concrete
24443 class, taking over the role of replace_event.
24444 (class insert_event): Delete.
24445 (class replace_event): Rename to class line_event. Convert to
24447 (edit_context::add_fixits): Reimplement.
24448 (edit_context::apply_insert): Delete.
24449 (edit_context::apply_replace): Rename to...
24450 (edit_context::apply_fixit): ...this. Convert to half-open range.
24451 (edited_file::apply_insert): Delete.
24452 (edited_file::apply_replace): Rename to...
24453 (edited_file::apply_fixit): ...this.
24454 (edited_line::~edited_line): Drop deletion of events.
24455 (edited_line::apply_insert): Delete.
24456 (edited_line::apply_replace): Rename to...
24457 (edited_line::apply_fixit): ...this. Convert to half-open range.
24458 Update for change to type of m_line_events.
24459 * edit-context.h (edit_context::apply_insert): Delete.
24460 (edit_context::apply_replace): Rename to...
24461 (edit_context::apply_fixit): ...this.
24463 2017-05-01 Martin Sebor <msebor@redhat.com>
24465 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
24468 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
24471 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
24472 __get_cpuid_max returns 0.
24473 (__get_cpuid_count): Ditto.
24475 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
24477 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
24478 replacement expression is another instance of one of its arguments.
24480 2017-05-01 Jakub Jelinek <jakub@redhat.com>
24483 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
24484 check for stack push/pop autoinc.
24485 * config/i386/i386.c (ix86_agi_dependent): Return false
24486 if the only reason why modified_in_p returned true is that
24487 addr is SP based and set_insn is a push or pop.
24489 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
24491 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
24494 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
24497 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
24498 (account_size_time): Use two predicates - exec_pred and
24500 (evaluate_conditions_for_known_args): Compute both clause and
24502 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
24503 (inline_summary_t::duplicate): Update.
24504 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
24506 (compute_inline_parameters): Likewise.
24507 (estimate_edge_size_and_time): Update caluclation of time.
24508 (estimate_node_size_and_time): Compute both time and nonspecialized
24510 (estimate_ipcp_clone_size_and_time): Update.
24511 (inline_merge_summary): Update.
24512 (do_estimate_edge_time): Update.
24513 (do_estimate_edge_size): Update.
24514 (do_estimate_edge_hints): Update.
24515 (inline_read_section, inline_write_summary): Stream both new predicates.
24516 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
24518 (compute_inlined_call_time): Cleanup.
24519 (big_speedup_p): Update.
24520 (edge_badness): Update.
24521 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
24522 (size_time_entry): Replace predicate by exec_predicate and
24523 nonconst_predicate.
24524 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
24525 (estimate_edge_time): Return also nonspec_time.
24526 (reset_edge_growth_cache): Update.
24528 2017-04-29 Jakub Jelinek <jakub@redhat.com>
24530 PR rtl-optimization/80491
24531 * ifcvt.c (noce_process_if_block): When looking for x setter
24532 with missing else_bb, don't check only the insn right before
24533 cond_earliest, but look for the last insn that x is modified in
24534 within the same bb.
24536 PR rtl-optimization/80491
24537 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
24539 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
24541 PR tree-optimization/80487
24542 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
24544 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24546 PR tree-optimization/79697
24547 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
24548 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
24549 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
24551 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
24552 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
24554 2017-04-28 Martin Sebor <msebor@redhat.com>
24556 PR tree-optimization/80523
24557 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
24558 (init_target_to_host_charmap, target_to_host, target_strtol10): New
24560 (maybe_warn, format_directive, parse_directive): Use new functions.
24561 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
24563 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
24565 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
24567 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
24569 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
24570 target_header_dir): Set correctly.
24571 * configure: Regenerated.
24572 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
24573 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
24574 instead of SYSTEM_HEADER_DIR.
24576 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
24578 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
24579 (estimate_local_effects): Likewise.
24580 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
24581 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
24582 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
24583 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
24584 do_estimate_edge_time, estimate_edge_time): Likewise.
24585 * ipa-inline-analysis.c (estimate_node_size_and_time,
24586 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
24587 (estimate_time_after_inlining): Remove.
24589 2017-04-28 Martin Liska <mliska@suse.cz>
24591 * doc/gcov.texi: Enhance documentation of gcov.
24593 2017-04-28 Martin Liska <mliska@suse.cz>
24595 * doc/gcov.texi: Sort options in alphabetic order.
24596 * doc/gcov-dump.texi: Likewise.
24597 * doc/gcov-tool.texi: Likewise.
24598 * gcov.c (print_usage): Likewise.
24599 * gcov-dump.c (print_usage): Likewise.
24600 * gcov-tool.c (print_merge_usage_message): Likewise.
24601 (print_rewrite_usage_message): Likewise.
24602 (print_overlap_usage_message): Likewise.
24604 2017-04-28 Martin Liska <mliska@suse.cz>
24606 PR gcov-profile/53915
24607 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
24609 2017-04-28 Martin Liska <mliska@suse.cz>
24611 PR gcov-profile/79891
24612 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
24613 is marked by compiler as living on a line.
24614 (get_cycles_count): Remove usage of the union.
24615 (output_intermediate_file): Likewise.
24616 (find_source): Fix GNU coding style.
24617 (accumulate_line_counts): Remove old non-all block mode.
24618 (output_lines): Remove usage of the union.
24619 * profile.c (output_location): Include all BBs, even if
24620 belonging to a same line (and file) as a previous BB.
24622 2017-04-28 Martin Liska <mliska@suse.cz>
24624 * gcov.c (process_args): Handle new argument 'w'.
24625 (read_graph_file): Assign ID to BBs.
24626 (output_branch_count): Display BB # if verbose flag is set.
24627 (output_lines): Likewise for arcs.
24628 (print_usage): Add '--verbose' option help.
24629 * doc/gcov.texi: Document --verbose (-w) option.
24631 2017-04-28 Martin Liska <mliska@suse.cz>
24633 * gcov.c (struct block_location_info): New struct.
24634 (process_file): Fill up the new structure.
24635 (read_graph_file): Replace usage of encoding by the newly added
24637 (add_line_counts): Likewise.
24638 (accumulate_line_counts): Remove usage of the union.
24639 (function_info::function_info): New function.
24640 (function_info::~function_info): Likewise.
24641 (process_file): Call delete instead of release_function.
24642 (release_function): Release the function.
24643 (release_structures): Call delete instead of release_function.
24644 (solve_flow_graph): Replace usage of num_blocks.
24645 (find_exception_blocks): Likewise.
24646 (output_lines): Fix GNU coding style.
24648 2017-04-28 Martin Liska <mliska@suse.cz>
24651 * coverage.c (coverage_remove_note_file): New function.
24652 * coverage.h: Declare the function.
24653 * toplev.c (finalize): Clean if an error has been seen.
24655 2017-04-28 Martin Liska <mliska@suse.cz>
24657 PR gcov-profile/80031
24658 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
24659 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
24660 * gcov.c (read_graph_file): Read just number of blocks.
24661 * profile.c (branch_prob): Do not stream 0 flags per a basic
24664 2017-04-28 Martin Liska <mliska@suse.cz>
24666 * gcov-dump.c (tag_*): Add new argument to declarations.
24667 (dump_gcov_file): Likewise.
24668 (tag_blocks): Add and use new argument depth.
24669 (tag_arcs): Likewise.
24670 (tag_lines): Likewise.
24671 (tag_counters): Likewise.
24672 (tag_summary): Likewise.
24673 (dump_working_sets): Use depth to do a proper indentation.
24675 2017-04-28 Jakub Jelinek <jakub@redhat.com>
24678 * cgraph.h (symtab_node::debug_symtab): No longer inline.
24679 * symtab.c (symtab_node::debug_symtab): Move definition here.
24681 2017-04-28 Richard Biener <rguenther@suse.de>
24683 * lto-streamer.h (LTO_major_version): Bump to 7.
24685 2017-04-28 Richard Biener <rguenther@suse.de>
24687 * tree-vrp.c (assert_info): New struct.
24688 (add_assert_info): New helper.
24689 (register_edge_assert_for_2): Refactor to add asserts to a vector
24691 (register_edge_assert_for_1): Likewise.
24692 (register_edge_assert_for): Likewise.
24693 (finish_register_edge_assert_for): New helper actually registering
24694 asserts where live on edge.
24695 (find_conditional_asserts): Adjust.
24696 (find_switch_asserts): Likewise.
24697 (evrp_dom_walker::try_find_new_range): Generalize.
24698 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
24700 2017-04-27 Marek Polacek <polacek@redhat.com>
24703 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
24704 arg10 and arg11 to itype.
24706 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
24708 * doc/extend.texi (Object Size Checking): Improve grammar.
24710 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
24713 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
24714 that the logic for permitting reciprocal estimates matches that
24717 2017-04-27 Jakub Jelinek <jakub@redhat.com>
24720 * tree.c (type_cache_hasher::equal): Only compare
24721 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
24722 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
24723 non-aggregate element types.
24724 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
24725 about the flag on ARRAY_TYPEs in the comment, formatting fix.
24727 2017-04-27 Richard Biener <rguenther@suse.de>
24729 PR middle-end/80533
24730 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
24731 stripping ARRAY_REFs from MEM_EXPR make sure we're not
24732 keeping a reference to a trailing array.
24734 2017-04-27 Richard Biener <rguenther@suse.de>
24736 PR middle-end/80539
24737 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
24738 being in loop-closed SSA form conservatively.
24739 (chrec_fold_multiply_poly_poly): Likewise.
24741 2017-04-27 Tamar Christina <tamar.christina@arm.com>
24743 PR middle-end/79665
24744 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
24745 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
24747 2017-04-27 Jakub Jelinek <jakub@redhat.com>
24750 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
24751 (aarch64_function_arg_alignment): Return unsigned int again, but still
24752 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
24753 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
24754 Don't emit -Wpsabi note.
24755 (aarch64_function_arg_boundary): Likewise.
24756 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
24759 2017-04-26 Nathan Sidwell <nathan@acm.org>
24761 * tree.h (crc32_unsigned_n): Declare.
24762 (crc32_unsigned, crc32_unsigned): Make inline.
24763 * tree.c (crc32_unsigned_bits): Replace with ...
24764 (crc32_unsigned_n): ... this.
24765 (crc32_unsigned, crc32_byte): Remove.
24766 (crc32_string): Remove unnecessary braces.
24768 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
24770 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
24771 * ipa-inline-analysis.c (MAX_TIME): Remove.
24772 (account_size_time): Use sreal for time.
24773 (dump_inline_summary): Update.
24774 (estimate_function_body_sizes): Update.
24775 (estimate_edge_size_and_time): Update.
24776 (estimate_calls_size_and_time): Update.
24777 (estimate_node_size_and_time): Update.
24778 (inline_merge_summary): Update.
24779 (inline_update_overall_summary): Update.
24780 (estimate_time_after_inlining): Update.
24781 (inline_read_section): Update.
24782 (inline_write_summary): Update.
24783 * ipa-inline.c (compute_uninlined_call_time): Update.
24784 (compute_inlined_call_time): Update.
24785 (recursive_inlining): Update.
24786 (inline_small_functions): Update.
24787 (dump_overall_stats): Update.
24788 * ipa-inline.h: Include sreal.h.
24789 (size_time_entry): Turn time to sreal.
24790 (inline_summary): Turn self_time nad time to sreal.
24792 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
24794 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
24796 (sreal::stream_out, sreal::stream_in): New.
24797 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
24799 2017-04-25 Jakub Jelinek <jakub@redhat.com>
24801 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
24804 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
24807 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
24809 (dimode_scalar_chain::compute_convert_gain): Ditto.
24810 (dimode_scalar_chain::make_vector_copies): Ditto.
24811 (dimode_scalar_chain::convert_reg): Ditto.
24812 (dimode_scalar_chain::convert_insn): Ditto.
24813 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
24814 (VI248_AVX512BW_1): New mode iterator.
24815 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
24816 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
24819 2017-04-25 Martin Sebor <msebor@redhat.com>
24821 PR tree-optimization/80497
24822 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
24823 constants are representable in HOST_WIDE_INT.
24824 (parse_directive): Ditto.
24826 2017-04-25 Martin Sebor <msebor@redhat.com>
24829 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
24830 (new_zero_array): Adjust signature.
24831 (dom_info::dom_init): Used unsigned rather that size_t.
24832 (dom_info::dom_info): Same.
24834 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24835 Jakub Jelinek <jakub@redhat.com>
24838 * config/arm/arm.c: Include gimple.h.
24839 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
24840 returns negative, increment ncrn only if it returned positive.
24841 (arm_needs_doubleword_align): Return int instead of bool,
24842 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
24843 members, but if there is any such non-FIELD_DECL
24844 > PARM_BOUNDARY aligned decl, return -1 instead of false.
24845 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
24846 returns negative, increment nregs only if it returned positive.
24847 (arm_setup_incoming_varargs): Likewise.
24848 (arm_function_arg_boundary): Emit -Wpsabi note if
24849 arm_needs_doubleword_align returns negative, return
24850 DOUBLEWORD_ALIGNMENT only if it returned positive.
24852 2017-04-25 Marek Polacek <polacek@redhat.com>
24855 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
24856 first argument to type.
24858 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
24861 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
24862 type checks to test for compatibility instead of equality.
24864 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24865 Jakub Jelinek <jakub@redhat.com>
24868 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
24870 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
24871 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
24872 the alignment computation, but return their maximum in warn_alignment.
24873 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
24874 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
24876 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
24877 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
24880 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24882 * config/arc/simdext.md (dmpyh): Fix typo.
24884 2017-04-25 Richard Biener <rguenther@suse.de>
24886 PR tree-optimization/80492
24887 * alias.c (compare_base_decls): Handle registers with asm
24888 specification conservatively.
24889 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
24890 compare_base_decls returning dont-know properly.
24892 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24894 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
24895 (legitimate_offset_address_p): New function.
24896 (arc_legitimate_address_p): Use above function.
24898 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24900 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
24902 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24904 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
24905 ACCH registers whenever they are available.
24907 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24909 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
24910 double regs fix when not used.
24912 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24914 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
24916 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
24917 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
24919 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24921 * config/arc/arc.c (arc_output_addsi): Check for h-register class
24922 when emitting short ADD instructions.
24924 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24926 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
24928 (cmpsi_cc_c_insn): Likewise.
24929 (cbranchsi4_scratch): Compute proper instruction length using
24930 compact_hreg_operand.
24931 * config/arc/predicates.md (compact_hreg_operand): New predicate.
24933 2017-04-25 Richard Biener <rguenther@suse.de>
24935 PR middle-end/80509
24936 * passes.c (pass_manager::pass_manager): Initialize
24937 m_name_to_pass_map.
24939 2017-04-25 Richard Biener <rguenther@suse.de>
24941 PR tree-optimization/79201
24942 * tree-ssa-sink.c (statement_sink_location): Handle calls.
24944 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24947 * config/s390/vector.md: Split MEM->GPR vector moves for
24948 non-s_operand addresses.
24950 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24953 * config/s390/predicates.md (reload_const_wide_int_operand): New
24955 * config/s390/s390.md ("movti"): Remove d/P alternative.
24956 ("movti_bigconst"): New pattern definition.
24958 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
24961 * s390-protos.h (s390_expand_cs_hqi): Removed.
24962 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
24963 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
24964 modes as well as CCZ1mode and CCZmode.
24965 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
24966 signature of s390_emit_compare_and_swap.
24967 (s390_expand_cs_hqi): Likewise, make static.
24968 (s390_expand_cs_tdsi): Generate an explicit compare before trying
24969 compare-and-swap, in some cases.
24970 (s390_expand_cs): Wrapper function.
24971 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
24973 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
24974 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
24975 patterns for small and large integers. Forbid symref memory operands.
24976 Move expander to s390.c. Require cc register.
24977 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
24978 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
24979 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
24980 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
24981 symref memory operands. Remove CC mode and call s390_match_ccmode
24983 ("atomic_exchange<mode>"): Allow and implement all integer modes.
24985 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
24987 * config/s390/s390.md (define_peephole2): New peephole to help
24988 combining the load-and-test pattern with volatile memory.
24990 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
24992 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
24993 with CCZmode for TARGET_Z196.
24995 2017-04-25 Jakub Jelinek <jakub@redhat.com>
24997 PR rtl-optimization/80501
24998 * combine.c (make_compound_operation_int): Set subreg_code to SET
24999 even for AND with mask of the sign bit of mode.
25001 PR rtl-optimization/80500
25002 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
25003 sum's initial value.
25005 2017-04-25 Julian Brown <julian@codesourcery.com>
25006 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25008 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
25010 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
25012 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
25014 2017-04-25 Julian Brown <julian@codesourcery.com>
25015 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25017 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
25018 (thunderx2t99_sha): New Reservation.
25020 2017-04-25 Julian Brown <julian@codesourcery.com>
25021 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25023 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
25024 type for 1-element load.
25026 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
25028 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
25030 2017-04-24 Martin Jambor <mjambor@suse.cz>
25032 PR tree-optimization/80293
25033 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
25034 char arrays not totally scalarizable if it is false.
25035 (analyze_all_variable_accesses): Pass correct value in the new
25036 parameter. Add a statistics counter.
25038 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
25040 PR middle-end/79931
25041 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
25043 2017-04-24 Richard Biener <rguenther@suse.de>
25045 PR tree-optimization/80494
25046 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
25047 out for complex types.
25049 2017-04-24 Richard Biener <rguenther@suse.de>
25051 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
25052 * tree-ssa-sccvn.c (print_scc): Print SCC size.
25053 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
25054 (DFS): Adjust and never fail.
25055 (sccvn_dom_walker::fail): Remove.
25056 (sccvn_dom_walker::before_dom_children): Adjust.
25057 (run_scc_vn): Likewise and never fail.
25058 * tree-ssa-pre.c (pass_pre::execute): Adjust.
25059 (pass_fre::execute): Likewise.
25061 2017-04-24 Richard Biener <rguenther@suse.de>
25063 PR tree-optimization/79725
25064 * tree-ssa-sink.c (statement_sink_location): Return whether
25065 failure reason was zero uses. Move that check later.
25066 (sink_code_in_bb): Deal with zero uses by removing the stmt
25069 2017-04-24 Richard Biener <rguenther@suse.de>
25072 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
25073 pointer-based references.
25075 2017-04-24 Richard Biener <rguenther@suse.de>
25078 * pass_manager.h (pass_manager::operator new): Remove.
25079 (pass_manager::operator delete): Likewise.
25080 * passes.c (pass_manager::operator new): Remove.
25081 (pass_manager::operator delete): Likewise.
25082 (pass_manager::pass_manager): Zero individual pass members.
25084 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
25087 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
25088 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
25089 Check "XEXP (src, 1)" operand here.
25090 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
25091 Check "XEXP (src, 1)" operand here.
25092 (dimode_scalar_chain::make_vector_copies): Detect count register
25093 of a shift instruction. Zero extend count register from QImode
25094 to DImode to satisfy vector shift pattern count operand predicate.
25095 Substitute vector shift count operand with a DImode copy.
25096 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
25099 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
25101 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
25102 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
25103 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
25104 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
25105 (UNSPEC_NOREX_MEM): Remove definition.
25107 2017-04-21 Richard Biener <rguenther@suse.de>
25109 PR tree-optimization/79547
25110 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
25111 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
25112 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
25113 without any constraints.
25115 2017-04-21 Richard Biener <rguenther@suse.de>
25117 PR tree-optimization/78847
25118 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
25120 2017-04-21 Richard Biener <rguenther@suse.de>
25122 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
25123 (build_distinct_type_copy): Likewise.
25124 (build_variant_type_copy): Likewise.
25125 * tree.c (build_qualified_type): Pass down mem-stat info.
25126 (build_distinct_type_copy): Likewise.
25127 (build_variant_type_copy): Likewise.
25129 2017-04-21 Richard Biener <rguenther@suse.de>
25131 PR tree-optimization/80237
25132 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
25134 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
25135 for a simplified result.
25137 2016-04-21 Richard Biener <rguenther@suse.de>
25139 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
25140 sth as strict as a simple_iv but a chrec without symbols and an
25141 operand defined in the loop we are peeling (and not some subloop).
25142 (propagate_constants_for_unrolling): Propagate all constants.
25144 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25147 * config/i386/i386.c (print_reg): Remove assert for disalowed
25148 regno values, call output_operand_lossage instead.
25150 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25153 * config/i386/constraints.md (Yc): New register constraint.
25154 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
25155 Use Yc constraint for alternative 2 of operand 0. Remove
25156 preferred_for_speed attribute.
25158 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
25160 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
25161 lastprivate clauses in SIMT case.
25163 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
25165 * doc/invoke.texi (-Wextra-semi): Document new warning option.
25167 2017-04-20 Richard Biener <rguenther@suse.de>
25169 PR tree-optimization/57796
25170 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
25171 as N scalar stores.
25172 (vect_model_load_cost): Cost gathers as N scalar loads.
25174 2017-04-20 Richard Biener <rguenther@suse.de>
25176 * ggc-page.c (ggc_allocated_p): Rename to ...
25177 (safe_lookup_page_table_entry): ... this and return the lookup
25179 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
25181 2017-04-20 Richard Biener <rguenther@suse.de>
25183 PR tree-optimization/80453
25184 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
25185 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
25186 from the conditions.
25187 (vn_phi_eq): Pass them down.
25188 (vn_phi_lookup): Record them.
25189 (vn_phi_insert): Likewise.
25191 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
25193 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
25194 uninitialized variable warning to avoid buffer overrun.
25196 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
25199 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
25200 is suppressed for '{ 0 }' in C.
25202 2017-04-20 Jakub Jelinek <jakub@redhat.com>
25204 * BASE-VER: Set to 8.0.0.
25206 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
25208 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
25209 priority .init_array and .fini_array section with SECTION_NOTYPE
25212 2017-04-20 Jakub Jelinek <jakub@redhat.com>
25214 PR middle-end/80423
25215 * tree.h (build_array_type): Add typeless_storage default argument.
25216 * tree.c (type_cache_hasher::equal): Also compare
25217 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
25218 (build_array_type): Add typeless_storage argument, set
25219 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
25221 (build_nonshared_array_type): Adjust build_array_type_1 caller.
25222 (build_array_type): Likewise. Add typeless_storage argument.
25224 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
25225 Jakub Jelinek <jakub@redhat.com>
25227 PR tree-optimization/80426
25228 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
25229 operation on symbolic operands, also compute the overflow for the
25230 invariant part when the operation degenerates into a negation.
25232 2017-04-19 Jakub Jelinek <jakub@redhat.com>
25235 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
25236 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
25239 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
25241 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
25244 * config/avr/avr.c (tree.h): Include it.
25245 (cgraph.h): Include it.
25246 (avr_encode_section_info): Don't warn for uninitialized progmem
25247 variable if it's just an alias.
25249 2017-04-19 Richard Biener <rguenther@suse.de>
25252 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
25253 when needed by AutoPGO.
25255 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
25258 * doc/lto.texi: Remove an extra 'that'.
25260 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
25262 PR rtl-optimization/80429
25263 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
25264 are only used in debug insns.
25266 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
25267 Vladimir Makarov <vmakarov@redhat.com>
25269 * config/sparc/predicates.md (input_operand): Add comment. Return
25270 true for any memory operand when LRA is in progress.
25271 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
25273 2017-04-18 Jeff Law <law@redhat.com>
25276 * mips.md ({return,simple_return}_internal): Do not overwrite
25279 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25281 PR tree-optimization/80443
25282 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
25283 instead of adding 1, subtract -1 and similarly instead of subtracting
25286 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
25288 PR rtl-optimization/80357
25289 * haifa-sched.c (tmp_bitmap): New variable.
25290 (model_recompute): Handle duplicate use records.
25291 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
25292 (free_global_sched_pressure_data): Free it.
25294 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
25297 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
25298 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25299 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25300 instead of SYSTEM_HEADER_DIR.
25302 2017-04-18 Jeff Law <law@redhat.com>
25304 PR middle-end/80422
25305 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
25306 predecessors after walking up the insn chain.
25308 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25311 * dwarf2out.c (modified_type_die): Try harder not to emit internal
25312 sizetype type into debug info.
25314 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
25317 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
25318 unneeded test for TARGET_UPPER_REGS_SF.
25319 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
25321 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25324 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
25325 instead of gsi_after_labels.
25327 2017-04-18 Jeff Law <law@redhat.com>
25329 * regcprop.c (maybe_mode_change): Avoid creating copies of the
25333 2017-04-13 Jeff Law <law@redhat.com>
25334 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
25335 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
25337 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
25340 * config/avr/avr.c (intl.h): Include it.
25341 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
25343 2017-04-18 Martin Liska <mliska@suse.cz>
25345 PR gcov-profile/78783
25346 * gcov-tool.c (gcov_output_files): Validate that destination
25347 file is either removed by the tool or by a user.
25349 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
25350 Guy Benyei <guybe@mellanox.com>
25352 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
25353 block, and do not negate it, the stored id is already negative.
25355 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
25357 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
25359 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
25362 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
25363 masks of options that should be turned off if the VSX vector
25364 options are turned off.
25365 (OTHER_P8_VECTOR_MASKS): Likewise.
25366 (OTHER_VSX_VECTOR_MASKS): Likewise.
25367 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
25368 rs6000_disable_incompatible_switches to validate no type switches
25370 (rs6000_incompatible_switch): New function to disallow turning on
25371 other vector options if -mno-vsx, -mno-power8-vector, or
25372 -mno-power9-vector are specified.
25374 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25376 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
25378 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25380 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
25381 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
25382 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
25383 (ARG_POINTER_CFA_OFFSET): Likewise.
25385 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25387 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
25388 conditions to take advantage of various optimizations.
25390 2017-04-13 Jeff Law <law@redhat.com>
25392 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
25393 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
25394 (zero_extendsidi2_dext): Likewise.
25396 2017-04-13 Jakub Jelinek <jakub@redhat.com>
25399 * fold-const.c (fold_ternary_loc): Revert
25400 use op0 instead of fold_convert_loc (loc, type, arg0) part of
25403 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
25405 PR rtl-optimization/80343
25406 * lra-remat.c (update_scratch_ops): Assign original hard reg to
25407 new scratch pseudo.
25409 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
25412 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
25413 to ubsan_encode_value.
25415 2017-04-13 Jeff Law <law@redhat.com>
25417 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
25418 appearing in DEBUG_INSNs.
25420 2017-04-13 Martin Liska <mliska@suse.cz>
25422 PR gcov-profile/80413
25423 * gcov-io.c (gcov_write_string): Copy to buffer just when
25424 allocated size is greater than zero.
25426 2017-04-13 Jakub Jelinek <jakub@redhat.com>
25429 * dwarf2out.c (decls_for_scope): Ignore declarations of
25430 current_function_decl in BLOCK_NONLOCALIZED_VARS.
25432 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
25435 * ipa-visibility.c (non_local_p): Fix typos.
25436 (localize_node): When localizing symbol in same comdat group,
25437 dissolve the group only when we know external symbols are going
25439 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
25441 2017-04-12 Jakub Jelinek <jakub@redhat.com>
25443 PR tree-optimization/79390
25444 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
25445 order does not result in usable sequence, retry with reversed operand
25451 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
25452 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
25453 op0 instead of fold_convert_loc (loc, type, arg0).
25455 2017-04-12 Jeff Law <law@redhat.com>
25457 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
25458 has a delay slot in the generated code.
25460 * config/cris/cris.md (cris_preferred_reload_class): Return
25461 GENNONACR_REGS rather than GENERAL_REGS.
25463 2017-04-12 Jakub Jelinek <jakub@redhat.com>
25466 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
25467 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
25468 signedness of the result type.
25470 2017-04-12 Richard Biener <rguenther@suse.de>
25471 Jeff Law <law@redhat.com>
25473 PR tree-optimization/80359
25474 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
25475 trim stores to TARGET_MEM_REFs.
25477 2017-04-12 Richard Biener <rguenther@suse.de>
25479 PR tree-optimization/79390
25480 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
25481 threading case even more.
25483 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
25486 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
25487 for quad_address_p for TImode, instead of just not indexed_address.
25489 2017-04-12 Richard Biener <rguenther@suse.de>
25490 Bernd Edlinger <bernd.edlinger@hotmail.de>
25492 PR middle-end/79671
25493 * alias.c (component_uses_parent_alias_set_from): Handle
25494 TYPE_TYPELESS_STORAGE.
25495 (get_alias_set): Likewise.
25496 * tree-core.h (tree_type_common): Add typeless_storage flag.
25497 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
25498 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
25499 for types containing members with TYPE_TYPELESS_STORAGE.
25500 (place_field): Likewise.
25501 (layout_type): Likewise for ARRAY_TYPE.
25502 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
25503 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
25504 TYPE_TYPELESS_STORAGE.
25505 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
25507 2017-04-12 Jakub Jelinek <jakub@redhat.com>
25510 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
25511 first argument to type.
25513 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25517 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
25518 CONST0_RTX (mode) rather than const0_rtx where appropriate.
25519 (rs6000_expand_binop_builtin): Likewise.
25520 (rs6000_expand_ternop_builtin): Likewise; also add missing
25521 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
25522 vshasigma built-ins.
25523 * doc/extend.texi: Document that vec_xxpermdi's third argument
25524 must be a constant.
25526 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
25528 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
25529 Use shift_const cost parameter when calculating gain of STV shifts.
25531 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
25533 PR rtl-optimization/70478
25534 * lra-constraints.c (process_alt_operands): Check memory for
25535 disfavoring memory insn operand.
25537 2017-04-11 Jakub Jelinek <jakub@redhat.com>
25539 PR middle-end/80100
25540 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
25541 left shift in unsigned HOST_WIDE_INT type.
25543 PR rtl-optimization/80385
25544 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
25545 (not (neg X)) into (plus X -1) for complex or non-integral modes.
25548 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
25549 if they have any depend clauses.
25551 2017-04-11 Martin Liska <mliska@suse.cz>
25554 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
25555 * ipa-split.c (split_function): Create a local comdat symbol
25556 if caller is in a comdat group.
25558 2017-04-11 Martin Liska <mliska@suse.cz>
25561 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
25564 2017-04-11 Martin Sebor <msebor@redhat.com>
25566 PR middle-end/80364
25567 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
25568 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
25570 (directive::set_width, directive::set_precision, format_character):
25572 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
25575 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
25578 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
25579 conflict, set target->arch_name instead of target->cpu_name.
25581 2017-04-11 Richard Biener <rguenther@suse.de>
25583 PR tree-optimization/80374
25584 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
25585 build_zero_cst, remove fold_convertible_p check again.
25587 2017-04-11 Martin Liska <mliska@suse.cz>
25590 * ubsan.c (instrument_object_size): Do not instrument register
25593 2017-04-11 Jakub Jelinek <jakub@redhat.com>
25596 * config/i386/i386-builtin-types.def
25597 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
25598 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
25599 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
25600 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
25601 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
25602 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
25603 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
25604 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
25605 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
25606 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
25607 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
25608 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
25609 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
25610 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
25611 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
25612 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
25613 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
25614 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
25615 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
25616 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
25617 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
25618 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
25619 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
25620 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
25621 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
25622 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
25623 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
25624 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
25625 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
25626 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
25627 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
25628 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
25629 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
25630 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
25631 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
25632 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
25633 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
25634 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
25635 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
25636 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
25637 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
25638 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
25639 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
25640 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
25641 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
25642 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
25644 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
25645 flag to second_arg_count, handle 4 argument function type _COUNT
25646 aliases, handle second_arg_count on second argument rather than last.
