1 2017-11-03 Jeff Law <law@redhat.com>
3 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
4 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
5 to probe at the start of a noreturn function.
7 2017-11-03 Jakub Jelinek <jakub@redhat.com>
9 PR tree-optimization/78821
10 * gimple-ssa-store-merging.c: Update the file comment.
11 (MAX_STORE_ALIAS_CHECKS): Define.
12 (struct store_operand_info): New type.
13 (store_operand_info::store_operand_info): New constructor.
14 (struct store_immediate_info): Add rhs_code and ops data members.
15 (store_immediate_info::store_immediate_info): Add rhscode, op0r
16 and op1r arguments to the ctor, initialize corresponding data members.
17 (struct merged_store_group): Add load_align_base and load_align
19 (merged_store_group::merged_store_group): Initialize them.
20 (merged_store_group::do_merge): Update them.
21 (merged_store_group::apply_stores): Pick the constant for
22 encode_tree_to_bitpos from one of the two operands, or skip
23 encode_tree_to_bitpos if neither operand is a constant.
24 (class pass_store_merging): Add process_store method decl. Remove
25 bool argument from terminate_all_aliasing_chains method decl.
26 (pass_store_merging::terminate_all_aliasing_chains): Remove
27 var_offset_p argument and corresponding handling.
28 (stmts_may_clobber_ref_p): New function.
29 (compatible_load_p): New function.
30 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
31 if there is overlap and rhs_code is not INTEGER_CST. For
32 non-overlapping stores terminate group if rhs is not mergeable.
33 (get_alias_type_for_stmts): Change first argument from
34 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
35 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
36 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
38 (get_location_for_stmts): Change first argument from
39 auto_vec<gimple *> & to vec<gimple *> &.
40 (struct split_store): Remove orig_stmts data member, add orig_stores.
41 (split_store::split_store): Create orig_stores rather than orig_stmts.
42 (find_constituent_stmts): Renamed to ...
43 (find_constituent_stores): ... this. Change second argument from
44 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
45 to info structures rather than the statements.
46 (split_group): Rename ALLOW_UNALIGNED argument to
47 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
48 it. Adjust find_constituent_stores caller.
49 (imm_store_chain_info::output_merged_store): Handle rhs_code other
50 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
51 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
52 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
53 (mem_valid_for_store_merging): New function.
54 (handled_load): New function.
55 (pass_store_merging::process_store): New method.
56 (pass_store_merging::execute): Use process_store method. Adjust
57 terminate_all_aliasing_chains caller.
59 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
61 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
62 Return true for more constants, symbols and label references.
63 (aarch64_valid_floating_const): Remove unused function.
65 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
68 * config/aarch64/aarch64.c (aarch64_layout_frame):
69 Undo forcing of LR at bottom of frame.
71 2017-11-03 Jeff Law <law@redhat.com>
74 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
75 for int_registers_saved.
77 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
78 extracted from tree-ssa-dom.c.
79 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
80 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
81 (record_equivalences_from_incoming_edge): Add additional argument
82 to single_pred_edge_ignoring_loop_edges call.
83 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
84 (uncprop_dom_walker::before_dom_children): Add additional argument
85 to single_pred_edge_ignoring_loop_edges call.
86 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
87 single_pred_edge_ignoring_loop_edges rather than open coding.
88 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
90 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
92 * match.pd (-(-A)): Rewrite.
94 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
96 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
97 (rs6000_emit_int_cmove): New declaration.
98 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
99 (rs6000_emit_sISEL): Delete.
100 (rs6000_emit_int_cmove): Make non-static.
101 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
102 instead of rs6000_emit_sISEL.
104 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
106 * asan.c (create_cond_insert_point): Maintain profile.
107 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
109 * basic-block.h (struct basic_block_def): Remove frequency.
110 (EDGE_FREQUENCY): Use to_frequency
111 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
113 (find_traces): Update to use to_frequency.
114 (find_traces_1_round): Likewise; use only IPA counts.
115 (bb_to_key): Likewise.
116 (connect_traces): Use IPA counts only.
117 (copy_bb_p): Update to use to_frequency.
118 (fix_up_crossing_landing_pad): Likewise.
119 (sanitize_hot_paths): Likewise.
120 * bt-load.c (basic_block_freq): Likewise.
121 * cfg.c (init_flow): Set count_max to uninitialized.
122 (check_bb_profile): Remove frequencies; check counts.
123 (dump_bb_info): Do not dump frequencies.
124 (update_bb_profile_for_threading): Update counts only.
125 (scale_bbs_frequencies_int): Likewise.
126 (MAX_SAFE_MULTIPLIER): Remove.
127 (scale_bbs_frequencies_gcov_type): Update counts only.
128 (scale_bbs_frequencies_profile_count): Update counts only.
129 (scale_bbs_frequencies): Update counts only.
130 * cfg.h (struct control_flow_graph): Add count-max.
131 (update_bb_profile_for_threading): Update prototype.
132 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
133 (find_many_sub_basic_blocks): Likewise.
134 * cfgcleanup.c (try_forward_edges): Likewise.
135 (try_crossjump_to_edge): Likewise.
136 * cfgexpand.c (expand_gimple_cond): Likewise.
137 (expand_gimple_tailcall): Likewise.
138 (construct_init_block): Likewise.
139 (construct_exit_block): Likewise.
140 * cfghooks.c (verify_flow_info): Check consistency of counts.
141 (dump_bb_for_graph): Do not dump frequencies.
142 (split_block_1): Do not update frequencies.
143 (split_edge): Do not update frequencies.
144 (make_forwarder_block): Do not update frequencies.
145 (duplicate_block): Do not update frequencies.
146 (account_profile_record): Do not update frequencies.
147 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
148 for global heuristics.
149 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
150 (expected_loop_iterations_unbounded): Use counts only.
151 * cfgloopmanip.c (scale_loop_profile): Simplify.
152 (create_empty_loop_on_edge): Simplify
154 (duplicate_loop_to_header_edge): Simplify
155 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
156 (update_br_prob_note): Take care of removing note when profile
158 (relink_block_chain): Do not dump frequency.
159 (rtl_account_profile_record): Use to_frequency.
160 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
161 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
162 (cgraph_update_edges_for_call_stmt_node): Likewise.
163 (cgraph_edge::verify_count_and_frequency): Update.
164 (cgraph_node::verify_node): Temporarily disable frequency verification.
165 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
167 (cgraph_edge::rebuild_edges): Convert to ipa counts.
168 * cgraphunit.c (init_lowered_empty_function): Do not initialize
170 (cgraph_node::expand_thunk): Update profile.
171 * except.c (dw2_build_landing_pads): Do not update frequency.
172 * final.c (compute_alignments): Use to_frequency.
173 (dump_basic_block_info): Do not dump frequency.
174 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
175 (dump_gimple_bb_header): Do not dump frequency.
176 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
178 * gimple-streamer-in.c (input_bb): Do not stream frequency.
179 * gimple-streamer-out.c (output_bb): Do not stream frequency.
180 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
181 (init_before_recovery): Do not update frequency.
182 (sched_create_recovery_edges): Do not update frequency.
183 * hsa-gen.c (convert_switch_statements): Do not update frequency.
184 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
185 (ipa_cp_c_finalize): Set max_count to uninitialized.
186 * ipa-fnsummary.c (get_minimal_bb): Use counts.
187 (param_change_prob): Use counts.
188 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
190 * ipa-split.c (consider_split): Use to_frequency.
191 (split_function): Use to_frequency.
192 * ira-build.c (loop_compare_func): Likewise.
193 (mark_loops_for_removal): Likewise.
194 (mark_all_loops_for_removal): Likewise.
195 * loop-doloop.c (doloop_modify): Do not update frequency.
196 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
198 * lto-streamer-in.c (input_function): Update count_max.
199 * omp-expand.c (expand_omp_taskreg): Update count_max.
200 * omp-simd-clone.c (simd_clone_adjust): Update profile.
201 * predict.c (maybe_hot_frequency_p): Use to_frequency.
202 (maybe_hot_count_p): Use ipa counts only.
203 (maybe_hot_bb_p): Simplify.
204 (maybe_hot_edge_p): Simplify.
205 (probably_never_executed): Do not take frequency argument.
206 (probably_never_executed_bb_p): Do not pass frequency.
207 (probably_never_executed_edge_p): Likewise.
208 (combine_predictions_for_bb): Check that profile is nonzero.
209 (propagate_freq): Do not set frequency.
210 (drop_profile): Simplify.
211 (counts_to_freqs): Simplify.
212 (expensive_function_p): Use to_frequency.
213 (propagate_unlikely_bbs_forward): Simplify.
214 (determine_unlikely_bbs): Simplify.
215 (estimate_bb_frequencies): Add hack to silence graphite issues.
216 (compute_function_frequency): Use ipa counts.
217 (pass_profile::execute): Update.
218 (rebuild_frequencies): Use counts only.
219 (force_edge_cold): Use counts only.
220 * profile-count.c (profile_count::dump): Dump new count types.
221 (profile_count::differs_from_p): Check compatiblity.
222 (profile_count::to_frequency): New function.
223 (profile_count::to_cgraph_frequency): New function.
224 * profile-count.h (struct function): Declare.
225 (enum profile_quality): Add profile_guessed_local and
226 profile_guessed_global0.
227 (class profile_proability): Decrease number of bits to 29;
228 update from_reg_br_prob_note and to_reg_br_prob_note.
229 (class profile_count: Update comment; decrease number of bits
230 to 61. Check compatibility.
231 (profile_count::compatible_p): New private member function.
232 (profile_count::ipa_p): New member function.
233 (profile_count::operator<): Handle global zero correctly.
234 (profile_count::operator>): Handle global zero correctly.
235 (profile_count::operator<=): Handle global zero correctly.
236 (profile_count::operator>=): Handle global zero correctly.
237 (profile_count::nonzero_p): New member function.
238 (profile_count::force_nonzero): New member function.
239 (profile_count::max): New member function.
240 (profile_count::apply_scale): Handle IPA scalling.
241 (profile_count::guessed_local): New member function.
242 (profile_count::global0): New member function.
243 (profile_count::ipa): New member function.
244 (profile_count::to_frequency): Declare.
245 (profile_count::to_cgraph_frequency): Declare.
246 * profile.c (OVERLAP_BASE): Delete.
247 (compute_frequency_overlap): Delete.
248 (compute_branch_probabilities): Do not use compute_frequency_overlap.
249 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
250 * sched-ebb.c (rank): Use counts only.
251 * shrink-wrap.c (handle_simple_exit): Use counts only.
252 (try_shrink_wrapping): Use counts only.
253 (place_prologue_for_one_component): Use counts only.
254 * tracer.c (find_best_predecessor): Use to_frequency.
255 (find_trace): Use to_frequency.
256 (tail_duplicate): Use to_frequency.
257 * trans-mem.c (expand_transaction): Do not update frequency.
258 * tree-call-cdce.c: Do not update frequency.
259 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
260 (gimple_merge_blocks): Likewise.
261 (gimple_split_edge): Likewise.
262 (gimple_duplicate_sese_region): Likewise.
263 (gimple_duplicate_sese_tail): Likewise.
264 (move_sese_region_to_fn): Likewise.
265 (gimple_account_profile_record): Likewise.
266 (insert_cond_bb): Likewise.
267 * tree-complex.c (expand_complex_div_wide): Likewise.
268 * tree-eh.c (lower_resx): Update profile.
269 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
271 (initialize_cfun): Do not initialize frequencies
272 (freqs_to_counts): Delete.
273 (copy_cfg_body): Ignore count parameter.
275 (expand_call_inline): Update count_max.
276 (optimize_inline_calls): Update count_max.
277 (tree_function_versioning): Update count_max.
278 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
279 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
281 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
282 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
283 (try_peel_loop): Likewise.
284 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
286 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
287 (tree_transform_and_unroll_loop): Do not use frequencies
288 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
289 Use reliable prediction only.
290 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
291 * tree-ssa-sink.c (select_best_block): Use to_frequency.
292 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
294 * tree-ssa-threadupdate.c (create_block_for_threading): Do
296 (any_remaining_duplicated_blocks): Likewise.
297 (update_profile): Likewise.
298 (estimated_freqs_path): Delete.
299 (freqs_to_counts_path): Delete.
300 (clear_counts_path): Delete.
301 (ssa_fix_duplicate_block_edges): Likewise.
302 (duplicate_thread_path): Likewise.
303 * tree-switch-conversion.c (gen_inbound_check): Use counts.
304 * tree-tailcall.c (decrease_profile): Do not update frequency.
305 (eliminate_tail_call): Likewise.
306 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
307 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
308 (optimize_mask_stores): Likewise.
309 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
310 * ubsan.c (ubsan_expand_null_ifn): Update profile.
311 (ubsan_expand_ptr_ifn): Update profile.
312 * value-prof.c (gimple_ic): Simplify.
313 * value-prof.h (gimple_ic): Update prototype.
314 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
315 * ipa-inline.c (compute_uninlined_call_time): Be sure that
317 (want_inline_self_recursive_call_p): Likewise.
318 (resolve_noninline_speculation): Only cummulate defined counts.
319 (inline_small_functions): Use nonzero_p.
320 (ipa_inline): Do not access freed node.
322 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
324 * config/aarch64/aarch64.c (aarch64_override_options_internal):
325 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
327 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
329 * config/riscv/riscv.c (riscv_legitimize_move): Handle
330 non-legitimate address.
332 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
334 * config/rs6000/rs6000.md (*lt0_disi): Delete.
335 (*lt0_<mode>di, *lt0_<mode>si): New.
337 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
339 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
342 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
343 Jim Wilson <jim.wilson@linaro.org>
345 * config/aarch64/aarch64-cores.def (saphira): New CPU.
346 * config/aarch64/aarch64-tune.md: Regenerated.
347 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
348 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
350 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
352 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
353 (arc_expand_prologue): Restore blink for millicode.
354 * config/arc/linux.h (LINK_EH_SPEC): Defined.
356 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
359 * config/i386/i386.c (ix86_vector_duplicate_value): Use
360 gen_vec_duplicate after forcing the scalar into a register.
362 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
364 * combine (try_combine): Print the insns input to try_combine to the
367 2017-11-02 Steve Ellcey <sellcey@cavium.com>
370 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
371 Remove second argument from aarch64_process_target_attr call.
372 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
374 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
376 (aarch64_handle_attr_arch): Remove second argument.
377 (aarch64_handle_attr_cpu): Ditto.
378 (aarch64_handle_attr_tune): Ditto.
379 (aarch64_handle_attr_isa_flags): Ditto.
380 (aarch64_process_one_target_attr): Ditto.
381 (aarch64_process_target_attr): Ditto.
382 (aarch64_option_valid_attribute_p): Remove second argument.
383 on aarch64_process_target_attr call.
385 2017-11-02 David Malcolm <dmalcolm@redhat.com>
387 * diagnostic.c: Include "selftest-diagnostic.h".
388 (selftest::assert_location_text): New function.
389 (selftest::test_diagnostic_get_location_text): New function.
390 (selftest::diagnostic_c_tests): Call it.
392 2017-11-02 David Malcolm <dmalcolm@redhat.com>
394 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
395 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
396 (class selftest::test_diagnostic_context): Move to...
397 * selftest-diagnostic.c: New file.
398 * selftest-diagnostic.h: New file.
400 2017-11-02 James Bowman <james.bowman@ftdichip.com>
402 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
403 offset range for FT32B.
404 * config/ft32/ft32.h: option "mcompress" enables relaxation.