25648 2017-04-10 Jeff Law <law@redhat.com>
25650 PR tree-optimization/80374
25651 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
25652 record anything if we can not convert integer_zero_node to the
25655 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
25658 * config/rs6000/rs6000.c (rs6000_option_override_internal):
25659 Enhance special handling given to the TARGET_P9_MINMAX option in
25660 relation to certain other options.
25662 2017-04-10 Bin Cheng <bin.cheng@arm.com>
25664 PR tree-optimization/80153
25665 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
25666 remove POINTER_PLUS_EXPR's base part directly, rather than through
25669 2017-04-10 Richard Biener <rguenther@suse.de>
25670 Bin Cheng <bin.cheng@arm.com>
25672 PR tree-optimization/80153
25673 * tree-affine.c (aff_combination_to_tree): Get base pointer from
25674 the first element of pointer type aff_tree. Build result expr in
25676 (add_elt_to_tree): Convert to type unconditionally. Remove other
25677 fold_convert calls.
25678 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
25679 (rewrite_use_nonlinear_expr): Check invariant using iv information.
25681 2017-04-10 Richard Biener <rguenther@suse.de>
25683 * tree-ssa-structalias.c (find_func_aliases): Properly handle
25686 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
25688 PR rtl-optimization/70478
25689 * lra-constraints.c (curr_small_class_check): New.
25690 (update_and_check_small_class_inputs): New.
25691 (process_alt_operands): Update curr_small_class_check. Disfavor
25692 alternative insn memory operands. Check available regs for small
25695 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
25698 * config/mips/mips.opt (-mvirt): Update description.
25699 * doc/invoke.texi (-mvirt): Likewise.
25701 2017-04-10 Richard Biener <rguenther@suse.de>
25703 PR middle-end/80362
25704 * fold-const.c (fold_binary_loc): Look at unstripped ops when
25705 looking for NEGATE_EXPR in -A / -B to A / B folding.
25707 2017-04-10 Martin Liska <mliska@suse.cz>
25709 PR gcov-profile/80224
25710 * gcov.c (print_usage): Fix usage string.
25711 (get_gcov_intermediate_filename): Remove.
25712 (output_gcov_file): Use both for normal and intermediate format.
25713 (generate_results): Do not initialize special file for
25714 intermediate format.
25716 2017-04-10 Richard Biener <rguenther@suse.de>
25718 PR tree-optimization/80304
25719 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
25722 2017-04-10 Nathan Sidwell <nathan@acm.org>
25725 * config/rs6000/rs6000.c (rs6000_vector_type): New.
25726 (rs6000_init_builtins): Use it.
25728 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25730 * config/arm/arm.md (<mrc>): Add mode to SET source.
25731 (<mrrc>): Likewise.
25733 2017-04-10 Richard Biener <rguenther@suse.de>
25735 PR middle-end/80344
25736 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
25738 2017-04-10 Jakub Jelinek <jakub@redhat.com>
25741 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
25742 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
25743 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
25744 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
25745 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
25746 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
25747 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
25748 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
25749 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
25750 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
25751 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
25752 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
25753 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
25754 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
25755 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
25756 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
25757 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
25758 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
25759 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
25760 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
25761 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
25762 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
25763 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
25765 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
25767 PR rtl-optimization/70478
25768 * lra-constraints.c: Reverse the last patch.
25770 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
25772 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
25773 Add comment for WCHAR_T.
25775 2017-04-08 Martin Liska <mliska@suse.cz>
25778 2017-04-07 Martin Liska <mliska@suse.cz>
25781 * ipa-split.c (split_function): Add function part to a same comdat
25784 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25787 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
25789 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
25791 * rs6000/rs6000.c (vec_load_pendulum): Rename...
25792 (vec_pairing): ...to this.
25793 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
25794 (rs6000_sched_init): Adjust for name change.
25795 (struct rs6000_sched_context): Likewise.
25796 (rs6000_init_sched_context): Likewise.
25797 (rs6000_set_sched_context): Likewise.
25799 2017-04-07 Jakub Jelinek <jakub@redhat.com>
25805 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
25807 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
25808 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
25809 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
25811 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
25813 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
25815 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
25817 PR rtl-optimization/70703
25818 * ira-color.c (update_conflict_hard_regno_costs): Use
25819 int64_t instead of HOST_WIDE_INT.
25821 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
25823 PR rtl-optimization/70478
25824 * lra-constraints.c (process_alt_operands): Disfavor alternative
25825 insn memory operands.
25827 2017-04-07 Jeff Law <law@redhat.com>
25829 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
25830 CALL and NOTE_INSN_CALL_ARG_LOCATION.
25832 2017-04-07 Martin Liska <mliska@suse.cz>
25835 * config/aarch64/aarch64.c (aarch64_process_target_attr):
25836 Show error message instead of an ICE.
25838 2017-04-07 Martin Liska <mliska@suse.cz>
25841 * ipa-split.c (split_function): Add function part to a same comdat
25844 2017-04-07 Richard Biener <rguenther@suse.de>
25846 PR middle-end/80341
25847 * tree.c (get_unwidened): Also handle ! for_type case for
25849 * convert.c (do_narrow): Split out from ...
25850 (convert_to_integer_1): ... here. Do not pass final truncation
25851 type to get_unwidened for TRUNC_DIV_EXPR.
25853 2017-04-07 Richard Biener <rguenther@suse.de>
25855 * tree-affine.c (wide_int_ext_for_comb): Take type rather
25857 (aff_combination_const): Adjust.
25858 (aff_combination_scale): Likewise.
25859 (aff_combination_add_elt): Likewise.
25860 (aff_combination_add_cst): Likewise.
25861 (aff_combination_convert): Likewise.
25862 (add_elt_to_tree): Likewise. Remove unused argument.
25863 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
25865 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
25867 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
25869 * config/arm/arm.c (arm_default_short_enums): Use
25870 ARM_DEFAULT_SHORT_ENUMS.
25871 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
25873 2017-04-06 Jakub Jelinek <jakub@redhat.com>
25876 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
25877 members with redundant out-of-class redeclaration.
25879 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
25882 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
25883 * config/i386/i386.md (*zero_extendsidi2):
25884 Add (?*x,*x) and (?*v,*v) alternatives.
25886 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
25889 * config/i386/i386.c (ix86_expand_builtin)
25890 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
25891 mode from insn data. Convert operands to insn operand mode.
25892 Copy operands that don't satisfy insn predicate to a register.
25894 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
25896 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
25899 2017-04-06 Richard Biener <rguenther@suse.de>
25901 PR tree-optimization/80334
25902 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
25903 preserve alignment of accesses.
25905 2017-04-06 Richard Biener <rguenther@suse.de>
25907 PR tree-optimization/80262
25908 * tree-sra.c (build_ref_for_offset): Preserve address-space
25910 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
25911 Drop useless address-space information on MEM_REF offsets.
25913 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
25915 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
25917 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
25919 PR rtl-optimization/70703
25920 * ira-color.c (update_conflict_hard_regno_costs): Use
25921 HOST_WIDE_INT instead of long.
25923 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
25926 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
25927 not defined for x86_64 target. Add -mmmx target option when __SSE2__
25929 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
25930 for x86_64 target. Handle -m3dnowa option.
25932 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
25934 PR rtl-optimization/70703
25935 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
25936 (update_conflict_hard_regno_costs): Use long instead of unsigned
25937 arithmetic for cost calculation.
25939 2017-04-05 Jakub Jelinek <jakub@redhat.com>
25940 Bernd Edlinger <bernd.edlinger@hotmail.de>
25943 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
25946 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
25949 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
25950 ptr_mode with Pmode throughout.
25951 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
25952 into probe_stack_range and use DImode.
25954 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
25957 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
25958 call_eh_return is true.
25960 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25962 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
25963 Initialize last_match_fntype_index.
25965 2017-04-05 Jakub Jelinek <jakub@redhat.com>
25968 * tree-nvr.c: Include internal-fn.h.
25969 (pass_return_slot::execute): Ignore internal calls without
25972 2017-04-04 Jakub Jelinek <jakub@redhat.com>
25973 Richard Biener <rguenther@suse.de>
25976 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
25977 captures used multiple times, except for the last use.
25978 * generic-match-head.c: Include gimplify.h.
25980 2017-04-04 Jakub Jelinek <jakub@redhat.com>
25982 PR tree-optimization/79390
25983 * target.h (struct noce_if_info): Declare.
25984 * targhooks.h (default_noce_conversion_profitable_p): Declare.
25985 * target.def (noce_conversion_profitable_p): New target hook.
25986 * ifcvt.h (struct noce_if_info): New type, moved from ...
25987 * ifcvt.c (struct noce_if_info): ... here.
25988 (noce_conversion_profitable_p): Renamed to ...
25989 (default_noce_conversion_profitable_p): ... this. No longer
25991 (noce_try_store_flag_constants, noce_try_addcc,
25992 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
25993 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
25994 instead of noce_conversion_profitable_p.
25995 * config/i386/i386.c: Include ifcvt.h.
25996 (ix86_option_override_internal): Don't override
25997 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
25998 (ix86_noce_conversion_profitable_p): New function.
25999 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
26000 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
26001 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
26002 * doc/tm.texi: Regenerated.
26004 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26006 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
26009 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26012 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
26013 instructions for small multiply cores.
26015 2017-04-04 Jeff Law <law@redhat.com>
26017 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
26019 (mips_expand_vec_perm_const): Initialize elements in orig_perm
26020 that are not set by the loop over the elements.
26022 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26025 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
26026 int mode, convert_modes it to mode as unsigned, otherwise use
26027 lowpart_subreg to mode rather than SImode.
26028 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
26029 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
26030 Use DImode instead of SImode for the shift count operand.
26031 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
26034 2017-04-04 Richard Biener <rguenther@suse.de>
26036 PR middle-end/80281
26037 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
26038 arithmetic done for the negate or the plus. Simplify.
26039 (A - (-B) -> A + B): Likewise.
26040 * fold-const.c (split_tree): Make sure to not negate pointers.
26042 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
26044 PR rtl-optimization/60818
26045 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
26046 a compare of comparisons with the thing compared if this results
26047 in a different machine mode.
26049 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
26051 * alias.c (base_alias_check): Fix typo in comment.
26052 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
26053 * cgraphunit.c (symbol_table::compile): Likewise.
26054 * collect2.c (maybe_run_lto_and_relink): Likewise.
26055 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
26056 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
26057 * config/avr/avr.c (avr_map_op_t): Likewise.
26058 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
26059 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
26060 * config/epiphany/epiphany.md (movcc): Likewise.
26061 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
26062 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
26064 * config/mips/mips.c (mips_save_restore_reg): Likewise.
26065 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
26066 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
26067 * config/sh/sh.c (sh_rtx_costs): Likewise.
26068 * fold-const.c (fold_truth_andor): Likewise.
26069 * genautomata.c (collapse_flag): Likewise.
26070 * gengtype.h (struct type::u::s): Likewise.
26071 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
26072 * input.c (FORMAT_AMOUNT): Likewise.
26073 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
26074 (known_aggs_to_agg_replacement_list): Likewise.
26075 * ipa-inline-analysis.c: Likewise.
26076 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
26077 * ipa-polymorphic-call.c
26078 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
26079 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
26080 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
26082 * modulo-sched.c (apply_reg_moves): Likewise.
26083 * omp-expand.c (build_omp_regions_1): Likewise.
26084 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
26085 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
26086 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
26087 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
26088 * value-prof.c: Likewise.
26089 * var-tracking.c (val_reset): Likewise.
26091 2017-04-03 Richard Biener <rguenther@suse.de>
26093 PR tree-optimization/80275
26094 * fold-const.c (split_address_to_core_and_offset): Handle
26097 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
26099 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
26100 descriptors is at least equal to that of functions.
26102 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
26104 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
26106 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
26109 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
26110 (mov<IMOD4:mode>): New expander.
26111 (*mov<IMOD4:mode>_internal): New insn and split pattern.
26113 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
26115 PR rtl-optimization/79405
26116 * fwprop.c (propagations_left): New variable.
26117 (forward_propagate_into): Decrement it.
26118 (fwprop_init): Initialize it.
26119 (fw_prop): If the variable has reached zero, stop propagating.
26120 (fwprop_addr): Ditto.
26122 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26125 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
26126 a FUNCTION_DECL, pass it as decl instead of origin to
26129 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
26131 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
26134 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
26137 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
26138 TARGET_VSX_SMALL_INTEGER.
26140 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26142 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
26143 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
26145 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26147 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
26148 extraction from odd-numbered MSA register.
26150 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26152 PR middle-end/80173
26153 * expmed.c (store_bit_field_1): Don't attempt to create
26154 a word subreg out of hard registers wider than word if they
26155 have HARD_REGNO_NREGS of 1 for their mode.
26157 PR middle-end/80163
26158 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
26159 conversions to integer types wider than word and pointer.
26162 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
26163 (rtx_equal_for_cselib_p): Pass 0 to it.
26164 * cselib.c (cselib_hasher::equal): Likewise.
26165 (rtx_equal_for_cselib_1): Add depth argument. If depth
26166 is 128, don't look up VALUE locs and punt. Increment
26167 depth in recursive calls when walking VALUE locs.
26169 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
26171 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
26172 (make_gcov_file_name): Use the canonical path name for generating
26174 (read_line): Fix handling of files with ascii null bytes.
26176 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
26178 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
26179 to initialise a vector register instead
26180 of using a const_int.
26182 2017-03-30 Jakub Jelinek <jakub@redhat.com>
26184 PR translation/80189
26185 * gimplify.c (omp_default_clause): Use %qs instead of %s in
26186 diagnostic messages.
26188 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
26191 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
26192 (dfp_diex_<mode>): Update mode of operand 1.
26193 * doc/extend.texi (dxex, dxexq): Document change to return type.
26194 (diex, diexq): Document change to argument type.
26196 2017-03-30 Martin Jambor <mjambor@suse.cz>
26199 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
26200 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
26201 it reflects the signature changes performed at the callee side.
26202 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
26203 to cgraph_build_function_type_skip_args.
26204 (build_function_decl_skip_args): Adjust call to the above function.
26206 2017-03-30 Jakub Jelinek <jakub@redhat.com>
26209 * config/i386/sse.md
26210 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
26211 register as dest whenever it is a MEM not rtx_equal_p to the
26212 corresponding dup operand, and when forcing into reg move the
26213 reg into the memory afterwards.
26214 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
26215 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
26216 for the force_reg mode.
26217 (avx512vl_vextractf128<mode>): Use register as dest either
26218 always when a MEM, or when it is a MEM not rtx_equal_p to the
26219 corresponding dup operand, or even not when it is a CONST_VECTOR
26220 depending on the mode and lo vs. hi.
26221 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
26223 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
26224 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
26225 Likewise. Require that operands[2] is even.
26226 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
26227 Remove extraneous parens. Require that operands[2] is a multiple
26229 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
26230 operands[0] is a MEM if <mask_applied>, the predicates/constraints
26231 disallow memory then.
26233 2017-03-30 Richard Biener <rguenther@suse.de>
26235 PR tree-optimization/77498
26236 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
26237 to non-constants over backedges.
26239 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
26241 PR rtl-optimization/80233
26242 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
26243 as last_combined_insn. Do not test for BARRIER_P separately.
26245 2017-03-29 Andreas Schwab <schwab@suse.de>
26248 * calls.c (prepare_call_address): Convert funexp to Pmode before
26249 copying to temp reg.
26251 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26253 PR tree-optimization/80158
26254 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
26255 Handle possible future case of more than one alternate
26257 (replace_rhs_if_not_dup): Likewise.
26258 (replace_one_candidate): Likewise.
26260 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
26262 PR rtl-optimization/80193
26263 * ira.c (ira): Do not check allocation for LRA.
26265 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
26267 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
26268 (nvptx_output_simt_exit): Declare.
26269 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
26270 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
26271 (init_softstack_frame): Move initialization of crtl->is_leaf to...
26272 (nvptx_declare_function_name): ...here. Emit declaration of local
26273 memory space buffer for omp_simt_enter insn.
26274 (nvptx_output_unisimt_switch): New.
26275 (nvptx_output_softstack_switch): New.
26276 (nvptx_output_simt_enter): New.
26277 (nvptx_output_simt_exit): New.
26278 * config/nvptx/nvptx.h (struct machine_function): New fields
26279 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
26280 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
26281 (UNSPECV_SIMT_EXIT): Ditto.
26282 (omp_simt_enter_insn): New insn.
26283 (omp_simt_enter): New expansion.
26284 (omp_simt_exit): New insn.
26285 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
26287 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
26288 (expand_GOMP_SIMT_ENTER_ALLOC): New.
26289 (expand_GOMP_SIMT_EXIT): New.
26290 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
26291 (GOMP_SIMT_ENTER_ALLOC): Ditto.
26292 (GOMP_SIMT_EXIT): Ditto.
26293 * target-insns.def (omp_simt_enter): New insn.
26294 (omp_simt_exit): Ditto.
26295 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
26297 (lower_rec_simd_input_clauses): Implement SIMT privatization.
26298 (lower_rec_input_clauses): Likewise.
26299 (lower_lastprivate_clauses): Handle SIMT privatization.
26301 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
26302 (ompdevlow_adjust_simt_enter): New.
26303 (find_simtpriv_var_op): New.
26304 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
26305 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
26307 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
26308 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
26309 (copy_decl_for_dup_finish): Ditto.
26311 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
26313 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
26316 * config/i386/i386.c (ix86_option_override_internal): Always
26317 allow -mpreferred-stack-boundary=3 for 64-bit targets.
26319 2017-03-28 Bin Cheng <bin.cheng@arm.com>
26321 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
26323 2017-03-28 Bin Cheng <bin.cheng@arm.com>
26325 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
26326 mark new edge's irreducible flag accordign to it.
26327 (vect_do_peeling): Check loop preheader edge's irreducible flag
26328 and pass it to function slpeel_add_loop_guard.
26330 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
26332 PR tree-optimization/80218
26333 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
26334 Update block frequencies and counts.
26336 2017-03-28 Richard Biener <rguenther@suse.de>
26338 PR tree-optimization/78644
26339 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
26340 of a simplification result we may not use it at all.
26342 2017-03-28 Richard Biener <rguenther@suse.de>
26345 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
26346 without arguments, generate default definition of a SSA name.
26348 2017-03-28 Richard Biener <rguenther@suse.de>
26350 PR middle-end/80222
26351 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
26352 TYPE_REF_CAN_ALIAS_ALL references.
26353 * fold-const.c (fold_indirect_ref_1): Likewise.
26355 2017-03-28 Martin Liska <mliska@suse.cz>
26358 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
26359 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
26361 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
26362 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
26364 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
26365 (EXTRA_SPECS): Define.
26366 (SUBTARGET_EXTRA_SPECS): Likewise.
26367 (SUBTARGET_CPP_SPEC): Likewise.
26368 * config/arc/elf.h (EXTRA_SPECS): Renamed to
26369 SUBTARGET_EXTRA_SPECS.
26370 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
26372 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
26374 * config/arc/simdext.md (vst64_insn): Update pattern.
26375 (vld32wh_insn): Likewise.
26376 (vld32wl_insn): Likewise.
26377 (vld64_insn): Likewise.
26378 (vld32_insn): Likewise.
26380 2017-03-28 Marek Polacek <polacek@redhat.com>
26383 * fold-const.c (fold_comparison): Use protected_set_expr_location
26384 instead of SET_EXPR_LOCATION.
26386 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
26388 * tree.c (add_expr): Avoid name lookup warning.
26390 2017-03-27 Jeff Law <law@redhat.com>
26392 PR tree-optimization/80216
26393 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
26394 function name. Limit recursion depth.
26395 (record_temporary_equivalences): Corresponding changes.
26397 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
26399 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
26402 2017-03-27 Jakub Jelinek <jakub@redhat.com>
26405 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
26407 * cfgcleanup.c (reg_note_cfa_p): New array.
26408 (insns_have_identical_cfa_notes): New function.
26409 (old_insns_match_p): Don't cross-jump in between /f
26410 and non-/f instructions. If both i1 and i2 are frame related,
26411 verify all CFA notes, their order and content.
26413 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
26416 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
26417 HImode and SImode with zero extend to DImode to one insn.
26418 (bswap<mode>2_extenddi): Likewise.
26419 (bswapsi2_extenddi): Likewise.
26420 (bswaphi2_extendsi): Likewise.
26421 (bswaphi2): Combine bswap HImode and SImode into one insn.
26422 Separate memory insns from swapping register.
26423 (bswapsi2): Likewise.
26424 (bswap<mode>2): Likewise.
26425 (bswaphi2_internal): Delete, no longer used.
26426 (bswapsi2_internal): Likewise.
26427 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
26428 store, and gpr<-gpr swap insns.
26429 (bswap<mode>2_store): Likewise.
26430 (bswaphi2_reg): Register only splitter, combine with the splitter.
26431 (bswaphi2 splitter): Likewise.
26432 (bswapsi2_reg): Likewise.
26433 (bswapsi2 splitter): Likewise.
26434 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
26435 the insns into load, store, and register/register insns.
26436 (bswapdi2_ldbrx): Likewise.
26437 (bswapdi2_load): Likewise.
26438 (bswapdi2_store): Likewise.
26439 (bswapdi2_reg): Likewise.
26441 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
26443 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
26444 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
26446 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
26449 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
26451 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
26452 special handling for target option conflicts between dform
26453 options (-mpower9-dform, -mpower9-dform-vector,
26454 -mpower9-dform-scalar) and -mno-direct-move.
26456 2017-03-27 Richard Biener <rguenther@suse.de>
26458 PR tree-optimization/80181
26459 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
26461 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26463 * config/arc/predicates.md (move_double_src_operand): Replace the
26464 call to move_double_src_operand with a call to address_operand.
26466 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26468 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
26469 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
26470 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
26472 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26474 * config/arc/predicates.md (long_immediate_loadstore_operand):
26475 Consider scaled addresses cases.
26477 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26479 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
26480 restored when in interrupt.
26481 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
26482 doesn't have delay slot.
26484 2017-03-27 Richard Biener <rguenther@suse.de>
26487 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
26488 inlined thunk clones.
26490 2017-03-27 Jakub Jelinek <jakub@redhat.com>
26493 * asan.c (instrument_derefs): Copy over last operand from
26494 original COMPONENT_REF to the new COMPONENT_REF with
26495 DECL_BIT_FIELD_REPRESENTATIVE.
26496 * ubsan.c (instrument_object_size): Likewise.
26498 2017-03-27 Richard Biener <rguenther@suse.de>
26500 PR tree-optimization/80170
26501 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
26502 sure DR/SCEV didnt fold in constants we do not see when looking
26503 at the reference base alignment.
26505 2017-03-27 Richard Biener <rguenther@suse.de>
26507 PR middle-end/80171
26508 * gimple-fold.c (fold_ctor_reference): Properly guard against
26509 NULL return value from canonicalize_constructor_val.
26511 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
26514 * config/i386/i386.c (ix86_expand_builtin)
26515 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
26516 flags reg setting and flags reg using instructions.
26517 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
26518 clobbering instructions to zero extend op2.
26520 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
26522 * doc/install.texi (Configuration) <--with-aix-soname>:
26523 Update link to AIX ld.
26525 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
26527 PR rtl-optimization/80160
26528 PR rtl-optimization/80159
26529 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
26530 reg_alternate_class into account.
26532 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
26535 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
26536 to consider in curr_insn_transform.
26538 2017-03-24 Jakub Jelinek <jakub@redhat.com>
26540 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
26541 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
26542 and emit_mode_inner.
26544 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26546 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
26547 argument to the overloaded builtin variants. Use the new flag to
26548 deprecate certain builtin variants.
26549 * config/s390/s390-builtin-types.def: Add new builtin types.
26550 * config/s390/s390-builtins.h: Support new flags field for
26551 overloaded builtins.
26552 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
26553 (s390_macro_to_expand): Enable vector float data type.
26554 (s390_cpu_cpp_builtins_internal): Indicate support of the new
26555 builtins by incrementing the __VEC__ version number.
26556 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
26558 (s390_resolve_overloaded_builtin): Emit error messages depending
26559 on the builtin flags.
26560 * config/s390/s390.c (s390_expand_builtin): Support additional
26561 flags argument. Change error message to match the messages
26562 emitted in s390-c.c.
26563 * config/s390/s390.md: New UNSPEC_* constants.
26564 (op_type): Add new instruction types.
26565 * config/s390/vecintrin.h: Add new builtins and test data class
26567 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
26568 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
26569 (VEC_INEXACT, VEC_NOINEXACT): New constants.
26570 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
26571 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
26572 ("vec_mergel<mode>"): V_HW -> VEC_HW.
26574 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
26575 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
26576 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
26577 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
26579 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
26580 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
26581 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
26582 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
26584 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
26585 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
26586 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
26587 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
26588 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
26589 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
26590 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
26592 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
26593 ("vec_scatter_element<V_HW_4:mode>_DI")
26594 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
26595 ("vec_fpint<mode>", "vflls")
26596 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
26597 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
26598 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
26599 ("*vec_cmphe<mode>_cc"): ... these.
26601 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
26602 mode constant instead of magic value.
26604 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26606 * config/s390/s390.c (s390_expand_vec_compare): Support other
26607 vector floating point modes than just V2DF.
26608 (s390_expand_vcond): Likewise.
26609 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
26610 (s390_cannot_change_mode_class): Prevent mode changes between TF
26611 and V1TF in vector registers.
26612 * config/s390/s390.md (DF, SF): New mode attributes.
26613 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
26614 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
26615 SFmode support for VRs.
26616 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
26618 (VFT, VF_HW): New mode iterators.
26619 (vw, sdx): New mode attributes.
26620 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
26621 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
26622 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
26623 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
26624 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
26625 also the new vector floating point modes. Renaming to ...
26627 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
26628 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
26629 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
26630 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
26631 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
26632 ("vec_unordered<mode>"): ... these.
26634 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
26635 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
26636 ("*vec_extendv2df"): New insn definitions.
26638 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26640 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
26641 ("mulditi3_2", "*muldi3_sign"): New patterns.
26642 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
26643 rename the pattern definition.
26645 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26647 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
26649 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
26651 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26653 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
26654 instruction if possible.
26655 * config/s390/vector.md (vec_halfnumelts): New mode
26657 ("*vec_vllezlf<mode>"): New pattern.
26659 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26661 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
26662 ("popcountv4si2", "popcountv2di2"): Rename to ...
26663 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
26664 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
26666 ("popcount<mode>2_vxe"): New pattern.
26668 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26670 * common/config/s390/s390-common.c (processor_flags_table): Add
26672 * config.gcc: Add arch12.
26673 * config/s390/driver-native.c (s390_host_detect_local_cpu):
26674 Default to arch12 for unknown CPU model numbers.
26675 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
26676 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
26677 PROCESSOR_max sanity check.
26678 * config/s390/s390-opts.h (enum processor_type): Add
26680 * config/s390/s390.c (processor_table): Add arch12.
26681 (s390_expand_builtin): Add check for B_VXE flag.
26682 (s390_issue_rate): Add PROCESSOR_ARCH12.
26683 (s390_get_sched_attrmask): Likewise.
26684 (s390_get_unit_mask): Likewise.
26685 (s390_sched_score): Enable z13 scheduling for arch12.
26686 (s390_sched_reorder): Likewise.
26687 (s390_sched_variable_issue): Likewise.
26688 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
26690 (s390_tune_attr): Use z13 scheduling also for arch12.
26691 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
26692 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
26693 (TARGET_VXE_P): New macros.
26694 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
26695 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
26696 * config/s390/s390.opt: Add arch12 as processor_type.
26698 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26700 * config/s390/s390.md
26701 ("fixuns_truncdddi2", "fixuns_trunctddi2")
26702 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
26703 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
26705 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
26706 Rename expanders to ...
26708 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
26709 ("fixuns_truncdddi2_emu"): ... these.
26711 ("fixuns_trunc<mode>si2_emu"): New expander.
26713 ("*fixuns_truncdfdi2_z13"): Rename to ...
26714 ("*fixuns_truncdfdi2_vx"): ... this.
26716 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26718 * config/s390/2964.md: Remove the single element vector compare
26719 instructions which are no longer used.
26720 * config/s390/s390.c (s390_select_ccmode): Remove handling of
26722 (s390_canonicalize_comparison): Remove handling of DFmode
26724 (s390_expand_vec_compare_scalar): Remove function.
26725 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
26726 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
26728 ("*cmp<mode>_ccs"): Add wfcdb instruction.
26730 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26732 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
26734 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
26735 will anyway by matched by mov<mode>_64dfp.
26737 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26739 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
26740 vlef/vstef. Add missing operand to vleif.
26742 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26744 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
26745 pair for all vector types with 64 bit elements.
26746 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
26747 * config/s390/vector.md (V_HW_64): ... here.
26748 (V_128_NOSINGLE): New mode iterator.
26749 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
26750 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
26751 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
26752 ("*vec_load_pairv2di"): Change to ...
26753 ("*vec_load_pair<mode>"): ... this one.
26755 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26757 * config/s390/constraints.md: Add comments.
26758 (jKK): Reject element sizes > 8 bytes.
26759 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
26761 * config/s390/s390.md: Add the s_operand checks formerly in
26762 s390_split_ok_p to various splitters where they are still
26764 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
26765 for 128 bit vectors. Plus two splitters.
26767 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26769 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
26772 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26775 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
26776 error if the boundary argument is not constant.
26778 2017-03-24 Jakub Jelinek <jakub@redhat.com>
26780 PR rtl-optimization/80112
26781 * loop-doloop.c (doloop_condition_get): Don't check condition
26782 if cmp isn't SET with IF_THEN_ELSE src.
26784 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26786 PR tree-optimization/80158
26787 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
26788 replacing a candidate statement, also replace it for the
26789 candidate's alternate interpretation.
26790 (replace_rhs_if_not_dup): Likewise.
26791 (replace_one_candidate): Likewise.
26793 2017-03-24 Richard Biener <rguenther@suse.de>
26795 PR tree-optimization/80167
26796 * graphite-isl-ast-to-gimple.c
26797 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
26799 (translate_isl_ast_to_gimple::get_rename): Likewise.
26801 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
26803 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
26804 handling of certain combinations of target options, including the
26805 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
26806 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
26808 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26811 * config/arm/arm.md (*load_multiple): Add reload_completed to
26812 matching condition.
26814 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26815 Richard Biener <rguenth@suse.de>
26817 PR tree-optimization/79908
26818 PR tree-optimization/80136
26819 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
26820 been cast away, gimplify_and_add suffices.
26822 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
26824 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
26826 2017-03-23 Richard Biener <rguenther@suse.de>
26828 PR tree-optimization/80032
26829 * gimplify.c (gimple_push_cleanup): Forced unconditional
26830 cleanups still have to go to the conditional_cleanups
26833 2017-03-22 Jakub Jelinek <jakub@redhat.com>
26835 PR tree-optimization/80072
26836 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
26838 (next_operand_entry_id): Change type to unsigned int.
26839 (sort_by_operand_rank): Make sure to return the right return value
26840 even if unsigned fields are bigger than INT_MAX.
26841 (struct oecount): Change cnt and id type to unsigned int.
26842 (oecount_hasher::equal): Formatting fix.
26843 (oecount_cmp): Make sure to return the right return value
26844 even if unsigned fields are bigger than INT_MAX.
26845 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
26848 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
26849 TREE_READONLY on result if writing it more than once.
26852 * doc/invoke.texi (-fsanitize=thread): Document that with
26853 -fnon-call-exceptions atomics are not able to throw
26857 * tsan.c: Include tree-eh.h.
26858 (instrument_builtin_call): Call maybe_clean_eh_stmt or
26859 maybe_clean_or_replace_eh_stmt where needed.
26860 (instrument_memory_accesses): Add cfg_changed argument.
26861 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
26862 if it returned true.
26863 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
26865 PR rtl-optimization/63191
26866 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
26867 wrapper function, moved the whole old content into ...
26868 (ix86_delegitimize_address_1): ... this. New inline function.
26869 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
26870 true as last argument instead of ix86_delegitimize_address.
26872 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
26874 * config/aarch64/aarch64.c (generic_branch_cost): Copy
26875 cortexa57_branch_cost.
26877 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
26879 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
26881 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26884 * doc/md.texi (Constraints): Document wA constraint.
26885 * config/rs6000/constraints.md (wA): New.
26886 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
26887 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
26888 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
26889 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
26891 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
26894 * gimplify.c (is_oacc_declared): New function.
26895 (oacc_default_clause): Use it to set default flags for acc declared
26896 variables inside parallel regions.
26897 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
26898 declared variables.
26899 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
26900 declare attribute to any decl as necessary.
26902 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
26905 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
26906 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
26907 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
26908 (arm_arch_lpae): This.
26909 * config/arm/arm.c (arm_arch7ve): Rename into ...
26910 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
26911 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
26914 2017-03-22 Martin Liska <mliska@suse.cz>
26917 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
26918 error message instead of an ICE.
26920 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26922 * doc/extend.texi (6.11 Additional Floating Types): Revise.
26924 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
26926 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
26928 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
26931 2017-03-21 Martin Sebor <msebor@redhat.com>
26933 * doc/extend.texi: Use "cannot" instead of "can't."
26934 * doc/hostconfig.texi: Same.
26935 * doc/install.texi: Same.
26936 * doc/invoke.texi: Same.
26937 * doc/loop.texi: Same.
26938 * doc/md.texi: Same.
26939 * doc/objc.texi: Same.
26940 * doc/rtl.texi: Same.
26941 * doc/tm.texi: Same.
26942 * doc/tm.texi.in: Same.
26943 * doc/trouble.texi: Same.
26945 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
26948 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
26949 (collect_checksum_attributes): Set it.
26950 (die_checksum_ordered): Use it.
26952 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26954 PR tree-optimization/79908
26955 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
26956 change: For a VA_ARG whose LHS has been cast away, use
26957 force_gimple_operand to construct the side effects.
26959 2017-03-21 David Malcolm <dmalcolm@redhat.com>
26961 PR translation/80001
26962 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
26963 more amenable to translation.
26964 (oacc_loop_auto_partitions): Likewise.
26966 2017-03-21 Marek Polacek <polacek@redhat.com>
26967 Martin Sebor <msebor@redhat.com>
26969 PR tree-optimization/80109
26970 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
26971 on INTEGRAL_TYPE_P.
26973 2017-03-21 Jakub Jelinek <jakub@redhat.com>
26974 Segher Boessenkool <segher@kernel.crashing.org>
26977 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
26978 check reg_used_between_p between insn and one of succ or succ2
26979 depending on if succ is artificial insn not inserted into insn
26982 2017-03-21 Martin Liska <mliska@suse.cz>
26984 PR gcov-profile/80081
26985 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
26986 * doc/gcc.texi: Include gcov-dump stuff.
26987 * doc/gcov-dump.texi: New file.
26989 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
26991 PR rtl-optimization/79150
26992 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
26993 conditional jump, if the jump is the last insn of the loop.
26995 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26996 Richard Biener <rguenth@suse.de>
26998 PR tree-optimization/79908
26999 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27000 been cast away, use force_gimple_operand to construct the side
27003 2017-03-21 Martin Liska <mliska@suse.cz>
27005 PR libfortran/79956
27006 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
27009 2017-03-21 Brad Spengler <spender@grsecurity.net>
27012 * plugin.c (htab_hash_plugin): New function.
27013 (add_new_plugin): Use it and adjust.
27014 (parse_plugin_arg_opt): Adjust.
27015 (init_one_plugin): Likewise.
27017 2017-03-21 Richard Biener <rguenther@suse.de>
27019 PR tree-optimization/80032
27020 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
27021 if set force the cleanup to happen unconditionally.
27022 (gimplify_target_expr): Push inserted clobbers with force_uncond
27023 to avoid them being removed by control-dependent DCE.
27025 2017-03-21 Richard Biener <rguenther@suse.de>
27027 PR tree-optimization/80122
27028 * tree-inline.c (copy_bb): Do not expans va-arg packs or
27029 va_arg_pack_len when the inlined call stmt requires pack
27031 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
27033 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27036 * tsan.c (instrument_builtin_call): If the memory model argument
27037 is not a constant, assume it is valid.
27040 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
27043 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27045 PR rtl-optimization/79910
27046 * combine.c (can_combine_p): Do not allow combining an I0 or I1
27047 if its dest is used by an insn before I2 (other than the combined
27048 insns themselves, which are properly handled already).
27050 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27053 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27055 * combine.c (record_used_regs): New static function.
27056 (try_combine): Handle situations where there is an additional
27057 instruction between I2 and I3 which needs to have a LOG_LINK
27061 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27063 * combine.c (try_combine): Delete redundant i1 test. Call
27064 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27066 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27069 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
27070 alternatives 13/14.
27072 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27074 PR tree-optimization/80054
27075 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
27076 the optimization if a PHI or any of its arguments is not dominated
27077 by the candidate's basis. Use gphi* rather than gimple* as
27079 (replace_profitable_candidates): Clean up a gimple* variable that
27080 should be a gphi* variable.
27082 2017-03-20 Martin Sebor <msebor@redhat.com>
27085 * doc/extend.texi (attribute constructor): Document present limitation.
27087 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
27090 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
27091 __POWER9_VECTOR__ #ifdef control, change template definition to
27092 use Power9-specific built-in function.
27093 (vec_any_eq): Likewise.
27094 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
27095 to control outcomes from this test.
27096 (vector_ae_<mode>p): For VEC_F modes, likewise.
27098 2017-03-20 Ian Lance Taylor <iant@google.com>
27100 * config/i386/i386.c (ix86_function_regparm): Save an extra
27101 register for -fsplit-stack with DECL_STATIC_CHAIN.
27103 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
27106 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
27107 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
27109 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
27111 * config/riscv/riscv.c (riscv_print_operand): Use "fence
27113 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
27116 2017-03-20 Marek Polacek <polacek@redhat.com>
27119 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
27121 2017-03-20 Richard Biener <rguenther@suse.de>
27123 PR tree-optimization/80113
27124 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
27125 allocate extra SSA name for PHI def.
27126 (add_close_phis_to_outer_loops): Likewise.
27127 (add_close_phis_to_merge_points): Likewise.
27128 (copy_loop_close_phi_args): Likewise.
27129 (copy_cond_phi_nodes): Likewise.
27131 2017-03-20 Martin Liska <mliska@suse.cz>
27133 PR middle-end/79753
27134 * tree-chkp.c (chkp_build_returned_bound): Do not build
27135 returned bounds for a LHS that's not a BOUNDED_P type.
27137 2017-03-20 Martin Liska <mliska@suse.cz>
27141 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
27142 COMPLEX_CST and VECTOR_CST.
27144 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27147 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
27148 target operand. A new splitter adds the clobber statement in case
27149 the target operand is dead anyway.
27151 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
27153 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
27154 to age-old versions of binutils and glibc.
27156 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
27158 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
27160 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27162 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
27164 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27166 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
27167 requirement for binutils 2.13.
27169 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27171 * combine.c (try_combine): Delete redundant i1 test. Call
27172 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27174 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
27176 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
27177 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
27179 <riscv64-*-elf>: Re-arrange section
27180 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
27181 <riscv32-*-linux>: Likewise.
27182 <riscv64-*-elf>: Likewise
27183 <riscv64-*-linux>: Likewise.
27185 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
27188 * aarch64.opt(verbose-cost-dump): Fix typo.
27190 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
27193 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
27194 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
27196 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27198 * reload.c (find_reloads): When reloading a nonoffsettable address,
27199 use RELOAD_OTHER for it and its address reloads.
27201 PR rtl-optimization/79910
27202 * combine.c (record_used_regs): New static function.
27203 (try_combine): Handle situations where there is an additional
27204 instruction between I2 and I3 which needs to have a LOG_LINK
27207 2017-03-17 Jeff Law <law@redhat.com>
27209 PR tree-optimization/71437
27210 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
27211 conditional in the hash table first.
27212 (vrp_dom_walker::before_dom_children): Extract condition from
27213 ASSERT_EXPR. Record condition, its inverion and any implied
27214 conditions as well.
27216 2017-03-17 Marek Polacek <polacek@redhat.com>
27217 Markus Trippelsdorf <markus@trippelsdorf.de>
27219 PR tree-optimization/80079
27220 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
27223 2017-03-17 Richard Biener <rguenther@suse.de>
27225 PR middle-end/80075
27226 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
27227 Properly verify the LHS before the RHS possibly claims to be
27229 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
27232 2017-03-17 Martin Jambor <mjambor@suse.cz>
27234 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
27235 (List of -O2 options): Likewise.
27236 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
27239 2017-03-17 Tom de Vries <tom@codesourcery.com>
27241 * gcov-dump.c (print_usage): Print bug_report_url.
27243 2017-03-17 Richard Biener <rguenther@suse.de>
27245 PR middle-end/80050
27246 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
27247 (parser::peek): Likewise.
27249 2017-03-17 Richard Biener <rguenther@suse.de>
27251 PR tree-optimization/80048
27252 * sese.c (free_sese_info): Properly release rename_map and
27253 copied_bb_map elements.
27255 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
27257 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
27258 Add linked-list forward and backlinks. Insert on
27259 construction, remove on destruction.
27260 (class pass_store_merging): Add m_stores_head field.
27261 (pass_store_merging::terminate_and_process_all_chains):
27262 Iterate over m_stores_head list.
27263 (pass_store_merging::terminate_all_aliasing_chains):
27265 (pass_store_merging::execute): Check for debug stmts first.
27266 Push new chains onto the m_stores_head stack.
27268 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
27271 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
27272 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
27273 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
27275 2017-03-16 Jeff Law <law@redhat.com>
27277 PR tree-optimization/71437
27278 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
27279 member function. Implementation moved into after_dom_children
27280 member function and into the threader's thread_outgoing_edges
27282 (dom_opt_dom_walker::after_dom_children): Simplify by moving
27283 some code into new thread_outgoing_edges.
27284 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
27285 definition. Simplify marker handling (do it here). Assume we always
27286 have the available expression and the const/copies tables.
27287 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
27289 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
27290 * tree-vrp.c (equiv_stack): No longer file scoped.
27291 (vrp_dom_walker): New class.
27292 (vrp_dom_walker::before_dom_children): New member function.
27293 (vrp_dom_walker::after_dom_children): Likewise.
27294 (identify_jump_threads): Setup domwalker. Use it rather than
27295 walking edges in a random order by hand. Simplify setup/finalization.
27296 (finalize_jump_threads): Remove.
27297 (vrp_finalize): Do not call identify_jump_threads here.
27298 (execute_vrp): Do it here instead and call thread_through_all_blocks
27301 PR tree-optimization/71437
27302 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
27304 (simplify_stmt_for_jump_threading): Add basic_block argument. All
27306 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
27307 (dom_opt_dom_walker::thread_across_edge): Remove
27308 handle_dominating_asserts argument. All callers changed.
27309 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
27310 changes. Remove calls to lhs_of_dominating_assert. Other
27311 uses of handle_dominating_asserts turn into unconditional code
27312 (simplify_control_stmt_condition_1): Likewise.
27313 (simplify_control_stmt_condition): Likewise.
27314 (thread_through_normal_block, thread_across_edge): Likewise.
27315 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
27316 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
27317 object if it is not an SSA_NAME.
27318 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
27319 before calling into the VRP specific simplifiers.
27320 (identify_jump_threads): Remove handle_dominating_asserts
27323 2017-03-16 Jakub Jelinek <jakub@redhat.com>
27326 * tree-diagnostic.c (default_tree_printer): No longer static.
27327 * tree-diagnostic.h (default_tree_printer): New prototype.
27329 2017-03-16 Tamar Christina <tamar.christina@arm.com>
27331 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
27332 Change ins into fmov.
27334 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27336 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
27337 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
27338 Use h_con constraint for operand 1.
27339 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
27340 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
27342 2017-03-15 Jeff Law <law@redhat.com>
27344 PR tree-optimization/71437
27345 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
27346 (record_temporary_equivalences): Use it.
27348 PR tree-optimization/71437
27349 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
27350 tree-ssa-scopedtables.
27351 (lookup_avail_expr, build_and_record_new_cond): Likewise.
27352 (record_conditions, record_cond, vuse_eq): Likewise.
27353 (record_edge_info): Adjust to API tweak of record_conditions.
27354 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
27355 (record_temporary_equivalences, optimize_stmt): Likewise.
27356 (eliminate_redundant_computations): Likewise.
27357 (record_equivalences_from_stmt): Likewise.
27358 * tree-ssa-scopedtables.c: Include options.h and params.h.
27359 (vuse_eq): New function, moved from tree-ssa-dom.c
27360 (build_and_record_new_cond): Likewise.
27361 (record_conditions): Likewise. Accept vector of conditions rather
27362 than edge_equivalence structure for first argument.
27363 for the first argument.
27364 (avail_exprs_stack::lookup_avail_expr): New member function, moved
27365 from tree-ssa-dom.c.
27366 (avail_exprs_stack::record_cond): Likewise.
27367 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
27368 from tree-ssa-dom.c.
27369 (avail_exprs_stack): Add new member functions lookup_avail_expr
27371 (record_conditions): Declare.
27373 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
27376 * lra-constraints.c (process_alt_operands): Increase reject for
27377 reloading an input/output operand.
27379 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
27382 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
27383 insns to convert from signed/unsigned char/short to IEEE 128-bit
27385 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
27387 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
27390 * config/i386/i386.c (ix86_vector_duplicate_value): Create
27391 subreg of inner mode for values already in registers.
27393 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
27395 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
27396 iteration reg is used after the loop.
27398 2017-03-14 Martin Sebor <msebor@redhat.com>
27400 PR tree-optimization/79800
27401 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
27402 precision in negative-positive range.
27403 (format_floating): Call non-const overload with adjusted precision.
27405 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
27408 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
27411 2017-03-14 Martin Sebor <msebor@redhat.com>
27413 PR middle-end/80020
27414 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
27415 * builtins.def (aligned_alloc): Use it.
27418 * Makefile.in (GTFILES): Add calls.c.
27419 * calls.c: Include "gt-calls.h".
27421 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
27423 PR rtl-optimization/79728
27424 * regs.h (struct target_regs): New field
27425 x_contains_allocatable_regs_of_mode.
27426 (contains_allocatable_regs_of_mode): New macro.
27427 * reginfo.c (init_reg_sets_1): Initialize it, and change
27428 contains_reg_of_mode so it includes global regs as well.
27429 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
27430 rather than contains_regs_of_mode.
27432 2017-03-14 Martin Liska <mliska@suse.cz>
27434 * doc/invoke.texi: Document options that can't be combined with
27435 -fcheck-pointer-bounds.
27437 2017-03-14 Martin Liska <mliska@suse.cz>
27439 PR middle-end/79831
27440 * doc/invoke.texi (-Wchkp): Document the option.
27442 2017-03-14 Martin Liska <mliska@suse.cz>
27444 * Makefile.in: Install gcov-dump.
27446 2017-03-14 Martin Liska <mliska@suse.cz>
27448 * multiple_target.c (expand_target_clones): Bail out for
27449 an invalid attribute.
27451 2017-03-14 Richard Biener <rguenther@suse.de>
27453 * alias.c (struct alias_set_entry): Pack properly.
27454 * cfgloop.h (struct loop): Likewise.
27455 * cse.c (struct set): Likewise.
27456 * ipa-utils.c (struct searchc_env): Likewise.
27457 * loop-invariant.c (struct invariant): Likewise.
27458 * lra-remat.c (struct cand): Likewise.
27459 * recog.c (struct change_t): Likewise.
27460 * rtl.h (struct address_info): Likewise.
27461 * symbol-summary.h (function_summary): Likewise.
27462 * tree-loop-distribution.c (struct partition): Likewise.
27463 * tree-object-size.c (struct object_size_info): Likewise.
27464 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
27465 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
27466 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
27467 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
27468 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
27469 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
27470 (struct _stmt_vec_info): Likewise.
27472 2017-03-14 Martin Liska <mliska@suse.cz>
27475 * multiple_target.c (create_dispatcher_calls): Check that
27476 a target can create a function dispatcher.
27478 2017-03-14 Martin Liska <mliska@suse.cz>
27481 * multiple_target.c (expand_target_clones): Drop local.local
27482 flag for default implementation.
27484 2017-03-14 Richard Biener <rguenther@suse.de>
27486 PR tree-optimization/80030
27487 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
27489 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
27491 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
27492 gcc_fallthrough() instead of __attribute__((fallthrough));
27494 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
27496 * doc/gcc.texi: Remove "up" link to (DIR).
27497 * doc/gccint.texi: Ditto.
27499 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
27501 * doc/install.texi (Specific) <avr>: Remove reference to
27504 2017-03-13 Jeff Law <law@redhat.com>
27506 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
27507 attribute rather than comments.
27509 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
27510 match_scratch operand is highest.
27512 2017-03-13 Martin Liska <mliska@suse.cz>
27514 PR middle-end/78339
27515 * ipa-pure-const.c (warn_function_noreturn): If the declarations
27516 is a CHKP clone, use original declaration.
27518 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27520 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
27521 (arc_conditional_register_usage): Use a different allocation order
27522 when optimizing for size.
27523 * common/config/arc/arc-common.c (arc_option_optimization_table):
27524 Section anchors default on when optimizing for size.
27526 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27528 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
27530 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27532 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
27533 * config/arc/arc.md (cpu_facility): Add cd variant.
27534 (*movqi_insn): Add code density variant.
27535 (*movhi_insn): Likewise.
27536 (*movqi_insn): Likewise.
27537 (*addsi3_mixed): Likewise.
27538 (subsi3_insn): Likewise.
27540 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27542 * config/arc/arc.md (movsi_cond_exec): Update constraint.
27544 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27546 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
27547 expressions with MINUS and UNARY ops.
27549 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27552 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
27554 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
27555 between vec_select and vector argument.
27556 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
27557 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
27558 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
27559 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
27560 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
27561 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
27563 2017-03-13 Richard Biener <rguenther@suse.de>
27566 * params.def (vect-max-peeling-for-alignment): Fix typo.
27568 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
27570 * doc/install.texi (Specific) <mips-*-*>: Remove description of
27571 issue that only occurred with binutils below 2.18.
27573 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
27575 * doc/install.texi (Specific) <cris-axis-elf>: No longer
27576 refer to binutils 2.11/2.12 minimum.
27578 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
27580 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
27581 ftp.kernel.org and simplify binutils requirement.
27583 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
27585 * doc/invoke.texi (Warning Options): Fix spelling of link-time
27587 (Optimize Options): Ditto. Also remove redundancy.
27589 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27591 PR translation/79848
27592 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
27594 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
27595 to G_ to avoid double translation.
27597 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27599 PR translation/79923
27600 * auto-profile.c (get_combined_location): Convert leading
27601 character of diagnostics to lower case and remove trailing period.
27602 (read_profile): Likewise for various diagnostics.
27603 * config/arm/arm.c (arm_option_override): Remove trailing period
27604 from various diagnostics.
27605 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
27606 (msp430_expand_delay_cycles): Likewise.
27608 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27611 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
27612 full command-line argument, rather than just "str".
27613 (aarch64_validate_march): Likewise.
27614 (aarch64_validate_mtune): Likewise.
27616 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
27618 PR rtl-optimization/78911
27619 * lra-assigns.c (must_not_spill_p): New function.
27620 (spill_for): Use it.
27622 2017-03-10 Jakub Jelinek <jakub@redhat.com>
27624 PR tree-optimization/79981
27625 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
27626 ATOMIC_COMPARE_EXCHANGE ifn result.
27627 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
27628 IFN_ATOMIC_COMPARE_EXCHANGE.
27630 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27633 * opts.c (parse_sanitizer_options): Add missing question mark to
27634 "did you mean" message.
27636 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27638 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
27640 (VMULEUH_UNS): Likewise.
27641 (VMULOUB_UNS): Likewise.
27642 (VMULOUH_UNS): Likewise.
27643 * config/rs6000/rs6000.c (builtin_function_type): Remove
27644 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
27646 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27649 * read-rtl-function.c (function_reader::read_rtx_operand): Update
27650 x with result of extra_parsing_for_operand_code_0.
27651 (function_reader::extra_parsing_for_operand_code_0): Convert
27652 return type from void to rtx, returning x. When reading
27653 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
27654 larger size containing struct block_symbol.
27656 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
27658 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
27659 -mfloat128-hardware without -m64.
27661 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
27664 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
27665 entries to the case statement that marks unsigned arguments to
27666 overloaded functions.
27668 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
27670 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
27671 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
27673 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
27676 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
27677 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
27679 2017-03-10 Martin Liska <mliska@suse.cz>
27683 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
27684 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
27687 2017-03-10 Olivier Hainque <hainque@adacore.com>
27689 * tree-switch-conversion (array_value_type): Start by resetting
27690 candidate type to it's main variant.
27692 2017-03-10 Jakub Jelinek <jakub@redhat.com>
27694 PR rtl-optimization/79909
27695 * combine.c (try_combine): Use simplify_replace_rtx on individual
27696 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
27697 of the whole CALL_INSN_FUNCTION_USAGE.
27699 PR tree-optimization/79972
27700 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
27701 get_range_info on SSA_NAMEs. Formatting fixes.
27703 2017-03-10 Richard Biener <rguenther@suse.de>
27704 Jakub Jelinek <jakub@redhat.com>
27706 PR tree-optimization/77975
27707 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
27708 edge to be constant.
27709 (get_val_for): For constant x return it. Formatting fix.
27710 (loop_niter_by_eval): Avoid pointless looping if the next iteration
27711 would use the same bases as the current one.
27713 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27715 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
27716 instead of vec_select for V1TImode.
27717 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
27719 (VSX_LE_128): Add V1TI to this mode iterator.
27720 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
27721 (*vsx_le_perm_store_<mode>): Likewise.
27722 (pre-reload splitter for VSX stores): Likewise.
27723 (post-reload splitter for VSX stores): Likewise.
27724 (*vsx_xxpermdi2_le_<mode>): Likewise.
27725 (*vsx_lxvd2x2_le_<mode>): Likewise.
27726 (*vsx_stxvd2x2_le_<mode>): Likewise.
27728 2017-03-09 Michael Eager <eager@eagercon.com>
27730 Correct failures with --enable-checking=yes,rtl.
27732 * config/microblaze/microblaze.c (microblaze_expand_shift):
27733 Replace GET_CODE test with CONST_INT_P and INTVAL test with
27734 test for const0_rtx.
27735 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
27736 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
27738 2017-03-09 Richard Biener <rguenther@suse.de>
27740 PR tree-optimization/79977
27741 * graphite-scop-detection.c (scop_detection::merge_sese):
27742 Handle the case of extra exits to blocks dominating the entry.
27744 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
27746 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
27749 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
27751 PR rtl-optimization/79949
27752 * lra-constraints.c (process_alt_operands): Check memory when
27753 trying to predict a cycle. Print about the overall increase.
27755 2017-03-09 Richard Biener <rguenther@suse.de>
27757 PR middle-end/79971
27758 * gimple-expr.c (useless_type_conversion_p): Preserve
27759 TYPE_SATURATING for fixed-point types.
27761 2017-03-09 Richard Biener <rguenther@suse.de>
27764 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
27765 alignment of BLKmode params.
27767 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27770 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
27771 (VALL_NO_V2Q): Likewise.
27773 * config/aarch64/aarch64-simd.md
27774 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
27776 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
27777 VALL_NO_V2Q mode iterator.
27778 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
27780 2017-03-09 Martin Liska <mliska@suse.cz>
27782 PR tree-optimization/79631
27783 * tree-chkp-opt.c (chkp_is_constant_addr): Call
27784 tree_int_cst_sign_bit just for INTEGER constants.
27786 2017-03-09 Martin Liska <mliska@suse.cz>
27790 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
27793 2017-03-09 Marek Polacek <polacek@redhat.com>
27796 * tree.c (inchash::add_expr): Handle TREE_VEC.
27798 2017-03-09 Martin Liska <mliska@suse.cz>
27801 (chkp_narrow_size_and_offset): New function.
27802 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
27803 (void chkp_parse_bit_field_ref): New function.
27804 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
27805 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
27807 2017-03-09 Martin Liska <mliska@suse.cz>
27810 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
27811 (chkp_find_bounds_1): Remove gcc_unreachable.
27813 2017-03-09 Jakub Jelinek <jakub@redhat.com>
27816 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
27817 BUILT_IN_SYNC*, determine the access type from the size suffix and
27818 always build a MEM_REF with that type. Handle forgotten
27819 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
27822 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
27823 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
27824 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
27825 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
27826 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
27827 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
27828 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
27829 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
27830 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
27831 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
27832 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
27833 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
27834 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
27835 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
27836 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
27837 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
27838 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
27839 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
27840 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
27841 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
27842 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
27843 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
27844 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
27845 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
27846 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
27847 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
27848 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
27849 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
27850 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
27851 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
27852 definitions outside of __OPTIMIZE__ guarded section.
27855 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
27856 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
27857 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
27858 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
27861 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27863 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
27864 ("vfenez<mode>"): Add missing constraints.
27866 2017-03-08 Martin Sebor <msebor@redhat.com>
27869 * config/nds32/nds32.c (nds32_option_override):
27870 Fix misspelled diagnostic.
27872 2017-03-08 Jakub Jelinek <jakub@redhat.com>
27875 * gimplify.c (gimplify_omp_for): Replace index var in outer
27876 taskloop statement with an artificial variable and add
27877 OMP_CLAUSE_PRIVATE clause for it.
27879 2017-03-08 Richard Biener <rguenther@suse.de>
27881 PR tree-optimization/79955
27882 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
27883 for accesses that are completely outside of the variable.
27885 2017-03-08 Andrew Haley <aph@redhat.com>
27887 PR tree-optimization/79943
27888 * tree-ssa-loop-split.c (compute_new_first_bound): When
27889 calculating the new upper bound, (END-BEG) should be added, not
27892 2017-03-08 Jakub Jelinek <jakub@redhat.com>
27894 * config/avr/avr.md (setmemhi): Make sure match_dup
27895 operand number comes before match_scratch.
27897 2017-03-08 Richard Biener <rguenther@suse.de>
27899 PR tree-optimization/79920
27900 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
27901 with ncopies == 1 to ...
27902 (vect_transform_slp_perm_load): ... here. Properly compute
27903 all element loads by iterating VF times over the group. Do
27904 not handle ncopies (computed in a broken way) in
27905 vect_create_mask_and_perm.
27907 2017-03-08 Jakub Jelinek <jakub@redhat.com>
27910 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
27911 is a uniform vector, use uniform_vector_p return value instead of
27912 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
27914 2017-03-07 Marek Polacek <polacek@redhat.com>
27916 PR middle-end/79809
27917 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
27918 (alloca_call_type): Likewise.
27920 2017-03-07 Martin Liska <mliska@suse.cz>
27922 * gcov.c (process_args): Put comment to correct location.
27924 2017-03-07 Martin Liska <mliska@suse.cz>
27926 PR middle-end/68270
27927 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
27928 Use array_at_struct_end_p instead of DECL_CHAIN (field).
27929 (chkp_narrow_bounds_for_field): Likewise.
27930 (chkp_parse_array_and_component_ref): Pass one more argument to
27933 2017-03-07 Richard Biener <rguenther@suse.de>
27935 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
27938 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
27940 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
27941 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
27943 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27946 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
27947 to end of description.
27948 (PARAM_MAX_STORES_TO_MERGE): Likewise.
27950 2017-03-07 Jakub Jelinek <jakub@redhat.com>
27952 PR rtl-optimization/79901
27953 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
27955 (*avx512f_<code><mode>3<mask_name>): ... this.
27956 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
27957 iterator instead of VI8_AVX2_AVX512BW.
27959 PR rtl-optimization/79901
27960 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
27961 min/max expander, expand it using expand_vec_cond_expr.
27964 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
27967 2017-03-06 Jakub Jelinek <jakub@redhat.com>
27970 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
27971 to void * for PCH reasons.
27972 * dwarf2out.c (output_loc_operands, output_die): Cast
27973 v.val_vec.array to unsigned char *.
27975 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
27978 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
27981 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
27983 PR rtl-optimization/79571
27984 * lra-constraints.c (process_alt_operands): Calculate static
27985 reject and subtract it from overall when only addresses will be
27988 2017-03-06 Julia Koval <julia.koval@intel.com>
27991 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
27992 incoming stack boundary to 128 for 64-bit targets.
27994 2017-03-06 Richard Biener <rguenther@suse.de>
27996 PR tree-optimization/79894
27997 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
27998 to NULL after folding it.
28000 2017-03-06 Richard Biener <rguenther@suse.de>
28002 PR tree-optimization/79824
28003 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
28004 check disabling peeling for gaps.
28006 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
28008 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28009 attributes): Document gettimeofday.
28011 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28013 * config/s390/s390.c (s390_option_override_internal): Set
28014 PARAM_MIN_VECT_LOOP_BOUND
28016 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28018 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
28019 * config/s390/s390.md: Likewise.
28021 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28024 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
28025 (<avx2_avx512>_perm<mode>): Rename to ...
28026 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
28028 (<avx512>_perm<mode>_mask): Rename to ...
28029 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
28031 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
28032 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
28033 instead of VI8F_256_512.
28034 (avx512f_perm<mode>): New define_expand.
28035 (avx512f_perm<mode>_mask): Likewise.
28036 (avx512f_perm<mode>_1<mask_name>): New define_insn.
28037 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
28039 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28041 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
28042 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
28044 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
28046 2017-03-06 Martin Liska <mliska@suse.cz>
28049 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
28050 when having a SSA NAME w/o VAR_DECL assigned to it.
28052 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28054 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
28055 msa_dpsub_<su>_d): Fix MODE for vec_select.
28057 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28059 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
28061 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
28063 2017-03-06 Richard Biener <rguenther@suse.de>
28065 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
28066 * plugin.c (register_plugin_info): Likewise.
28067 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
28069 2017-03-05 Jakub Jelinek <jakub@redhat.com>
28071 * config/i386/sse.md (sse_storehps, sse_storelps,
28072 avx_<castmode><avxsizesuffix>_<castmode>,
28073 avx512f_<castmode><avxsizesuffix>_<castmode>,
28074 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
28075 in condition that at least one operand is not a MEM.