405 * config/ft32/ft32.md: Add TARGET_NOPM.
406 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
407 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
409 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
411 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
413 2017-11-02 Jeff Law <law@redhat.com>
415 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
416 virtual keyword on FINAL OVERRIDE members.
418 * tree-ssa-propagate.h (ssa_propagation_engine): Group
419 virtuals together. Add virtual destructor.
420 (substitute_and_fold_engine): Similarly.
422 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
424 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
426 2017-11-02 Richard Biener <rguenther@suse.de>
428 PR tree-optimization/82795
429 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
431 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
433 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
435 (gcc_GAS_CHECK_FEATURE): Remove.
436 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
437 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
438 * configure: Regenerate.
440 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
442 * config/arc/arc.c (hwloop_optimize): Account for empty
445 2017-11-02 Richard Biener <rguenther@suse.de>
448 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
449 Truncate ARRAY_REF index and element size.
451 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
453 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
455 2017-11-01 Jeff Law <law@redhat.com>
457 * tree-ssa-ccp.c (ccp_folder): New class derived from
458 substitute_and_fold_engine.
459 (ccp_folder::get_value): New member function.
460 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
461 (ccp_fold_stmt): Remove prototype.
462 (ccp_finalize): Call substitute_and_fold from the ccp_class.
463 * tree-ssa-copy.c (copy_folder): New class derived from
464 substitute_and_fold_engine.
465 (copy_folder::get_value): Renamed from get_value.
466 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
467 * tree-vrp.c (vrp_folder): New class derived from
468 substitute_and_fold_engine.
469 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
470 (vrp_folder::get_value): New member function.
471 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
472 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
473 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
474 provide a class interface to folder/substitute routines.
475 (ssa_prop_fold_stmt_fn): Remove typedef.
476 (ssa_prop_get_value_fn): Likewise.
477 (subsitute_and_fold): Remove prototype.
478 (replace_uses_in): Likewise.
479 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
480 Renamed from replace_uses_in. Call the virtual member function
481 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
482 (substitute_and_fold_dom_walker): Remove initialization of
483 data member entries for calbacks. Add substitute_and_fold_engine
484 member and initialize it.
485 (substitute_and_fold_dom_walker::before_dom_children0: Use the
486 member functions for get_value, replace_phi_args_in c
487 replace_uses_in, and fold_stmt calls.
488 (substitute_and_fold_engine::substitute_and_fold): Renamed from
489 substitute_and_fold. Remove assert. Update ctor call.
491 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
492 (ssa_prop_visit_phi_fn): Likewise.
493 (class ssa_propagation_engine): New class to provide an interface
495 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
497 (ssa_prop_visit_phi): Likewise.
498 (ssa_propagation_engine::simulate_stmt): Moved into class.
499 Call visit_phi/visit_stmt from the class rather than via
500 file scoped static variables.
501 (ssa_propagation_engine::simulate_block): Moved into class.
502 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
503 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
504 set file scoped statics for the visit_stmt/visit_phi callbacks.
505 * tree-complex.c (complex_propagate): New class derived from
506 ssa_propagation_engine.
507 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
508 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
509 (tree_lower_complex): Call ssa_propagate via the complex_propagate
511 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
512 ssa_propagation_engine.
513 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
514 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
515 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
516 * tree-ssa-copy.c (copy_prop): New class derived from
517 ssa_propagation_engine.
518 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
519 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
520 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
521 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
522 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
523 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
524 (execute_vrp): Call ssa_propagate from the vrp_prop class.
526 2017-11-01 Jakub Jelinek <jakub@redhat.com>
528 PR rtl-optimization/82778
529 PR rtl-optimization/82597
530 * compare-elim.c (struct comparison): Add in_a_setter field.
531 (find_comparison_dom_walker::before_dom_children): Remove killed
532 bitmap and df_simulate_find_defs call, instead walk the defs.
533 Compute last_setter and initialize in_a_setter. Merge definitions
534 with first initialization for a few variables.
535 (try_validate_parallel): Use insn_invalid_p instead of
536 recog_memoized. Return insn rather than just the pattern.
537 (try_merge_compare): Fix up comment. Don't uselessly test if
538 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
540 (execute_compare_elim_after_reload): Remove df_chain_add_problem
543 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
544 Alan Hayward <alan.hayward@arm.com>
545 David Sherwood <david.sherwood@arm.com>
547 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
548 aarch64_hard_regno_nregs to get the number of registers
551 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
552 Alan Hayward <alan.hayward@arm.com>
553 David Sherwood <david.sherwood@arm.com>
555 * config/aarch64/constraints.md (Upl): Rename to...
557 * config/aarch64/aarch64.md
558 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
561 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
562 Alan Hayward <alan.hayward@arm.com>
563 David Sherwood <david.sherwood@arm.com>
565 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
566 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
569 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
570 Alan Hayward <alan.hayward@arm.com>
571 David Sherwood <david.sherwood@arm.com>
573 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
574 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
575 (aarch64_evpc_dup): Generate rtl direcly, rather than using
577 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
579 * config/aarch64/iterators.md: Add a comment above the permute
580 unspecs to say that they are generated directly by
581 aarch64_expand_vec_perm_const.
582 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
584 2017-11-01 Nathan Sidwell <nathan@acm.org>
586 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
588 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
590 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
591 and medany code models, and describe what they do.
593 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
595 Revert accidental duplicate:
597 * combine.c (can_change_dest_mode): Reject changes in
598 REGMODE_NATURAL_SIZE.
600 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
602 PR rtl-optimization/64682
603 PR rtl-optimization/69567
604 PR rtl-optimization/69737
605 PR rtl-optimization/82683
606 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
607 register mentioned in the note, drop the note, unless it came from I3,
608 in which case it should go to I3 again.
610 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
612 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
613 and return false if not.
614 (clear_bytes_written_by, live_bytes_read): Update accordingly.
616 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
618 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
619 range is known to be empty.
621 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
622 Alan Hayward <alan.hayward@arm.com>
623 David Sherwood <david.sherwood@arm.com>
625 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
626 and CONST_VECTOR_NUNITS instead of computing the number of units from
627 the byte sizes of the vector and element.
628 (simplify_binary_operation_1): Likewise.
629 (simplify_const_binary_operation): Likewise.
630 (simplify_ternary_operation): Likewise.
632 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
633 Alan Hayward <alan.hayward@arm.com>
634 David Sherwood <david.sherwood@arm.com>
636 * var-tracking.c (INT_MEM_OFFSET): Replace with...
637 (int_mem_offset): ...this new function.
638 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
639 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
640 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
643 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
644 Alan Hayward <alan.hayward@arm.com>
645 David Sherwood <david.sherwood@arm.com>
647 * lower-subreg.c (interesting_mode_p): New function.
648 (compute_costs, find_decomposable_subregs, decompose_register)
649 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
650 (resolve_clobber, dump_choices): Use it.
652 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
653 Alan Hayward <alan.hayward@arm.com>
654 David Sherwood <david.sherwood@arm.com>
656 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
658 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
659 Alan Hayward <alan.hayward@arm.com>
660 David Sherwood <david.sherwood@arm.com>
662 * alias.c (find_base_value, find_base_term): Only process integer
663 truncations. Check the precision rather than the size.
665 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
666 Alan Hayward <alan.hayward@arm.com>
667 David Sherwood <david.sherwood@arm.com>
669 * machmode.h (is_narrower_int_mode): New function
670 * optabs.c (expand_float, expand_fix): Use it.
671 * dwarf2out.c (rotate_loc_descriptor): Likewise.
673 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
674 Alan Hayward <alan.hayward@arm.com>
675 David Sherwood <david.sherwood@arm.com>
677 * rtl.h (narrower_subreg_mode): New function.
678 * ira-color.c (update_costs_from_allocno): Use it.
680 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
681 Alan Hayward <alan.hayward@arm.com>
682 David Sherwood <david.sherwood@arm.com>
684 * optabs-query.h (convert_optab_p): New function, split out from...
685 (convert_optab_handler): ...here.
686 (widening_optab_handler): Delete.
687 (find_widening_optab_handler): Remove permit_non_widening parameter.
688 (find_widening_optab_handler_and_mode): Likewise. Provide an
689 override that operates on mode class wrappers.
690 * optabs-query.c (widening_optab_handler): Delete.
691 (find_widening_optab_handler_and_mode): Remove permit_non_widening
692 parameter. Assert that the two modes are the same class and that
693 the "from" mode is narrower than the "to" mode. Use
694 convert_optab_handler instead of widening_optab_handler.
695 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
696 instead of widening_optab_handler.
697 * expr.c (expand_expr_real_2): Update calls to
698 find_widening_optab_handler.
699 * optabs.c (expand_widen_pattern_expr): Likewise.
700 (expand_binop_directly): Take the insn_code as a parameter.
701 (expand_binop): Only call find_widening_optab_handler for
702 conversion optabs; use optab_handler otherwise. Update calls
703 to find_widening_optab_handler and expand_binop_directly.
704 Use convert_optab_handler instead of widening_optab_handler.
705 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
706 find_widening_optab_handler and use scalar_mode rather than
708 (convert_plusminus_to_widen): Likewise.
710 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
711 Alan Hayward <alan.hayward@arm.com>
712 David Sherwood <david.sherwood@arm.com>
714 * machmode.h (fixed_size_mode): New class.
715 * rtl.h (get_pool_mode): Return fixed_size_mode.
716 * gengtype.c (main): Add fixed_size_mode.
717 * target.def (get_raw_result_mode): Return a fixed_size_mode.
718 (get_raw_arg_mode): Likewise.
719 * doc/tm.texi: Regenerate.
720 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
721 * targhooks.c (default_get_reg_raw_mode): Likewise.
722 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
723 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
724 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
725 (msp430_get_raw_result_mode): Likewise.
726 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
727 * dbxout.c (dbxout_parms): Require fixed-size modes.
728 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
729 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
730 * omp-low.c (lower_oacc_reductions): Likewise.
731 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
732 (simplify_subreg): Update accordingly.
733 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
734 (force_const_mem): Update accordingly. Return NULL_RTX for modes
735 that aren't fixed-size.
736 (get_pool_mode): Return a fixed_size_mode.
737 (output_constant_pool_2): Take a fixed_size_mode.
739 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
740 Alan Hayward <alan.hayward@arm.com>
741 David Sherwood <david.sherwood@arm.com>
743 * doc/rtl.texi (vec_series): Document.
744 (const): Say that the operand can be a vec_series.
745 * rtl.def (VEC_SERIES): New rtx code.
746 * rtl.h (const_vec_series_p_1): Declare.
747 (const_vec_series_p): New function.
748 * emit-rtl.h (gen_const_vec_series): Declare.
749 (gen_vec_series): Likewise.
750 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
751 (gen_vec_series): Likewise.
752 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
753 * simplify-rtx.c (simplify_unary_operation): Handle negations
755 (simplify_binary_operation_series): New function.
756 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
757 (test_vector_ops_series): New function.
758 (test_vector_ops): Call it.
759 * config/powerpcspe/altivec.md (altivec_lvsl): Use
760 gen_const_vec_series.
761 (altivec_lvsr): Likewise.
762 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
764 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
765 Alan Hayward <alan.hayward@arm.com>
766 David Sherwood <david.sherwood@arm.com>
768 * doc/rtl.texi (const): Update description of address constants.
769 Say that vector constants are allowed too.
770 * common.md (E, F): Use CONSTANT_P instead of checking for
772 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
773 checking for CONST_VECTOR.
774 * expmed.c (make_tree): Use build_vector_from_val for a CONST
776 * expr.c (expand_expr_real_2): Check for vector modes instead
777 of checking for CONST_VECTOR.
778 * rtl.h (const_vec_p): New function.
779 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
780 (unwrap_const_vec_duplicate): Handle them here too.
782 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
783 David Malcolm <dmalcolm@redhat.com>
784 Alan Hayward <alan.hayward@arm.com>
785 David Sherwood <david.sherwood@arm.com>
787 * rtl.h (vec_duplicate_p): New function.
788 * selftest-rtl.c (assert_rtx_eq_at): New function.
789 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
790 (assert_rtx_eq_at): Declare.
791 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
792 * selftest-run-tests.c (selftest::run_tests): Call it.
793 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
794 (simplify_unary_operation_1): Recursively handle vector duplicates.
795 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
797 (simplify_subreg): Handle subregs of vector duplicates.
798 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
799 (selftest::simplify_rtx_c_tests): New functions.
801 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
802 Alan Hayward <alan.hayward@arm.com>
803 David Sherwood <david.sherwood@arm.com>
805 * emit-rtl.h (gen_const_vec_duplicate): Declare.
806 (gen_vec_duplicate): Likewise.
807 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
809 (gen_const_vector): ...here.
810 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
811 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
812 whose elements are all equal.
813 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
814 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
815 (simplify_relational_operation): Likewise.
816 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
818 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
819 (aarch64_expand_vector_init): Likewise.
820 * config/arm/arm.c (neon_vdup_constant): Likewise.
821 (neon_expand_vector_init): Likewise.
822 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
823 (arm_block_set_unaligned_vect): Likewise.
824 (arm_block_set_aligned_vect): Likewise.
825 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
826 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
827 (expand_vec_perm_even_odd_pack): Likewise.
828 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
829 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
830 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
831 gen_const_vec_duplicate.
832 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
833 * config/mips/mips.c (mips_gen_const_int_vector): Use
834 gen_const_vec_duplicate.
835 (mips_expand_vector_init): Use CONST0_RTX.
836 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
837 (define_split): Use gen_const_vec_duplicate.
838 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
839 (define_split): Use gen_const_vec_duplicate.
840 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
841 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
842 * config/spu/spu.c (spu_const): Likewise.
844 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
845 Alan Hayward <alan.hayward@arm.com>
846 David Sherwood <david.sherwood@arm.com>
848 * combine.c (can_change_dest_mode): Reject changes in
849 REGMODE_NATURAL_SIZE.
851 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
853 * configure.ac (--enable-libssp): New.
854 (gcc_cv_libc_provides_ssp): Check for explicit setting before
855 trying to determine target-specific default. Adjust indentation.
856 * configure: Regenerated.
857 * doc/install.texi (Configuration): Expand --disable-libssp
860 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
862 config/i386/i386.c (ix86_expand_epilogue): Correct stack
865 2017-10-31 Martin Jambor <mjambor@suse.cz>
868 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
870 2017-10-31 David Malcolm <dmalcolm@redhat.com>
872 * auto-profile.c (autofdo_source_profile::read): Use
873 UNKNOWN_LOCATION rather than 0.
874 * diagnostic-core.h (warning_at_rich_loc): Rename to...
875 (warning_at): ...this overload.
876 (warning_at_rich_loc_n): Rename to...
877 (warning_n): ...this overload.
878 (error_at_rich_loc): Rename to...
879 (error_at): ...this overload.
880 (pedwarn_at_rich_loc): Rename to...
881 (pedwarn): ...this overload.
882 (permerror_at_rich_loc): Rename to...
883 (permerror): ...this overload.
884 (inform_at_rich_loc): Rename to...
885 (inform): ...this overload.
886 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
887 (diagnostic_n_impl_richloc): Rename to...
888 (diagnostic_n_impl): ...this rich_location *-based decl.
889 (inform_at_rich_loc): Rename to...
890 (inform): ...this, and add an assertion.
891 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
892 (warning_at_rich_loc): Rename to...
893 (warning_at): ...this, and add an assertion.
894 (warning_at_rich_loc_n): Rename to...