28077 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28079 PR middle-end/79805
28080 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
28081 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
28083 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
28084 gimple_call_nothrow_p flag based on whether original builtin can throw.
28085 If it can, emit following stmts on the fallthrough edge.
28086 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
28087 don't create new bb if inserting just debug stmts on the edge, try to
28088 insert them on the fallthru bb or just reset debug stmts.
28090 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
28093 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
28094 restore recog_data (including the operand rtxes inside it) around
28095 the call to get_insn_template.
28097 2017-03-03 Martin Sebor <msebor@redhat.com>
28099 PR tree-optimization/79699
28100 * context.c (context::~context): Free MPFR caches to avoid
28101 a memory leak on program exit.
28103 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28105 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
28106 Use wide_int::ulow () instead of .elt (0).
28108 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28110 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
28111 (*pushxf): Limit oF constraint to 32bit targets and add oC
28112 constraint for 64bit targets.
28113 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
28114 (*pushdf): Change rmF constraint to rmC.
28116 2017-03-03 Martin Liska <mliska@suse.cz>
28118 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
28119 Remove unused variable.
28121 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28124 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
28125 is a memory operand, increase num_memory.
28126 (ix86_expand_args_builtin): Likewise.
28128 2017-03-03 Jan Hubicka <jh@suse.cz>
28131 * ipa-devirt.c (maybe_record_node): Properly handle
28132 __cxa_pure_virtual visibility.
28134 2017-03-03 Martin Liska <mliska@suse.cz>
28136 PR tree-optimization/79803
28137 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
28139 (pass_loop_prefetch::execute): Disabled optimization if an
28140 assumption about L1 cache size is not met.
28142 2017-03-03 Martin Liska <mliska@suse.cz>
28144 PR rtl-optimization/79574
28145 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
28146 (hash_scan_set): Likewise.
28147 (dump_hash_table): Likewise.
28148 (hoist_code): Likewise.
28150 2017-03-03 Richard Biener <rguenther@suse.de>
28152 * fixed-value.c (fixed_from_string): Restore use of elt (1)
28153 in place of uhigh ().
28154 (fixed_convert_from_real): Likewise.
28156 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28159 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
28161 2017-03-03 Richard Biener <rguenther@suse.de>
28163 PR middle-end/79818
28164 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
28165 TYPE_OVERFLOW_UNDEFINED check.
28167 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28169 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
28171 (vector_ae_<mode>_p): Likewise.
28172 (vector_nez_<mode>_p): Likewise.
28173 (vector_ne_v2di_p): Likewise.
28174 (vector_ae_v2di_p): Likewise.
28175 (vector_ne_<mode>_p): Likewise.
28176 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
28178 (vsx_tsqrt<mode>2_fe): Likewise.
28180 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
28183 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
28185 2017-03-02 Jakub Jelinek <jakub@redhat.com>
28187 PR rtl-optimization/79780
28188 * cprop.c (one_cprop_pass): When second and further conditional trap
28189 in a single basic block is turned into an unconditional trap, turn it
28190 into a deleted note to avoid RTL verification failures.
28192 2017-03-02 Richard Biener <rguenther@suse.de>
28194 * fold-const.c (const_binop): Use ulow () instead of elt (0).
28196 2017-03-02 Richard Biener <rguenther@suse.de>
28198 PR tree-optimization/79345
28200 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
28201 param and abort the walk, returning -1 if it is hit.
28202 (walk_aliased_vdefs): Take a limit param and pass it on.
28203 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
28204 defaulting to 0 and return a signed int.
28205 * tree-ssa-uninit.c (struct check_defs_data): New struct.
28206 (check_defs): New helper.
28207 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
28208 about uninitialized memory.
28209 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
28210 bogus uninitialized warning.
28211 (fixed_convert_from_real): Likewise.
28213 2017-03-02 Bin Cheng <bin.cheng@arm.com>
28215 PR tree-optimization/66768
28216 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
28217 iv_use if base object can't be determined.
28219 2017-03-02 Jakub Jelinek <jakub@redhat.com>
28221 PR tree-optimization/79345
28222 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
28223 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
28224 (get_pattern_stats): Initialize it.
28225 * genemit.c (gen_expand): Verify match_scratch numbers come after
28226 match_operand/match_dup numbers.
28227 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
28228 match_scratch numbers.
28229 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
28231 * config/s390/s390.md (trunctdsd2): Likewise.
28233 2017-03-02 Richard Biener <rguenther@suse.de>
28235 * wide-int.h (wide_int_storage::operator=): Implement in terms
28238 2017-03-02 Richard Biener <rguenther@suse.de>
28240 PR tree-optimization/79777
28241 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
28242 the to insert expression to sth existing.
28244 2017-03-01 Martin Sebor <msebor@redhat.com>
28246 PR middle-end/79692
28247 * gimple-ssa-sprintf.c
28248 (directive::known_width_and_precision): New function.
28249 (format_integer): Use it.
28250 (get_mpfr_format_length): Consider the full range of precision
28251 when computing %g output with the # flag. Set the likely byte
28252 count to 3 rather than 1 when precision is indeterminate.
28253 (format_floating): Correct the lower bound of precision.
28255 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28257 * doc/invoke.texi: Document default code model for 64-bit Linux.
28259 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28262 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
28263 udiv rather than div since input pattern is unsigned.
28265 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
28267 * config/i386/i386.c (print_reg): Warn for values of
28268 unsupported size in integer register.
28270 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
28273 * config/rs6000/predicates.md (current_file_function_operand): Do
28274 not allow self calls to be local if the function is replaceable.
28276 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
28279 * config/rs6000/altivec.h (vec_ctz and others): Change the
28280 preprocessor macro that controls conditional compilation from
28281 _ARCH_PWR9 to __POWER9_VECTOR__.
28282 (vec_all_ne): Change parameterization of __altivec_scalar_pred
28283 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
28284 control (instead of _ARCH_PWR9 control) so that template
28285 definition uses power9-specific function.
28286 (vec_any_eq): Likewise.
28287 (vec_all_ne): Change macro definition to use a power9-specific
28288 expansion under #ifdef __POWER9_VECTOR__ control (instead of
28289 _ARCH_PWR9 control).
28290 (vec_any_eq) Likewise.
28291 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
28292 expansion for CMPNEF to remove support for xvcmpnesp instruction.
28293 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
28294 support for xvcmpnedp instruction.
28295 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
28296 macro expansion so that Power9 implementation of vec_all_ne does
28297 not use the AltiVec predicate framework.
28298 (VCMPNEH_P): Likewise.
28299 (VCMPNEW_P): Likewise.
28300 (VCMPNED_P): Likewise.
28301 (VCMPNEFP_P): Likewise.
28302 (VCMPNEDP_P): Likewise.
28303 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
28304 implementation of vec_any_eq to not use AltiVec predicate
28306 (VCMPAEH_P): Likewise.
28307 (VCMPAEW_P): Likewise.
28308 (VCMPAED_P): Likewise.
28309 (VCMPAEFP_P): Likewise.
28310 (VCMPAEDP_P): Likewise.
28311 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
28312 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
28313 not use the AltiVec predicate framework.
28314 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
28315 of vec_any_eq to not use AltiVec predicate framework.
28316 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
28317 support for predefined __POWER9_VECTOR__ macro to indicate that
28318 Power9 instruction selection is enabled.
28319 (altivec_overloaded_builtins): Remove extraneous
28320 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
28321 function argument types RS6000_BTI_bool_V16QI and
28322 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
28323 entry for overloaded function argument types RS6000_BTI_bool_V4SI
28324 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
28325 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
28326 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
28327 Power9 for implementations of vec_cmpne. Change the signature for
28328 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
28329 (representing vec_all_ne) to remove the previously described first
28330 argument of type RS6000_BTI_INTSI, as this was an artifact of
28331 reliance on the AltiVec predicate framework, which is no longer
28332 used in the implementation of these functions. Add
28333 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
28334 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
28335 since, unlike the AltiVec predicate framework implementation, we
28336 do not share function descriptors between vec_alle and vec_anyeq.
28337 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
28338 set of modes that receive special treatment even when
28339 TARGET_P9_VECTOR is true. The special treatment emits code that
28340 does not depend on Power9 instructions.
28341 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
28342 define_expand to not rely on AltiVec predicate framework.
28343 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
28345 (vector_ne_v2di_p): Change this define_expand to not rely on
28346 AltiVec predicate framework.
28347 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
28349 (vector_ne_<mode>_p): Change this define_expand to not rely on
28350 AltiVec predicate framework.
28351 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
28353 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
28354 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
28355 define_insn pattern.
28356 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
28357 define_insn pattern because the xvcmpne<VSs>. instruction is not
28359 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
28360 instruction is not supported.
28362 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28364 * config/nvptx/nvptx.c: Include intl.h.
28366 2017-03-01 Martin Jambor <mjambor@suse.cz>
28369 * ipa-prop.h (ipa_bits): Removed field known.
28370 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
28371 to pointers. Adjusted their comments to warn about their sharing.
28372 (ipcp_transformation_summary): Change bits to a vector of pointers.
28373 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
28374 (ipa_get_ipa_bits_for_value): Declare.
28375 * tree-vrp.h (value_range): Mark as GTY((for_user)).
28376 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
28377 (ipa_bits_hash_table): Likewise.
28378 (ipa_vr_ggc_hash_traits): Likewise.
28379 (ipa_vr_hash_table): Likewise.
28380 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
28381 being pointers and vr_known being removed.
28382 (ipa_set_jf_unknown): Likewise.
28383 (ipa_get_ipa_bits_for_value): New function.
28384 (ipa_set_jfunc_bits): Likewise.
28385 (ipa_get_value_range): New overloaded functions.
28386 (ipa_set_jfunc_vr): Likewise.
28387 (ipa_compute_jump_functions_for_edge): Use the above functions to
28388 construct bits and vr parts of jump functions.
28389 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
28390 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
28392 (ipcp_grow_transformations_if_necessary): Also allocate
28393 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
28395 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
28396 them. Fix too long lines.
28397 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
28398 vr_known being removed.
28399 (ipa_read_jump_function): Use new setter functions to construct bits
28400 and vr parts of jump functions or set them to NULL.
28401 (write_ipcp_transformation_info): Adjust for bits being pointers.
28402 (read_ipcp_transformation_info): Likewise.
28403 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
28405 Include gt-ipa-prop.h.
28406 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
28408 (ipcp_store_bits_results): Likewise.
28409 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
28410 Do not write to existing jump functions but use a temporary instead.
28412 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28415 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
28416 attempt to use its first operand as BIT_FIELD_REF base.
28418 2017-03-01 Richard Biener <rguenther@suse.de>
28420 PR middle-end/79721
28421 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
28422 interpolating formula in wrapping arithmetic.
28423 (chrec_apply): Convert chrec_evaluate return value to wanted type.
28425 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28427 PR tree-optimization/79734
28428 * tree-vect-generic.c (expand_vector_condition): Optimize
28429 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
28430 Handle VEC_COND_EXPR where comparison has different inner width from
28431 type's inner width.
28433 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
28435 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
28436 markup, and similar issues. Remove @opindex entries for things
28437 that aren't options. Add missing -mmpy-option entries.
28439 2017-02-28 Jakub Jelinek <jakub@redhat.com>
28441 PR tree-optimization/79737
28442 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
28443 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
28444 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
28445 instead of byte_size. Formatting fix.
28446 (shift_bytes_in_array_right): Formatting fix.
28448 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
28451 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
28452 condition on optimize for the leaf function test.
28454 2017-02-28 Martin Liska <mliska@suse.cz>
28457 * read-rtl-function.c (function_reader::handle_unknown_directive):
28458 Bail out when one uses -flto.
28460 2017-02-28 Martin Liska <mliska@suse.cz>
28462 * common.opt: Replace space with tabular for options of <number>
28464 * config/i386/i386.opt: Show <number> value for
28465 -mlarge-data-threshold.
28466 * opts.c (print_filtered_help): Do not display number in hexadecimal
28469 2017-02-28 Martin Liska <mliska@suse.cz>
28471 * common.opt: Fix --help=option -Q for options which are of
28474 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
28476 * config/i386/i386.c (print_reg): Error out for values
28477 of 8-bit size in invalid integer register.
28479 2017-02-28 Martin Sebor <msebor@redhat.com>
28481 PR tree-optimization/79691
28482 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
28484 2017-02-28 Jakub Jelinek <jakub@redhat.com>
28487 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
28488 gcc_unreachable with output_operand_lossage.
28490 2017-02-28 Richard Biener <rguenther@suse.de>
28492 PR tree-optimization/79740
28493 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
28495 (visit_nary_op): Insert the nary into the hashtable if we
28496 pattern-matched sth.
28497 * tree-ssa-pre.c (eliminate_insert): Robustify.
28499 2017-02-28 Richard Biener <rguenther@suse.de>
28501 PR middle-end/79731
28502 * fold-const.c (decode_field_reference): Reject out-of-bound
28505 2017-02-28 Jakub Jelinek <jakub@redhat.com>
28507 * config/i386/i386.c: Include intl.h.
28508 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
28509 instead of just cond ? "..." : "...".
28510 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
28511 * coverage.c (read_counts_file): Likewise.
28512 * omp-offload.c: Include intl.h.
28513 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
28514 of just cond ? "..." : "...".
28515 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
28516 of just cond ? "..." : "...".
28518 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
28521 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
28523 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
28524 'tune for' CPU name.
28525 * config/arm/arm-cpu-data.h: Regenerated.
28527 2017-02-28 Richard Biener <rguenther@suse.de>
28529 PR tree-optimization/79732
28530 * tree-inline.c (expand_call_inline): Do not shadow var.
28532 2017-02-28 Richard Biener <rguenther@suse.de>
28534 PR tree-optimization/79723
28535 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
28536 address-space properly.
28538 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
28540 * doc/optinfo.texi (Optimization groups): Fix option used for
28542 * doc/invoke.texi (-fopt-info): Document "omp".
28543 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
28544 (OPTGROUP_ALL): Add OPTGROUP_OMP.
28545 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
28546 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
28547 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
28549 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
28551 * dumpfile.c (optgroup_options): Instead of "openmp", associate
28552 OPTGROUP_OMP with "omp".
28554 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
28557 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
28558 for arithmetic shift of unsigned V2DI.
28560 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
28562 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
28563 arc/linux.h headers.
28564 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
28565 (LINK_SPEC): Likewise.
28566 (ARC_TLS_EXTRA_START_SPEC): Likewise.
28567 (EXTRA_SPECS): Likewise.
28568 (STARTFILE_SPEC): Likewise.
28569 (ENDFILE_SPEC): Likewise.
28570 (LIB_SPEC): Likewise.
28571 (TARGET_SDATA_DEFAULT): Likewise.
28572 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
28573 (MULTILIB_DEFAULTS): Likewise.
28574 (DWARF2_UNWIND_INFO): Likewise.
28575 * config/arc/big.h: New file.
28576 * config/arc/elf.h: Likewise.
28577 * config/arc/linux.h: Likewise.
28578 * config/arc/t-uClibc: Remove.
28580 2017-02-27 Bin Cheng <bin.cheng@arm.com>
28582 PR tree-optimization/77536
28583 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
28584 (tree_transform_and_unroll_loop): Use above function to compute the
28585 estimated niter of unrolled loop and use it when scaling profile.
28586 Also use count info rather than frequency if it's non-zero.
28587 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
28588 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
28589 (vect_transform_loop): Call above function.
28591 2017-02-27 Richard Biener <rguenther@suse.de>
28593 PR tree-optimization/45397
28594 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
28595 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
28596 (visit_nary_op): Add pattern matching for CSEing sign-changed
28597 or truncated operations with wider ones.
28599 2017-02-27 Richard Biener <rguenther@suse.de>
28601 PR tree-optimization/79690
28602 * tree-vect-stmts.c (vectorizable_store): Use vector type
28603 built from the DR with address-space.
28605 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
28607 * doc/invoke.texi (Optimize Options): Refine the description
28608 of asan-use-after-return.
28610 2017-02-25 Alan Modra <amodra@gmail.com>
28612 PR rtl-optimization/79584
28613 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
28614 base, not ad->base_term, the reg within base. Remove assertion
28615 that ad->base == ad->base_term. Replace gen_int_mode using
28616 bogus mode with const0_rtx.
28618 2017-02-25 Jakub Jelinek <jakub@redhat.com>
28620 PR middle-end/79396
28621 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
28622 FMA_EXPR like tcc_binary or tcc_unary.
28624 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
28627 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
28629 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
28630 (output_loc_operands): Handle DW_OP_call_ref and
28631 DW_OP_GNU_variable_value.
28632 (struct variable_value_struct): New type.
28633 (struct variable_value_hasher): Likewise.
28634 (variable_value_hash): New variable.
28635 (string_types): Remove.
28636 (copy_loc_descr): New function.
28637 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
28638 (prepend_loc_descr_to_each): New function.
28639 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
28640 instead of add_loc_descr_to_each if the first argument is single
28641 location list and the second has multiple.
28642 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
28643 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
28644 when looking for variable value which doesn't have other location info.
28645 (loc_list_from_tree): Formatting fix.
28646 (gen_array_type_die): Simplify DW_AT_string_length handling.
28647 (adjust_string_types): Remove.
28648 (gen_subprogram_die): Don't call adjust_string_types nor test/set
28649 string_types. Call resolve_variable_values.
28650 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
28651 (resolve_addr_in_expr): Likewise. Add A argument.
28652 (copy_deref_exprloc): Remove deref argument. Adjust for the
28653 original expression being DW_OP_GNU_variable_value with optionally
28654 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
28655 optionally after it.
28656 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
28657 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
28658 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
28659 (variable_value_hasher::hash, variable_value_hasher::equal): New
28661 (resolve_variable_value_in_expr, resolve_variable_value,
28662 resolve_variable_values, note_variable_value_in_expr,
28663 note_variable_value): New functions.
28664 (dwarf2out_early_finish): Call note_variable_value on all toplevel
28667 2017-02-24 Jakub Jelinek <jakub@redhat.com>
28670 * opts.h (handle_generated_option): Add GENERATED_P argument.
28671 * opts-common.c (handle_option): Adjust function comment.
28672 (handle_generated_option): Add GENERATED_P argument, pass it to
28674 (control_warning_option): Pass false to handle_generated_option
28676 * opts.c (maybe_default_option): Pass true to handle_generated_option
28678 * optc-gen.awk: Likewise.
28680 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28682 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
28683 a REG, look at the REG it is a SUBREG of.
28684 (splitter for cmpeqsi_t): Ditto.
28686 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28688 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
28689 the special USEs with the pattern of the insn, not the insn itself.
28691 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
28694 * doc/invoke.texi: Document -mload-store-pairs.
28696 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28697 Sandra Loosemore <sandra@codesourcery.com>
28699 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
28700 argument isn't a CONST_INT.
28701 (nios2_alternate_compare_const): Assert op is a CONST_INT.
28702 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
28703 (nios2_validate_compare): Bypass alternate compare logic if *op2
28704 is not a CONST_INT.
28705 (ldstwm_operation_p): Return false if first_base is not a REG or
28706 if first_offset is not a CONST_INT.
28708 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28710 * config/cris/cris.md: Use correct operand in a define_peephole2.
28712 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28714 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
28716 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28718 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
28719 this_insn if it is an INSN or JUMP_INSN.
28720 (force_offsettable): Look at base, not at addr.
28721 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
28722 on things that aren't necessarily CONST_INTs.
28724 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
28726 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
28727 -mfpmath=sse is the default also for x86-32 targets with SSE2
28728 instruction set when @option{-ffast-math} is enabled
28730 2017-02-24 Jeff Law <law@redhat.com>
28732 PR rtl-optimizatoin/79286
28733 * ira.c (update_equiv_regs): Drop may_trap_p exception to
28736 2017-02-24 Richard Biener <rguenther@suse.de>
28738 PR tree-optimization/79389
28739 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
28742 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
28744 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
28745 function comment to reflect reality.
28746 (loop_exits_before_overflow): Fix typo in function description.
28748 2017-02-24 Richard Biener <rguenther@suse.de>
28750 PR tree-optimization/79389
28751 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
28752 properly that a threading opportunity exists. Detect conditional
28753 copy/constant propagation opportunities.
28755 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
28757 * config/visium/visium.md (type): Add trap.
28758 (b): New mode attribute.
28759 (*btst): Rename into...
28760 (*btst<mode>): ...this and adjust.
28761 (*cbranchsi4_btst_insn): Rename into...
28762 (*cbranch<mode>4_btst_insn): ...this and adjust.
28763 (trap): New define_insn.
28765 2017-02-23 Jakub Jelinek <jakub@redhat.com>
28767 PR tree-optimization/79389
28768 * ifcvt.c (struct noce_if_info): Add rev_cond field.
28769 (noce_reversed_cond_code): New function.
28770 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
28771 reversed_comparison_code. Formatting fix.
28772 (noce_try_store_flag): Test rev_cond != NULL in addition to
28773 reversed_comparison_code.
28774 (noce_try_store_flag_constants): Likewise.
28775 (noce_try_store_flag_mask): Likewise.
28776 (noce_try_addcc): Use rev_cond if non-NULL instead of
28777 reversed_comparison_code.
28778 (noce_try_cmove_arith): Likewise. Formatting fixes.
28779 (noce_try_minmax, noce_try_abs): Clear rev_cond.
28780 (noce_find_if_block): Initialize rev_cond.
28781 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
28782 instead of false as last argument never attempt to reverse it
28785 2017-02-23 Bin Cheng <bin.cheng@arm.com>
28787 PR tree-optimization/79663
28788 * tree-predcom.c (combine_chains): Process refs in reverse order
28789 only for ZERO length chains, and add explaining comment.
28791 2017-02-23 Jeff Law <law@redhat.com>
28793 PR tree-optimization/79578
28794 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
28795 in call to operand_equal_p.
28797 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
28800 * config/i386/cpuid.h: Fix another undefined behavior.
28802 2017-02-23 Richard Biener <rguenther@suse.de>
28804 PR tree-optimization/79683
28805 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
28806 vector types for data-refs.
28808 2017-02-23 Martin Liska <mliska@suse.cz>
28810 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
28812 2017-02-23 Jakub Jelinek <jakub@redhat.com>
28814 PR middle-end/79665
28815 * internal-fn.c (get_range_pos_neg): Moved to ...
28816 * tree.c (get_range_pos_neg): ... here. No longer static.
28817 * tree.h (get_range_pos_neg): New prototype.
28818 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
28819 are known to be in between 0 and signed maximum inclusive, try to
28820 expand both unsigned and signed divmod and use the cheaper one from
28823 2017-02-22 Jeff Law <law@redhat.com>
28825 PR tree-optimization/79578
28826 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
28827 to compare base operands.
28829 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
28832 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
28833 gpc_reg_operand instead of fpr_reg_operand.
28835 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
28837 * config/mips/mips.c (mips_return_in_memory): Force FP
28838 vector types to be returned in memory for o32 ABI.
28840 2017-02-22 Jakub Jelinek <jakub@redhat.com>
28842 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
28843 instead of DW_TAG_member for static data member declarations and don't
28844 set no_linkage_name for static inline data members.
28845 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
28848 2017-02-22 Martin Liska <mliska@suse.cz>
28850 * doc/invoke.texi: Replace inequality signs with square brackets
28853 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
28856 * lra-constraints.c (simplify_operand_subreg): Handle
28857 WORD_REGISTER_OPERATIONS targets.
28859 2017-02-22 Jakub Jelinek <jakub@redhat.com>
28862 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
28863 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
28864 elimination by swapping fld*.
28866 2017-02-22 Richard Biener <rguenther@suse.de>
28868 PR tree-optimization/79673
28869 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
28870 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
28871 irrelevant address-space qualifiers and avoiding a
28872 ADDR_SPACE_CONVERT_EXPR from fold_convert.
28874 2017-02-22 Richard Biener <rguenther@suse.de>
28876 PR tree-optimization/79666
28877 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
28878 to not symbolically negate if that may introduce undefined
28881 2017-02-22 Martin Liska <mliska@suse.cz>
28884 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
28885 * data-streamer-out.c (streamer_write_gcov_count_stream):
28887 * value-prof.c (stream_out_histogram_value): Make assert more
28888 precise based on type of counter.
28890 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
28893 * config/i386/i386.md (standard_x87sse_constant_load splitter):
28894 Use nonimmediate_operand instead of memory_operand for operand 1.
28895 (float-extend standard_x87sse_constant_load splitter): Ditto.
28897 2017-02-21 Jeff Law <law@redhat.com>
28899 PR tree-optimization/79621
28900 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
28901 blocks with edges to themselves.
28903 2017-02-21 Jakub Jelinek <jakub@redhat.com>
28906 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
28907 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
28908 Use gimple_call_builtin_p.
28911 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
28912 on temporarily removed DEBUG_INSNs.
28914 PR tree-optimization/79649
28915 * tree-loop-distribution.c (classify_partition): Give up on
28916 non-generic address space loads/stores.
28918 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
28920 * doc/loop.texi (Loop manipulation): Remove nonexistent
28921 tree_ssa_loop_version from the documentation.
28922 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
28924 2017-02-21 Jakub Jelinek <jakub@redhat.com>
28927 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
28928 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
28929 * config/rs6000/rs6000.c: Include except.h.
28930 (rs6000_expand_split_stack_prologue): Call
28931 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
28933 2017-02-21 Martin Jambor <mjambor@suse.cz>
28936 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
28937 have been analyzed.
28939 2017-02-21 Martin Jambor <mjambor@suse.cz>
28941 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
28942 for backward compatibility only.
28943 * doc/invoke.texi (Option Summary): Remove all references to
28944 -fipa-cp-alignment.
28946 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
28950 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
28952 * lra-constraints.c (curr_insn_transform): Handle
28953 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
28955 2017-02-21 Martin Liska <mliska@suse.cz>
28957 * config/i386/i386.opt: Replace -masm-dialect with -masm.
28959 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
28961 PR translation/79638
28962 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
28964 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
28967 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
28968 (arm_function_ok_for_sibcall): Return false for an indirect call by
28969 descriptor if all the argument registers are used.
28970 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
28971 alignment of the function.
28973 2017-02-21 Jakub Jelinek <jakub@redhat.com>
28975 PR tree-optimization/61441
28976 * simplify-rtx.c (simplify_const_unary_operation): For
28977 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
28978 the sNaN unmodified.
28980 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
28982 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
28983 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
28984 instead of SYSTEM_HEADER_DIR.
28986 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
28987 Martin Liška <mliska@suse.cz>
28989 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
28990 Fix typos and grammar, use active voice, and clarify.
28992 2017-02-20 Marek Polacek <polacek@redhat.com>
28994 PR middle-end/79537
28995 * gimplify.c (gimplify_expr): Handle unused *&&L;.
28998 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
29000 2017-02-20 Jakub Jelinek <jakub@redhat.com>
29003 * config/i386/i386.c (ix86_expand_builtin): Handle
29004 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
29005 ix86_builtins_isa[fcode].isa as a requirement of those
29006 flags and any other flag in the bitmask.
29007 (ix86_init_mmx_sse_builtins): Use 0 instead of
29008 ~OPTION_MASK_ISA_64BIT as mask.
29009 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
29010 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
29011 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
29012 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
29014 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29017 * lra-constraints.c (split_reg): Check requested split mode
29018 is supported by the register.
29020 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29022 * lra-constraints.c (simplify_operand_subreg): Remove early
29025 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29028 * lra-constraints.c (curr_insn_transform): Tighten condition
29029 for converting SUBREG reloads from OP_OUT to OP_INOUT.
29031 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29034 * lra-constraints.c (curr_insn_transform): Handle
29035 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29037 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
29040 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
29042 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
29044 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
29047 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
29050 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29052 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
29055 2017-02-18 Jakub Jelinek <jakub@redhat.com>
29057 * final.c (last_columnnum, override_columnnum): New variables.
29058 (final_start_function): Set last_columnnum, pass it to begin_prologue
29059 hook and pass 0 to dwarf2out_begin_prologue.
29060 (final_scan_insn): Update override_columnnum. Pass last_columnnum
29061 to source_line debug hook.
29062 (notice_source_line): Compute last_columnnum and for debug_column_info
29063 return true on column changes.
29064 * debug.h (struct gcc_debug_hooks): Add column argument to
29065 source_line and begin_prologue hooks.
29066 (debug_nothing_int_charstar_int_bool): Remove prototype.
29067 (debug_nothing_int_int_charstar,
29068 debug_nothing_int_int_charstar_int_bool): New prototypes.
29069 (dwarf2out_begin_prologue): Add column argument.
29070 * debug.c (do_nothing_debug_hooks): Adjust source_line and
29071 begin_prologue hooks.
29072 (debug_nothing_int_charstar_int_bool): Remove.
29073 (debug_nothing_int_int_charstar,
29074 debug_nothing_int_int_charstar_int_bool): New functions.
29075 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
29076 through to dwarf2out_source_line.
29077 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
29078 (dwarf2out_source_line): Add column argument, emit it if requested.
29079 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
29081 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29082 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29083 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
29084 through to dwarf2out_begin_prologue.
29085 (vmsdbgout_source_line): Add column argument, pass it through to
29086 dwarf2out_source_line.
29087 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
29088 dbxout_source_line caller.
29089 (dbxout_source_line): Add column argument.
29091 * common.opt (gno-column-info, gcolumn-info): New options.
29092 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
29093 (check_die): Also test for multiple DW_AT_decl_column attributes.
29094 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
29095 DW_AT_decl_column if requested.
29096 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
29098 (gen_variable_die): Likewise.
29099 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
29100 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
29103 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
29104 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
29105 (ix86_handle_option): Handle OPT_m3dnowa.
29106 * doc/invoke.texi (-m3dnowa): Document.
29107 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
29108 -m3dnowa instead of -m3dnow -march=athlon.
29111 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
29112 instead of gcc_assert for K, r and R code checks. Formatting fixes.
29114 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29117 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
29118 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
29119 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
29120 generator for vsx_xxpermdi_<mode>_be.
29121 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
29122 force big-endian semantics.
29123 (vsx_xxpermdi_<mode>_be): New define_expand with same
29124 implementation as previous version of vsx_xxpermdi_<mode>.
29126 2017-02-17 Jakub Jelinek <jakub@redhat.com>
29128 PR tree-optimization/79327
29129 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
29130 variable, its initialization and use.
29132 2017-02-17 Julia Koval <julia.koval@intel.com>
29134 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
29135 (OPTION_MASK_ISA_PKU_UNSET): New.
29136 (ix86_handle_option): Handle -mrdpid.
29137 * config/i386/cpuid.h (bit_RDPID): New.
29138 * config/i386/driver-i386.c (host_detect_local_cpu):
29139 Detect RDPID feature.
29140 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
29141 * config/i386/i386-c.c (ix86_target_macros_internal):
29143 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
29144 (ix86_valid_target_attribute_inner_p): Add "rdpid".
29145 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
29146 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
29147 * config/i386/i386.md (define_insn "rdpid"): New.
29148 * config/i386/i386.opt Add -mrdpid.
29149 * config/i386/immintrin.h (_rdpid_u32): New.