895 (warning_n): ...this, and add an assertion.
896 (warning_n): Update location_t-based implementation for removal of
897 location_t-based diagnostic_n_impl.
898 (pedwarn_at_rich_loc): Rename to...
899 (pedwarn): ...this, and add an assertion.
900 (permerror_at_rich_loc): Rename to...
901 (permerror): ...this, and add an assertion.
902 (error_n): Update for removal of location_t-based diagnostic_n_impl.
903 (error_at_rich_loc): Rename to...
904 (error_at): ...this, and add an assertion.
905 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
906 (driver::do_spec_on_infiles): Likewise.
907 * substring-locations.c (format_warning_va): Update for renaming
908 of inform_at_rich_loc.
910 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
912 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
913 _Float<N>X built-in functions so that the variant without the
914 "__builtin_" prefix is only enabled for the GNU C and Objective C
915 languages when they are in non-strict ANSI/ISO mode.
916 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
917 * target.def (floatn_builtin_p): Add a target hook to control
918 whether _Float<N> and _Float<N>X built-in functions without the
919 "__builtin_" prefix are enabled, and return true for C and
920 Objective C in the default hook. Include langhooks.h in
922 * targhooks.h (default_floatn_builtin_p): Likewise.
923 * targhooks.c (default_floatn_builtin_p): Likewise.
924 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
925 floatn_builtin_p target hook.
926 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
928 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
929 Eric Botcazou <ebotcazou@adacore.com>
931 PR rtl-optimization/81803
932 * lra-constraints.c (curr_insn_transform): Also reload the whole
933 register for a strict subreg no wider than a word if this is for
934 a WORD_REGISTER_OPERATIONS target.
936 2017-10-31 Jason Merrill <jason@redhat.com>
938 * gdbinit.in: Skip over inlines from timevar.h.
940 2017-10-31 Martin Liska <mliska@suse.cz>
942 * doc/gcov.texi: Document new option.
943 * gcov.c (print_usage): Likewise print it.
944 (process_args): Support the argument.
945 (format_count): New function.
946 (format_gcov): Use the function.
948 2017-10-31 Martin Liska <mliska@suse.cz>
950 * gcov.c (struct name_map): do not use typedef.
951 Define operator== and operator<.
952 (name_search): Remove.
954 (main): Do not allocate names.
955 (process_file): Add vertical space.
956 (generate_results): Use std::find.
957 (release_structures): Do not release memory.
958 (find_source): Use std::find.
960 2017-10-31 Martin Liska <mliska@suse.cz>
962 * gcov.c (struct line_info): Remove it's typedef.
963 (line_info::line_info): Add proper ctor.
964 (line_info::has_block): Do not use a typedef.
965 (struct source_info): Do not use typedef.
967 (get_cycles_count): Likewise.
968 (output_intermediate_file): Iterate via vector iterator.
969 (add_line_counts): Use std::vector methods.
970 (accumulate_line_counts): Likewise.
971 (output_lines): Likewise.
973 2017-10-31 Martin Liska <mliska@suse.cz>
975 * gcov.c (struct source_info): Remove typedef.
976 (source_info::source_info): Add proper ctor.
977 (accumulate_line_counts): Use struct, not it's typedef.
978 (output_gcov_file): Likewise.
979 (output_lines): Likewise.
980 (main): Do not allocate an array.
981 (output_intermediate_file): Use size of vector container.
982 (process_file): Resize the vector.
983 (generate_results): Do not preallocate, use newly added vector
985 (release_structures): Do not release sources.
986 (find_source): Use vector methods.
987 (add_line_counts): Do not use typedef.
989 2017-10-31 Martin Liska <mliska@suse.cz>
991 * doc/gcov.texi: Document that.
992 * gcov.c (add_line_counts): Mark lines with a non-executed
994 (output_line_beginning): Handle such lines.
995 (output_lines): Pass new argument.
996 (output_intermediate_file): Print it in intermediate format.
998 2017-10-31 Martin Liska <mliska@suse.cz>
1000 * color-macros.h: New file.
1001 * diagnostic-color.c: Factor out color related to macros to
1003 * doc/gcov.texi: Document -k option.
1004 * gcov.c (INCLUDE_STRING): Include string.h.
1005 (print_usage): Add -k option.
1006 (process_args): Parse it.
1007 (pad_count_string): New function.
1008 (output_line_beginning): Likewise.
1009 (DEFAULT_LINE_START): New macro.
1010 (output_lines): Support color output.
1012 2017-10-31 Martin Liska <mliska@suse.cz>
1014 PR gcov-profile/82633
1015 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
1016 their interaction with GCOV infrastructure.
1017 * configure.ac: Add -fkeep-{inline,static}-functions to
1019 * configure: Regenerate.
1021 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
1024 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
1026 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
1029 * config/rs6000/rs6000.md (allocate_stack): Force update interval
1030 into a register if it does not fit into an immediate offset field.
1032 2017-10-31 Olivier Hainque <hainque@adacore.com>
1034 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
1036 2017-10-31 Julia Koval <julia.koval@intel.com>
1038 * config.gcc: Add gfniintrin.h.
1039 * config/i386/gfniintrin.h: New.
1040 * config/i386/i386-builtin-types.def
1041 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1042 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1043 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1044 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1045 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1046 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
1047 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
1048 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
1049 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
1050 V64QI_FTYPE_V64QI_V64QI_INT): New types.
1051 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1052 * config/i386/immintrin.h: Include gfniintrin.h.
1053 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
1055 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
1057 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
1059 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1061 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
1062 (arm_ashldi3_1bit): Remove pattern.
1063 (ashrdi3): Remove shift by 1 expansion.
1064 (arm_ashrdi3_1bit): Remove pattern.
1065 (lshrdi3): Remove shift by 1 expansion.
1066 (arm_lshrdi3_1bit): Remove pattern.
1067 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
1068 cost of ashldi3 by 1.
1069 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
1070 (<shift>di3_neon): Likewise.
1072 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
1074 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
1075 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
1076 and (*aarch64_simd_mov<VQ:mode>).
1077 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
1078 pattern alternative.
1079 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
1080 attributes to match pattern alternative.
1082 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
1084 * config.gcc (powerpc*-*-*): Add emmintrin.h.
1085 * config/rs6000/emmintrin.h: New file.
1086 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
1088 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1090 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
1091 * (movdi_vfp_cortexa8): Remove pattern.
1093 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1095 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
1097 (Specific, alpha*-dec-osf5.1): Remove.
1098 (Specific, mips-sgi-irix5): Remove.
1099 (Specific, mips-sgi-irix6): Remove.
1101 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1104 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
1105 arguments to clear_bit_region_be.
1107 2017-10-30 Jim Wilson <wilson@tuliptree.org>
1109 * gimplify.c: Include memmodel.h.
1111 2017-10-30 Martin Jambor <mjambor@suse.cz>
1113 * omp-grid.c (grid_attempt_target_gridification): Also insert a
1114 condition whether loop should be executed at all.
1116 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
1118 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1119 gimple folding of vec_madd() intrinsics.
1120 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
1121 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
1122 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
1124 2017-10-30 Richard Biener <rguenther@suse.de>
1126 PR tree-optimization/82762
1128 2017-10-23 Richard Biener <rguenther@suse.de>
1130 PR tree-optimization/82129
1132 2017-08-01 Richard Biener <rguenther@suse.de>
1134 PR tree-optimization/81181
1135 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
1136 (compute_antic): ... end of iteration here.
1138 2017-10-30 Joseph Myers <joseph@codesourcery.com>
1140 * doc/invoke.texi (C Dialect Options): Document -std=c17,
1141 -std=iso9899:2017 and -std=gnu17.
1142 * doc/standards.texi (C Language): Document C17 support.
1143 * doc/cpp.texi (Overview): Mention -std=c17.
1144 (Standard Predefined Macros): Document C11 and C17 values of
1145 __STDC_VERSION__. Do not refer to C99 support as incomplete.
1146 * doc/extend.texi (Inline): Do not list individual options for
1147 standards newer than C99.
1148 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
1150 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
1153 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
1155 * asan.c (asan_finish_file): Align asan globals array by shadow
1158 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1161 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
1162 (struct store_immediate_info): Add bitregion_start and bitregion_end
1164 (store_immediate_info::store_immediate_info): Add brs and bre
1165 arguments and initialize bitregion_{start,end} from those.
1166 (struct merged_store_group): Add bitregion_start, bitregion_end,
1167 align_base and mask fields. Drop unnecessary struct keyword from
1168 struct store_immediate_info. Add do_merge method.
1169 (clear_bit_region_be): Use memset instead of loop storing zeros.
1170 (merged_store_group::do_merge): New method.
1171 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
1172 stores as long as the surrounding bitregions have no gaps.
1173 (merged_store_group::merge_overlapping): Use do_merge.
1174 (merged_store_group::apply_stores): Test that bitregion_{start,end}
1175 is byte aligned, rather than requiring that start and width are
1176 byte aligned. Drop unnecessary struct keyword from
1177 struct store_immediate_info. Allocate and populate also mask array.
1178 Make start of the arrays relative to bitregion_start rather than
1179 start and size them according to bitregion_{end,start} difference.
1180 (struct imm_store_chain_info): Drop unnecessary struct keyword from
1181 struct store_immediate_info.
1182 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
1183 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
1184 struct keyword from struct store_immediate_info.
1185 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
1186 between stores as long as the surrounding bitregions have no gaps.
1188 (struct split_store): Add orig non-static data member.
1189 (split_store::split_store): Initialize orig to false.
1190 (find_constituent_stmts): Return store_immediate_info *, non-NULL
1191 if there is exactly a single original stmt. Change stmts argument
1192 to pointer from reference, if NULL, don't push anything to it. Add
1193 first argument, use it to optimize skipping over orig stmts that
1194 are known to be before bitpos already. Simplify.
1195 (split_group): Return unsigned int count how many stores are or
1196 would be needed rather than a bool. Add allow_unaligned argument.
1197 Change split_stores argument from reference to pointer, if NULL,
1198 only do a dry run computing how many stores would be produced.
1199 Rewritten algorithm to use both alignment and misalign if
1200 !allow_unaligned and handle bitfield stores with gaps.
1201 (imm_store_chain_info::output_merged_store): Set start_byte_pos
1202 from bitregion_start instead of start. Compute allow_unaligned
1203 here, if true, do 2 split_group dry runs to compute which one
1204 produces fewer stores and prefer aligned if equal. Punt if
1205 new count is bigger or equal than original before emitting any
1206 statements, rather than during that. Remove no longer needed
1207 new_ssa_names tracking. Replace num_stmts with
1208 split_stores.length (). Use 32-bit stack allocated entries
1209 in split_stores auto_vec. Try to reuse original store lhs/rhs1
1210 if possible. Handle bitfields with gaps.
1211 (pass_store_merging::execute): Ignore bitsize == 0 stores.
1212 Compute bitregion_{start,end} for the stores and construct
1213 store_immediate_info with that. Formatting fixes.
1215 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
1218 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
1219 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
1221 2017-10-29 Jim Wilson <wilson@tuliptree.org>
1223 * gimplify.c: Include tm_p.h.
1225 * common.opt (gcoff): Re-add as ignored option.
1226 (gcoff1, gcoff2, gcoff3): Likewise.
1228 * Makefile.in (OBJS): Delete sdbout.o.
1229 (GTFILES): Delete $(srcdir)/sdbout.c.
1230 * debug.h: Delete sdb_debug_hooks.
1231 * final.c: Delete sdbout.h include.
1232 (final_scan_insn): Delete SDB_DEBUG check.
1233 (rest_of_clean_state): Likewise.
1234 * output.h: Delete sdb_begin_function_line.
1237 * toplev.c: Delete sdbout.h include.
1238 (process_options): Delete SDB_DEBUG check.
1239 * tree-core.h (tree_type_common): Delete pointer field of
1241 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
1242 TYPE_SYMTAB_POINTER.
1243 * tree.h (TYPE_SYMTAB_POINTER): Delete.
1244 (TYPE_SYMTAB_IS_POINTER): Delete.
1245 (TYPE_SYMTAB_IS_DIE): Renumber.
1246 * xcoffout.c: Refer to former sdbout.c file.
1247 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
1249 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
1250 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
1251 (Debugging Options): Delete -gcoff.
1252 (-gstabs): Delete SDB reference.
1254 (-gcoff@var{level}): Delete.
1255 * doc/passes.texi (Debugging information output): Delete SDB and
1256 sdbout.c references.
1257 * doc/tm.texi: Regenerate.
1258 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
1259 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
1261 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
1262 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
1263 (SDB_DEBUGGING_INFO): Delete.
1264 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
1265 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
1266 * target.def (output_source_filename): Delete COFF reference.
1268 * common.opt (gcoff): Delete.
1269 (gxcoff+): Update Negative chain.
1270 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
1272 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
1273 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
1274 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
1276 (expand_function_start): Change sdb reference to past tense.
1277 (expand_function_end): Change sdb reference to past tense.
1278 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
1279 * opts.c (debug_type_names): Delete coff entry.
1280 (common_handle_option): Delete OPT_gcoff case.
1281 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
1283 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
1284 * config/cris/cris.h: Delete SDB reference in comment.
1285 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
1286 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
1287 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
1288 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
1290 (ix86_expand_prologue): Likewise.
1291 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
1292 * config/ia64/ia64.h: Likewise.
1293 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
1294 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
1296 * config/mmix/mmix.h: Likewise.
1297 * config/nds32/nds32.c: Likewise.
1298 * config/stormy/storym16.h: Likewise.
1299 * config/visium/visium.h: Likewise.
1300 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
1302 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
1304 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
1305 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
1306 FRAME_POINTER_REGNUM point at high end of local var area.
1308 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
1310 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
1311 Move comment around. Do not reset best_edge for a copiable
1312 destination if the copy would cause a partition change.
1313 (better_edge_p): Remove redundant check.
1315 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
1317 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
1319 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
1321 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
1322 for math functions that have _Float<N> and _Float<N>X variants.
1323 (mathfn_built_in_2): Add support for math functions that have
1324 _Float<N> and _Float<N>X variants.
1325 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
1326 (expand_builtin_mathfn_ternary): Add support for fma with
1327 _Float<N> and _Float<N>X variants.
1328 (expand_builtin): Likewise.
1329 (fold_builtin_3): Likewise.
1330 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
1331 create math function _Float<N> and _Float<N>X variants as external
1333 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
1334 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
1335 the __builtin_ prefix and if not strict ansi, without the prefix.
1336 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
1337 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
1338 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
1339 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
1340 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
1341 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
1342 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
1343 function signatures for fma _Float<N> and _Float<N>X variants.
1344 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
1345 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
1346 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
1347 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
1348 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
1349 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
1350 * gencfn-macros.c (print_case_cfn): Add support for math functions
1351 that have _Float<N> and _Float<N>X variants.
1352 (print_define_operator_list): Likewise.
1353 (fltfn_suffixes): Likewise.
1355 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
1356 for math functions that have _Float<N> and _Float<N>X variants.
1357 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
1358 and _Float<N>X variants.
1359 (COPYSIGN): Likewise.
1362 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
1363 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
1365 (integer_valued_read_call_p): Likewise.
1366 * fold-const-call.c (fold_const_call_ss): Likewise.
1367 (fold_const_call_sss): Add support for copysign, fmin, and fmax
1368 _Float<N> and _Float<N>X variants.
1369 (fold_const_call_ssss): Add support for fma _Float<N> and
1370 _Float<N>X variants.