29151 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
29153 PR rtl-optimization/79541
29154 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
29155 instead of transforming it into USE.
29157 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
29159 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
29160 If HONOR_SNANS (SFmode) force the input to a register.
29161 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
29162 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
29163 an frsp or similar insn.
29165 2017-02-17 Martin Liska <mliska@suse.cz>
29167 PR rtl-optimization/79577
29168 * params.def (selsched-max-sched-times): Increase minimum to 1.
29170 2017-02-17 Martin Liska <mliska@suse.cz>
29172 PR rtl-optimization/79574
29173 * gcse.c (want_to_gcse_p): Prevent integer overflow.
29175 2017-02-17 Martin Liska <mliska@suse.cz>
29177 PR tree-optimization/79529
29178 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
29179 ssa_defined_default_def_p to handle cases which are implicitly
29181 * tree-ssa.c (ssa_defined_default_def_p): New function.
29182 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
29183 which are implicitly defined.
29184 * tree-ssa.h (ssa_defined_default_def_p): Declare.
29186 2017-02-17 Richard Biener <rguenther@suse.de>
29188 PR middle-end/79576
29189 * params.def (max-ssa-name-query-depth): Limit to 10.
29191 2017-02-17 Richard Biener <rguenther@suse.de>
29193 PR tree-optimization/79552
29194 * tree-ssa-structalias.c (visit_loadstore): Properly verify
29197 2017-02-17 Richard Biener <rguenther@suse.de>
29200 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
29202 2017-02-17 Marek Polacek <polacek@redhat.com>
29204 PR middle-end/79536
29205 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
29206 (fold_negate_expr): New wrapper.
29208 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
29210 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
29211 Correct terminology and de-emphasize pre-standard behavior.
29213 2017-02-16 Alan Modra <amodra@gmail.com>
29215 PR rtl-optimization/79286
29216 * ira.c (def_dominates_uses): New function.
29217 (update_equiv_regs): Don't create an equivalence for insns that
29218 may trap where the register def does not dominate the use.
29220 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
29222 PR rtl-optimization/78127
29223 * lra.c (lra): Call lra_eliminate before finish the loop after
29226 2017-02-16 Richard Biener <rguenther@suse.de>
29228 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
29230 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
29231 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
29232 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
29233 (isl_val_int_from_wi): New function.
29234 (extract_affine_gmp): Rename to ...
29235 (extract_affine_wi): ... this, take a widest_int.
29236 (extract_affine_int): Just wrap extract_affine_wi.
29237 (add_param_constraints): Use isl_val_int_from_wi.
29238 (add_loop_constraints): Likewise, and extract_affine_wi.
29240 2017-02-15 Jeff Law <law@redhat.com>
29242 PR middle-end/79521
29243 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
29244 ira_init_register_move_cost_if_necessary.
29246 2017-02-15 Martin Sebor <msebor@redhat.com>
29248 PR middle-end/32003
29249 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
29250 removed in a prior commit.
29252 2017-02-15 Bin Cheng <bin.cheng@arm.com>
29254 PR tree-optimization/79347
29255 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
29256 counters during peeling.
29258 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
29260 * Makefile.in (site.exp): Remove "set ISLVER".
29262 2017-02-15 Jakub Jelinek <jakub@redhat.com>
29265 * real.c (real_from_integer): Call real_convert even for decimal.
29267 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
29270 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
29272 2017-02-14 Andrew Pinski <apinski@cavium.com>
29274 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
29275 cores and change the partno/implementer to be correct.
29276 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
29277 the 'B" as the implementer.
29278 * config/aarch64/aarch64-tune.md: Regenerate.
29280 2017-02-14 Carl Love <cel@us.ibm.com>
29282 * config/rs6000/rs6000.c: Add case statement entry to make the
29283 xvcvuxdsp built-in argument unsigned.
29284 * config/rs6000/vsx.md: Fix the source and return operand types so they
29285 match the instruction definitions from the ISA document. Fix typo
29286 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
29289 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
29292 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
29293 member early_clobber_alts.
29294 * lra-lives.c (reg_early_clobber_p): New.
29295 (process_bb_lives): Use it.
29296 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
29297 (debug_operand_data): Initialize early_clobber_alts.
29298 (setup_operand_alternative): Set up early_clobber_alts.
29299 (collect_non_operand_hard_regs): Ditto. Pass early clobber
29300 alternatives to new_insn_reg.
29301 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
29303 (lra_update_insn_regno_info): Pass the new arg.
29305 2017-02-14 Jakub Jelinek <jakub@redhat.com>
29307 PR middle-end/79505
29308 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
29309 (new_oacc_loop_raw): Don't clear already cleared fields.
29312 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
29313 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
29314 _mm512_prefetch_i64gather_ps): New inline functions and macros.
29316 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
29319 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
29321 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
29324 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
29325 the extra instruction to the right place to store 128-bit constant
29328 2017-02-14 Martin Sebor <msebor@redhat.com>
29330 PR middle-end/79448
29331 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
29332 warning for strings of unknown length.
29334 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
29336 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
29338 2017-02-14 Jeff Law <law@redhat.com>
29341 * ira-costs.c (scan_one_insn): Initialize register move costs
29342 for pseudos seen in USE/CLOBBER insns.
29344 PR tree-optimization/79095
29345 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
29346 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
29347 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
29348 if the operands are known to be not equal, then the resulting range
29350 (intersect_ranges): If the new range is ~[0,0] and the old range is
29351 wide, then prefer ~[0,0].
29352 * tree-vrp.c (overflow_comparison_p_1): New function.
29353 (overflow_comparison_p): New function.
29354 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
29355 if NAME is used in an overflow test.
29356 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
29357 overflow check that can be expressed as an equality test, then adjust
29358 ops to be that equality test.
29360 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29362 * config/s390/s390-builtin-types.def: Remove flags argument.
29363 * config/s390/s390.c (s390_init_builtins): Likewise.
29365 2017-02-14 Martin Liska <mliska@suse.cz>
29367 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
29368 vector. Fix trailing white spaces.
29370 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
29372 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
29375 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29377 PR rtl-optimization/68664
29378 * config/arm/arm.c (arm_sched_can_speculate_insn):
29379 New function. Declare prototype.
29380 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
29382 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29384 PR rtl-optimization/68664
29385 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
29387 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
29389 2017-02-14 Amit Pawar <amit.pawar@amd.com>
29391 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
29392 max skip bytes for function, loop and jump.
29394 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29396 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
29397 ABS_EXPR for gimple dump.
29399 2017-02-14 Jakub Jelinek <jakub@redhat.com>
29402 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
29404 PR tree-optimization/79408
29405 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
29406 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
29407 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
29408 also if rhs1 is INTEGER_CST.
29410 2017-02-14 Richard Biener <rguenther@suse.de>
29412 PR middle-end/79432
29413 * tree-into-ssa.c (insert_phi_nodes): When the function can
29414 have abnormal edges rewrite SSA names with broken use-def
29415 dominance out of SSA and register them for PHI insertion.
29417 2017-02-13 Martin Sebor <msebor@redhat.com>
29419 PR middle-end/79496
29420 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
29421 clearing info.nowrite flag when snprintf size argument is a range.
29423 2017-02-13 Jakub Jelinek <jakub@redhat.com>
29425 * cprop.c (cprop_jump): Add missing space in string literal.
29426 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
29427 (get_constraint_for_component_ref): Likewise.
29428 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
29429 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
29430 * lra-constraints.c (process_alt_operands): Likewise.
29431 * ipa-inline.c (inline_small_functions): Likewise.
29432 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
29433 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
29434 * trans-mem.c (diagnose_tm_1_op): Likewise.
29435 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
29436 (grid_parallel_clauses_gridifiable): Likewise.
29438 * config/nvptx/mkoffload.c (process): Add space in between
29441 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
29442 "MOD4_SSE_REGS" and "ALL_REGS".
29444 * spellcheck.c (test_data): Add , in between "foo" and "food".
29446 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29449 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
29450 boundary crossing check and subsequent code generation agree.
29452 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29454 * config/aarch64/aarch64.c (has_memory_op): Delete.
29455 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
29458 2017-02-13 Jakub Jelinek <jakub@redhat.com>
29460 PR rtl-optimization/79388
29461 PR rtl-optimization/79450
29462 * combine.c (distribute_notes): When removing TEM_INSN for which
29463 corresponding dest has last value recorded, invalidate that last
29466 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29468 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
29469 of explicit '@'. Add missing assembly comment marker on branch costs
29472 2017-02-13 Nathan Sidwell <nathan@acm.org>
29474 * gengtype-lex.l (<in_struct>): Add '/'.
29476 2017-02-13 Martin Liska <mliska@suse.cz>
29479 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
29481 2017-02-13 Richard Biener <rguenther@suse.de>
29483 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
29485 * configure: Re-generate.
29486 * config.in: Likewise.
29487 * graphite-dependences.c: Simplify as if
29488 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
29489 * graphite-isl-ast-to-gimple.c: Likewise.
29490 * graphite-optimize-isl.c: Likewise.
29491 * graphite-poly.c: Likewise.
29492 * graphite-sese-to-poly.c: Likewise.
29493 * graphite.h: Likewise.
29494 * toplev.c: Include isl/version.h and use isl_version () for
29495 printing the ISL version.
29496 * doc/install.texi: Update ISL requirement.
29498 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
29500 * doc/standards.texi (Standards): Update reference to
29503 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
29505 * doc/extend.texi (Named Address Spaces): sourceware.org now
29507 * doc/install.texi (Binaries): Ditto.
29510 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
29512 * doc/cpp.texi: Replace "stringify"/"stringification" with C
29513 standard terminology "stringize"/"stringizing" throughout.
29514 * doc/cppinternals.texi: Likewise.
29516 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
29518 * doc/extend.texi: Fix some spelling mistakes and typos.
29519 * doc/invoke.texi: Likewise.
29521 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
29524 * params.def (inline-min-speedup) Change from 10 to 8.
29526 2017-02-11 Jakub Jelinek <jakub@redhat.com>
29528 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
29531 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
29534 * ipa-inline-analysis.c (get_minimal_bb): New function.
29535 (record_modified): Use it.
29536 (remap_edge_change_prob): Handle also ancestor functions.
29538 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
29540 * doc/contrib.texi (Contributors): Remove broken link into
29541 the Mauve CVS repository.
29543 2017-02-11 Jakub Jelinek <jakub@redhat.com>
29545 PR middle-end/79454
29546 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
29547 result computation whenever lhs doesn't have vector mode, not
29548 just when it has BLKmode.
29550 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
29552 * doc/makefile.texi (profiledbootstrap): Refer to the
29553 installation instructions only in textual form.
29555 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29558 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
29560 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
29562 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
29563 (Specific): Update mingw-w64 reference.
29565 (Specific): Remove broken link to Renesas RX processor.
29567 2017-02-10 Richard Biener <rguenther@suse.de>
29569 * toplev.c (process_options): Do not mention obsolete graphite
29570 options when printing sorry message about missing graphite support.
29571 Mention -floop-nest-optimize.
29573 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
29575 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
29576 (vtst_p16): Likewise.
29577 (vtstq_p8): Likewise.
29578 (vtstq_p16): Likewise.
29580 (vtstq_p64): Likewise.
29581 * config/arm/arm_neon.h (vgetq_lane_p64): New.
29582 (vset_lane_p64): New.
29583 (vsetq_lane_p64): New.
29585 2017-02-10 Jakub Jelinek <jakub@redhat.com>
29587 PR tree-optimization/79411
29588 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
29589 stmt operands are SSA_NAMEs used in abnormal phis.
29590 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
29593 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
29596 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
29599 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
29601 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
29603 2017-02-09 Jakub Jelinek <jakub@redhat.com>
29605 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
29609 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
29610 not arbitrary TREE_CONSTANT.
29613 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
29614 "omp declare target link" attribute unless is_global_var.
29615 * omp-offload.c (find_link_var_op): Likewise.
29617 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
29618 Chung-Lin Tang <cltang@codesourcery.com>
29620 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
29622 (gimplify_adjust_omp_clauses): Don't delete TILE.
29623 (gimplify_omp_for): Deal with TILE.
29624 * internal-fn.c (expand_GOACC_TILE): New function.
29625 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
29627 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
29628 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
29630 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
29631 avoid DIV for outermost collapse var.
29632 (expand_oacc_for): Insert tile element loop as needed. Adjust.
29633 Remove out of date comments, fix whitespace.
29634 * omp-general.c (omp_extract_for_data): Deal with tiling.
29635 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
29636 adjust OLF_DIM_BASE value.
29637 (struct omp_for_data): Add tiling field.
29638 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
29639 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
29640 for auto loops. Remove default auto determining, moved to
29641 oacc_loop_fixed_partitions.
29642 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
29643 stmts, add e_mask field.
29644 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
29645 (oacc_thread_numbers): Use oacc_dim_call.
29646 (oacc_xform_tile): New.
29647 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
29648 (finish_oacc_loop): Adjust for ifns vector.
29649 (oacc_loop_discover_walk): Append loop abstraction sites to list,
29650 add case for GOACC_TILE fns.
29651 (oacc_loop_xform_loop): Delete.
29652 (oacc_loop_process): Iterate over call list directly, and add
29653 handling for GOACC_TILE fns.
29654 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
29656 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
29657 vector partitioning to outer loops. Assign 2 partitions to loops
29658 when available. Add TILE handling.
29659 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
29660 (execite_oacc_device_lower): Process GOACC_TILE fns,
29661 ignore unknown specs.
29662 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
29663 * tree.c (omp_clause_num_ops): Adjust TILE ops.
29664 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
29666 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
29668 * configure.ac (ACX_BUGURL): Update.
29669 * configure: Regenerate.
29671 2017-02-09 Richard Biener <rguenther@suse.de>
29673 PR tree-optimization/69823
29674 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
29675 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
29677 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
29679 * config/arc/arc-c.def: Add __NPS400__ definition.
29680 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
29681 (TARGET_NPS400): Define.
29683 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
29685 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
29687 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
29688 pointer, arch_info.
29689 (arc_cpu_types): Fill the arch_info field with a pointer into the
29690 arc_arch_types table.
29691 (arc_selected_cpu): Declare.
29692 * config/arc/arc.c (arc_selected_cpu): Make global.
29693 (arc_selected_arch): Delete.
29694 (arc_base_cpu): Delete.
29695 (arc_override_options): Remove references to deleted variables,
29696 update access to arch information.
29697 (ARC_OPT): Update access to arch information.
29698 (ARC_OPTX): Likewise.
29699 * config/arc/arc.h (arc_base_cpu): Remove declaration.
29700 (TARGET_ARC600): Update access to arch information.
29701 (TARGET_ARC601): Likewise.
29702 (TARGET_ARC700): Likewise.
29703 (TARGET_EM): Likewise.
29704 (TARGET_HS): Likewise.
29705 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
29708 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
29711 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
29712 condition/operands for integer GE/LE/GEU/LEU operations.
29714 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
29716 PR translation/79397
29717 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
29720 2017-02-08 Martin Jambor <mjambor@suse.cz>
29723 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
29724 whether allocation happened.
29725 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
29726 nothing was allocated.
29728 2017-02-08 Jakub Jelinek <jakub@redhat.com>
29730 PR tree-optimization/79408
29731 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
29732 constant, but SSA_NAME with a known integer range, use the minimum
29733 of that range instead of op1 to determine if modulo can be replaced
29734 with its first operand.
29736 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29738 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
29740 2017-02-08 Richard Biener <rguenther@suse.de>
29742 PR tree-optimization/71824
29743 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
29744 Check all loops contained in the merged region.
29746 2017-02-07 Andrew Pinski <apinski@cavium.com>
29748 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
29750 2017-02-07 Andrew Pinski <apinski@cavium.com>
29752 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
29753 (thunderxt88): Likewise.
29754 (thunderxt81): Disable LSE and change v8.1 to v8.
29755 (thunderxt83): Likewise.
29757 2017-02-07 Jakub Jelinek <jakub@redhat.com>
29758 Richard Biener <rguenther@suse.de>
29760 PR middle-end/79399
29761 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
29762 type from int to size_t.
29763 * ira-costs.c (struct_costs_size): Change type from int to size_t.
29765 2017-02-07 Jakub Jelinek <jakub@redhat.com>
29767 PR rtl-optimization/79386
29768 * cprop.c (bypass_conditional_jumps): Initialize
29769 bypass_last_basic_block already before splitting bbs after
29770 unconditional traps...
29771 (bypass_conditional_jumps): ... rather than here.
29774 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
29775 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
29776 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
29777 fix -masm=intel patterns.
29779 2017-02-07 Richard Biener <rguenther@suse.de>
29781 PR tree-optimization/79256
29782 PR middle-end/79278
29783 * builtins.c (get_object_alignment_2): Use min_align_of_type
29784 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
29785 and ADJUST_FIELD_ALIGN.
29787 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
29789 * doc/tm.texi: Regenerate.
29790 * stor-layout.c (layout_decl): Adjust.
29791 (update_alignment_for_field): Likewise.
29792 (place_field): Likewise.
29793 (min_align_of_type): Likewise.
29794 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
29795 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
29796 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
29797 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
29798 * config/frv/frv.c (frv_adjust_field_align): Likewise.
29799 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
29800 * config/i386/i386.c (x86_field_alignment): Likewise.
29801 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
29802 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
29803 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
29804 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
29805 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
29806 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
29810 2017-01-30 Richard Biener <rguenther@suse.de>
29812 PR tree-optimization/79256
29813 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
29814 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
29817 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
29819 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
29820 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
29821 builtins to SImode and emit a zero-extend, if necessary.
29823 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
29825 * docs/invoke.texi (RISC-V Options): Alphabetize.
29827 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
29829 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
29832 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
29834 * config/riscv/riscv.c: New file.
29835 * common/config/riscv/riscv-common.c: Likewise.
29836 * config.gcc: Likewise.
29837 * config/riscv/constraints.md: Likewise.
29838 * config/riscv/elf.h: Likewise.
29839 * config/riscv/generic.md: Likewise.
29840 * config/riscv/linux.h: Likewise.
29841 * config/riscv/multilib-generator: Likewise.
29842 * config/riscv/peephole.md: Likewise.
29843 * config/riscv/pic.md: Likewise.
29844 * config/riscv/predicates.md: Likewise.
29845 * config/riscv/riscv-builtins.c: Likewise.
29846 * config/riscv/riscv-c.c: Likewise.
29847 * config/riscv/riscv-ftypes.def: Likewise.
29848 * config/riscv/riscv-modes.def: Likewise.
29849 * config/riscv/riscv-opts.h: Likewise.
29850 * config/riscv/riscv-protos.h: Likewise.
29851 * config/riscv/riscv.h: Likewise.
29852 * config/riscv/riscv.md: Likewise.
29853 * config/riscv/riscv.opt: Likewise.
29854 * config/riscv/sync.md: Likewise.
29855 * config/riscv/t-elf-multilib: Likewise.
29856 * config/riscv/t-linux: Likewise.
29857 * config/riscv/t-linux-multilib: Likewise.
29858 * config/riscv/t-riscv: Likewise.
29859 * configure.ac: Likewise.
29860 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
29861 Waterman as RISC-V maintainers.
29862 * doc/install.texi: Add RISC-V entries.
29863 * doc/invoke.texi: Add RISC-V options section.
29864 * doc/md.texi: Add RISC-V constraints section.
29865 * configure: Regenerated.
29867 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
29870 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
29871 false values to be constant vectors with all 0 or all 1 bits set.
29872 (vcondu<mode><mode>): Likewise.
29873 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
29875 (fpmask_comparison_operator): Update comment.
29876 (vecint_comparison_operator): New predicate.
29877 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
29878 vector conditionals when the true and false values are constant
29879 vectors with all 0 bits or all 1 bits set.
29881 2017-02-06 Martin Sebor <msebor@redhat.com>
29883 PR tree-optimization/79376
29884 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
29886 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
29888 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
29889 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
29890 to simplify split condition.
29892 2017-02-06 Jakub Jelinek <jakub@redhat.com>
29894 * omp-expand.c (oxpand_omp_atomic_fetch_op,
29895 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
29898 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
29900 PR rtl-optimization/68664
29901 * target.def (can_speculate_insn): New hook.
29902 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
29903 * doc/tm.texi: Regenerate.
29904 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
29905 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
29906 (rs6000_sched_can_speculate_insn): New function.
29908 2017-02-06 Jakub Jelinek <jakub@redhat.com>
29910 PR tree-optimization/79284
29911 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
29912 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
29913 vectorizable_mask_load_store, vectorizable_operation,
29914 vect_is_simple_cond, get_same_sized_vectype): Use it instead
29915 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
29916 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
29917 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
29918 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
29919 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
29920 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
29921 is_gimple_assign (stmt). Replace another such test with
29922 is_gimple_assign (stmt).
29924 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
29927 * config/avr/avr.c (rtl-iter.h): Include it.
29928 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
29929 (avr_legitimate_combined_insn): ...and implementation.
29931 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
29933 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
29934 * config/s390/s390.c (s390_const_operand_ok)
29935 (s390_canonicalize_comparison, s390_extract_part)
29936 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
29937 (s390_contiguous_bitmask_p, s390_rtx_costs)
29938 (legitimize_pic_address): Likewise.
29939 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
29940 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
29941 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
29942 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
29943 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
29945 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
29947 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
29948 REGNO($0) == REGNO($1).
29950 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29952 * config/s390/linux.h(SIZE_TYPE): Add comment.
29954 2017-02-06 Julian Brown <julian@codesourcery.com>
29955 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29956 Virendra Pathak <virendra.pathak@broadcom.com>
29958 * config/aarch64/aarch64-cores.def: Change the scheduler
29960 * config/aarch64/aarch64.md: Include thunderx2t99.md.
29961 * config/aarch64/thunderx2t99.md: New file.
29963 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
29965 * doc/standards.texi (Go Language): Update link to language
29968 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
29970 * tree-eh.c (lower_resx): Sanitize profile.
29971 (cleanup_empty_eh_move_lp): Likewise.
29973 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
29976 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
29978 * cfgloopmanip.h (loop_version): Update prototype.
29979 * modulo-sched.c (sms_schedule): Update call of loop_version.
29980 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
29981 * tree-parloops.c (gen_parallel_loop): Likewise.
29982 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
29983 * tree-ssa-loop-split.c (split_loop): Likewise.
29984 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
29985 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
29987 2017-02-05 Martin Liska <mliska@suse.cz>
29990 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
29992 (print_operand_address): Initialize a struct to zero.
29994 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
29996 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
29997 garbage collector only in textual form.
29999 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30001 * doc/extend.texi (x86 specific memory model extensions for
30002 transactional memory): Simplify a phrase.
30004 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
30007 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
30008 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
30009 (atomic_storedi_1): Likewise.
30011 2017-02-04 Jakub Jelinek <jakub@redhat.com>
30013 PR tree-optimization/79338
30014 * tree-parloops.c (gather_scalar_reductions): Don't call
30015 vect_analyze_loop_form for loop->inner before destroying loop's
30018 2017-02-03 Martin Sebor <msebor@redhat.com>
30020 PR tree-optimization/79327
30021 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
30022 when precision has resulted in leading zeros.
30023 (format_integer): Adjust the likely counter to assume an unknown
30024 argument that may be zero is non-zero.
30026 2017-02-03 Jason Merrill <jason@redhat.com>
30029 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
30030 avoid copying non-taken branch.
30032 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30034 PR tree-optimization/79340
30035 * tree-vect-loop.c (vectorizable_reduction): Release
30036 vec_defs elements after safe_splicing them into other vectors.
30039 PR tree-optimization/79327
30040 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
30041 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
30043 (format_integer): Use wide_int_to_tree instead of build_int_cst
30044 + to_?hwi. If argmin is NULL, just set argmin and argmax to
30045 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
30046 of shortest and longest sequence.
30048 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
30050 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
30051 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
30053 2017-02-03 Walter Lee <walt@tilera.com>
30056 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
30057 after initial stackframe link reg save.
30058 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
30060 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30063 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
30064 wu for stxssp alternative.
30066 2017-02-03 Martin Sebor <msebor@redhat.com>
30068 PR tree-optimization/79352
30069 * gimple-fold.c (get_range_strlen): Add argument.
30070 (get_range_strlen): Change return type to bool.
30071 (get_maxval_strlen): Pass in a dummy argument.
30072 * gimple-fold.h (get_range_strlen): Change return type to bool.
30073 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
30074 * tree.h (array_at_struct_end_p): Add argument.
30075 * tree.c (array_at_struct_end_p): Handle it.
30077 2017-02-03 Martin Liska <mliska@suse.cz>
30080 * multiple_target.c (create_dispatcher_calls): Redirect edge
30081 from a caller of a dispatcher.
30082 (expand_target_clones): Make the clones local.
30083 (ipa_target_clone): Do both target clones and resolvers.
30084 (ipa_dispatcher_calls): Remove the pass.
30085 (pass_dispatcher_calls::gate): Likewise.
30086 (make_pass_dispatcher_calls): Likewise.
30087 * passes.def (pass_target_clone): Put as very first IPA early
30090 2017-02-03 Martin Liska <mliska@suse.cz>
30092 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
30093 in case of a function with ifunc attribute.
30095 2017-02-03 Martin Liska <mliska@suse.cz>
30097 * cgraph.c (cgraph_node::dump): Dump function version info.
30098 * symtab.c (symtab_node::dump_base): Add missing new line.
30100 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30102 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
30103 (ifcombine_ifandif): Use it.
30105 2017-02-03 Martin Liska <mliska@suse.cz>
30107 * doc/invoke.texi: Document default value for
30108 use-after-scope-direct-emission-threshold.
30110 2017-02-03 Martin Liska <mliska@suse.cz>
30112 PR tree-optimization/79339
30113 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
30114 (format_floating): Likewise.
30116 2017-02-03 Martin Liska <mliska@suse.cz>
30119 * ipa-prop.c (ipa_node_params_t::insert): Remove current
30121 (ipa_node_params_t::remove): Likewise.
30122 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
30123 initialization from removed ipa_node_params_t::insert.
30124 (ipa_node_params::~ipa_node_params): Move from removed
30125 ipa_node_params_t::release.
30126 * symbol-summary.h (symbol_summary::m_released): New member.
30127 Do not release a summary twice. Do not allow to call finalizer
30128 for types of a summary that live in GGC memory.
30130 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30132 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
30135 2017-02-02 Martin Sebor <msebor@redhat.com>
30137 PR middle-end/79275
30138 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
30139 (format_string): Tighten up the range of output for non-constant
30140 strings and correct the expected range for wide non-constant strings.
30142 2017-02-02 Martin Sebor <msebor@redhat.com>
30144 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
30146 PR middle-end/32003
30147 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
30149 (-fdump-tree-@var): Add to index and document how to come up
30150 with pass-specific option and dump file names.
30151 (-fdump-passes): Clarify where to look for output.
30153 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30155 PR middle-end/77445
30156 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
30157 statistics of the analyzed path; allow threading for speed when
30158 any of BBs along the path are optimized for speed.
30160 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
30162 PR middle-end/78468
30163 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
30164 settings of the virtual registers.
30167 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
30169 * explow.c (get_dynamic_stack_size): Take known alignment of stack
30170 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
30173 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30175 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
30176 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
30178 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30180 * config/s390/s390.md: Add missing comments with the expanded
30182 * config/s390/vector.md: Likewise.
30183 * config/s390/vx-builtins.md: Likewise.
30185 2017-02-02 Jakub Jelinek <jakub@redhat.com>
30188 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
30189 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
30190 conditions on a single line.
30192 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30194 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
30195 __S390_VX__ to __VX__.
30197 2017-02-01 Andrew Pinski <apinski@cavium.com>
30199 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
30200 stmt_info to record_stmt_cost.
30201 (vect_get_known_peeling_cost): Pass stmt_info if known to
30203 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
30204 cpu_vector_cost field into
30205 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
30206 field into vec_int_stmt_cost and vec_fp_stmt_cost.
30207 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
30208 splitting of scalar_stmt_cost and vec_stmt_cost.
30209 (thunderx_vector_cost): Likewise.
30210 (cortexa57_vector_cost): LIkewise.
30211 (exynosm1_vector_cost): Likewise.
30212 (xgene1_vector_cost): Likewise.
30213 (thunderx2t99_vector_cost): Improve after the splitting of the two
30215 (aarch64_builtin_vectorization_cost): Update for the splitting of
30216 scalar_stmt_cost and vec_stmt_cost.
30218 2017-02-01 Torvald Riegel <triegel@redhat.com>
30219 Richard Henderson <rth@redhat.com>
30221 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
30222 conditional on existance of a fast atomic load.
30223 * optabs-query.c (can_atomic_load_p): New function.
30224 * optabs-query.h (can_atomic_load_p): Declare it.
30225 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
30226 no fast atomic load is available for the particular size of access.
30227 (expand_atomic_compare_and_swap): Likewise.
30228 (expand_atomic_load): Likewise.
30229 (expand_atomic_store): Likewise.
30230 (expand_atomic_fetch_op): Likewise.
30231 * testsuite/lib/target-supports.exp
30232 (check_effective_target_sync_int_128): Remove x86 because it provides
30233 no fast atomic load.
30234 (check_effective_target_sync_int_128_runtime): Likewise.
30236 2017-02-01 Richard Biener <rguenther@suse.de>
30238 * graphite.c: Include tree-vectorizer.h for find_loop_location.
30239 (graphite_transform_loops): Provide opt-info for optimized nests.
30240 * tree-parloop.c (parallelize_loops): Provide opt-info for
30241 parallelized loops.
30243 2017-02-01 Richard Biener <rguenther@suse.de>
30245 PR middle-end/79315
30246 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
30247 was not set before.
30249 2017-02-01 Richard Biener <rguenther@suse.de>
30251 PR tree-optimization/71824
30252 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
30253 Verify the loops are valid in the merged SESE region.
30254 (scop_detection::can_represent_loop_1): Check analyzing the
30255 evolution of the number of iterations in the region succeeds.
30257 2017-01-31 Ian Lance Taylor <iant@golang.org>
30259 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
30260 REG_ARGS_SIZE note to 32-bit push insns and call insn.
30262 2017-01-31 David Malcolm <dmalcolm@redhat.com>
30264 PR preprocessor/79210
30265 * input.c (get_substring_ranges_for_loc): Replace line_width
30266 assertion with error-handling.
30268 2017-01-31 Richard Biener <rguenther@suse.de>
30270 PR tree-optimization/77318
30271 * graphite-sese-to-poly.c (extract_affine): Fix assert.
30272 (create_pw_aff_from_tree): Take loop parameter.
30273 (add_condition_to_pbb): Pass loop of the condition to
30274 create_pw_aff_from_tree.
30276 2017-01-31 Jakub Jelinek <jakub@redhat.com>
30278 * config/s390/s390.c (s390_asan_shadow_offset): New function.
30279 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
30281 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
30285 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
30287 (convert_int_to_float128): Likewise.
30288 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
30289 (convert_int_to_float128): Likewise.
30290 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
30291 (UNSPEC_IEEE128_CONVERT): Likewise.
30292 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
30293 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
30294 Use local variables for IBM extended format.
30295 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
30296 (fix_trunc<mode>si2_fprs): Likewise.