1371 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
1372 support for copysign and fma _Float<N> and _Float<N>X variants.
1373 (backprop::process_builtin_call_use): Likewise.
1374 * tree-call-cdce.c (can_test_argument_range); Add support for
1375 sqrt _Float<N> and _Float<N>X variants.
1376 (edom_only_function): Likewise.
1377 (get_no_error_domain): Likewise.
1378 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
1379 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
1380 copysign _Float<N> and _Float<N>X variants.
1381 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
1382 handled by machine independent code.
1383 (FMAF128): Likewise.
1384 * doc/cpp.texi (Common Predefined Macros): Document defining
1385 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
1386 fma _Float<N> and _Float<N>X variants.
1388 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
1391 * config/i386/i386-modes.def (CCFPU): Remove definition.
1392 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
1393 (ix86_cc_modes_compatible): Ditto.
1394 (ix86_expand_carry_flag_compare): Ditto.
1395 (ix86_expand_int_movcc): Ditto.
1396 (ix86_expand_int_addcc): Ditto.
1397 (ix86_reverse_condition): Ditto.
1398 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
1399 Return true/false for unordered/ordered fp comparisons.
1400 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
1401 (ix86_prepare_fp_compare_args): Update for rename.
1402 (ix86_expand_fp_compare): Update for rename. Generate unordered
1403 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
1404 (ix86_expand_sse_compare_and_jump): Ditto.
1405 * config/i386/predicates.md (fcmov_comparison_operator):
1406 Remove CCFPU mode handling.
1407 (ix86_comparison_operator): Ditto.
1408 (ix86_carry_flag_operator): Ditto.
1409 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
1410 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
1411 (*cmpu<mode>_cc_i387): Ditto.
1412 (FPCMP): Remove mode iterator.
1413 (unord): Remove mode attribute.
1414 (unord_subst): New define_subst transformation
1415 (unord): New define_subst attribute.
1417 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
1418 (*cmpi<unord>xf_i387): Ditto.
1419 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
1420 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
1421 using unord_subst transformation.
1422 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
1423 (round_saeonly): Also handle CCFP mode.
1424 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
1425 Remove UNSPEC_SAHF unspec handling.
1427 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
1429 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
1431 2017-10-27 Jeff Law <law@redhat.com>
1433 * gimple-ssa-sprintf.c: Include domwalk.h.
1434 (class sprintf_dom_walker): New class, derived from dom_walker.
1435 (sprintf_dom_walker::before_dom_children): New function.
1436 (struct call_info): Moved into sprintf_dom_walker class
1437 (compute_formath_length, handle_gimple_call): Likewise.
1438 (sprintf_length::execute): Call the dominator walker rather
1439 than walking the statements.
1441 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
1442 gimple statement locations.
1443 (check_array_bounds): Corresponding changes. Get the statement's
1444 location directly from wi->stmt.
1446 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
1449 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
1451 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
1453 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
1454 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
1456 2017-10-27 Jakub Jelinek <jakub@redhat.com>
1459 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
1460 * config/i386/i386.c (maybe_get_pool_constant): Removed.
1461 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
1462 maybe_get_pool_constant.
1463 * config/i386/predicates.md (zero_extended_scalar_load_operand):
1466 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1468 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
1469 2.26 caveat. Update gas and gld versions.
1470 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
1473 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1475 * cgraph.h (set_malloc_flag): Declare.
1476 * cgraph.c (set_malloc_flag_1): New function.
1477 (set_malloc_flag): Likewise.
1478 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
1479 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
1481 (read_ipa_call_summary): Add support for reading is_return_callee.
1482 (write_ipa_call_summary): Stream is_return_callee.
1483 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
1484 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
1485 ipa-prop.h, ipa-fnsummary.h.
1486 (pure_const_names): Change to static.
1487 (malloc_state_e): Define.
1488 (malloc_state_names): Define.
1489 (funct_state_d): Add field malloc_state.
1490 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
1491 (check_retval_uses): New function.
1492 (malloc_candidate_p): Likewise.
1493 (analyze_function): Add support for malloc attribute.
1494 (pure_const_write_summary): Stream malloc_state.
1495 (pure_const_read_summary): Add support for reading malloc_state.
1496 (dump_malloc_lattice): New function.
1497 (propagate_malloc): New function.
1498 (warn_function_malloc): New function.
1499 (ipa_pure_const::execute): Call propagate_malloc and
1500 ipa_free_fn_summary.
1501 (pass_local_pure_const::execute): Add support for malloc attribute.
1502 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
1503 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
1505 2017-10-27 Martin Liska <mliska@suse.cz>
1507 PR gcov-profile/82457
1508 * doc/invoke.texi: Document that one needs a non-strict ISO mode
1509 for fork-like functions to be properly instrumented.
1511 2017-10-27 Richard Biener <rguenther@suse.de>
1514 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
1515 info when we redirected EH.
1517 2017-10-26 Michael Collison <michael.collison@arm.com>
1519 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
1521 (<optab>_trunchf<GPI:mode>2: New pattern.
1522 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
1523 * config/aarch64/iterators.md (wv): New mode attribute.
1524 (vf, VF): New mode attributes.
1525 (vgp, VGP): New mode attributes.
1526 (s): Update attribute with SImode and DImode prefixes.
1528 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
1530 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
1531 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
1532 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
1533 (nios2_option_overide): Initialize it. Don't allow R0-relative
1534 addressing with PIC.
1535 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
1536 (nios2_symbolic_constant_p): Likewise.
1537 (nios2_legitimate_address_p): Likewise.
1538 (nios2_r0rel_section_name_p): New.
1539 (nios2_symbol_ref_in_r0rel_data_p): New.
1540 (nios2_emit_move_sequence): Handle r0rel_constant_p.
1541 (r0rel_constant_p): New.
1542 (nios2_print_operand_address): Handle r0rel_constant_p.
1543 (nios2_cdx_narrow_form_p): Likewise.
1544 * config/nios2/nios2.opt (mr0rel-sec=): New option.
1545 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
1546 (Nios II Options): Document -mr0rel-sec.
1548 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
1550 * config/nios2/nios2.c: Include xregex.h.
1551 (nios2_gprel_sec_regex): New.
1552 (nios2_option_overide): Initialize it. Don't allow GP-relative
1553 addressing with PIC.
1554 (nios2_small_section_name_p): Check for regex match.
1555 * config/nios2/nios2.opt (mgprel-sec=): New option.
1556 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
1557 (Nios II Options): Document -mgprel-sec.
1559 2017-10-26 Jim Wilson <wilson@tuliptree.org>
1561 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
1563 2017-10-26 Tom de Vries <tom@codesourcery.com>
1565 PR tree-optimization/82707
1566 * gimple.c (gimple_copy): Fix unsharing of
1567 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
1569 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
1571 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
1572 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
1573 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
1574 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
1575 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
1576 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
1577 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
1578 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
1579 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
1580 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
1581 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
1582 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
1583 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
1584 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
1585 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
1586 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
1587 _mm512_mask_cmpunord_ps_mask): New intrinsics.
1589 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
1591 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
1593 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
1594 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
1595 warning to rs6000.c. Remove the Undocumented flag, since it has
1597 (-mabi=ibmlongdouble): Likewise.
1598 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
1599 already set, set the default format for long double.
1600 (rs6000_debug_reg_global): Print whether long double is IBM or
1602 (rs6000_option_override_internal): Rework setting long double
1603 format. Only warn if the user is changing the long double default
1604 and they did not use -Wno-psabi.
1605 * doc/invoke.texi (PowerPC options): Update the documentation for
1606 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
1608 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
1609 Alan Hayward <alan.hayward@arm.com>
1610 David Sherwood <david.sherwood@arm.com>
1612 * rtl.h (wider_subreg_mode): New function.
1613 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
1614 rather than an unsigned int *.
1615 * ira-color.c (regno_max_ref_width): Replace with...
1616 (regno_max_ref_mode): ...this new variable.
1617 (coalesced_pseudo_reg_slot_compare): Update accordingly.
1618 Use wider_subreg_mode.
1619 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
1620 rather than an unsigned int *.
1621 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
1622 (process_alt_operands): Likewise.
1623 (invariant_p): Likewise.
1624 * lra-spills.c (assign_mem_slot): Likewise.
1625 (add_pseudo_to_slot): Likewise.
1626 * lra.c (collect_non_operand_hard_regs): Likewise.
1627 (add_regs_to_insn_regno_info): Likewise.
1628 * reload1.c (regno_max_ref_width): Replace with...
1629 (regno_max_ref_mode): ...this new variable.
1630 (reload): Update accordingly. Update call to
1631 ira_sort_regnos_for_alter_reg.
1632 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
1633 (init_eliminable_invariants): Update to use regno_max_ref_mode.
1634 (scan_paradoxical_subregs): Likewise.
1636 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
1638 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
1639 (aarch64_frame): Add emit_frame_chain boolean.
1640 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1641 Move eh_return case to aarch64_layout_frame.
1642 (aarch64_layout_frame): Initialize emit_frame_chain.
1643 (aarch64_expand_prologue): Use emit_frame_chain.
1645 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
1647 * config/aarch64/aarch64.c (aarch64_layout_frame):
1648 Ensure LR is always stored at the bottom of the callee-saves.
1649 Remove rarely used frame layout which saves callee-saves at top of
1650 frame, so the store of LR can be used as a valid probe in all cases.
1652 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
1654 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
1655 Improve unaligned TImode/TFmode base/offset split.
1657 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
1658 Alan Hayward <alan.hayward@arm.com>
1659 David Sherwood <david.sherwood@arm.com>
1661 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
1662 * combine.c (find_single_use_1): Likewise.
1663 (expand_field_assignment): Likewise.
1664 (move_deaths): Likewise.
1665 * lra-constraints.c (simplify_operand_subreg): Likewise.
1666 (curr_insn_transform): Likewise.
1667 * lra.c (collect_non_operand_hard_regs): Likewise.
1668 (add_regs_to_insn_regno_info): Likewise.
1669 * rtlanal.c (reg_referenced_p): Likewise.
1670 (covers_regno_no_parallel_p): Likewise.
1672 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
1674 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
1675 Don't print any bits outside the precision of the value.
1676 * wide-int.cc (test_printing): Add some new tests.
1678 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1680 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
1681 supports -xbrace_comment option.
1682 * configure: Regenerate.
1683 * config.in: Regenerate.
1684 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
1685 (ASM_CPU_SPEC): Use it.
1687 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
1689 * target.def (static_rtx_alignment): New hook.
1690 * targhooks.h (default_static_rtx_alignment): Declare.
1691 * targhooks.c (default_static_rtx_alignment): New function.
1692 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
1693 * doc/tm.texi: Regenerate.
1694 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
1695 instead of targetm.constant_alignment. Remove call to
1697 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
1698 (cris_preferred_mininum_alignment): New function, split out from...
1699 (cris_constant_alignment): ...here.
1700 (cris_static_rtx_alignment): New function.
1701 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
1703 (ix86_constant_alignment): ...here.
1704 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
1705 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
1706 (mmix_static_rtx_alignment): New function.
1707 * config/spu/spu.c (spu_static_rtx_alignment): New function.
1708 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
1710 2017-10-26 Tamar Christina <tamar.christina@arm.com>
1713 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
1714 Add flag_trapping_math and flag_fp_int_builtin_inexact.
1716 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
1718 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
1719 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
1720 mark as a sign-extending load.
1721 (local_pic_load_u): Define.
1723 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
1726 * fold-const.c (operand_equal_for_comparison_p): Also return true
1727 if ARG0 is a simple variant of ARG1 with narrower precision.
1728 (fold_ternary_loc): Always pass unstripped operands to the predicate.
1730 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
1732 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
1734 * i386.h (processor_costs): Add gather_static, gather_per_elt,
1735 scatter_static, scatter_per_elt.
1736 * x86-tune-costs.h: Add new cost entries.
1738 2017-10-25 Richard Biener <rguenther@suse.de>
1740 * tree-ssa-sccvn.h (vn_eliminate): Declare.
1741 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
1742 class pass_fre): Move to ...
1743 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
1744 class pass_fre): ... here and adjust for statistics.
1746 2017-10-25 Jakub Jelinek <jakub@redhat.com>
1749 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
1750 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
1751 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
1752 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
1755 2017-10-25 Richard Biener <rguenther@suse.de>
1757 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
1758 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
1759 eliminate_push_avail, eliminate_insert): Move inside...
1760 (class eliminate_dom_walker): ... this class in preparation
1762 (fini_eliminate): Remove by merging with ...
1763 (eliminate): ... this function. Adjust for class changes.
1764 (pass_pre::execute): Remove fini_eliminate call.
1765 (pass_fre::execute): Likewise.
1767 2017-10-24 Jakub Jelinek <jakub@redhat.com>
1770 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
1771 (VPERMI2, VPERMI2I): New mode iterators.
1772 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
1773 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
1775 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
1776 mode iterator. Remove 3 old define_insn patterns.
1777 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
1778 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
1779 VPERMI2 mode iterator, remove the other two expanders.
1780 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
1781 to use VPERMI2 mode iterator, add another alternative for vpermi2*
1782 instructions, remove the other two patterns.
1783 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
1784 mode iterator, remove the other two patterns.
1785 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
1786 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
1787 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
1788 and adjust argument order accordingly.
1789 (ix86_expand_vec_perm): Adjust caller.
1790 (expand_vec_perm_1): Likewise.
1791 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
1792 (expand_vec_perm_vpermt2_vpshub2): ... this.
1793 (ix86_expand_vec_perm_const_1): Adjust caller.
1794 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
1797 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
1798 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
1799 (vec_shl_<mode>): Remove unused expander.
1800 (avx512bw_<shift_insn><mode>3): New define_insn.
1801 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
1802 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
1804 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
1807 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
1810 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
1812 PR rtl-optimization/82396
1813 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
1814 (autopref_multipass_init): Simplify initialization.
1815 (autopref_rank_data): Simplify sort order.
1816 * gcc/sched-int.h (autopref_multipass_data_): Remove
1817 multi_mem_insn_p, min_offset and max_offset.
1819 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
1822 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1823 Check special value of flag_omit_frame_pointer.
1824 (aarch64_can_eliminate): Likewise.
1825 (aarch64_override_options_after_change_1): Simplify handling of
1826 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
1828 2017-10-24 Richard Biener <rguenther@suse.de>
1830 PR tree-optimization/82697
1831 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
1832 zero for conditional load and unconditional store.
1834 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
1836 * doc/install.texi: Document bootstrap-cet.
1838 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
1841 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
1842 ENDBR instruction at function entrance if function is only
1845 2017-10-24 Jakub Jelinek <jakub@redhat.com>
1848 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
1849 patterns to better describe from which operation the CF is computed.
1850 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
1851 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
1852 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
1853 is 0, use _0 suffixed expanders instead of emitting a comparison
1856 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
1858 * config/i386/i386.md(*movsf_internal, *movdf_internal):
1859 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
1861 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
1864 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
1865 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
1866 * loop-iv.c (iv_get_reaching_def): Likewise.
1867 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
1868 variable is promoted and the partition contains undefined values.
1870 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
1872 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
1874 (nios2_address_cost): Define.
1875 (nios2_legitimize_address): Recognize (exp + constant) directly.
1876 (TARGET_ADDRESS_COST): Define.
1878 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
1880 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
1881 (nios2_symbolic_memory_operand_p): Declare.
1882 (nios2_split_large_constant): Declare.
1883 (nios2_split_symbolic_memory_operand): Declare.
1884 * config/nios2/nios2.c: Adjust includes.
1885 (nios2_symbolic_constant_allowed): New.
1886 (nios2_symbolic_constant_p): New.
1887 (nios2_plus_symbolic_constant_p): New.
1888 (nios2_valid_addr_expr_p): Recognize addresses involving
1890 (nios2_legitimate_address_p): Likewise, also LO_SUM.
1891 (nios2_symbolic_memory_operand_p): New.
1892 (nios2_large_constant_p): New.
1893 (nios2_split_large_constant): New.
1894 (nios2_split_plus_large_constant): New.
1895 (nios2_split_symbolic_memory_operand): New.
1896 (nios2_legitimize_address): Code refactoring. Handle addresses
1897 involving symbolic constants.
1898 (nios2_emit_move_sequence): Likewise.
1899 (nios2_print_operand): Improve error output.
1900 (nios2_print_operand_address): Handle LO_SUM.
1901 (nios2_cdx_narrow_form_p): Likewise.
1902 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
1903 operands involving symbolic constants.
1904 (movhi_internal, movsi_internal): Likewise.
1905 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
1906 (extendhisi2, extendqi<mode>2): Likewise.
1908 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
1910 * tree-pass.h (PROP_rtl_split_insns): Define.
1911 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
1913 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
1915 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
1917 2017-10-23 Jakub Jelinek <jakub@redhat.com>
1920 * target.def (const_not_ok_for_debug_p): Default to
1921 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
1922 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
1923 * targhooks.c (default_const_not_ok_for_debug_p): New function.
1924 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
1925 which targetm.const_not_ok_for_debug_p returned true.
1926 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
1928 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
1930 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
1931 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
1932 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
1934 (ix86_const_not_ok_for_debug_p): New function.
1935 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
1936 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
1938 2017-10-23 David Malcolm <dmalcolm@redhat.com>
1941 * system.h: Conditionally include "unique-ptr.h" if
1942 INCLUDE_UNIQUE_PTR is defined.
1943 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
1944 of defining INCLUDE_UNIQUE_PTR before including "system.h".
1946 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
1948 * config/rl78/rl78.md: New define_expand "subdi3".
1950 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
1953 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
1954 DF_REF_INSN if DF_REF_INSN_INFO is false.
1956 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
1958 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
1959 xmm_move instead of sse_move.
1960 (sse_store_index): New function.
1961 (ix86_register_move_cost): Be more sensible about mismatch stall;
1962 model AVX moves correctly; make difference between sse->integer and
1964 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
1965 moves; make difference between SSE and AVX.
1966 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
1967 and zmm_move. Increase size of sse load and store tables;
1968 add unaligned load and store tables; add ssemmx_to_integer.
1969 * x86-tune-costs.h: Update all entries according to real
1970 move latencies from Agner Fog's manual and chip documentation.
1972 2017-10-23 Jakub Jelinek <jakub@redhat.com>
1975 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
1976 * config/i386/constraints.md (Wf): New constraint.
1977 * config/i386/i386.md (UNSPEC_SBB): New unspec.
1978 (cmp<dwi>_doubleword): Removed.
1979 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
1980 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
1981 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
1982 expand with cmp<dwi>_doubleword. For LTU and GEU use
1983 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
1985 * common.opt (gcolumn-info): Enable by default.
1986 * doc/invoke.texi (gcolumn-info): Document new default.
1988 2017-10-23 Richard Biener <rguenther@suse.de>
1990 PR tree-optimization/82672
1991 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
1992 Fold the stmt if we propagated into it.
1994 2017-10-23 Richard Biener <rguenther@suse.de>
1996 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
1997 (bitmap_remove_expr_from_set): ... this. All callers call this
1998 for non-constant values.
1999 (bitmap_set_subtract): Rename to...
2000 (bitmap_set_subtract_expressions): ... this. Adjust and
2002 (bitmap_set_contains_value): Remove superfluous check.
2003 (bitmap_set_replace_value): Inline into single caller ...
2004 (bitmap_value_replace_in_set): ... here and simplify.
2005 (dependent_clean): Merge into ...
2006 (clean): ... this using an overload. Adjust.
2007 (prune_clobbered_mems): Adjust.
2008 (compute_antic_aux): Likewise.
2009 (compute_partial_antic_aux): Likewise.
2011 2017-10-23 Richard Biener <rguenther@suse.de>
2013 PR tree-optimization/82129
2015 2017-08-01 Richard Biener <rguenther@suse.de>
2017 PR tree-optimization/81181
2018 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2019 (compute_antic): ... end of iteration here.
2021 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2023 * target.def (starting_frame_offset): New hook.
2024 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
2025 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
2026 * doc/tm.texi.in: Regenerate.
2027 * hooks.h (hook_hwi_void_0): Declare.
2028 * hooks.c (hook_hwi_void_0): New function.
2029 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
2030 STARTING_FRAME_OFFSET.
2031 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
2032 * reload1.c (reload): Likewise.
2033 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
2034 instead of STARTING_FRAME_OFFSET.
2035 * function.c (try_fit_stack_local): Likewise.
2036 (assign_stack_local_1): Likewise
2037 (instantiate_virtual_regs): Likewise.
2038 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
2039 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
2040 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
2041 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
2042 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
2043 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
2044 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
2045 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
2046 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
2047 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
2048 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
2049 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
2050 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
2051 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
2052 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
2053 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
2054 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
2055 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
2056 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
2057 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
2058 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
2059 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
2060 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
2061 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
2062 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
2063 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
2064 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
2065 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
2066 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
2067 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
2068 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
2069 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
2070 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
2071 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
2072 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
2073 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
2074 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
2075 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
2076 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
2077 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
2078 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
2079 * config/avr/avr.c (avr_starting_frame_offset): Make static and
2080 return a HOST_WIDE_INT.
2081 (avr_builtin_setjmp_frame_value): Use it instead of
2082 STARTING_FRAME_OFFSET.
2083 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2084 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
2085 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
2087 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2088 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
2089 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
2090 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2091 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
2092 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
2093 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2094 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
2095 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
2096 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2097 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
2098 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
2100 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2101 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
2102 * config/mips/mips.c (mips_compute_frame_info): Refer to
2103 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2104 (mips_starting_frame_offset): New function.
2105 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2106 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
2107 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
2108 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
2109 and return a HOST_WIDE_INT.
2110 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2111 (mmix_initial_elimination_offset): Refer to
2112 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2113 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
2114 * config/pa/pa.c (pa_starting_frame_offset): New function.
2115 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
2116 (pa_expand_prologue): Likewise.
2117 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2118 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
2119 !FRAME_GROWS_DOWNWARD handling to...
2120 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2121 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
2122 !FRAME_GROWS_DOWNWARD handling to...
2123 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2124 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
2125 !FRAME_GROWS_DOWNWARD handling to...
2126 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2127 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
2129 (rs6000_starting_frame_offset): New function.
2130 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
2131 !FRAME_GROWS_DOWNWARD handling to...
2132 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2133 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
2134 !FRAME_GROWS_DOWNWARD handling to...
2135 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2136 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
2137 !FRAME_GROWS_DOWNWARD handling to...
2138 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2139 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
2140 (rs6000_starting_frame_offset): New function.
2141 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
2142 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
2143 * config/vax/vax.c (vax_starting_frame_offset): New function.
2144 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
2145 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2146 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
2147 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
2148 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2149 * system.h (STARTING_FRAME_OFFSET): Poison.
2151 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2153 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
2154 SCALAR_TYPE_MODE instead of TYPE_MODE.
2156 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2157 Alan Hayward <alan.hayward@arm.com>
2158 David Sherwood <david.sherwood@arm.com>
2160 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
2162 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2163 Alan Hayward <alan.hayward@arm.com>
2164 David Sherwood <david.sherwood@arm.com>
2166 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
2168 2017-10-23 Richard Biener <rguenther@suse.de>
2170 PR tree-optimization/82129
2171 * tree-ssa-pre.c (bitmap_set_and): Remove.
2172 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
2173 canonicalizing expressions in the set to those with lowest
2174 ID rather than taking that from the first edge.
2176 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2178 * combine.c (rtx_equal_for_field_assignment_p): Use
2179 byte_lowpart_offset.
2181 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2182 Alan Hayward <alan.hayward@arm.com>
2183 David Sherwood <david.sherwood@arm.com>
2185 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
2186 to a SUBREG_PROMOTED_VAR.
2188 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2189 Alan Hayward <alan.hayward@arm.com>
2190 David Sherwood <david.sherwood@arm.com>
2192 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
2193 (expand_debug_source_expr): Likewise.
2194 * combine.c (combine_simplify_rtx): Likewise.
2195 * cse.c (fold_rtx): Likewise.
2196 * optabs.c (expand_float): Likewise.
2197 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2198 (simplify_binary_operation_1): Likewise.
2200 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2201 Alan Hayward <alan.hayward@arm.com>
2202 David Sherwood <david.sherwood@arm.com>
2204 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
2205 (record_promoted_value): Likewise.
2206 * expr.c (expand_expr_real_2): Likewise.
2207 * ree.c (update_reg_equal_equiv_notes): Likewise.
2208 (combine_set_extension): Likewise.
2209 * rtlanal.c (low_bitmask_len): Likewise.
2210 * simplify-rtx.c (neg_const_int): Likewise.
2211 (simplify_binary_operation_1): Likewise.
2213 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2214 Alan Hayward <alan.hayward@arm.com>
2215 David Sherwood <david.sherwood@arm.com>
2217 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
2218 * regcprop.c (maybe_mode_change): Likewise.
2219 * reload1.c (alter_reg): Likewise.
2221 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2223 * inchash.h (inchash::hash::add_wide_int): New function.
2224 * lto-streamer-out.c (hash_tree): Use it.
2226 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2228 * inchash.h (inchash::hash::add_wide_int): Rename to...
2229 (inchash::hash::add_hwi): ...this.
2230 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
2231 (polymorphic_call_target_hasher::hash): Likewise.
2232 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
2233 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
2234 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
2235 * lto-streamer-out.c (hash_tree): Likewise.
2236 * optc-save-gen.awk: Likewise.
2237 * tree.c (add_expr): Likewise.
2239 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
2242 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
2243 for ordered inequality comparisons even with TARGET_IEEE_FP.
2245 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
2248 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
2249 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
2250 Expand with cmp<dwi>_doubleword.
2252 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2254 * extend.texi: Add x86 specific to 'nocf_check' attribute.
2255 List CET intrinsics.
2256 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
2257 specific to -fcf-protection option.
2259 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2261 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
2262 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
2263 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
2264 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
2265 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
2266 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
2267 (extra_objs): Add cet.o for Linux/x86 targets.
2268 (tmake_file): Add i386/t-cet for Linux/x86 targets.
2269 * config/i386/cet.c: New file.
2270 * config/i386/cetintrin.h: Likewise.
2271 * config/i386/t-cet: Likewise.
2272 * config/i386/cpuid.h (bit_SHSTK): New.
2273 (bit_IBT): Likewise.
2274 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
2275 pass IBT and SHSTK bits.
2276 * config/i386/i386-builtin-types.def
2277 (VOID_FTYPE_UNSIGNED_PVOID): New.
2278 (VOID_FTYPE_UINT64_PVOID): Likewise.
2279 * config/i386/i386-builtin.def: Add CET intrinsics.
2280 * config/i386/i386-c.c (ix86_target_macros_internal): Add
2281 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
2282 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
2283 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
2285 * config/i386/i386.c (rest_of_insert_endbranch): New.
2286 (pass_data_insert_endbranch): Likewise.
2287 (pass_insert_endbranch): Likewise.
2288 (make_pass_insert_endbranch): Likewise.
2289 (ix86_notrack_prefixed_insn_p): Likewise.
2290 (ix86_target_string): Add -mibt, -mshstk flags.
2291 (ix86_option_override_internal): Add flag_cf_protection
2293 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
2294 (ix86_print_operand): Add 'notrack' prefix output.
2295 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
2296 (ix86_expand_builtin): Expand CET intrinsics.
2297 (x86_output_mi_thunk): Add 'endbranch' instruction.
2298 * config/i386/i386.h (TARGET_IBT): New.
2299 (TARGET_IBT_P): Likewise.
2300 (TARGET_SHSTK): Likewise.
2301 (TARGET_SHSTK_P): Likewise.
2302 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
2303 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
2304 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
2305 (builtin_setjmp_setup): New pattern.
2306 (builtin_longjmp): Likewise.
2307 (rdssp<mode>): Likewise.
2308 (incssp<mode>): Likewise.
2309 (saveprevssp): Likewise.
2310 (rstorssp): Likewise.
2311 (wrss<mode>): Likewise.
2312 (wruss<mode>): Likewise.
2313 (setssbsy): Likewise.
2314 (clrssbsy): Likewise.
2315 (nop_endbr): Likewise.
2316 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
2318 * config/i386/immintrin.h: Include <cetintrin.h>.
2319 * config/i386/linux-common.h
2320 (file_end_indicate_exec_stack_and_cet): New prototype.
2321 (TARGET_ASM_FILE_END): New.
2323 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
2325 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
2326 latencies instead of having separate table; make difference between
2327 integer and float costs.
2328 * i386.h (processor_costs): Remove scalar_stmt_cost,
2329 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
2330 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
2332 * x86-tune-costs.h: Remove entries which has been removed in
2333 procesor_costs from all tables; make cond_taken_branch_cost
2334 and cond_not_taken_branch_cost COST_N_INSNS based.
2336 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
2338 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
2340 2017-10-20 Jakub Jelinek <jakub@redhat.com>
2342 * config/i386/i386.md (isa): Remove fma_avx512f.
2343 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
2344 <avx512>_fmadd_<mode>_mask3<round_name>,
2345 <avx512>_fmsub_<mode>_mask<round_name>,
2346 <avx512>_fmsub_<mode>_mask3<round_name>,
2347 <avx512>_fnmadd_<mode>_mask<round_name>,
2348 <avx512>_fnmadd_<mode>_mask3<round_name>,
2349 <avx512>_fnmsub_<mode>_mask<round_name>,
2350 <avx512>_fnmsub_<mode>_mask3<round_name>,
2351 <avx512>_fmaddsub_<mode>_mask<round_name>,
2352 <avx512>_fmaddsub_<mode>_mask3<round_name>,
2353 <avx512>_fmsubadd_<mode>_mask<round_name>,
2354 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
2355 (*vec_widen_umult_even_v16si<mask_name>,
2356 *vec_widen_smult_even_v16si<mask_name>): Likewise.
2357 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
2359 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2361 * extend.texi: Add 'nocf_check' documentation.
2362 * gimple.texi: Add second parameter to
2363 gimple_build_call_from_tree.
2364 * invoke.texi: Add -fcf-protection documentation.
2365 * rtl.texi: Add REG_CALL_NOTRACK documenation.
2367 2017-10-20 Richard Biener <rguenther@suse.de>
2369 PR tree-optimization/82473
2370 * tree-vect-loop.c (vectorizable_reduction): Properly get at
2371 the largest input type.
2373 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2375 * c-attribs.c (handle_nocf_check_attribute): New function.
2376 (c_common_attribute_table): Add 'nocf_check' handling.
2377 * gimple-parser.c: Add second argument NULL to
2378 gimple_build_call_from_tree.
2379 * attrib.c (comp_type_attributes): Check nocf_check attribute.
2380 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
2382 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
2383 * common.opt: Add fcf-protection flag.
2384 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
2385 * flag-types.h: Add enum cf_protection_level.