30297 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
30298 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
30299 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
30300 to know that we can now have integers of all sizes in vector
30302 (fix<uns>_<mode>di2_hw): Likewise.
30303 (float<uns>_<mode>si2_hw): Likewise.
30304 (fix_<mode>si2_hw): Likewise.
30305 (fixuns_<mode>si2_hw): Likewise.
30306 (float<uns>_<mode>di2_hw): Likewise.
30307 (float_<mode>di2_hw): Likewise.
30308 (float_<mode>si2_hw): Likewise.
30309 (floatuns_<mode>di2_hw): Likewise.
30310 (floatuns_<mode>si2_hw): Likewise.
30311 (xscvqp<su>wz_<mode>): Delete, no longer used.
30312 (xscvqp<su>dz_<mode>): Likewise.
30313 (xscv<su>dqp_<mode>): Likewise.
30314 (ieee128_mfvsrd_64bit): Likewise.
30315 (ieee128_mfvsrd_32bit): Likewise.
30316 (ieee128_mfvsrwz): Likewise.
30317 (ieee128_mtvsrw): Likewise.
30318 (ieee128_mtvsrd_64bit): Likewise.
30319 (ieee128_mtvsrd_32bit): Likewise.
30321 2017-01-31 Martin Liska <mliska@suse.cz>
30324 * ipa-prop.c (ipa_free_all_node_params): Call release method
30325 instead of ~sumbol_summary to not to trigger double times
30328 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
30330 PR tree-optimization/71691
30331 * bitmap.h (class auto_bitmap): New.
30332 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
30333 is_maybe_undefined instead of ssa_undefined_value_p.
30335 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30337 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
30338 __S390_ARCH_LEVEL__ to __ARCH__.
30340 2017-01-31 Jakub Jelinek <jakub@redhat.com>
30342 PR tree-optimization/79267
30343 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
30344 if should_remove_lhs_p is true.
30346 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
30349 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
30350 (add_alignment_attribute): New.
30351 (base_type_die): Add alignment attribute.
30352 (subrange_type_die): Likewise.
30353 (modified_type_die): Likewise.
30354 (gen_array_type_die): Likewise.
30355 (gen_descr_array_type_die: Likewise.
30356 (gen_enumeration_type_die): Likewise.
30357 (gen_subprogram_die): Likewise.
30358 (gen_variable_die): Likewise.
30359 (gen_field_die): Likewise.
30360 (gen_ptr_to_mbr_type_die): Likewise.
30361 (gen_struct_or_union_type_die): Likewise.
30362 (gen_subroutine_type_die): Likewise.
30363 (gen_typedef_die): Likewise.
30364 (base_type_cmp): Compare alignment attribute.
30366 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30369 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
30370 (setb_unsigned) New pattern for setb with CCUNS.
30371 * config/rs6000/rs6000.c (expand_block_compare): Use a different
30372 subfc./subfe sequence to avoid overflow problems. Generate a
30373 shorter sequence with cmpld/setb for power9.
30374 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
30375 for generating subfc. instruction.
30376 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
30377 now uses this instruction.
30379 2017-01-30 Ian Lance Taylor <iant@google.com>
30382 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
30383 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
30385 2017-01-30 Martin Sebor <msebor@redhat.com>
30387 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
30388 Move constant to the right of a relational operator.
30389 (get_mpfr_format_length, format_character, format_string): Ditto.
30390 (should_warn_p, maybe_warn): Same.
30392 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
30394 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
30397 * asan.c (get_translation_unit_decl): Remove function.
30398 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
30400 2017-01-30 Martin Liska <mliska@suse.cz>
30402 PR gcov-profile/79259
30403 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
30404 -fprofile-generate.
30406 2017-01-30 Martin Liska <mliska@suse.cz>
30409 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
30410 Initialize variables with NULL value.
30412 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
30415 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
30417 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
30419 2017-01-30 Richard Biener <rguenther@suse.de>
30421 PR tree-optimization/79276
30422 * tree-vrp.c (process_assert_insertions): Properly adjust common
30423 when removing a duplicate.
30425 2017-01-30 Richard Biener <rguenther@suse.de>
30427 PR tree-optimization/79256
30428 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
30429 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
30431 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
30433 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
30436 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
30437 ("*r<noxa>sbg_<mode>_sll_bitmask")
30438 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
30439 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
30440 Use contiguous_bitmask_nowrap_operand.
30442 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30445 * config/rs6000/altivec.h (vec_xl): Revise #define.
30446 (vec_xst): Likewise.
30448 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
30450 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
30452 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
30454 PR rtl-optimization/79194
30455 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
30456 traps before call to bypass_conditional_jumps.
30458 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
30460 PR tree-optimization/71374
30461 * lra-constraints.c (check_conflict_input_operands): New.
30462 (match_reload): Use it.
30464 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
30467 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
30468 account to calculate conflict_set.
30470 2017-01-27 Bin Cheng <bin.cheng@arm.com>
30472 PR rtl-optimization/78559
30473 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
30474 other_insn in combine.
30476 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
30478 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
30479 uint16_type_node for BT_UINT16.
30481 2017-01-27 David Malcolm <dmalcolm@redhat.com>
30483 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
30484 "RTL Tests" to menu.
30485 (GIMPLE Tests): New node.
30486 (RTL Tests): New node.
30488 2017-01-27 Richard Biener <rguenther@suse.de>
30490 PR tree-optimization/79245
30491 * tree-loop-distribution.c (distribute_loop): Apply cost
30492 modeling also to detected patterns.
30494 2017-01-27 Richard Biener <rguenther@suse.de>
30496 PR tree-optimization/71433
30497 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
30498 (compare_assert_loc): New function.
30499 (process_assert_insertions): Sort and optimize assert locations
30500 to remove duplicates and push down identical assertions on
30501 edges to their destination block.
30503 2017-01-27 Richard Biener <rguenther@suse.de>
30505 PR tree-optimization/79244
30506 * tree-vrp.c (remove_range_assertions): Forcefully propagate
30507 out SSA names even if abnormal.
30509 2017-01-27 Jakub Jelinek <jakub@redhat.com>
30511 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
30512 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
30513 instead of MPFR_RNDN.
30515 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
30518 * arm.c (arm_option_override): Don't call build_target_option_node
30519 until after doing all option overrides.
30520 (arm_valid_target_attribute_tree): Likewise.
30522 2017-01-27 Martin Liska <mliska@suse.cz>
30524 * doc/invoke.texi (-fprofile-arcs): Document profiling support
30525 for {cd}tors and C++ {cd}tors.
30527 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
30529 * config/s390/s390.md ("*setmem_long_and")
30530 ("*setmem_long_and_31z"): Use zero_extend instead of and.
30532 2017-01-26 Martin Sebor <msebor@redhat.com>
30534 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
30537 2017-01-26 Martin Sebor <msebor@redhat.com>
30539 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
30540 HAVE_DFmode before using XFmode or DFmode.
30541 (parse_directive): Avoid using the z length modifier to avoid
30542 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
30544 PR middle-end/78703
30545 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
30546 to accept adjustment as an array.
30547 (get_int_range): New function.
30548 (struct directive): Make width and prec arrays.
30549 (directive::set_width, directive::set_precision): Call get_int_range.
30550 (format_integer, format_floating): Handle width and precision ranges.
30551 (format_string, parse_directive): Same.
30553 2017-01-26 Jakub Jelinek <jakub@redhat.com>
30556 * dwarf2out.c (generate_skeleton_bottom_up): For children with
30557 comdat_type_p set, just clone them, but keep the children in the
30561 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
30562 which have direct callers with -fvar-tracking-assignments enabled
30564 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
30565 inside of type units.
30567 2017-01-26 Martin Sebor <msebor@redhat.com>
30569 PR middle-end/78703
30570 * gimple-ssa-sprintf.c (struct result_range): Add likely and
30572 (struct format_result): Replace number_chars, number_chars_min,
30573 and number_chars_max with a single member of struct result_range.
30575 (format_result::operator+=): Adjust.
30576 (struct fmtresult): Remove bounded. Handle likely and unlikely
30578 (fmtresult::adjust_for_width_or_precision): New function.
30579 (fmtresult:type_max_digits): New function.
30580 (bytes_remaining): Handle likely and unlikely counters.
30581 (min_bytes_remaining): Remove.
30582 (format_percent): Simplify.
30583 (format_integer, format_floating): Set likely and unlikely counters.
30584 (get_string_length, format_character, format_string): Same.
30585 (format_plain, should_warn_p): New function.
30586 (maybe_warn): Call should_warn_p. Update diagnostic messages
30587 and handle those for all directives, including plain strings.
30588 (format_directive): Handle likely and unlikely counters.
30589 Remove unnecessary quoting from diagnostics. Add an informational
30591 (add_bytes): Remove.
30592 (pass_sprintf_length::compute_format_length): Simplify.
30593 (try_substitute_return_value): Handle likely and unlikely counters.
30595 2017-01-26 Carl Love <cel@us.ibm.com>
30597 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
30598 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
30600 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
30603 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
30604 endianess for subregs into account.
30605 * lra-constraints.c (lra_constraints): Do risky transformations
30606 always on the first iteration.
30607 * lra-lives.c (check_pseudos_live_through_calls): Add arg
30608 last_call_used_reg_set.
30609 (process_bb_lives): Define and use last_call_used_reg_set.
30610 * lra.c (lra): Always continue after lra_constraints on the first
30613 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
30615 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
30617 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
30619 2017-01-26 Jakub Jelinek <jakub@redhat.com>
30621 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
30622 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
30623 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
30624 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
30625 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
30626 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
30627 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
30628 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
30629 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
30631 2017-01-26 Marek Polacek <polacek@redhat.com>
30634 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
30635 for the third operand.
30637 2017-01-26 Jakub Jelinek <jakub@redhat.com>
30639 PR middle-end/79236
30640 * omp-low.c (struct omp_context): Add simt_stmt field.
30641 (scan_omp_for): Return omp_context *.
30642 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
30643 context to the _simt_ SIMD stmt.
30644 (lower_omp_for): For combined SIMD with sibling _simt_
30645 SIMD, make sure to use the same decls in _looptemp_
30646 clauses as in the sibling.
30648 2017-01-26 David Sherwood <david.sherwood@arm.com>
30650 PR middle-end/79212
30651 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
30654 2017-01-26 Jakub Jelinek <jakub@redhat.com>
30657 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
30658 emit fld b; fld a; if possible.
30660 * brig-builtins.def: Update copyright years.
30661 * config/arm/arm_acle_builtins.def: Update copyright years.
30663 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
30666 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
30667 constraint instead of o for the stxsd instruction.
30669 2017-01-25 Carl Love <cel@us.ibm.com>
30671 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
30672 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
30674 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
30676 * doc/invoke.texi (C++ Dialect Options): Fix typo.
30678 2017-01-25 Richard Biener <rguenther@suse.de>
30680 PR tree-optimization/69264
30681 * target.def (vector_alignment_reachable): Improve documentation.
30682 * doc/tm.texi: Regenerate.
30683 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
30685 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
30686 earlier changes with respect to TYPE_USER_ALIGN.
30687 (vector_alignment_reachable_p): Likewise. Improve dumping.
30689 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30692 * config/arm/arm.md (xordi3): Force constant operand into a register
30695 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30697 * doc/invoke.texi (-fstore-merging): Correct default optimization
30698 levels at which it is enabled.
30699 (-O): Move -fstore-merging from list to...
30702 2017-01-25 Richard Biener <rguenther@suse.de>
30705 * omp-expand.c: Include debug.h.
30706 (expand_omp_taskreg): Make sure to generate early debug before
30707 outlining anything from a function.
30708 (expand_omp_target): Likewise.
30709 (grid_expand_target_grid_body): Likewise.
30711 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
30714 * asan.c (get_translation_unit_decl): New function.
30715 (asan_add_global): Extract modules file name from globals
30716 TRANSLATION_UNIT_DECL name.
30718 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
30721 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
30722 for long calls with APCS frame and VFP.
30724 2017-01-24 David Malcolm <dmalcolm@redhat.com>
30726 * cfg.c (original_copy_tables_initialized_p): New function.
30727 * cfg.h (original_copy_tables_initialized_p): New decl.
30728 * cfgrtl.c (relink_block_chain): Guard the call to
30729 free_original_copy_tables with a call to
30730 original_copy_tables_initialized_p.
30731 * cgraph.h (symtab_node::native_rtl_p): New decl.
30732 * cgraphunit.c (symtab_node::native_rtl_p): New function.
30733 (symtab_node::needed_p): Don't assert for early assembly output
30734 for __RTL functions.
30735 (cgraph_node::finalize_function): Set "force_output" for __RTL
30737 (cgraph_node::analyze): Bail out early for __RTL functions.
30738 (analyze_functions): Update assertion to support __RTL functions.
30739 (cgraph_node::expand): Bail out early for __RTL functions.
30740 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
30742 * function.h (struct function): Update comment for field
30744 * gimple-expr.c: Include "tree-pass.h".
30745 (gimple_has_body_p): Return false for __RTL functions.
30746 * Makefile.in (OBJS): Add run-rtl-passes.o.
30747 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
30749 (gcc::pass_manager::get_clean_slate): New accessor.
30750 * passes.c: Include "insn-addr.h".
30751 (should_skip_pass_p): Add logging. Update logic for running
30752 "expand" to be compatible with both __GIMPLE and __RTL. Guard
30753 property-provider override so it is only done for gimple passes.
30755 (skip_pass): New function.
30756 (execute_one_pass): Call skip_pass when skipping passes.
30757 * read-md.c (md_reader::read_char): Support filtering
30758 the input to a subset of line numbers.
30759 (md_reader::md_reader): Initialize fields
30760 m_first_line and m_last_line.
30761 (md_reader::read_file_fragment): New function.
30762 * read-md.h (md_reader::read_file_fragment): New decl.
30763 (md_reader::m_first_line): New field.
30764 (md_reader::m_last_line): New field.
30765 * read-rtl-function.c (function_reader::create_function): Only
30766 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
30767 curr_properties. Set DECL_INITIAL to a dummy block.
30768 (read_rtl_function_body_from_file_range): New function.
30769 * read-rtl-function.h (read_rtl_function_body_from_file_range):
30771 * run-rtl-passes.c: New file.
30772 * run-rtl-passes.h: New file.
30774 2017-01-24 Jeff Law <law@redhat.com>
30776 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
30779 2017-01-24 Bin Cheng <bin.cheng@arm.com>
30781 PR tree-optimization/79159
30782 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
30783 (record_nonwrapping_iv): Improve boundary using above function if no
30784 value range information.
30786 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
30787 Martin Jambor <mjambor@suse.cz>
30789 * brig-builtins.def: New file.
30790 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
30791 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
30792 (DEF_HSAIL_SAT_BUILTIN): Likewise.
30793 (DEF_HSAIL_INTR_BUILTIN): Likewise.
30794 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
30795 * builtin-types.def (BT_INT8): New.
30796 (BT_INT16): Likewise.
30797 (BT_UINT8): Likewise.
30798 (BT_UINT16): Likewise.
30799 (BT_FN_ULONG): Likewise.
30800 (BT_FN_UINT_INT): Likewise.
30801 (BT_FN_UINT_ULONG): Likewise.
30802 (BT_FN_UINT_LONG): Likewise.
30803 (BT_FN_UINT_PTR): Likewise.
30804 (BT_FN_ULONG_PTR): Likewise.
30805 (BT_FN_INT8_FLOAT): Likewise.
30806 (BT_FN_INT16_FLOAT): Likewise.
30807 (BT_FN_UINT32_FLOAT): Likewise.
30808 (BT_FN_UINT16_FLOAT): Likewise.
30809 (BT_FN_UINT8_FLOAT): Likewise.
30810 (BT_FN_UINT64_FLOAT): Likewise.
30811 (BT_FN_UINT16_UINT32): Likewise.
30812 (BT_FN_UINT32_UINT16): Likewise.
30813 (BT_FN_UINT16_UINT16_UINT16): Likewise.
30814 (BT_FN_INT_PTR_INT): Likewise.
30815 (BT_FN_UINT_PTR_UINT): Likewise.
30816 (BT_FN_LONG_PTR_LONG): Likewise.
30817 (BT_FN_ULONG_PTR_ULONG): Likewise.
30818 (BT_FN_VOID_UINT64_UINT64): Likewise.
30819 (BT_FN_UINT8_UINT8_UINT8): Likewise.
30820 (BT_FN_INT8_INT8_INT8): Likewise.
30821 (BT_FN_INT16_INT16_INT16): Likewise.
30822 (BT_FN_INT_INT_INT): Likewise.
30823 (BT_FN_UINT_FLOAT_UINT): Likewise.
30824 (BT_FN_FLOAT_UINT_UINT): Likewise.
30825 (BT_FN_ULONG_UINT_UINT): Likewise.
30826 (BT_FN_ULONG_UINT_PTR): Likewise.
30827 (BT_FN_ULONG_ULONG_ULONG): Likewise.
30828 (BT_FN_UINT_UINT_UINT): Likewise.
30829 (BT_FN_VOID_UINT_PTR): Likewise.
30830 (BT_FN_UINT_UINT_PTR: Likewise.
30831 (BT_FN_UINT32_UINT64_PTR): Likewise.
30832 (BT_FN_INT_INT_UINT_UINT): Likewise.
30833 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
30834 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
30835 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
30836 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
30837 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
30838 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
30839 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
30840 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
30841 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
30842 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
30843 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
30844 * doc/frontends.texi: List BRIG FE.
30845 * doc/install.texi (Testing): Add BRIG tesring requirements.
30846 * doc/invoke.texi (Overall Options): Mention BRIG.
30847 * doc/standards.texi (Standards): Doucment BRIG HSA version.
30849 2017-01-24 Richard Biener <rguenther@suse.de>
30851 PR translation/79208
30852 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
30854 2017-01-24 Martin Jambor <mjambor@suse.cz>
30857 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
30858 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
30859 and known_contexts.
30861 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
30863 PR middle-end/79123
30864 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
30865 casts from signed to unsigned really don't have a range.
30867 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
30869 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
30870 GMP_RNDx for compatiblity.
30872 2017-01-24 Martin Liska <mliska@suse.cz>
30875 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
30876 that would prevent us to call alloca with -1 as argument.
30878 2017-01-24 Jakub Jelinek <jakub@redhat.com>
30880 * dwarf2out.c (output_compilation_unit_header, output_file_names):
30881 Avoid -Wformat-security warning.
30883 2017-01-23 Andrew Pinski <apinski@cavium.com>
30885 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
30888 2017-01-23 Martin Sebor <msebor@redhat.com>
30890 PR middle-end/78703
30891 * gimple-ssa-sprintf.c (warn_level): New global.
30892 (format_integer): Use it here and throughout the rest of the file.
30893 Use the same switch to compute sign as base.
30894 (maybe_warn): New function.
30895 (format_directive): Factor out warnings into maybe_warn.
30896 Add debugging output. Use warn_level.
30897 (add_bytes): Use warn_level.
30898 (pass_sprintf_length::compute_format_length): Add debugging output.
30899 (try_substitute_return_value): Same.
30900 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
30902 PR middle-end/78703
30903 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
30904 (struct fmtresult, format_integer, format_floating): Adjust.
30905 (fmtresult::fmtresult): Set max correctly in two argument ctor.
30906 (get_string_length, format_string,format_directive): Same.
30907 (pass_sprintf_length::compute_format_length): Same.
30908 (try_substitute_return_value): Simplify slightly.
30910 PR middle-end/78703
30911 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
30912 (fmtresult::operator+=): Outlined.
30913 (struct fmtresult): Add ctors.
30914 (struct conversion_spec): Rename...
30915 (struct directive): ...to this. Add and remove data members.
30916 (directive::set_width, directive::set_precision): New functions.
30917 (format_percent): Use fmtresult ctor.
30918 (get_width_and_precision): Remove.
30919 (format_integer): Make naming changes. Avoid computing width and
30921 (format_floating): Same. Adjust indentation.
30922 (format_character, format_none): New functions.
30923 (format_string): Moved character handling to format_character.
30924 (format_directive): Remove arguments, change return type.
30925 (parse_directive): New function.
30926 (pass_sprintf_length::compute_format_length): Move directive
30927 parsing to parse_directive.
30929 2017-01-23 Jakub Jelinek <jakub@redhat.com>
30931 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
30932 (assign_assembler_name_if_needed): ... this.
30933 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
30934 (assign_assembler_name_if_needed): ... this.
30935 (free_lang_data_in_cgraph): Adjust callers.
30936 * cgraphunit.c (cgraph_node::analyze): Likewise.
30937 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
30940 2017-01-23 Richard Biener <rguenther@suse.de>
30942 PR tree-optimization/79088
30943 PR tree-optimization/79188
30944 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
30945 resetting loop bounds after last path deletion. Reset loop
30946 bounds of the target loop, make code match the comments.
30947 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
30948 Make sure loops need no fixups.
30950 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
30952 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
30953 exponent support with double type for first argument.
30954 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
30955 type returned by __builtin_vec_extract_sig,
30956 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
30957 functions from "vector int" to "vector unsigned int" or from
30958 "vector long long int" to "vector unsigned long long int".
30959 Changed type returned by __builtin_vec_extract_exp,
30960 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
30961 functions from "vector int" to "vector unsigned int" or from
30962 "vector long long int" to "vector unsigned long long int".
30963 Changed return type of __builtin_vec_test_data_class,
30964 __builtin_vec_test_data_class_sp, and
30965 __builtin_vec_test_data_class_dp from "vector int" to
30966 "vector bool int" or from "vector long long int" to "vector bool
30967 long long int" and changed second argument type from "unsigned
30968 int" to "int". Added new overloaded function forms "vector float
30969 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
30970 "vector float __builtin_vec_insert_exp_sp (vector float, vector
30971 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
30972 double, vector unsigned long long int)" and "vector double
30973 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
30974 long int)". Changed return type of
30975 __builtin_scalar_test_data_class and
30976 __builtin_scalar_test_data_class_sp and
30977 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
30978 int" and changed second argument from "unsigned int" to "int".
30979 Changed type returned by __builtin_scalar_test_neg,
30980 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
30981 from "int" to "bool int". Added new overloaded function form
30982 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
30983 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
30984 exponent double-precision with floating point first argument.
30985 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
30986 documentation of scalar_test_data_class, scalar_test_neg,
30987 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
30988 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
30989 vec_test_data_class built-in functions to reflect refinements in
30990 their type signatures.
30992 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
30994 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
30996 (aarch64_elf_asm_destructor): Likewise.
30998 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
31000 PR rtl-optimization/78634
31001 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
31002 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
31003 * ifcvt.c (noce_try_cmove): Add missing cost check.
31005 PR rtl-optimization/71724
31006 * combine.c (if_then_else_cond): Look for situations where it is
31007 beneficial to undo the work of one of the recursive calls.
31009 2017-01-23 Bin Cheng <bin.cheng@arm.com>
31011 PR tree-optimization/70754
31012 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
31013 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
31014 combined stmt before it if not NULL.
31015 (combine_chains): Process refs reversely and compute dominance point
31018 2017-01-23 Martin Liska <mliska@suse.cz>
31020 PR tree-optimization/79196
31021 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
31022 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
31024 (strlen_optimize_stmt): Call the renamed function.
31026 2017-01-23 Michael Matz <matz@suse.de>
31028 PR tree-optimization/78384
31029 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
31031 2017-01-23 Richard Biener <rguenther@suse.de>
31033 PR tree-optimization/79186
31034 * tree-vrp.c (register_new_assert_for): Make sure we've seen
31035 both incoming edges before moving an assert.
31037 2017-01-23 Martin Jambor <mjambor@suse.cz>
31039 * ipa-prop.c (load_from_param_1): Removed.
31040 (load_from_unmodified_param): Bits from load_from_param_1 put back
31042 (load_from_param): Removed.
31043 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
31044 with stmt. Reverted back to use of load_from_unmodified_param.
31046 2017-01-23 Martin Jambor <mjambor@suse.cz>
31049 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
31050 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
31051 field a pointer to garbage collected vector, mark lattices and
31052 ipcp_orig_node with GTY((skip)).
31053 (ipa_get_param_count): Adjust to descriptors being a pointer.
31054 (ipa_get_param): Likewise.
31055 (ipa_get_type): Likewise.
31056 (ipa_get_param_move_cost): Likewise.
31057 (ipa_set_param_used): Likewise.
31058 (ipa_get_controlled_uses): Likewise.
31059 (ipa_set_controlled_uses): Likewise.
31060 (ipa_is_param_used): Likewise.
31061 (ipa_node_params_t): Move into garbage collector. New methods insert
31063 (ipa_node_params_sum): Annotate wth GTY(()).
31064 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
31066 (ipa_load_from_parm_agg): Adjust declaration.
31067 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
31068 * ipa-profile.c (ipa_profile): Likewise.
31069 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
31070 (ipa_populate_param_decls): Make descriptors parameter garbage
31072 (ipa_dump_param): Adjust to descriptors being a pointer.
31073 (ipa_alloc_node_params): Likewise.
31074 (ipa_initialize_node_params): Likewise.
31075 (load_from_param_1): Make descriptors parameter garbage collected.
31076 (load_from_unmodified_param): Likewise.
31077 (load_from_param): Likewise.
31078 (ipa_load_from_parm_agg): Likewise.
31079 (ipa_node_params::~ipa_node_params): Removed.
31080 (ipa_free_all_node_params): Remove call to delete operator.
31081 (ipa_node_params_t::insert): New.
31082 (ipa_node_params_t::remove): Likewise.
31083 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
31084 copy known_csts and known_contexts vectors.
31085 (ipa_read_node_info): Adjust to descriptors being a pointer.
31086 (ipcp_modif_dom_walker): Make m_descriptors field garbage
31088 (ipcp_transform_function): Make descriptors variable garbage
31091 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
31093 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
31094 * config/i386/avx512dqintrin.h: Ditto.
31095 * config/i386/avx512fintrin.h: Ditto.
31096 * config/i386/i386.c: Handle new builtins.
31097 * config/i386/i386-builtin.def: Add new builtins.
31098 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
31099 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
31101 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31102 Martin Liska <mliska@suse.cz>
31104 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
31105 * asan.c (asan_expand_poison_ifn): Support stores and use
31106 appropriate ASAN report function.
31107 * internal-fn.c (expand_ASAN_POISON_USE): New function.
31108 * internal-fn.def (ASAN_POISON_USE): Declare.
31109 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
31110 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
31111 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
31112 ASAN_POISON calls w/o LHS.
31113 * tree-ssa.c (execute_update_addresses_taken): Create clobber
31114 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
31115 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
31116 * gimplify.c (asan_poison_variables): Add attribute
31117 use_after_scope_memory to variables that really needs to live
31119 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
31120 having the attribute.
31122 2017-01-23 Martin Liska <mliska@suse.cz>
31124 * asan.c (create_asan_shadow_var): New function.
31125 (asan_expand_poison_ifn): Likewise.
31126 * asan.h (asan_expand_poison_ifn): New declaration.
31127 * internal-fn.c (expand_ASAN_POISON): Likewise.
31128 * internal-fn.def (ASAN_POISON): New builtin.
31129 * sanopt.c (pass_sanopt::execute): Expand
31130 asan_expand_poison_ifn.
31131 * tree-inline.c (copy_decl_for_dup_finish): Make function
31133 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
31134 * tree-ssa.c (is_asan_mark_p): New function.
31135 (execute_update_addresses_taken): Rewrite local variables
31136 (identified just by use-after-scope as addressable) into SSA.
31138 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31140 * doc/install.texi (Specific): opensource.apple.com uses https
31141 now. Remove trailing slash.
31143 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31145 * README.Portability: Remove note on an Irix compatibility issue.
31147 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
31149 * gcov.c (INCLUDE_ALGORITHM): Define.
31150 (INCLUDE_VECTOR): Define.
31151 No longer include <vector> and <algorithm> directly.
31153 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
31155 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
31157 * doc/invoke.texi (Code Gen Options): Ditto.
31159 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
31162 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
31164 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
31166 rtl-optimization/79125
31167 * cprop.c (local_cprop_pass): Handle cases where we make an
31168 unconditional trap.
31170 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
31174 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
31175 read from, for big endian.
31177 2017-01-20 Jiong Wang <jiong.wang@arm.com>
31179 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
31180 register pauth builtins for LP64 only.
31182 2017-01-20 Marek Polacek <polacek@redhat.com>
31185 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
31188 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
31190 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
31192 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
31193 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
31194 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
31196 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31199 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
31200 in big-endian mode when they are not a single duplicated value.
31202 2017-01-20 Richard Biener <rguenther@suse.de>
31204 * BASE-VER: Bump to 7.0.1.
31206 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
31208 * omp-low.c (omplow_simd_context): New struct. Use it...
31209 (lower_rec_simd_input_clauses): ...here and...
31210 (lower_rec_input_clauses): ...here to hold common data. Adjust all
31211 references to idx, lane, max_vf, is_simt.
31213 2017-01-20 Graham Markall <graham.markall@embecosm.com>
31215 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
31218 2017-01-20 Martin Jambor <mjambor@suse.cz>
31220 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
31221 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
31223 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
31224 (GTFILES): Rename hsa.c to hsa-common.c.
31225 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
31226 * hsa-dump.c: Likewise.
31227 * hsa-gen.c: Likewise.
31228 * hsa-regalloc.c: Likewise.
31229 * ipa-hsa.c: Likewise.
31230 * omp-expand.c: Likewise.
31231 * omp-low.c: Likewise.
31232 * toplev.c: Likewise.
31234 2017-01-20 Marek Polacek <polacek@redhat.com>
31237 * doc/invoke.texi: Document -Wduplicated-branches.
31238 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
31239 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
31240 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
31241 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
31242 return 0 only when not OEP_LEXICOGRAPHIC.
31243 (fold_build_cleanup_point_expr): Use the expression
31244 location when building CLEANUP_POINT_EXPR.
31245 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
31246 * tree.c (add_expr): Handle error_mark_node.
31248 2017-01-20 Martin Liska <mliska@suse.cz>
31251 * tree-profile.c (init_ic_make_global_vars): Do not call
31253 (gimple_init_gcov_profiler): Likewise.
31255 2017-01-20 Martin Liska <mliska@suse.cz>
31258 * cgraph.h (maybe_create_reference): Remove argument and
31260 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
31262 * ipa-cp.c (create_specialized_node): Likewise.
31263 * symtab.c (symtab_node::maybe_create_reference): Handle
31264 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
31266 2017-01-20 Martin Liska <mliska@suse.cz>
31268 * read-rtl-function.c (function_reader::create_function): Use
31269 build_decl instread of build_decl_stat.
31271 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
31273 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
31274 * config/i386/avx512dqintrin.h: Ditto.
31275 * config/i386/avx512fintrin.h: Ditto.
31276 * config/i386/i386-builtin-types.def: Add new types.
31277 * config/i386/i386.c: Handle new types.
31278 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
31279 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
31280 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
31281 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
31282 (__builtin_ia32_kshiftridi): New.
31283 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
31285 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
31289 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
31290 define to rs6000_init_stack_protect_guard.
31291 (rs6000_init_stack_protect_guard): New function.
31293 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
31294 Yunqiang Su <yunqiang.su@imgtec.com>
31296 * config.gcc (supported_defaults): Add madd4.
31297 (with_madd4): Add validation.
31298 (all_defaults): Add madd4.
31299 * config/mips/mips.opt (mmadd4): New option.
31300 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
31302 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
31304 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
31305 (ISA_HAS_FUSED_MADD4): Likewise.
31306 * doc/invoke.texi (-mmadd4): Document the new option.
31307 * doc/install.texi (--with-madd4): Document the new option.
31309 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31311 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
31312 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
31313 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
31314 (aarch64_init_pauth_hint_builtins): New.
31315 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
31316 (aarch64_expand_builtin): Expand new builtins.
31318 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31320 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
31321 * combine-stack-adj.c (no_unhandled_cfa): Handle
31322 REG_CFA_TOGGLE_RA_MANGLE.
31323 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
31324 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
31325 info for return address signing.
31326 (aarch64_expand_epilogue): Likewise.
31328 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31330 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
31331 * config/aarch64/aarch64-protos.h
31332 (aarch64_return_address_signing_enabled): New declaration.
31333 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
31335 (aarch64_expand_prologue): Sign return address before it's pushed onto
31337 (aarch64_expand_epilogue): Authenticate return address fetched from
31339 (aarch64_override_options): Sanity check for ILP32 and ISA level.
31340 (aarch64_attributes): New function attributes for "sign-return-address".
31341 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
31342 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
31343 ("*do_return"): Generate combined instructions according to key index.
31344 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
31345 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
31347 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
31348 * config/aarch64/aarch64.opt (msign-return-address=): New.
31349 * doc/extend.texi (AArch64 Function Attributes): Documents
31350 "sign-return-address=".
31351 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
31353 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
31355 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
31356 overall option summary.
31358 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31360 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
31361 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
31362 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
31363 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
31365 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
31367 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
31368 -mpower9-minmax by default for -mcpu=power9.
31369 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
31370 128-bit floating point.
31372 2017-01-20 Alan Modra <amodra@gmail.com>
31374 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
31375 optimizing for size.
31377 2017-01-20 Alan Modra <amodra@gmail.com>
31380 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
31381 for strcmp and strncmp from corresponding builtin decl.
31383 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
31385 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
31386 instead of i386/rtems-64.h.
31387 * config/i386/rtems-64.h: Remove.
31389 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
31393 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
31395 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
31397 2017-01-19 Tamar Christina <tamar.christina@arm.com>
31399 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
31400 Change int to HOST_WIDE_INT.
31401 * config/aarch64/aarch64-protos.h
31402 (aarch64_simd_gen_const_vector_dup): Likewise.
31403 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
31405 2017-01-19 David Malcolm <dmalcolm@redhat.com>
31407 * langhooks-def.h (lhd_type_for_size): New decl.
31408 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
31409 * langhooks.c (lhd_type_for_size): New function, taken from
31412 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
31414 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
31415 define_bypass for CR latency.
31416 (power9-cracked-alu): Update bypass latency and remove power9-branch.
31417 (power9-alu2): Add define_bypass for CR latency.
31419 (power9-mul): Update insn latency.
31420 (power9-mul-compare): Update insn latency, bypass latency and remove
31423 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31425 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
31427 * config/aarch64/aarch64.md
31428 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
31429 aarch64_nopcrelative_literal_loads.
31430 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
31432 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
31434 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
31435 TARGET_LOONGSON_3A.
31436 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
31438 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
31441 * config.gcc (supported_defaults): Add lxc1-sxc1.
31442 (with_lxc1_sxc1): Add validation.
31443 (all_defaults): Add lxc1-sxc1.
31444 * config/mips/mips.opt (mlxc1-sxc1): New option.
31445 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
31447 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
31448 __mips_no_lxc1_sxc1.
31449 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
31450 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
31451 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
31453 2017-01-19 Richard Biener <rguenther@suse.de>
31455 PR tree-optimization/72488
31456 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
31457 sure to restore SSA info.
31458 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
31460 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
31462 PR rtl-optimization/79121
31463 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
31464 of the inner type when shifting an extended value.
31466 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
31469 * symtab.c (symtab_node::equal_address_to): Fix comparing of
31470 interposable aliases.
31472 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
31475 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
31476 Use the evmergelohi instruction.
31477 (mov_si<mode>_e500_subreg4_2_le): Likewise.
31478 (mov_sitf_e500_subreg8_2_be): Likewise.
31479 (mov_sitf_e500_subreg12_2_le): Likewise.
31480 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
31481 (mov_si<mode>_e500_subreg4_2_be): Likewise.
31482 (mov_sitf_e500_subreg8_2_le): Likewise.
31483 (mov_sitf_e500_subreg12_2_be): Likewise.
31485 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31487 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
31488 attribute from vecsimple to vecperm.
31489 (altivec_vbpermq2): Likewise.
31491 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31494 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
31496 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31497 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
31498 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
31499 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
31500 case where N arg is SIZE_MAX.
31501 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
31502 (cmpstrsi): Add pattern.
31504 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
31506 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
31507 __builtin_vec_revb builtins.
31508 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
31509 built-in functions to support generation of the ISA 3.0 XXBR<x>
31510 vector byte reverse instructions.
31511 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
31512 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
31513 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
31514 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
31515 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
31516 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
31517 (P9V_BUILTIN_VEC_REVB): Likewise.
31518 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
31519 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
31520 (p9_xxbrq_v16qi): Likewise.
31521 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
31522 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
31523 (p9_xxbrh_v8hi): Likewise.
31524 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
31525 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
31526 vec_revb built-in functions.
31528 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
31530 PR rtl-optimization/78952
31531 * config/i386/i386.md (any_extract): New code iterator.
31532 (*insvqi_2): Use any_extract for source operand.
31533 (*insvqi_3): Use any_shiftrt for source operand.
31535 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
31537 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
31539 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
31541 2017-01-18 Matthias Klose <doko@ubuntu.com>
31543 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
31545 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31547 * config/rs6000/altivec.h (vec_bperm): Change #define.
31548 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
31549 (altivec_vbpermq2): New define_insn.
31550 (altivec_vbpermd): Likewise.
31551 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
31552 function interface.
31553 (VBPERMD): Likewise.
31554 (VBPERM): New polymorphic function interface.
31555 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
31556 Add entries for P9V_BUILTIN_VEC_VBPERM.
31557 * doc/extend.texi: Add interfaces for vec_bperm.
31559 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31561 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
31562 first letter of error messages.
31563 (s390_resolve_overloaded_builtin): Likewise.
31564 * config/s390/s390.c (s390_expand_builtin): Likewise.
31565 (s390_invalid_arg_for_unprototyped_fn): Likewise.
31566 (s390_valid_target_attribute_inner_p): Likewise.
31567 * config/s390/s390.md ("tabort"): Likewise.
31569 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
31571 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
31572 (ISA_AVOID_DIV_HILO): New macro.
31573 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
31574 (ISA_HAS_DDIV): Likewise.
31576 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
31578 * doc/invoke.texi (fabi-version): Correct number of occurrences.
31580 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
31582 * doc/invoke.texi (fabi-version): Spelling fix.
31584 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
31587 * doc/invoke.texi (fabi-version): Mention mangling fix for
31590 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
31593 * doc/invoke.texi (fabi-version): Document discriminator mangling.
31595 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
31598 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
31599 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
31601 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
31603 (stack_protect_test): Ditto.
31604 * config/rs6000/rs6000.opt (mstack-protector-guard=,
31605 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
31607 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
31608 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
31609 -mstack-protector-guard-offset=.
31610 (RS/6000 and PowerPC Options): Ditto.
31612 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
31614 * config/i386/i386.h (MASK_CLASS_P): New define.
31615 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
31616 there are no registers from different register sets also when
31617 mask registers are used. Update function comment.
31618 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
31619 to (*k/*r) and (*k/*km) alternatives.
31621 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
31623 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
31624 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
31625 (EH_RETURN_HANDLER_RTX): New define.
31626 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
31627 Force frame pointer in EH return functions.
31628 (aarch64_expand_epilogue): Add barrier for eh_return.
31629 (aarch64_final_eh_return_addr): Remove.
31630 (aarch64_eh_return_handler_rtx): New function.
31631 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
31633 (aarch64_eh_return_handler_rtx): New prototype.
31635 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31637 * config/rs6000/altivec.h (vec_rlmi): New #define.
31638 (vec_vrlnm): Likewise.
31639 (vec_rlnm): Likewise.
31640 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
31641 (UNSPEC_VRLNM): Likewise.
31642 (VIlong): New mode iterator.
31643 (altivec_vrl<VI_char>mi): New define_insn.
31644 (altivec_vrl<VI_char>nm): Likewise.
31645 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
31647 (VRLDNM): Likewise.
31648 (RLNM): New polymorphic function entry.
31649 (VRLWMI): New monomorphic function entry.
31650 (VRLDMI): Likewise.
31651 (RLMI): New polymorphic function entry.
31652 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
31653 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
31654 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
31657 2017-01-17 Jakub Jelinek <jakub@redhat.com>
31660 * dwarf2out.c (field_byte_offset): Restore the
31661 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
31662 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
31663 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
31666 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
31669 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
31671 2017-01-17 Jakub Jelinek <jakub@redhat.com>
31674 * dwarf2out.c (add_data_member_location_attribute): For constant
31675 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
31676 instead of DW_AT_data_member_location, DW_AT_bit_offset and
31677 DW_AT_byte_size attributes.
31679 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
31681 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
31682 after forcing to constant memory when the code model is medium.
31684 2017-01-17 Julia Koval <julia.koval@intel.com>
31687 * config/i386/avx512fintrin.h
31688 (_mm512_i32gather_ps): Change __addr type to void const*.
31689 (_mm512_mask_i32gather_ps): Ditto.
31690 (_mm512_i32gather_pd): Ditto.
31691 (_mm512_mask_i32gather_pd): Ditto.
31692 (_mm512_i64gather_ps): Ditto.
31693 (_mm512_mask_i64gather_ps): Ditto.
31694 (_mm512_i64gather_pd): Ditto.
31695 (_mm512_mask_i64gather_pd): Ditto.
31696 (_mm512_i32gather_epi32): Ditto.
31697 (_mm512_mask_i32gather_epi32): Ditto.
31698 (_mm512_i32gather_epi64): Ditto.
31699 (_mm512_mask_i32gather_epi64): Ditto.
31700 (_mm512_i64gather_epi32): Ditto.
31701 (_mm512_mask_i64gather_epi32): Ditto.
31702 (_mm512_i64gather_epi64): Ditto.
31703 (_mm512_mask_i64gather_epi64): Ditto.
31704 (_mm512_i32scatter_ps): Change __addr type to void*.
31705 (_mm512_mask_i32scatter_ps): Ditto.
31706 (_mm512_i32scatter_pd): Ditto.
31707 (_mm512_mask_i32scatter_pd): Ditto.
31708 (_mm512_i64scatter_ps): Ditto.
31709 (_mm512_mask_i64scatter_ps): Ditto.
31710 (_mm512_i64scatter_pd): Ditto.
31711 (_mm512_mask_i64scatter_pd): Ditto.
31712 (_mm512_i32scatter_epi32): Ditto.
31713 (_mm512_mask_i32scatter_epi32): Ditto.
31714 (_mm512_i32scatter_epi64): Ditto.
31715 (_mm512_mask_i32scatter_epi64): Ditto.
31716 (_mm512_i64scatter_epi32): Ditto.
31717 (_mm512_mask_i64scatter_epi32): Ditto.
31718 (_mm512_i64scatter_epi64): Ditto.
31719 (_mm512_mask_i64scatter_epi64): Ditto.
31720 * config/i386/avx512pfintrin.h
31721 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
31722 (_mm512_mask_prefetch_i32gather_ps): Ditto.
31723 (_mm512_mask_prefetch_i64gather_pd): Ditto.
31724 (_mm512_mask_prefetch_i64gather_ps): Ditto.
31725 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
31726 (_mm512_prefetch_i32scatter_ps): Ditto.
31727 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
31728 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
31729 (_mm512_prefetch_i64scatter_pd): Ditto.
31730 (_mm512_prefetch_i64scatter_ps): Ditto.
31731 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
31732 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
31733 * config/i386/avx512vlintrin.h
31734 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
31735 (_mm_mmask_i32gather_ps): Ditto.
31736 (_mm256_mmask_i32gather_pd): Ditto.
31737 (_mm_mmask_i32gather_pd): Ditto.
31738 (_mm256_mmask_i64gather_ps): Ditto.
31739 (_mm_mmask_i64gather_ps): Ditto.
31740 (_mm256_mmask_i64gather_pd): Ditto.
31741 (_mm_mmask_i64gather_pd): Ditto.
31742 (_mm256_mmask_i32gather_epi32): Ditto.
31743 (_mm_mmask_i32gather_epi32): Ditto.
31744 (_mm256_mmask_i32gather_epi64): Ditto.
31745 (_mm_mmask_i32gather_epi64): Ditto.
31746 (_mm256_mmask_i64gather_epi32): Ditto.
31747 (_mm_mmask_i64gather_epi32): Ditto.
31748 (_mm256_mmask_i64gather_epi64): Ditto.
31749 (_mm_mmask_i64gather_epi64): Ditto.
31750 (_mm256_i32scatter_ps): Change __addr type to void*.
31751 (_mm256_mask_i32scatter_ps): Ditto.
31752 (_mm_i32scatter_ps): Ditto.
31753 (_mm_mask_i32scatter_ps): Ditto.
31754 (_mm256_i32scatter_pd): Ditto.
31755 (_mm256_mask_i32scatter_pd): Ditto.
31756 (_mm_i32scatter_pd): Ditto.
31757 (_mm_mask_i32scatter_pd): Ditto.
31758 (_mm256_i64scatter_ps): Ditto.
31759 (_mm256_mask_i64scatter_ps): Ditto.
31760 (_mm_i64scatter_ps): Ditto.
31761 (_mm_mask_i64scatter_ps): Ditto.
31762 (_mm256_i64scatter_pd): Ditto.
31763 (_mm256_mask_i64scatter_pd): Ditto.
31764 (_mm_i64scatter_pd): Ditto.
31765 (_mm_mask_i64scatter_pd): Ditto.
31766 (_mm256_i32scatter_epi32): Ditto.
31767 (_mm256_mask_i32scatter_epi32): Ditto.
31768 (_mm_i32scatter_epi32): Ditto.
31769 (_mm_mask_i32scatter_epi32): Ditto.
31770 (_mm256_i32scatter_epi64): Ditto.
31771 (_mm256_mask_i32scatter_epi64): Ditto.
31772 (_mm_i32scatter_epi64): Ditto.
31773 (_mm_mask_i32scatter_epi64): Ditto.
31774 (_mm256_i64scatter_epi32): Ditto.
31775 (_mm256_mask_i64scatter_epi32): Ditto.
31776 (_mm_i64scatter_epi32): Ditto.
31777 (_mm_mask_i64scatter_epi32): Ditto.
31778 (_mm256_i64scatter_epi64): Ditto.
31779 (_mm256_mask_i64scatter_epi64): Ditto.
31780 (_mm_i64scatter_epi64): Ditto.
31781 (_mm_mask_i64scatter_epi64): Ditto.
31782 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
31783 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
31784 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
31785 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
31786 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
31787 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
31788 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
31789 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
31790 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
31791 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
31792 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
31793 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
31794 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
31795 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
31796 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
31797 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
31798 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
31799 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
31800 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
31801 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
31802 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
31803 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
31804 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
31805 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
31806 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
31807 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
31808 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
31809 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
31810 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
31811 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
31812 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
31813 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
31814 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
31815 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
31816 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
31817 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
31818 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
31819 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
31820 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
31821 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
31822 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
31823 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
31824 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
31825 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
31826 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
31827 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
31828 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
31829 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
31830 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
31831 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
31832 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
31833 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
31834 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
31835 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
31836 definitions accordingly.
31838 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
31839 Kuan-Lin Chen <kuanlinchentw@gmail.com>
31842 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
31845 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
31848 * ira-conflicts.c (ira_build_conflicts): Update total conflict
31849 hard regs for inner regno.
31851 2017-01-17 Martin Liska <mliska@suse.cz>
31854 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
31855 assumption and add comment.
31857 2017-01-17 Nathan Sidwell <nathan@acm.org>
31859 * ipa-visibility.c (localize_node): New function, broken out of ...
31860 (function_and_variable_visibility): ... here. Call it.
31862 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
31864 PR middle-end/77445
31865 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
31866 correctly set frequency of oudgoing edge.
31867 (duplicate_thread_path): Fix profile updating.
31869 2017-01-17 Jakub Jelinek <jakub@redhat.com>
31872 * configure.ac: Add GCC_BASE_VER.
31873 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
31874 version from BASE-VER file.
31875 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
31876 (gcc.o): Depend on $(BASEVER).
31877 * common.opt (dumpfullversion): New option.
31878 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
31879 * doc/invoke.texi: Document -dumpfullversion.
31880 * doc/install.texi: Document --with-gcc-major-version-only.
31881 * configure: Regenerated.
31883 2017-01-17 Richard Biener <rguenther@suse.de>
31885 PR tree-optimization/71433
31886 * tree-vrp.c (register_new_assert_for): Merge same asserts
31887 on all incoming edges.
31888 (process_assert_insertions_for): Handle insertions at the
31891 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
31893 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
31894 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
31896 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
31899 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
31902 2017-01-17 Alan Modra <amodra@gmail.com>
31905 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
31906 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
31907 symbolic stack limit when pic.
31909 2017-01-16 Martin Sebor <msebor@redhat.com>
31911 PR tree-optimization/78608
31912 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
31914 2017-01-16 Jeff Law <law@redhat.com>
31917 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
31918 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
31919 for several include directories that may be relative to sysroot.
31920 * config/i386/x-mingw32 (gplus_includedir): Define.
31921 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
31922 (native_system_includedir): Likewise.
31923 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
31924 override if TARGET_SYSTEM_ROOT is defined.
31925 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
31927 PR tree-optimization/79090
31928 PR tree-optimization/33562
31929 PR tree-optimization/61912
31930 PR tree-optimization/77485
31931 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
31932 and computed trims into the dump file.
31934 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
31936 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
31938 2017-01-16 Jakub Jelinek <jakub@redhat.com>
31941 * gimplify.c (gimplify_init_constructor): If want_value and
31942 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
31946 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
31947 sequence. Formatting fixes.
31948 (doloop_optimize): Formatting fixes.
31951 * gcc.c (debug_level_greater_than_spec_func): New function.
31952 (static_spec_functions): Add debug-level-gt spec function.
31953 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
31955 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
31956 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
31957 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
31958 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
31959 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
31960 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
31962 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
31964 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
31965 QImode fixups to general and mask registers only.
31967 2017-01-16 Carl Love <cel@us.ibm.com>
31969 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
31970 for built-in functions
31971 vector signed char vec_nabs (vector signed char)
31972 vector signed short vec_nabs (vector signed short)
31973 vector signed int vec_nabs (vector signed int)
31974 vector signed long long vec_nabs (vector signed long long)
31975 vector float vec_nabs (vector float)
31976 vector double vec_nabs (vector double)
31977 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
31979 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
31980 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
31981 * doc/extend.texi: Update the documentation file for the new built-in
31984 2017-01-16 Martin Sebor <msebor@redhat.com>
31986 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
31989 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31991 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
31992 UNSPEC_VSX__XXSPLTD to require special splat handling.
31994 2017-01-16 David Malcolm <dmalcolm@redhat.com>
31997 * system.h: Poison strndup.
31999 2017-01-16 Alan Modra <amodra@gmail.com>
32002 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
32005 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
32007 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
32009 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
32011 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
32012 call recog here. Assert that INSN_CODE (insn) is non-negative.
32014 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32017 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
32019 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
32020 in the currently scheduled RTL fragment.
32022 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32024 PR rtl-optimization/78751
32025 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
32028 2017-01-14 Jeff Law <law@redhat.com>
32030 PR tree-optimization/79090
32031 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
32032 variable length stores.
32033 (compute_trims): Delete dead assignment to *trim_tail.
32034 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
32037 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
32039 PR rtl-optimization/78626
32040 PR rtl-optimization/78727
32041 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
32042 of a block, and split such blocks after everything else is finished.
32044 2017-01-14 Alan Modra <amodra@gmail.com>
32047 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
32048 target legitimate_combined_insn.
32049 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
32050 (rs6000_legitimate_combined_insn): New function.
32051 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
32053 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
32054 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
32055 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
32057 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
32059 * doc/frontends.texi (G++ and GCC): Remove references to Java.
32061 2017-01-13 Jeff Law <law@redhat.com>
32063 PR tree-optimization/33562
32064 PR tree-optimization/61912
32065 PR tree-optimization/77485
32066 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
32068 (delete_dead_assignment): Likewise.
32069 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
32070 statement to delete_dead_call and delete_dead_assignment.
32072 2017-01-13 David Malcolm <dmalcolm@redhat.com>
32075 * substring-locations.c (format_warning_va): Strengthen case 1 so
32076 that both endpoints of the substring must be within the format
32077 range for just the substring to be printed.
32079 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
32081 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
32082 * config/i386/i386.c (ix86_target_string): Add missing options
32083 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
32084 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
32085 flags_other and ix86_target_other to flags2_other. Display unknown
32087 (ix86_valid_target_attribute_inner_p): Add missing options and
32088 reorder options by implied ISAs, as in ix86_target_string.
32090 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32092 * hash-table.h (hash_table::too_empty_p): New function.
32093 (hash_table::expand): Use it.
32094 (hash_table::traverse): Likewise.
32095 (hash_table::empty_slot): Use sizeof (value_type) instead of
32096 sizeof (PTR) to convert bytes to elements. Shrink the table
32097 if the current size is excessive for the current number of
32100 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32102 * ira-costs.c (record_reg_classes): Break from the inner loop
32103 early once alt_fail is known to be true. Update outer loop
32104 handling accordingly.
32106 2017-01-13 Jeff Law <law@redhat.com>
32108 * tree-ssa-dse.c (decrement_count): New function.
32109 (increment_start_addr, maybe_trim_memstar_call): Likewise.
32110 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
32111 when we know the partially dead statement is a mem* function.
32113 PR tree-optimization/61912
32114 PR tree-optimization/77485
32115 * tree-ssa-dse.c: Include expr.h.
32116 (maybe_trim_constructor_store): New function.
32117 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
32119 PR tree-optimization/33562
32120 PR tree-optimization/61912
32121 PR tree-optimization/77485
32122 * doc/invoke.texi: Document new dse-max-object-size param.
32123 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
32124 * tree-ssa-dse.c: Include params.h.
32125 (dse_store_status): New enum.
32126 (initialize_ao_ref_for_dse): New, partially extracted from
32128 (valid_ao_ref_for_dse, normalize_ref): New.
32129 (setup_live_bytes_from_ref, compute_trims): Likewise.
32130 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
32131 (maybe_trim_partially_dead_store): Likewise.
32132 (maybe_trim_complex_store): Likewise.
32133 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
32134 Track what bytes live from the original store. Return tri-state
32135 for dead, partially dead or live.
32136 (dse_dom_walker): Add constructor, destructor and new private members.
32137 (delete_dead_call, delete_dead_assignment): New extracted from
32139 (dse_optimize_stmt): Make a member of dse_dom_walker.
32140 Use initialize_ao_ref_for_dse.
32142 PR tree-optimization/33562
32143 PR tree-optimization/61912
32144 PR tree-optimization/77485
32145 * sbitmap.h (bitmap_count_bits): Prototype.
32146 (bitmap_clear_range, bitmap_set_range): Likewise.
32147 * sbitmap.c (bitmap_clear_range): New function.
32148 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
32150 2017-01-13 Martin Liska <mliska@suse.cz>
32153 * function.c (set_cfun): Add new argument force.
32154 * function.h (set_cfun): Likewise.
32155 * ipa-inline-transform.c (inline_call): Use the function when
32156 strict alising from is dropped for function we inline to.
32158 2017-01-13 Richard Biener <rguenther@suse.de>
32160 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
32161 for dumping GIMPLE INTEGER_CSTs.
32163 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32165 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
32166 to 201112L since C++17.
32168 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
32171 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
32172 if -fsanitize=kernel-address is present.
32174 2017-01-13 Richard Biener <rguenther@suse.de>
32176 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
32177 as _Literal ( type ) number in case usual suffixes do not
32178 preserve all information.
32180 2017-01-13 Richard Biener <rguenther@suse.de>
32182 PR tree-optimization/77283
32183 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
32184 and ssa-iterators.h.
32185 (is_feasible_trace): Implement a cost model based on joiner
32188 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
32191 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
32192 char or short to __float128/_Float128 directly.
32194 2017-01-12 Martin Sebor <msebor@redhat.com>
32196 to -Wformat-overflow.
32197 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
32198 (min_bytes_remaining): Same.
32199 (get_string_length): Same.
32200 (format_string): Same.
32201 (format_directive): Same.
32203 (pass_sprintf_length::handle_gimple_call): Same.
32205 2017-01-12 Jakub Jelinek <jakub@redhat.com>
32207 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
32208 info.nowrite calls with no lhs that can't throw. Return bool
32209 whether gsi_remove has been called or not.
32210 (pass_sprintf_length::handle_gimple_call): Return bool whether
32211 try_substitute_return_value called gsi_remove. Formatting fix.
32212 (pass_sprintf_length::execute): Don't use gsi_remove if
32213 handle_gimple_call returned true.
32216 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
32217 be removed due to side-effects, don't remove following barrier nor
32218 turn the successor edge into fallthru edge.
32220 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32223 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
32224 element-reversing loads and stores as not swappable.
32226 2017-01-12 Nathan Sidwell <nathan@acm.org>
32227 Nicolai Stange <nicstange@gmail.com>
32229 * combine.c (try_combine): Don't ignore result of overlap checking
32230 loop. Combine overlap & asm check into single loop.
32232 2017-01-12 Richard Biener <rguenther@suse.de>
32234 * tree-pretty-print.c (dump_generic_node): Provide -gimple
32235 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
32237 2017-01-12 Richard Biener <rguenther@suse.de>
32239 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
32240 and TS_TARGET_OPTION directly derive from TS_BASE.
32241 * tree-core.h (tree_optimization_option): Derive from tree_base.
32242 (tree_target_option): Likewise.
32244 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
32246 * config/i386/i386.c (memory_address_length): Increase len
32247 only when rip_relative_addr_p returns false.
32249 2017-01-11 Julia Koval <julia.koval@intel.com>
32251 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
32252 (OPTION_MASK_ISA_SGX_SET): New.
32253 (ix86_handle_option): Handle OPT_msgx.
32254 * config.gcc: Added sgxintrin.h.
32255 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
32256 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
32257 * config/i386/i386.c (ix86_target_string): Add -msgx.
32259 (ix86_option_override_internal): Handle new options.
32260 (ix86_valid_target_attribute_inner_p): Add sgx.
32261 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
32262 * config/i386/i386.opt: Add msgx.
32263 * config/i386/sgxintrin.h: New file.
32264 * config/i386/x86intrin.h: Add sgxintrin.h.
32266 2017-01-11 Jakub Jelinek <jakub@redhat.com>
32269 * fold-const.c (maybe_nonzero_address): Return 1 for function
32271 (tree_single_nonzero_warnv_p): Don't handle function local objects
32275 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
32278 2017-01-11 David Malcolm <dmalcolm@redhat.com>
32281 * opts.c: Include "spellcheck.h"
32282 (struct string_fragment): New struct.
32283 (struct edit_distance_traits<const string_fragment &>): New
32285 (get_closest_sanitizer_option): New function.
32286 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
32288 2017-01-11 Jakub Jelinek <jakub@redhat.com>
32290 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
32292 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
32293 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
32294 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
32295 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
32296 for initial die_offset if dwarf_split_debug_info.
32297 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
32298 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
32300 (output_skeleton_debug_sections): Formatting fix. Use
32301 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
32302 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
32304 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
32306 * config/arm/cortex-a53.md: Add bypasses for
32307 cortex_a53_r2f_cvt.
32308 (cortex_a53_r2f): Only use for transfers.
32309 (cortex_a53_f2r): Likewise.
32310 (cortex_a53_r2f_cvt): Add reservation for conversions.
32311 (cortex_a53_f2r_cvt): Likewise.
32313 2017-01-11 Tamar Christina <tamar.christina@arm.com>
32315 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
32316 to all inlined functions, change static to extern.
32318 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
32321 * config/arm/arm.c (legitimize_pic_address): Handle reference to
32323 (arm_assemble_integer): Likewise.
32325 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
32327 * config.gcc: Use new awk script to check CPU, FPU and architecture
32328 parameters for --with-... options.
32329 * config/arm/parsecpu.awk: New file
32330 * config/arm/arm-cpus.in: New file.
32331 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
32333 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
32335 * config/arm/t-arm: Update dependency rules.
32336 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
32337 of processing .def files.
32338 * config/arm/genopt.sh: Deleted.
32339 * config/arm/gentune.sh: Deleted.
32340 * config/arm/arm-cores.def: Deleted.
32341 * config/arm/arm-arches.def: Deleted.
32342 * config/arm/arm-fpus.def: Deleted.
32343 * config/arm/arm-tune.md: Regenerated.
32344 * config/arm/arm-tables.opt: Regenerated.
32345 * config/arm/arm-cpu.h: New generated file.
32346 * config/arm/arm-cpu-data.h: New generated file.
32347 * config/arm/arm-cpu-cdata.h: New generated file.
32349 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
32352 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
32354 (input_varpool_node): Unpack dynamically_initialized bit.
32356 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
32358 PR rtl-optimization/79032
32359 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
32360 the alignment of the adjusted memory reference against that of MODE,
32361 instead of the alignment of the original memory reference.
32363 2017-01-11 Martin Jambor <mjambor@suse.cz>
32365 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
32367 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
32368 decorated functions.
32370 2017-01-11 Richard Biener <rguenther@suse.de>
32372 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
32373 set range/nonnull info for PHI results. Do not set it on
32374 stmts marked for removal.
32376 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
32378 * expr.c (store_field): In the bitfield case, fetch the return value
32379 from the registers before applying a single big-endian adjustment.
32380 Always do a final load for a BLKmode value not larger than a word.
32382 2017-01-10 David Malcolm <dmalcolm@redhat.com>
32385 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
32386 that we correctly handle column numbers greater than
32387 LINE_MAP_MAX_COLUMN_NUMBER.
32389 2017-01-10 Martin Sebor <msebor@redhat.com>
32391 PR middle-end/78245
32392 * gimple-ssa-sprintf.c (get_destination_size): Call
32393 {init,fini}object_sizes.
32394 * tree-object-size.c (addr_object_size): Adjust.
32395 (pass_through_call): Adjust.
32396 (pass_object_sizes::execute): Adjust.
32397 * tree-object-size.h (fini_object_sizes): Declare.
32399 2017-01-10 Martin Sebor <msebor@redhat.com>
32401 PR tree-optimization/78775
32402 * builtins.c (get_size_range): Move...
32403 * calls.c: ...to here.
32404 (alloc_max_size): Accept zero argument.
32405 (operand_signed_p): Remove.
32406 (maybe_warn_alloc_args_overflow): Call get_size_range.
32407 * calls.h (get_size_range): Declare.
32409 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
32411 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
32412 from TI's devices.csv file as of September 2016.
32413 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
32415 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
32417 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
32418 * doc/invoke.texi: Likewise.