2386 * gimple.c (gimple_build_call_from_tree): Add second parameter.
2387 Add 'nocf_check' attribute propagation to gimple call.
2388 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
2389 (gimple_build_call_from_tree): Update prototype.
2390 (gimple_call_nocf_check_p): New function.
2391 (gimple_call_set_nocf_check): Likewise.
2392 * gimplify.c: Add second argument to gimple_build_call_from_tree.
2393 * ipa-icf.c: Add nocf_check attribute in statement hash.
2394 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
2395 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
2396 * toplev.c (process_options): Add flag_cf_protection handling.
2398 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
2400 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
2402 2017-10-20 Richard Biener <rguenther@suse.de>
2404 PR tree-optimization/82603
2405 * tree-if-conv.c (predicate_mem_writes): Make sure to only
2406 remove false predicated stores.
2408 2017-10-20 Richard Biener <rguenther@suse.de>
2410 * graphite-isl-ast-to-gimple.c
2411 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
2412 Remove return value and simplify, dump copied stmt after lhs
2414 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
2415 Reduce dump verbosity.
2416 (gsi_insert_earliest): Likewise.
2417 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
2418 * graphite.c (print_global_statistics): Adjust dumping.
2419 (print_graphite_scop_statistics): Likewise.
2420 (print_graphite_statistics): Do not dump loops here.
2421 (graphite_transform_loops): But here.
2423 2017-10-20 Nicolas Roche <roche@adacore.com>
2425 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
2426 * configure: Regenerate.
2428 2017-10-20 Jakub Jelinek <jakub@redhat.com>
2431 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
2432 functions when optimizing replace GIMPLE_RETURN stmts with
2433 calls to __builtin_unreachable ().
2436 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
2437 for -fsanitize=thread link of executables.
2438 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
2439 link of executables.
2442 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
2444 (<shift_insn><mode>3<mask_name>): Change the last of the 3
2445 define_insns for logical vector shifts to use VI248_AVX512BW
2446 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
2447 condition, useless isa and prefix attributes. Change the first
2448 2 of these define_insns to ...
2449 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
2450 define_insn for avx512vl.
2451 (<shift_insn><mode>3): ... and this, new define_insn without
2452 masking for non-avx512vl.
2455 * config/i386/sse.md (*andnot<mode>3,
2456 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
2457 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
2458 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
2459 not applied use empty suffix even for TARGET_AVX512VL.
2460 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
2461 is applied, supply evex,evex or evex,evex,evex instead of just
2464 2017-10-20 Julia Koval <julia.koval@intel.com>
2466 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
2467 (OPTION_MASK_ISA_GFNI_UNSET): New.
2468 (ix86_handle_option): Handle OPT_mgfni.
2469 * config/i386/cpuid.h (bit_GFNI): New.
2470 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
2471 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
2472 * config/i386/i386.c (ix86_target_string): Add -mgfni.
2473 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
2474 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
2475 * config/i386/i386.opt: Add mgfni.
2477 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
2479 * config/msp430/msp430.c (msp430_option_override): Disable
2480 -fdelete-null-pointer-checks.
2481 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
2483 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
2485 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
2486 of x87 and SSE instructions.
2488 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
2490 * asan.c (create_cond_insert_point): Do not update edge count.
2491 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
2492 (afdo_propagate_circuit): Likewise.
2493 (afdo_calculate_branch_prob): Likewise.
2494 (afdo_annotate_cfg): Likewise.
2495 * basic-block.h (struct edge_def): Remove count.
2496 (edge_def::count): New accessor.
2497 * bb-reorder.c (rotate_loop): Update.
2498 (find_traces_1_round): Update.
2499 (connect_traces): Update.
2500 (sanitize_hot_paths): Update.
2501 * cfg.c (unchecked_make_edge): Update.
2502 (make_single_succ_edge): Update.
2503 (check_bb_profile): Update.
2504 (dump_edge_info): Update.
2505 (update_bb_profile_for_threading): Update.
2506 (scale_bbs_frequencies_int): Update.
2507 (scale_bbs_frequencies_gcov_type): Update.
2508 (scale_bbs_frequencies_profile_count): Update.
2509 (scale_bbs_frequencies): Update.
2510 * cfganal.c (connect_infinite_loops_to_exit): Update.
2511 * cfgbuild.c (compute_outgoing_frequencies): Update.
2512 (find_many_sub_basic_blocks): Update.
2513 * cfgcleanup.c (try_forward_edges): Update.
2514 (try_crossjump_to_edge): Update
2515 * cfgexpand.c (expand_gimple_cond): Update
2516 (expand_gimple_tailcall): Update
2517 (construct_exit_block): Update
2518 * cfghooks.c (verify_flow_info): Update
2519 (redirect_edge_succ_nodup): Update
2520 (split_edge): Update
2521 (make_forwarder_block): Update
2522 (duplicate_block): Update
2523 (account_profile_record): Update
2524 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
2525 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
2526 * cfgloopmanip.c (scale_loop_profile): Update.
2528 (lv_adjust_loop_entry_edge): Update.
2529 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
2530 (force_nonfallthru_and_redirect): Update.
2531 (purge_dead_edges): Update.
2532 (rtl_flow_call_edges_add): Update.
2533 * cgraphunit.c (init_lowered_empty_function): Update.
2534 (cgraph_node::expand_thunk): Update.
2535 * gimple-pretty-print.c (dump_probability): Update.
2536 (dump_edge_probability): Update.
2537 * gimple-ssa-isolate-paths.c (isolate_path): Update.
2538 * haifa-sched.c (sched_create_recovery_edges): Update.
2539 * hsa-gen.c (convert_switch_statements): Update.
2540 * ifcvt.c (dead_or_predicable): Update.
2541 * ipa-inline-transform.c (inline_transform): Update.
2542 * ipa-split.c (split_function): Update.
2543 * ipa-utils.c (ipa_merge_profiles): Update.
2544 * loop-doloop.c (add_test): Update.
2545 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
2546 * lto-streamer-in.c (input_cfg): Update.
2547 (input_function): Update.
2548 * lto-streamer-out.c (output_cfg): Update.
2549 * modulo-sched.c (sms_schedule): Update.
2550 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
2551 * predict.c (maybe_hot_edge_p): Update.
2552 (unlikely_executed_edge_p): Update.
2553 (probably_never_executed_edge_p): Update.
2554 (dump_prediction): Update.
2555 (drop_profile): Update.
2556 (propagate_unlikely_bbs_forward): Update.
2557 (determine_unlikely_bbs): Update.
2558 (force_edge_cold): Update.
2559 * profile.c (compute_branch_probabilities): Update.
2560 * reg-stack.c (better_edge): Update.
2561 * shrink-wrap.c (handle_simple_exit): Update.
2562 * tracer.c (better_p): Update.
2563 * trans-mem.c (expand_transaction): Update.
2564 (split_bb_make_tm_edge): Update.
2565 * tree-call-cdce.c: Update.
2566 * tree-cfg.c (gimple_find_sub_bbs): Update.
2567 (gimple_split_edge): Update.
2568 (gimple_duplicate_sese_region): Update.
2569 (gimple_duplicate_sese_tail): Update.
2570 (gimple_flow_call_edges_add): Update.
2571 (insert_cond_bb): Update.
2572 (execute_fixup_cfg): Update.
2573 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
2574 * tree-complex.c (expand_complex_div_wide): Update.
2575 * tree-eh.c (lower_resx): Update.
2576 (unsplit_eh): Update.
2577 (cleanup_empty_eh_move_lp): Update.
2578 * tree-inline.c (copy_edges_for_bb): Update.
2579 (freqs_to_counts): Update.
2580 (copy_cfg_body): Update.
2581 * tree-ssa-dce.c (remove_dead_stmt): Update.
2582 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
2583 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
2584 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
2585 (unloop_loops): Update.
2586 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
2587 * tree-ssa-loop-split.c (connect_loops): Update.
2588 (split_loop): Update.
2589 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
2590 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
2591 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
2592 * tree-ssa-reassoc.c (branch_fixup): Update.
2593 * tree-ssa-tail-merge.c (replace_block_by): Update.
2594 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
2595 (compute_path_counts): Update.
2596 (update_profile): Update.
2597 (recompute_probabilities): Update.
2598 (update_joiner_offpath_counts): Update.
2599 (estimated_freqs_path): Update.
2600 (freqs_to_counts_path): Update.
2601 (clear_counts_path): Update.
2602 (ssa_fix_duplicate_block_edges): Update.
2603 (duplicate_thread_path): Update.
2604 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
2605 (case_bit_test_cmp): Update.
2606 (collect_switch_conv_info): Update.
2607 (gen_inbound_check): Update.
2608 (do_jump_if_equal): Update.
2609 (emit_cmp_and_jump_insns): Update.
2610 * tree-tailcall.c (decrease_profile): Update.
2611 (eliminate_tail_call): Update.
2612 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
2613 (vect_do_peeling): Update.
2614 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
2615 * ubsan.c (ubsan_expand_null_ifn): Update.
2616 (ubsan_expand_ptr_ifn): Update.
2617 * value-prof.c (gimple_divmod_fixed_value): Update.
2618 (gimple_mod_pow2): Update.
2619 (gimple_mod_subtract): Update.
2620 (gimple_ic): Update.
2621 (gimple_stringop_fixed_value): Update.
2623 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
2626 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
2628 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
2630 PR rtl-optimization/82395
2631 * ira-color.c (allocno_priority_compare_func): Fix comparison step
2632 based on non_spilled_static_chain_regno_p.
2634 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
2636 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
2637 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
2638 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
2640 2017-10-19 Martin Sebor <msebor@redhat.com>
2642 PR tree-optimization/82596
2643 * tree.c (array_at_struct_end_p): Handle STRING_CST.
2645 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
2647 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
2648 (get_mem_refs_of_builtin_call): Likewise.
2649 * builtins.c (expand_builtin_apply): Adjust call to
2650 allocate_dynamic_stack_space.
2651 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
2652 the third argument to allocate_dynamic_stack_space, otherwise -1.
2653 (expand_builtin): Deal with all alloca variants.
2654 (is_inexpensive_builtin): Likewise.
2655 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
2656 * calls.c (special_function_p): Deal with all alloca variants.
2657 (initialize_argument_information): Adjust call to
2658 allocate_dynamic_stack_space.
2659 (expand_call): Likewise.
2660 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
2661 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
2662 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
2663 use it for the stack usage computation.
2664 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
2665 * function.c (gimplify_parameters): Call build_alloca_call_expr.
2666 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
2667 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
2668 (in_loop_p): Remove first argument and useless check.
2669 (pass_walloca::execute): Remove useless test and adjust call to above.
2670 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
2671 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
2672 (gimplify_call_expr): Deal with all alloca variants.
2673 * hsa-gen.c (gen_hsa_alloca): Likewise.
2674 (gen_hsa_insns_for_call): Likewise.
2675 * ipa-pure-const.c (special_builtin_state): Likewise.
2676 * tree-chkp.c (chkp_build_returned_bound): Likewise.
2677 * tree-object-size.c (alloc_object_size): Likewise.
2678 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
2679 (call_may_clobber_ref_p_1): Likewise.
2680 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
2681 (ccp_fold_stmt): Likewise.
2682 (optimize_stack_restore): Likewise.
2683 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
2684 (mark_all_reaching_defs_necessary_1): Likewise.
2685 (propagate_necessity): Likewise.
2686 (eliminate_unnecessary_stmts): Likewise.
2687 * tree.c (build_common_builtin_nodes): Build
2688 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
2689 (build_alloca_call_expr): New function.
2690 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
2691 (CASE_BUILT_IN_ALLOCA): Likewise.
2692 (build_alloca_call_expr): Declare.
2693 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
2695 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
2698 * dwarf2out.c (new_die_raw): New static inline function.
2699 (new_die): Use it to create the DIE.
2700 (add_AT_external_die_ref): Likewise.
2701 (clone_die): Likewise.
2702 (clone_as_declaration): Likewise.
2703 (dwarf2out_vms_debug_main_pointer): Likewise.
2704 (base_type_die): Likewise. Remove early return for corner cases.
2705 Do not call add_pubtype on the DIE here.
2706 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
2707 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
2708 typedefs for base types with DW_AT_endianity. Make sure a DIE with
2709 native order exists for base types, attach the DIE manually and call
2710 add_pubtype on it. Do not equate a reverse order DIE to the type.
2712 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
2714 * config/arm/arm.c (align_ok_ldrd_strd): New function.
2715 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
2717 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
2719 2017-10-19 Jakub Jelinek <jakub@redhat.com>
2721 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
2722 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
2723 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
2724 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
2725 * opts.c (sanitizer_opts): Add builtin.
2726 * ubsan.c (instrument_builtin): New function.
2727 (pass_ubsan::execute): Call it.
2728 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
2729 * doc/invoke.texi: Document -fsanitize=builtin.
2731 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
2732 builtins, store max (log2 (align), 0) into uchar field instead of
2733 align into uptr field.
2734 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
2735 store uchar 0 field instead of uptr 0 field.
2736 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
2737 instead of passing one address of struct with 2 locations pass
2738 two addresses of structs with 1 location each.
2739 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
2740 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
2741 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
2742 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
2743 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
2744 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
2745 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
2746 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
2748 2017-10-19 Martin Liska <mliska@suse.cz>
2751 * file-find.c (remove_prefix): Remove.
2752 * file-find.h (remove_prefix): Likewise.
2753 * gcc-ar.c: Remove smartness of lookup.
2755 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
2757 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
2758 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
2759 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
2761 2017-10-19 Jakub Jelinek <jakub@redhat.com>
2764 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
2765 (setcc + and to xor + setcc): New peephole2.
2767 2017-10-19 Tom de Vries <tom@codesourcery.com>
2769 * doc/sourcebuild.texi (Test Directives, Variants of
2770 dg-require-support): Add dg-require-stack-size.
2772 2017-10-19 Martin Liska <mliska@suse.cz>
2775 * gimplify.c (gimplify_decl_expr): Do not instrument variables
2776 that have a large alignment.
2777 (gimplify_target_expr): Likewise.
2779 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
2781 PR rtl-optimization/82602
2782 * ira.c (rtx_moveable_p): Return false for volatile asm.
2784 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
2787 * config/i386/i386-modes.def (CCGZ): New CC mode.
2788 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
2789 * config/i386/predicates.md (ix86_comparison_operator):
2791 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
2792 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
2793 with double-word subtraction.
2794 (put_condition_code): Handle CCGZmode.
2796 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
2798 * wide-int.cc (debug (const wide_int &)): New.
2799 (debug (const wide_int *)): New.
2800 (debug (const widest_int &)): New.
2801 (debug (const widest_int *)): New.
2803 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
2806 * lra-constraints.c (curr_insn_transform): Use non-input operand
2807 instead of output one for matched reload.
2809 2017-10-18 Bin Cheng <bin.cheng@arm.com>
2811 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
2812 (tree-ssa-loop-ivopts.h): New header file.
2813 (struct builtin_info): New fields.
2814 (classify_builtin_1): Compute and record base and offset parts for
2815 memset builtin partition by calling strip_offset.
2816 (offset_cmp, fuse_memset_builtins): New functions.
2817 (finalize_partitions): Fuse adjacent memset partitions by calling
2819 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
2820 Expose the interface.
2821 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
2823 2017-10-18 Bin Cheng <bin.cheng@arm.com>
2825 PR tree-optimization/82574
2826 * tree-loop-distribution.c (find_single_drs): New parameter. Check
2827 that data reference must be executed exactly once per iteration
2828 against the outermost loop in nest.
2829 (classify_partition): Update call to above function.
2831 2017-10-18 Richard Biener <rguenther@suse.de>
2833 PR tree-optimization/82591
2834 * graphite.c (graphite_transform_loops): Move code gen message
2836 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
2837 Here. Handle scop_to_isl_ast failing.
2838 (scop_to_isl_ast): Limit the number of ISL operations.
2840 2017-10-18 Richard Biener <rguenther@suse.de>
2842 * graphite-isl-ast-to-gimple.c
2843 (translate_isl_ast_to_gimple::set_rename): Simplify.
2844 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
2845 (graphite_copy_stmts_from_block): ... here.
2846 (copy_bb_and_scalar_dependences): Simplify.
2847 (add_parameters_to_ivs_params): Canonicalize.
2848 (generate_entry_out_of_ssa_copies): Simplify.
2849 * graphite-sese-to-poly.c (extract_affine_name): Simplify
2850 by passing in ISL dimension.
2851 (parameter_index_in_region_1): Rename to ...
2852 (parameter_index_in_region): ... this.
2853 (extract_affine): Adjust assert, pass down parameter index.
2854 (add_param_constraints): Use range-info when available.
2855 (build_scop_context): Adjust.
2856 * sese.c (new_sese_info): Adjust.
2857 (free_sese_info): Likewise.
2858 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
2859 Remove unused typedefs.
2860 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
2862 2017-10-18 Martin Liska <mliska@suse.cz>
2864 * combine.c (simplify_compare_const): Add gcc_fallthrough.
2866 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
2868 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
2869 (s390_sched_init): Do not reset s390_sched_state if we entered the
2870 current basic block via a fallthru edge and all others are unlikely.
2872 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
2874 * config/s390/s390.c (NUM_SIDES): New variable.
2875 (LONGRUNNING_THRESHOLD): New variable.
2876 (LATENCY_FACTOR): New variable.
2877 (s390_sched_score): Decrease score for long-running instructions on
2879 (s390_sched_variable_issue): Perform bookkeeping for long-running
2882 2017-10-18 Richard Biener <rguenther@suse.de>
2884 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
2885 Simplify with removal of the parameter rename map.
2886 (set_rename): Likewise.
2887 (should_copy_to_new_region): Likewise.
2888 (graphite_copy_stmts_from_block): Likewise.
2889 (copy_bb_and_scalar_dependences): Remove initialization of
2890 unused copied_bb_map.
2892 (copy_internal_parameters): Likewise.
2893 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
2894 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
2895 Use INTEGRAL_TYPE_P.
2896 (parameter_index_in_region_1): Rename to ...
2897 (assign_parameter_index_in_region): ... this. Assert we have
2898 a parameter we handle.
2899 (scan_tree_for_params): Adjust.
2900 * sese.h (parameter_rename_map_t): Remove.
2901 (struct sese_info_t): Remove unused parameter_rename_map and
2902 copied_bb_map members.
2903 * sese.c (new_sese_info): Adjust.
2904 (free_sese_info): Likewise.
2906 2017-10-18 Martin Liska <mliska@suse.cz>
2909 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
2910 on an abnormal edge.
2912 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
2914 * doc/invoke.texi (ffunction-sections and fdata-sections):
2917 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
2919 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
2920 the use statement can throw internally.
2922 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
2924 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
2925 any RTX present on the RHS of a SET.
2926 * compare-elim.c (try_eliminate_compare): Restore comment.
2928 2017-10-17 Jakub Jelinek <jakub@redhat.com>
2930 * langhooks.h (struct lang_hooks): Document that tree_size langhook
2931 may be also called on tcc_type nodes.
2932 * langhooks.c (lhd_tree_size): Likewise.
2934 2017-10-17 David Malcolm <dmalcolm@redhat.com>
2936 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
2937 format_warning_at_substring.
2938 (maybe_warn): Convert source_range * param to a location_t. Pass
2939 UNKNOWN_LOCATION rather than NULL to fmtwarn.
2940 (format_directive): Remove code to extract source_ranges and
2941 source_range * in favor of just a location_t.
2942 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
2944 * substring-locations.c (format_warning_va): Convert
2945 source_range * param to a location_t.
2946 (format_warning_at_substring): Likewise.
2947 * substring-locations.h (format_warning_va): Likewise.
2948 (format_warning_at_substring): Likewise.
2950 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
2952 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
2954 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
2955 and scatter/gather ops.
2957 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
2958 vec_gather_load and vec_scatter_store.
2959 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
2960 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
2961 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
2962 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
2963 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
2964 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
2966 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
2968 * reg-stack.c (compare_for_stack_reg): Add bool argument.
2969 Detect FTST instruction and handle its register pops. Only pop
2970 second operand if can_pop_second_op is true.
2971 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
2972 set can_pop_second_op to false in the compare_for_stack_reg call.
2974 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
2975 output_fp_compare for stack register operands.
2976 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
2977 instructions here. Do not emit stack register pops here. Assert
2978 that FCOMPP pops next to top stack register. Rewrite function.
2980 2017-10-17 Nathan Sidwell <nathan@acm.org>
2983 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
2984 use DECL_ASSEMBLER_NAME_RAW.
2987 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
2990 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
2991 Wilco Dijkstra <wilco.dijkstra@arm.com>
2993 * builtins.c (expand_builtin_update_setjmp_buf): Add a
2994 converstion to Pmode from the buf_addr.
2996 2017-10-17 Richard Biener <rguenther@suse.de>
2998 * graphite-dependences.c (scop_get_reads_and_writes): Change
2999 output parameters to references.
3001 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
3003 PR 71026/tree-optimization
3004 * fold-const.c (distribute_real_division): Removed.
3005 (fold_binary_loc): Remove calls to distribute_real_divison.
3007 2017-10-17 Richard Biener <rguenther@suse.de>
3009 * graphite-scop-detection.c
3010 (scop_detection::stmt_has_simple_data_refs_p): Always use
3011 the full nest as region.
3012 (try_generate_gimple_bb): Likewise.
3013 * sese.c (scalar_evolution_in_region): Simplify now that
3014 SCEV can handle instantiation in regions.
3015 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
3016 in the non-loop part of a function if requested.
3018 2017-10-17 Richard Biener <rguenther@suse.de>
3020 PR tree-optimization/82563
3021 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
3023 (graphite_regenerate_ast_isl): Call it.
3024 * graphite-scop-detection.c (build_scops): Remove entry edge split.
3026 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3028 PR tree-optimization/82549
3029 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
3030 Formatting fixes. Instead of calling make_bit_field_ref with negative
3033 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
3035 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
3036 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
3037 _mm_maskz_reduce_ss): New.
3038 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
3039 __builtin_ia32_reducess_mask): Ditto..
3040 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
3041 * config/i386/sse.md (reduces<mode>): Renamed to ...
3042 (reduces<mode><mask_scalar_name>): ... this.
3043 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
3045 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
3046 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
3048 2017-10-16 David Malcolm <dmalcolm@redhat.com>
3050 * Makefile.in (OBJS): Add unique-ptr-tests.o.
3051 * selftest-run-tests.c (selftest::run_tests): Call
3052 selftest::unique_ptr_tests_cc_tests.
3053 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
3054 * unique-ptr-tests.cc: New file.
3056 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
3059 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3061 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3062 (make_hard_regno_born, make_hard_regno_dead): Update
3063 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
3065 2017-10-16 Jeff Law <law@redhat.com>
3067 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
3069 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
3071 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
3073 2017-10-16 Olivier Hainque <hainque@adacore.com>
3075 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
3076 with_cpu if we were configured for an e500v2 target cpu name.
3078 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
3080 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
3081 * doc/invoke.texi: Document +nodsp as a valid extension for
3084 2017-10-16 Martin Liska <mliska@suse.cz>
3086 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
3087 (test_set_range): Likewise.
3088 (test_range_functions): Rename to ...
3089 (test_bit_in_range): ... this.
3090 (sbitmap_c_tests): Add new test.
3092 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3094 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
3096 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
3097 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
3099 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3101 * config/aarch64/aarch64-builtins.c
3102 (aarch64_types_quadopu_lane_qualifiers): New.
3103 (TYPES_QUADOPU_LANE): New.
3104 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
3105 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
3106 (aarch64_<sur>dot_laneq<vsi2qi>): New.
3107 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
3108 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
3109 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
3110 (Vdottype, DOTPROD): New.
3111 (sur): Add SDOT and UDOT.
3113 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3115 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
3116 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
3117 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
3119 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
3120 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
3121 Enable TARGET_DOTPROD.
3122 (cortex-a75.cortex-a55): Likewise.
3123 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
3125 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3127 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
3128 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
3130 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
3132 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
3133 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
3134 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
3135 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
3136 * config/arm/types.md (neon_dot, neon_dot_q): New.
3137 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
3139 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3141 * config/arm/arm.h (TARGET_DOTPROD): New.
3142 * config/arm/arm.c (arm_arch_dotprod): New.
3143 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
3144 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
3145 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
3146 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
3147 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
3148 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
3149 * doc/invoke.texi (armv8.2-a): Document dotprod
3151 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3153 * i386.c (ix86_vec_cost): New function.
3154 (ix86_rtx_costs): Handle vector operations better.
3155 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
3156 * x86-tune-costs.h: Add new costs to all tables.
3158 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3160 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
3162 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
3163 divsd, sqrtss and sqrtsd
3164 * x86-tune-costs.h: Add new entries to all costs.
3165 (znver1_cost): Fix to match real instruction latencies.
3167 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3168 Michael Collison <michael.collison@arm.com>
3170 * compare-elim.c: Include emit-rtl.h.
3171 (can_merge_compare_into_arith): New function.
3172 (try_validate_parallel): Likewise.
3173 (try_merge_compare): Likewise.
3174 (try_eliminate_compare): Call the above when no previous clobber
3176 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
3179 2017-10-14 Jakub Jelinek <jakub@redhat.com>
3183 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
3184 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
3188 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
3189 to be any operand_equal_p operands. For & (B - 1) require
3190 B to be power of 2. Recognize
3191 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
3193 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3196 * optabs.c (expand_memory_blockage): Fix call of
3197 targetm.have_memory_blockage.
3199 2017-10-14 Jakub Jelinek <jakub@redhat.com>
3202 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
3203 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
3204 objects to extra_objs instead of overwriting it.
3206 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3208 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
3209 Use any_fp_register_operand as operand[3] predicate. Simplify
3210 equality test for operands[2] and operands[4] memory location.
3211 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
3212 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
3213 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
3214 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
3215 any_fp_register_operand as operand[1] predicate. Simplify
3216 equality test for operands[0] and operands[3] memory location.
3217 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
3218 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
3219 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
3221 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3223 * target-insns.def: Add memory_blockage.
3224 * optabs.c (expand_memory_blockage): New function.
3225 (expand_asm_memory_barrier): Rename ...
3226 (expand_asm_memory_blockage): ... to this.
3227 (expand_mem_thread_fence): Call expand_memory_blockage
3228 instead of expand_asm_memory_barrier.
3229 (expand_mem_singnal_fence): Ditto.
3230 (expand_atomic_load): Ditto.
3231 (expand_atomic_store): Ditto.
3232 * doc/md.texi (Standard Pattern Names For Generation):
3233 Document memory_blockage instruction pattern.
3235 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
3237 * config/rl78/rl78.c (rl78_emit_libcall): New function.
3238 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
3239 * config/rl78/rl78.md: New define_expand "adddi3".
3241 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
3243 * cfghooks.c (verify_flow_info): Disable check that all probabilities
3246 2017-10-13 Jeff Law <law@redhat.com>
3248 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
3250 2017-10-13 Jakub Jelinek <jakub@redhat.com>
3253 * internal-fn.c (expand_mul_overflow): If both operands have
3254 the same highpart of -1 or 0 and the topmost bit of lowpart
3255 is different, overflow is if res <= 0 rather than res < 0.
3257 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
3259 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
3260 TARGET_P9_VECTOR code for unaligned_load case.
3262 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
3264 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
3266 2017-10-13 Nathan Sidwell <nathan@acm.org>
3268 * tree-core.h (tree_contains_struct): Make bool.
3269 * tree.c (tree_contains_struct): Likewise.
3270 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
3271 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
3272 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
3273 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
3274 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
3276 2017-10-13 Richard Biener <rguenther@suse.de>
3278 * graphite-isl-ast-to-gimple.c
3279 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
3280 parameters and dominance check.
3281 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
3282 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
3283 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
3284 Do not update SSA form here or do intermediate IL verification.
3285 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
3286 (graphite_initialize): Remove check on the number of loops in
3287 the function and inline into graphite_transform_loops.
3288 (graphite_finalize): Inline into graphite_transform_loops.
3289 (graphite_transform_loops): Perform SSA update and IL verification
3291 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
3293 2017-10-13 Richard Biener <rguenther@suse.de>
3295 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
3296 graphite_expression_type_precision): Avoid global constructor
3298 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
3299 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
3300 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
3301 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
3302 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
3303 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
3305 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
3308 * config/i386/i386.h (ix86_red_zone_size): New.
3309 * config/i386/i386.md (push peephole2s): Replace
3310 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
3312 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3313 Alan Hayward <alan.hayward@arm.com>
3314 David Sherwood <david.sherwood@arm.com>
3316 * combine.c (can_change_dest_mode): Reject changes in
3317 REGMODE_NATURAL_SIZE.
3319 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3320 Alan Hayward <alan.hayward@arm.com>
3321 David Sherwood <david.sherwood@arm.com>
3323 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
3324 (expand_debug_source_expr): Likewise.
3325 * combine.c (combine_simplify_rtx): Likewise.
3326 * cse.c (fold_rtx): Likewise.
3327 * fwprop.c (canonicalize_address): Likewise.
3328 * targhooks.c (default_shift_truncation_mask): Likewise.
3330 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3331 Alan Hayward <alan.hayward@arm.com>
3332 David Sherwood <david.sherwood@arm.com>
3334 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
3335 (widened_mode): Likewise.
3336 (expand_unop): Likewise.
3337 * ree.c (transform_ifelse): Likewise.
3338 (merge_def_and_ext): Likewise.
3339 (combine_reaching_defs): Likewise.
3340 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3342 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3343 Alan Hayward <alan.hayward@arm.com>
3344 David Sherwood <david.sherwood@arm.com>
3346 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
3347 * combine.c (gen_lowpart_for_combine): Likewise.
3348 * dwarf2out.c (rtl_for_decl_location): Likewise.
3349 * final.c (alter_subreg): Likewise.
3350 * rtlhooks.c (gen_lowpart_general): Likewise.
3351 (gen_lowpart_if_possible): Likewise.
3353 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3354 Alan Hayward <alan.hayward@arm.com>
3355 David Sherwood <david.sherwood@arm.com>
3357 * calls.c (expand_call): Use subreg_lowpart_offset.
3358 * cse.c (cse_insn): Likewise.
3359 * regcprop.c (copy_value): Likewise.
3360 (copyprop_hardreg_forward_1): Likewise.
3362 2017-10-13 Jakub Jelinek <jakub@redhat.com>
3365 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
3366 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
3367 =Q constraints to +Q and into insn condition add check
3368 that operands[0] and operands[1] are equal.
3369 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
3370 =Q constraints to +Q and into insn condition add check
3371 that operands[0] is equal to either operands[1] or operands[2].