32419 * doc/md.texi: Likewise.
32420 * doc/objc.texi: Likewise.
32422 2017-01-10 Joshua Conner <joshconner@google.com>
32424 * config/arm/fuchsia-elf.h: New file.
32425 * config/fuchsia.h: New file.
32426 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
32427 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
32429 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
32431 2016-01-10 Richard Biener <rguenther@suse.de>
32433 PR tree-optimization/79034
32434 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
32435 Propagate out degenerate PHIs in the joiner.
32437 2017-01-10 Martin Liska <mliska@suse.cz>
32439 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
32440 (sort_congruence_classes_by_decl_uid): Likewise.
32441 (sort_congruence_class_groups_by_decl_uid): Likewise.
32442 (sem_item_optimizer::merge_classes): Sort class, groups in these
32443 classes and members in the groups by DECL_UID of declarations.
32444 This would make merge operations stable.
32446 2017-01-10 Martin Liska <mliska@suse.cz>
32448 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
32449 usage of m_classes_vec.
32450 (sem_item_optimizer::~sem_item_optimizer): Likewise.
32451 (sem_item_optimizer::get_group_by_hash): Likewise.
32452 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
32453 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
32454 (sem_item_optimizer::verify_classes): Likewise.
32455 (sem_item_optimizer::process_cong_reduction): Likewise.
32456 (sem_item_optimizer::dump_cong_classes): Likewise.
32457 (sem_item_optimizer::merge_classes): Likewise.
32458 * ipa-icf.h (congruence_class_hash): Rename from
32459 congruence_class_group_hash. Remove declaration of m_classes_vec.
32461 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
32463 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
32464 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
32465 * config.gcc: Add avx512vpopcntdqintrin.h.
32466 * config/i386/avx512vpopcntdqintrin.h: New.
32467 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
32468 * config/i386/i386-builtin-types.def: Add new types.
32469 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
32470 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
32471 __builtin_ia32_vpopcountq_v8di_mask): New.
32472 * config/i386/i386-c.c (ix86_target_macros_internal): Define
32473 __AVX512VPOPCNTDQ__.
32474 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
32475 (PTA_AVX512VPOPCNTDQ): Define.
32476 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
32477 TARGET_AVX512VPOPCNTDQ_P): Define.
32478 * config/i386/i386.opt: Add mavx512vpopcntdq.
32479 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
32480 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
32482 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32484 PR middle-end/77484
32485 * predict.def (PRED_CALL): Set to 67.
32487 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
32489 * expr.c (store_field): In the bitfield case, if the value comes from
32490 a function call and is of an aggregate type returned in registers, do
32491 not modify the field mode; extract the value in all cases if the mode
32492 is BLKmode and the size is not larger than a word.
32494 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
32497 * config/i386/cpuid.h: Fix undefined behavior.
32499 2017-01-04 Jeff Law <law@redhat.com>
32501 PR tree-optimization/79007
32502 PR tree-optimization/67955
32503 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
32504 conservative for pt.null when flag_non_call_exceptions is on.
32506 2017-01-09 Jakub Jelinek <jakub@redhat.com>
32508 PR translation/79019
32509 PR translation/79020
32510 * params.def (PARAM_INLINE_MIN_SPEEDUP,
32511 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
32512 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
32514 * config/avr/avr.opt (maccumulate-args): Likewise.
32515 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
32516 * common.opt (freport-bug): Likewise.
32517 * cif-code.def (CIF_FINAL_ERROR): Likewise.
32518 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
32519 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
32520 translatable string.
32521 * config/i386/i386.c (function_value_32): Likewise.
32522 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
32523 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
32525 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
32526 * common/config/msp430/msp430-common.c (msp430_handle_option):
32528 * symtab.c (symtab_node::verify_base): Likewise.
32529 * opts.c (set_debug_level): Likewise.
32530 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
32531 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
32532 missing whitespace to translatable strings.
32533 * config/avr/avr.md (bswapsi2): Fix typo in comment.
32534 * config/sh/superh.h: Likewise.
32535 * config/i386/xopintrin.h: Likewise.
32536 * config/i386/znver1.md: Likewise.
32537 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
32538 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
32539 * double-int.h (struct double_int): Likewise.
32540 * double-int.c (div_and_round_double): Likewise.
32541 * wide-int.cc: Likewise.
32542 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
32543 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
32544 * cfgcleanup.c (crossjumps_occured): Renamed to ...
32545 (crossjumps_occurred): ... this.
32546 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
32549 PR tree-optimization/78899
32550 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
32551 returning bool return struct loop *, NULL for failure and the new
32553 (versionable_outer_loop_p): Don't version outer loop if it has
32554 dont_vectorized bit set.
32555 (tree_if_conversion): When versioning outer loop, ensure
32556 tree_if_conversion is performed also on the inner loop of the
32557 non-vectorizable outer loop copy.
32558 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
32559 LOOP_VECTORIZED in inner loop of the scalar outer loop and
32560 prevent vectorization of it.
32561 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
32562 the outer loop vectorization of the non-scalar version is attempted
32563 before vectorization of the inner loop in scalar version. If
32564 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
32565 vectorization of its inner loop.
32566 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
32567 has 2 inner loops, rename also on edges from bb whose single pred
32568 is outer_loop->header. Fix typo in function comment.
32570 2017-01-09 Martin Sebor <msebor@redhat.com>
32573 * asan.c (asan_emit_stack_protection): Increase local buffer size
32574 to avoid snprintf truncation warning.
32576 2017-01-09 Andrew Pinski <apinski@cavium.com>
32578 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
32579 to reference thunderx2t99 for the tuning structure
32580 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
32582 (thunderx2t99_extra_costs): This.
32583 * config/aarch64/aarch64-tune.md: Regenerate.
32584 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
32585 (vulcan_addrcost_table): This.
32586 (vulcan_regmove_cost): Rename to ...
32587 (thunderx2t99_regmove_cost): This.
32588 (vulcan_vector_cost): Rename to ...
32589 (thunderx2t99_vector_cost): this.
32590 (vulcan_branch_cost): Rename to ...
32591 (thunderx2t99_branch_cost): This.
32592 (vulcan_tunings): Rename to ...
32593 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
32594 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
32596 2017-01-09 Martin Jambor <mjambor@suse.cz>
32600 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
32601 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
32602 (propagate_vr_accross_jump_function): Use the above function for all
32603 value range computations for pass-through jump functions and type
32604 converasion from explicit value range values.
32605 (ipcp_propagate_stage): Do not attempt to deduce types of formal
32606 parameters from TYPE_ARG_TYPES.
32607 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
32608 (ipa_write_node_info): Stream type of the actual argument.
32609 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
32611 2017-01-09 Martin Liska <mliska@suse.cz>
32614 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
32615 (lookup_compiler): Do not show error message with have_E.
32617 2017-01-09 Jakub Jelinek <jakub@redhat.com>
32619 PR tree-optimization/78938
32620 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
32621 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
32622 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
32623 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
32626 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32628 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
32631 2017-01-09 Richard Biener <rguenther@suse.de>
32633 PR tree-optimization/78997
32634 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
32635 name condition properly.
32637 2017-01-09 Richard Biener <rguenther@suse.de>
32640 * dwarf2out.c (is_cxx): New overload with context.
32641 (is_naming_typedef_decl): Use it.
32643 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
32645 * invoke.texi (Option Summary): Correct spacing in option lists
32646 and add line breaks to fix over-long lines.
32648 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
32650 PR middle-end/17660
32652 * extend.texi (Common Variable Attributes): Add xref to GCC
32653 Internals manual to explain mode attribute keywords.
32655 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
32658 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
32659 and Preprocessor Options.
32660 (Options for Linking): Document -pthread here....
32661 (RS/6000 and PowerPC Options): ...not here.
32662 (Solaris 2 Options): ...or here.
32663 * doc/cppopts.texi: Document -pthread.
32665 2017-01-08 Martin Sebor <msebor@redhat.com>
32667 PR middle-end/77708
32668 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
32669 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
32670 New member functions.
32671 (format_directive): Used them.
32673 (pass_sprintf_length::handle_gimple_call): Same.
32674 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
32675 to avoid truncation for any argument.
32676 (extract_affine_mul): Same.
32677 * tree.c (get_file_function_name): Same.
32679 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32681 PR middle-end/77484
32682 * predict.def (PRED_INDIR_CALL): Set to 86.
32684 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
32686 PR preprocessor/54124
32687 * doc/cppopts.texi: Reformat -d subtable to list the full name
32688 of the options. Add cross-reference to the docs for the general
32689 compiler -d options.
32690 * doc/invoke.texi (Developer Options): Add cross-reference to the
32691 preprocessor-specific -d option documentation.
32693 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
32695 PR preprocessor/13498
32696 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
32697 redudant material, and reflect new command-line options.
32698 (System Headers): Likewise.
32700 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
32702 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
32703 -isystem, and -idirafter. Copy-edit.
32704 * doc/cppopts.texi: Copy-edit. Remove contradiction about
32705 default for -ftrack-macro-expansion. Delete obsolete and
32706 badly-formatted implementation details about -fdebug-cpp output.
32707 * doc/cppwarnopts.texi: Copy-edit.
32709 2017-01-07 David Malcolm <dmalcolm@redhat.com>
32712 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
32713 that the transition from a max line width >= 1<<10 to narrower
32714 lines works correctly.
32716 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
32718 * doc/options.texi (PerFunction): New.
32719 * opt-functions.awk (switch_flags): Map both Optimization and
32720 PerFunction to CL_OPTIMIZATION.
32721 * opth-gen.awk: Test for PerFunction flag along with
32723 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
32724 it only when the latter is present. Skip those that don't in
32725 the hash function generator.
32726 * common.opt (fvar-tracking): Mark as PerFunction instead of
32728 (fvar-tracking-assignments): Likewise.
32729 (fvar-tracking-assignments-toggle): Likewise.
32730 (fvar-tracking-uninit): Likewise.
32732 2017-01-07 Jakub Jelinek <jakub@redhat.com>
32734 PR translation/79018
32735 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
32738 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
32741 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
32742 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
32743 TARGET_LONG_JUMP_TABLE_OFFSETS.
32744 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
32745 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
32746 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
32747 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
32748 * config/m68k/m68k.md (tablejump expander): Likewise.
32749 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
32750 TARGET_LONG_JUMP_TABLE_OFFSETS.
32751 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
32752 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
32754 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
32755 David Holsgrove <david.holsgrove@xilinx.com>
32757 * common/config/microblaze/microblaze-common.c
32758 (TARGET_EXCEPT_UNWIND_INFO): Remove.
32759 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
32761 * config/microblaze/microblaze.c (microblaze_must_save_register)
32762 (microblaze_expand_epilogue, microblaze_return_addr): Handle
32764 (microblaze_eh_return): New function.
32765 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
32766 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
32767 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
32768 * config/microblaze/microblaze.md (eh_return): New pattern.
32770 2017-01-06 Jakub Jelinek <jakub@redhat.com>
32772 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
32773 GCC_DIAGNOSTIC_STRINGIFY): Define.
32775 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
32777 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32779 * config/arm/arm.md (<mcrr>): New.
32781 * config/arm/arm.c (arm_arch5te): New.
32782 (arm_option_override): Set arm_arch5te.
32783 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
32785 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
32786 (arm_mcrr_qualifiers): ... this. New.
32787 (MRRC_QUALIFIERS): Define to...
32788 (arm_mrrc_qualifiers): ... this. New.
32789 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
32791 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
32792 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
32793 (MRRCI, mrrc, MRRC): New.
32794 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
32795 VUNSPEC_MRRC2): New.
32797 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32799 * config/arm/arm.md (<mcr>): New.
32801 * config/arm/arm.c (arm_coproc_builtin_available): Add
32802 support for mcr, mrc, mcr2 and mrc2.
32803 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
32804 (arm_mcr_qualifiers): ... this. New.
32805 (MRC_QUALIFIERS): Define to ...
32806 (arm_mrc_qualifiers): ... this. New.
32807 (MCR_QUALIFIERS): Define to ...
32808 (arm_mcr_qualifiers): ... this. New.
32809 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
32811 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
32812 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
32813 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
32814 VUNSPEC_MRC2): New.
32816 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32818 * config/arm/arm.md (*ldc): New.
32822 * config/arm/arm.c (arm_coproc_builtin_available): Add
32823 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
32824 (arm_coproc_ldc_stc_legitimate_address): New.
32825 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
32826 'qualifier_const_pointer'.
32827 (LDC_QUALIFIERS): Define to...
32828 (arm_ldc_qualifiers): ... this. New.
32829 (STC_QUALIFIERS): Define to...
32830 (arm_stc_qualifiers): ... this. New.
32831 * config/arm/arm-protos.h
32832 (arm_coproc_ldc_stc_legitimate_address): New.
32833 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
32834 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
32835 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
32836 stc2, stcl, stc2l): New.
32837 * config/arm/constraints.md (Uz): New.
32838 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
32839 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
32840 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
32841 VUNSPEC_STC2L): New.
32843 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32845 * config/arm/arm.md (<cdp>): New.
32846 * config/arm/arm.c (neon_const_bounds): Rename this ...
32847 (arm_const_bounds): ... this.
32848 (arm_coproc_builtin_available): New.
32849 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
32850 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
32851 (CDP_QUALIFIERS): Define to...
32852 (arm_cdp_qualifiers): ... this. New.
32854 (arm_expand_builtin_args): Add case for 6 arguments.
32855 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
32856 (arm_const_bounds): ... this.
32857 (arm_coproc_builtin_available): New.
32858 * config/arm/arm_acle.h (__arm_cdp): New.
32860 * config/arm/arm_acle_builtins.def (cdp): New.
32862 * config/arm/iterators.md (CDPI,CDP,cdp): New.
32863 * config/arm/neon.md: Rename all 'neon_const_bounds' to
32864 'arm_const_bounds'.
32865 * config/arm/types.md (coproc): New.
32866 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
32867 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
32868 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
32869 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
32871 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32873 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
32874 (UBINOP_QUALIFIERS): New.
32876 (acle_builtin_data): New. Change comment.
32877 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
32878 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
32879 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
32880 arm_acle_builtins.def.
32881 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
32882 (arm_init_acle_builtins): New.
32883 (CRC32_BUILTIN): Remove.
32884 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
32885 crc32cb, crc32ch and crc32cw.
32886 (arm_init_crc32_builtins): Remove.
32887 (arm_init_builtins): Use arm_init_acle_builtins rather
32888 than arm_init_crc32_builtins.
32889 (arm_expand_acle_builtin): New.
32890 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
32891 * config/arm/arm_acle_builtins.def: New.
32893 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32895 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
32896 (arm_builtin_datum): ... this.
32897 (arm_init_neon_builtin): Rename to ...
32898 (arm_init_builtin): ... this. Add a new parameters PREFIX
32899 and USE_SIG_IN_NAME.
32900 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
32901 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
32902 'arm_builtin_datum'.
32903 (arm_init_vfp_builtins): Likewise.
32904 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
32905 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
32906 (arm_expand_neon_args): Rename to ...
32907 (arm_expand_builtin_args): ... this. Rename builtin_arg
32908 enum values and differentiate between ARG_BUILTIN_MEMORY
32909 and ARG_BUILTIN_NEON_MEMORY.
32910 (arm_expand_neon_builtin_1): Rename to ...
32911 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
32912 values, arm_expand_builtin_args and add bool parameter NEON.
32913 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
32914 (arm_expand_vfp_builtin): Likewise.
32915 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
32917 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32919 PR middle-end/77484
32920 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
32921 * predict.c (tree_estimate_probability_bb): Reverse direction of
32922 polymorphic call predictor.
32924 2017-01-06 David Malcolm <dmalcolm@redhat.com>
32926 * passes.c (execute_one_pass): Split out pass-skipping logic into...
32927 (determine_pass_name_match): ...this new function and...
32928 (should_skip_pass_p): ...this new function.
32930 2017-01-06 Nathan Sidwell <nathan@acm.org>
32932 * ipa-visibility.c (function_and_variable_visibility): Reformat
32933 comments and long lines. Remove extrneous if.
32934 * symtab.c (symtab_node::make_decl_local): Fix code format.
32935 (symtab_node::set_section_for_node): Fix comment typo.
32937 2017-01-06 Martin Liska <mliska@suse.cz>
32940 * lra-constraints.c: Rename invariant to lra_invariant.
32941 * predict.c (set_even_probabilities): Initialize e to NULL.
32943 2017-01-05 Martin Sebor <msebor@redhat.com>
32945 PR tree-optimization/78910
32946 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
32947 (format_integer): Correct off-by-one error in the handling
32948 of precision with negative numbers in signed conversions..
32950 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
32952 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
32954 2017-01-05 Jakub Jelinek <jakub@redhat.com>
32956 PR tree-optimization/71016
32957 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
32958 factor_out_conditional_conversion. Formatting fix.
32959 (factor_out_conditional_conversion): Add cond_stmt argument.
32960 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
32961 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
32964 2017-01-05 David Malcolm <dmalcolm@redhat.com>
32966 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
32967 read-rtl-function.o, and selftest-rtl.o.
32968 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
32969 (selftest::aarch64_test_loading_full_dump): New function.
32970 (selftest::aarch64_run_selftests): New function.
32971 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
32972 selftest::aarch64_run_selftests.
32973 * config/i386/i386.c
32974 (selftest::ix86_test_loading_dump_fragment_1): New function.
32975 (selftest::ix86_test_loading_call_insn): New function.
32976 (selftest::ix86_test_loading_full_dump): New function.
32977 (selftest::ix86_test_loading_unspec): New function.
32978 (selftest::ix86_run_selftests): Call the new functions.
32979 * emit-rtl.c (maybe_set_max_label_num): New function.
32980 * emit-rtl.h (maybe_set_max_label_num): New decl.
32981 * function.c (instantiate_decls): Guard call to
32982 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
32983 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
32985 * gensupport.c (gen_reader::gen_reader): Pass "false"
32986 for new "compact" param of rtx_reader.
32987 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
32988 rather than an empty string for NULL strings.
32989 * read-md.c: Potentially include config.h rather than bconfig.h.
32990 Wrap include of errors.h with #ifdef GENERATOR_FILE.
32991 (have_error): New global, copied from errors.c.
32992 (md_reader::read_name): Rename to...
32993 (md_reader::read_name_1): ...this, adding "out_loc" param,
32994 and converting "missing name or number" to returning false, rather
32996 (md_reader::read_name): Reimplement in terms of read_name_1.
32997 (md_reader::read_name_or_nil): New function.
32998 (md_reader::read_string): Handle "(nil)" by returning NULL.
32999 (md_reader::md_reader): Add new param "compact".
33000 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
33001 (md_reader::read_file): New method.
33002 * read-md.h (md_reader::md_reader): Add new param "compact".
33003 (md_reader::read_file): New method.
33004 (md_reader::is_compact): New accessor.
33005 (md_reader::read_name): Convert return type from void to file_location.
33006 (md_reader::read_name_or_nil): New decl.
33007 (md_reader::read_name_1): New decl.
33008 (md_reader::m_compact): New field.
33009 (noop_reader::noop_reader): Pass "false" for new "compact" param
33011 (rtx_reader::rtx_reader): Add new "compact" param.
33012 (rtx_reader::read_rtx_operand): Make virtual and convert return
33013 type from void to rtx.
33014 (rtx_reader::read_until): New decl.
33015 (rtx_reader::handle_any_trailing_information): New virtual function.
33016 (rtx_reader::postprocess): New virtual function.
33017 (rtx_reader::finalize_string): New virtual function.
33018 (rtx_reader::m_in_call_function_usage): New field.
33019 (rtx_reader::m_reuse_rtx_by_id): New field.
33020 * read-rtl-function.c: New file.
33021 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
33022 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
33023 (selftest::verify_three_block_rtl_cfg): New decl.
33024 * read-rtl-function.h: New file.
33025 * read-rtl.c: Potentially include config.h rather than bconfig.h.
33026 For host, include function.h, memmodel.h, and emit-rtl.h.
33027 (one_time_initialization): New function.
33028 (struct compact_insn_name): New struct.
33029 (compact_insn_names): New array.
33030 (find_code): Handle insn codes in compact dumps.
33031 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
33032 (bind_subst_iter_and_attr): Likewise.
33033 (add_condition_to_string): Likewise.
33034 (add_condition_to_rtx): Likewise.
33035 (apply_attribute_uses): Likewise.
33036 (add_current_iterators): Likewise.
33037 (apply_iterators): Likewise.
33038 (initialize_iterators): Guard usage of apply_subst_iterator with
33039 #ifdef GENERATOR_FILE.
33040 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
33041 (md_reader::read_mapping): Likewise.
33042 (add_define_attr_for_define_subst): Likewise.
33043 (add_define_subst_attr): Likewise.
33044 (read_subst_mapping): Likewise.
33045 (check_code_iterator): Likewise.
33046 (rtx_reader::read_rtx): Likewise. Move one-time initialization
33048 (one_time_initialization): New function.
33049 (rtx_reader::read_until): New method.
33050 (read_flags): New function.
33051 (parse_reg_note_name): New function.
33052 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
33053 Handle reuse_rtx ids.
33054 Wrap iterator lookup within #ifdef GENERATOR_FILE.
33055 Add parsing support for RTL dumps, mirroring the special-cases in
33056 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
33057 values, and calling handle_any_trailing_information.
33058 (rtx_reader::read_rtx_operand): Convert return type from void
33059 to rtx, returning return_rtx. Handle case 'e'. Call
33060 finalize_string on XSTR and XTMPL fields.
33061 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
33062 "(nil)" values were omitted. Call the postprocess vfunc on the
33064 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
33065 class ctor. Initialize m_in_call_function_usage. Call
33066 one_time_initialization.
33067 * rtl-tests.c (selftest::test_uncond_jump): Call
33068 set_new_first_and_last_insn.
33069 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
33070 * selftest-rtl.c: New file.
33071 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
33072 (selftest::get_insn_by_uid): New decl.
33073 * selftest-run-tests.c (selftest::run_tests): Call
33074 read_rtl_function_c_tests.
33075 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
33076 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
33079 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
33081 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
33082 operands in a special way. Assert that pos+len <= mode precision.
33084 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33086 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
33087 3 argument Alias with unlimited for the negative form.
33088 (fno-vect-cost-model): Removed.
33090 2017-01-05 Martin Liska <mliska@suse.cz>
33092 * hsa-gen.c (gen_hsa_divmod): New function.
33093 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
33095 2017-01-05 Martin Liska <mliska@suse.cz>
33098 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
33101 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33103 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
33104 small constant length operands.
33106 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33108 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
33109 between loop iterations.
33111 2017-01-05 Martin Liska <mliska@suse.cz>
33114 * gimplify.c (gimplify_decl_expr): Compare to
33115 asan_poisoned_variables instread of checking flags.
33116 (gimplify_target_expr): Likewise.
33117 (gimplify_expr): Likewise.
33118 (gimplify_function_tree): Conditionally initialize
33119 asan_poisoned_variables.
33121 2017-01-04 Jeff Law <law@redhat.com>
33123 PR tree-optimizatin/78812
33124 * rtl.h (contains_mem_rtx_p): Prototype.
33125 * ifcvt.c (containts_mem_rtx_p): Move from here to...
33126 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
33127 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
33128 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
33129 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
33131 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33133 * input.c (assert_char_at_range): Default-initialize actual_range.
33135 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33137 * df-scan.c (df_ref_create_structure): Make regno unsigned,
33138 to match the caller.
33140 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33142 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
33143 insns after final jump in test to emit dummy move.
33145 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33147 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
33148 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
33150 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33152 * multiple_target.c (create_dispatcher_calls): Init e_next.
33153 * tree-ssa-loop-split.c (split_loop): Init border.
33154 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
33157 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
33162 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
33163 (altivec_register_operand): Do not return true if the operand
33164 contains a SUBREG mixing SImode and SFmode.
33165 (vsx_register_operand): Likewise.
33166 (vsx_reg_sfsubreg_ok): New predicate.
33167 (vfloat_operand): Do not return true if the operand contains a
33168 SUBREG mixing SImode and SFmode.
33169 (vint_operand): Likewise.
33170 (vlogical_operand): Likewise.
33171 (gpc_reg_operand): Likewise.
33172 (int_reg_operand): Likewise.
33173 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
33174 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
33175 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
33177 (rs6000_emit_move_si_sf_subreg): New helper function.
33178 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
33179 fixup SUBREGs involving SImode and SFmode.
33180 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
33181 numbers for the new peephole2 optimization.
33182 (peephole2 for SFmode unions): New peephole2 to optimize cases in
33183 the GLIBC math library that do AND/IOR/XOR operations on single
33184 precision floating point.
33185 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
33186 target macros to say whether we need to avoid SUBREGs mixing
33188 (TARGET_ALLOW_SF_SUBREG): Likewise.
33189 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
33190 (UNSPEC_SI_FROM_SF): Likewise.
33191 (iorxor): Change spacing.
33192 (and_ior_xor): New iterator for AND, IOR, and XOR.
33193 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
33194 (movdi_from_sf_zero_ext): Likewise.
33195 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
33196 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
33197 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
33198 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
33199 (fms<mode>4): Likewise.
33200 (fnma<mode>4): Likewise.
33201 (fnms<mode>4): Likewise.
33202 (nfma<mode>4): Likewise.
33203 (nfms<mode>4): Likewise.
33205 2017-01-04 Marek Polacek <polacek@redhat.com>
33208 * doc/invoke.texi: Document -Wpointer-compare.
33210 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33212 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
33215 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
33216 descriptions for -gdwarf-5 and emit them as uleb128 instead of
33219 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
33222 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
33223 documentation of the powerpc_popcntb_ok attribute.
33224 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
33225 code to issue warning messages if a requested CPU configuration is
33226 not supported by the binary (assembler and loader) toolchain.
33227 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
33228 made to define a built-in function that has been disabled.
33229 (paired_init_builtins): Add assertion to prevent ICE if attempt is
33230 made to define a built-in function that has been disabled.
33231 (altivec_init_builtins): Add comment explaining why definition
33232 of the DST built-in functions is not preceded by an assertion
33233 check. Add assertions to prevent ICE if attempts are made to
33234 define an altivec predicate or an abs* built-in function that has
33236 (htm_init_builtins): Add comment explaining why definition of the
33237 htm built-in functions is not preceded by an assertion check.
33239 2017-01-04 Jeff Law <law@redhat.com>
33241 PR tree-optimizatin/67955
33242 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
33243 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
33244 the points-to solution does not include pt_null. Use DECL_PT_UID
33247 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
33249 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
33250 Use gen_int_mode instead of gen_lopwart for const_int operands.
33252 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33254 PR tree-optimization/71563
33255 * match.pd: Simplify X << Y into X if Y is known to be 0 or
33256 out of range value - has low bits known to be zero.
33258 2017-01-04 Alan Modra <amodra@gmail.com>
33260 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
33261 * configure: Regenerate.
33262 * config.in: Regenerate.
33264 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33267 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
33268 a substring of the message, but strcmp with the whole message. Ifdef
33269 ENABLE_NLS, translate the message first using dgettext.
33271 2017-01-03 Jeff Law <law@redhat.com>
33273 PR tree-optimizatin/78856
33274 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
33275 (mark_threaded_blocks): Remove code to truncate thread paths that
33276 cross multiple loop headers. Instead invalidate the cached loop
33277 iteration information and handle case of a thread path walking
33278 into an irreducible region.
33280 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
33283 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
33284 assertions. Add support for doing the signbit if the IEEE 128-bit
33285 floating point value is in a GPR.
33286 * config/rs6000/rs6000.md (Fsignbit): Delete.
33287 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
33288 Update the length attribute if the value is in a GPR.
33289 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
33290 the sign or zero extension instruction, since the value is always 0/1.
33291 (signbit<mode>2_dm2): Delete using <Fsignbit>.
33294 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
33295 extracting SImode to a GPR register so that we can generate a
33296 store, limit the vector to be in a traditional Altivec register
33297 for the vextuwrx instruction.
33299 2017-01-03 Ian Lance Taylor <iant@google.com>
33301 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
33303 2017-01-03 Martin Sebor <msebor@redhat.com>
33305 PR tree-optimization/78696
33306 * gimple-ssa-sprintf.c (format_floating): Correct handling of
33307 precision. Use MPFR for %f for greater fidelity. Correct handling
33309 (pass_sprintf_length::compute_format_length): Set width and precision
33310 specified by asrerisk to void_node for vararg functions.
33311 (try_substitute_return_value): Adjust dump output.
33313 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
33315 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
33317 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
33319 * doc/invoke.texi (SPARC options): Document -mlra as the default.
33320 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
33321 -mlra/-mno-lra was passed to the compiler.
33323 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
33325 PR rtl-optimization/65618
33326 * emit-rtl.c (try_split): Move initialization of "before" and
33327 "after" to just before the call to emit_insn_after_setloc.
33329 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
33331 * doc/md.texi (Standard Names): Remove reference to Java frontend.
33333 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
33335 * dwarf2out.c (gen_enumeration_type_die): When
33336 -gno-strict-dwarf, add a DW_AT_encoding attribute.
33338 2017-01-03 Jakub Jelinek <jakub@redhat.com>
33340 PR tree-optimization/78965
33341 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
33342 Change first argument from const call_info & to call_info &. For %n
33343 set info.nowrite to false.
33345 PR middle-end/78901
33346 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
33347 possibly throwing calls.
33349 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
33350 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
33351 and fns handling, rather than in a separate case SSA_NAME.
33353 2017-01-02 Jeff Law <law@redhat.com>
33355 * config/darwin-driver.c (darwin_driver_init): Const-correctness
33356 fixes for first_period and second_period variables.
33358 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
33361 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
33362 (*insvqi_1): New insn pattern.
33363 (*insvqi_1_mem_rex64): Ditto.
33364 (*insvqi_2): Ditto.
33365 (*insvqi_3): Rename from *insvqi.
33367 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
33369 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
33371 * doc/cfg.texi (Edges): Remove reference to Java.
33372 (Maintaining the CFG): Ditto.
33374 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33376 PR middle-end/77674
33377 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
33378 transparent aliases.
33380 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33382 PR middle-end/77484
33383 * predict.def (PRED_CALL): Update hitrate.
33384 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
33385 * predict.c (tree_estimate_probability_bb): Split CALL predictor
33386 into direct/indirect/polymorphic variants.
33388 2017-01-01 Jakub Jelinek <jakub@redhat.com>
33390 Update copyright years.
33392 * gcc.c (process_command): Update copyright notice dates.
33393 * gcov-dump.c (print_version): Ditto.
33394 * gcov.c (print_version): Ditto.
33395 * gcov-tool.c (print_version): Ditto.
33396 * gengtype.c (create_file): Ditto.
33397 * doc/cpp.texi: Bump @copying's copyright year.
33398 * doc/cppinternals.texi: Ditto.
33399 * doc/gcc.texi: Ditto.
33400 * doc/gccint.texi: Ditto.
33401 * doc/gcov.texi: Ditto.
33402 * doc/install.texi: Ditto.
33403 * doc/invoke.texi: Ditto.
33405 Copyright (C) 2017 Free Software Foundation, Inc.
33407 Copying and distribution of this file, with or without modification,
33408 are permitted in any medium without royalty provided the copyright
33409 notice and this notice are preserved.