3374 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
3375 instead of handling MINUS_EXPR twice (once for each argument),
3376 canonicalize operand order and handle just once, use rtype where
3377 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
3380 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
3381 any values of __C while still being pattern recognizable as a simple
3384 2017-10-13 Richard Biener <rguenther@suse.de>
3386 PR tree-optimization/82451
3388 2017-10-02 Richard Biener <rguenther@suse.de>
3390 PR tree-optimization/82355
3391 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
3392 a mapping for the enclosing loop but avoid generating one for
3394 (copy_bb_and_scalar_dependences): Remove premature codegen
3395 error on PHIs in blocks duplicated into multiple places.
3396 * graphite-scop-detection.c
3397 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
3398 in the region use it as loop and nest to analyze the DR in.
3399 (try_generate_gimple_bb): Likewise.
3400 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
3401 (add_loop_constraints): For blocks in a loop not in the region
3402 create a dimension with a single iteration.
3403 * sese.h (gbb_loop_at_index): Remove assert.
3405 * cfgloop.c (loop_preheader_edge): For the loop tree root
3406 return the single successor of the entry block.
3407 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3408 Reset the SCEV hashtable and niters.
3409 * graphite-scop-detection.c
3410 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
3411 assert that we only have POLYNOMIAL_CHREC that vary in loops
3412 contained in the region.
3413 (scop_detection::graphite_can_represent_expr): Adjust.
3414 (scop_detection::stmt_has_simple_data_refs_p): For loops
3415 not in the region set loop to NULL. The nest is now the
3416 entry edge to the region.
3417 (try_generate_gimple_bb): Likewise.
3418 * sese.c (scalar_evolution_in_region): Adjust for
3419 instantiate_scev change.
3420 * tree-data-ref.h (graphite_find_data_references_in_stmt):
3421 Make nest parameter the edge into the region.
3422 (create_data_ref): Likewise.
3423 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
3424 entry edge into a region and adjust instantiate_scev calls.
3425 (create_data_ref): Likewise.
3426 (graphite_find_data_references_in_stmt): Likewise.
3427 (find_data_references_in_stmt): Pass the loop preheader edge
3428 from the nest argument.
3429 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
3430 parameter the edge into the region.
3431 (instantiate_parameters): Use the loop preheader edge as entry.
3432 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
3434 (get_instantiated_value_entry): Make instantiate_below parameter
3435 the edge into the region.
3436 (instantiate_scev_name): Likewise. Adjust dominance checks,
3437 when we cannot use loop-based instantiation instantiate by
3438 walking use-def chains.
3439 (instantiate_scev_poly): Adjust.
3440 (instantiate_scev_binary): Likewise.
3441 (instantiate_scev_convert): Likewise.
3442 (instantiate_scev_not): Likewise.
3443 (instantiate_array_ref): Remove.
3444 (instantiate_scev_3): Likewise.
3445 (instantiate_scev_2): Likewise.
3446 (instantiate_scev_1): Likewise.
3447 (instantiate_scev_r): Do not blindly handle N-operand trees.
3448 Do not instantiate array-refs. Handle all constants and invariants.
3449 (instantiate_scev): Make instantiate_below parameter
3450 the edge into the region.
3451 (resolve_mixers): Use the loop preheader edge for the region
3452 parameter to instantiate_scev_r.
3453 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
3455 2017-10-13 Richard Biener <rguenther@suse.de>
3457 PR tree-optimization/82525
3458 * graphite-isl-ast-to-gimple.c
3459 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
3461 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
3462 Fail code generation when we cannot represent the isl integer.
3463 (binary_op_to_tree): Elide modulo operations that are no-ops
3464 in the type we code generate. Remove now superfluous code
3467 2017-10-13 Richard Biener <rguenther@suse.de>
3469 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
3470 (scop_detection::harmful_loop_in_region): Remove premature
3471 IV type restriction.
3472 (scop_detection::graphite_can_represent_scev): We can handle
3473 pointer IVs just fine.
3475 2017-10-13 Alan Modra <amodra@gmail.com>
3477 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
3478 "Clobbers and Scratch Registers". Add paragraph on
3479 alternative to clobbers for scratch registers and OpenBLAS
3482 2017-10-13 Alan Modra <amodra@gmail.com>
3484 * doc/extend.texi (Clobbers): Correct vax example. Delete old
3485 example of a memory input for a string of known length. Move
3486 commentary out of table. Add a number of new examples
3487 covering array memory inputs.
3489 2017-10-12 Martin Liska <mliska@suse.cz>
3491 PR tree-optimization/82493
3492 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
3493 (test_range_functions): New function.
3494 (sbitmap_c_tests): Likewise.
3495 * selftest-run-tests.c (selftest::run_tests): Run new tests.
3496 * selftest.h (sbitmap_c_tests): New function.
3498 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
3500 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
3502 * config/rs6000/amo.h: Fix spacing issue.
3504 2017-10-12 Jakub Jelinek <jakub@redhat.com>
3507 * config/i386/i386.md (*ashl<mode>3_mask_1,
3508 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
3509 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
3512 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
3514 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
3515 (profile_probability): Set max_probability
3516 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
3518 * profile-count.c (profile_probability::differs_from_p): Do not
3519 rely on max_probaiblity == 10000
3521 2017-10-12 Jeff Law <law@redhat.com>
3523 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
3526 2017-10-12 Martin Sebor <msebor@redhat.com>
3530 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
3531 (handle_alias_pairs): Call it.
3532 * common.opt (-Wattribute-alias): New option.
3533 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
3534 * doc/invoke.texi (-Wattribute-alias): Document.
3536 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
3539 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
3541 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3543 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3544 (make_hard_regno_born, make_hard_regno_dead): Update
3545 bb_killed_pseudos and bb_gen_pseudos.
3547 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
3549 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
3551 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
3553 * config/alpha/alpha.c (alpha_split_conditional_move):
3554 Use std::swap instead of manually swapping.
3555 (alpha_stdarg_optimize_hook): Ditto.
3556 (alpha_canonicalize_comparison): Ditto.
3558 2017-10-12 Bin Cheng <bin.cheng@arm.com>
3560 * tree-loop-distribution.c (struct builtin_info): New struct.
3561 (struct partition): Refactor fields into struct builtin_info.
3562 (partition_free): Free struct builtin_info.
3563 (build_size_arg_loc, build_addr_arg_loc): Delete.
3564 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
3565 information from struct builtin_info.
3566 (find_single_drs): New function refactored from classify_partition.
3567 Also moved builtin validity checks to this function.
3568 (compute_access_range, alloc_builtin): New functions.
3569 (classify_builtin_st, classify_builtin_ldst): New functions.
3570 (classify_partition): Refactor code into functions find_single_drs,
3571 classify_builtin_st and classify_builtin_ldst.
3572 (distribute_loop): Don't do runtime alias check when distributing
3574 (find_seed_stmts_for_distribution): New function.
3575 (pass_loop_distribution::execute): Refactor code finding seed
3576 stmts into above function. Support distribution for the innermost
3577 two-level loop nest. Adjust dump information.
3579 2017-10-12 Bin Cheng <bin.cheng@arm.com>
3581 * tree-loop-distribution.c: Adjust the general comment.
3582 (NUM_PARTITION_THRESHOLD): New macro.
3583 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
3584 (classify_partition): Skip builtin pattern of loop nest's inner loop.
3585 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
3586 in call to build_partition_graph.
3587 (finalize_partitions): New parameter. Make loop distribution more
3588 conservative by fusing more partitions.
3589 (distribute_loop): Don't do runtime alias check in case of loop nest
3591 (find_seed_stmts_for_distribution): New function.
3592 (prepare_perfect_loop_nest): New function.
3593 (pass_loop_distribution::execute): Refactor code finding seed stmts
3594 and loop nest into above functions. Support loop nest distribution.
3595 Adjust dump information accordingly.
3597 2017-10-12 Bin Cheng <bin.cheng@arm.com>
3599 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
3600 and set PTYPE_SEQUENTIAL for merged partition.
3602 2017-10-12 Richard Biener <rguenther@suse.de>
3604 PR tree-optimization/69728
3606 2017-09-19 Richard Biener <rguenther@suse.de>
3608 PR tree-optimization/69728
3609 * graphite-sese-to-poly.c (schedule_error): New global.
3610 (add_loop_schedule): Handle empty domain by failing the
3612 (build_original_schedule): Handle schedule_error.
3614 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
3615 domain by returning an unchanged schedule.
3617 2017-10-12 Jakub Jelinek <jakub@redhat.com>
3619 * genrecog.c (validate_pattern): For VEC_SELECT verify that
3620 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
3622 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
3624 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
3627 2017-10-12 Jakub Jelinek <jakub@redhat.com>
3630 * expr.c (store_field): Don't optimize away bitsize == 0 store
3631 from CALL_EXPR with addressable return type.
3633 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
3635 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
3636 * config/rs6000/rs6000.md (sel): Delete mode attribute.
3637 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
3638 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
3639 TARGET_ISEL instead of TARGET_ISEL<sel>.
3641 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
3643 * config/rs6000/rs6000.c
3644 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
3646 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
3648 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
3650 (reg_or_cint_operand): Fix comment.
3651 (reg_or_zero_operand): New predicate.
3652 * config/rs6000/rs6000-protos.h (output_isel): Delete.
3653 * config/rs6000/rs6000.c (output_isel): Delete.
3654 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
3655 instead of reg_or_cint_operand. Output instruction directly (not via
3657 (isel_unsigned_<mode>): Ditto.
3658 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
3659 gpc_reg_operand. Add an instruction alternative for this. Output
3660 instruction directly.
3661 (*isel_reversed_unsigned_<mode>): Ditto.
3663 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
3665 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
3666 (TARGET_CANONICALIZE_COMPARISON): Define.
3668 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
3671 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
3672 Check whether the dest is REG before adding REG_EQUIV note.
3674 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
3677 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3679 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3680 (make_hard_regno_born, make_hard_regno_dead): Update
3681 bb_killed_pseudos and bb_gen_pseudos.
3683 2017-10-11 Nathan Sidwell <nathan@acm.org>
3685 * incpath.h (enum incpath_kind): Name enum, prefix values.
3686 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
3687 * incpath.c (heads, tails): Use INC_MAX.
3688 (add_env_var_paths, add_standard_paths): Use incpath_kind.
3689 (merge_include_chains, split_quote_chain,
3690 register_include_chains): Update incpath_kind names.
3691 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
3692 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
3694 (add_framework_path, darwin_register_objc_includes): Likewise.
3695 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
3697 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
3699 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
3700 Do not use float_operator operator predicate.
3701 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
3702 * config/i386/predicates.md (float_operator): Remove predicate.
3704 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
3706 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
3707 (*jccxf_i387): Ditto.
3708 (*jcc<mode>_i387): Ditto.
3709 (*jccu<mode>_i387): Ditto.
3710 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
3711 (*jcc_*_i387 splitters): Remove.
3712 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
3713 * config/i386/i386.c (ix86_split_fp_branch): Remove.
3714 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
3717 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
3719 * profile-count.h (slow_safe_scale_64bit): New function.
3720 (safe_scale_64bit): New inline.
3721 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
3722 * profile-count.c: Include wide-int.h
3723 (slow_safe_scale_64bit): New.
3725 2017-10-11 Nathan Sidwell <nathan@acm.org>
3727 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
3728 HAS_DECL_ASSEMBLER_NAME_P.
3729 * gimple-expr.c (gimple_decl_printable_name: Check
3730 HAS_DECL_ASSEMBLER_NAME_P too.
3731 * ipa-utils.h (type_in_anonymous_namespace_p): Check
3732 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
3733 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
3734 * passes.c (rest_of_decl_compilation): Check
3735 HAS_DECL_ASSEMBLER_NAME_P too.
3736 * recog.c (verify_changes): Likewise.
3737 * tree-pretty-print.c (dump_decl_name): Likewise.
3738 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
3740 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
3741 (SET_DECL_ASSEMBLER_NAME): Use it.
3742 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
3743 (COPY_DECL_ASSEMBLER_NAME): Likewise.
3744 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
3746 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
3748 * config.gcc (i386, x86_64): Add extra objects.
3749 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
3750 (ix86_min_insn_size): Declare.
3751 (ix86_issue_rate): Declare.
3752 (ix86_adjust_cost): Declare.
3753 (ia32_multipass_dfa_lookahead): Declare.
3754 (ix86_macro_fusion_p): Declare.
3755 (ix86_macro_fusion_pair_p): Declare.
3756 (ix86_bd_has_dispatch): Declare.
3757 (ix86_bd_do_dispatch): Declare.
3758 (ix86_core2i7_init_hooks): Declare.
3759 (ix86_atom_sched_reorder): Declare.
3760 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
3761 (COSTS_N_BYTES): Move to x86-tune-costs.h.
3762 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
3763 (rip_relative_addr_p): Rename to ...
3764 (ix86_rip_relative_addr_p): ... this one; export.
3765 (memory_address_length): Update.
3766 (ix86_issue_rate): Move to x86-tune-sched.c.
3767 (ix86_flags_dependent): Move to x86-tune-sched.c.
3768 (ix86_agi_dependent): Move to x86-tune-sched.c.
3769 (exact_dependency_1): Move to x86-tune-sched.c.
3770 (exact_store_load_dependency): Move to x86-tune-sched.c.
3771 (ix86_adjust_cost): Move to x86-tune-sched.c.
3772 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
3773 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
3774 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
3775 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
3776 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
3777 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
3778 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
3779 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
3780 (min_insn_size): Rename to ...
3781 (ix86_min_insn_size): ... this one; export.
3782 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
3783 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
3784 (core2i7_first_cycle_multipass_backtrack): Move to
3785 x86-tune-sched-core.c.
3786 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
3787 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
3788 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
3789 (ix86_avoid_jump_mispredicts): Update.
3790 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
3791 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
3792 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
3793 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
3794 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
3795 (MAX_INSN): Move to ix86-tune-sched-bd.c.
3796 (MAX_IMM): Move to ix86-tune-sched-bd.c.
3797 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
3798 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
3799 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
3800 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
3801 (MAX_STORE): Move to ix86-tune-sched-bd.c.
3802 (BIG): Move to ix86-tune-sched-bd.c.
3803 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
3804 (enum insn_path): Move to ix86-tune-sched-bd.c.
3805 (get_mem_group): Move to ix86-tune-sched-bd.c.
3806 (is_cmp): Move to ix86-tune-sched-bd.c.
3807 (dispatch_violation): Move to ix86-tune-sched-bd.c.
3808 (is_branch): Move to ix86-tune-sched-bd.c.
3809 (is_prefetch): Move to ix86-tune-sched-bd.c.
3810 (init_window): Move to ix86-tune-sched-bd.c.
3811 (allocate_window): Move to ix86-tune-sched-bd.c.
3812 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
3813 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
3814 (process_end_window): Move to ix86-tune-sched-bd.c.
3815 (allocate_next_window): Move to ix86-tune-sched-bd.c.
3816 (find_constant): Move to ix86-tune-sched-bd.c.
3817 (get_num_immediates): Move to ix86-tune-sched-bd.c.
3818 (has_immediate): Move to ix86-tune-sched-bd.c.
3819 (get_insn_path): Move to ix86-tune-sched-bd.c.
3820 (get_insn_group): Move to ix86-tune-sched-bd.c.
3821 (count_num_restricted): Move to ix86-tune-sched-bd.c.
3822 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
3823 (add_insn_window): Move to ix86-tune-sched-bd.c.
3824 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
3825 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
3826 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
3827 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
3828 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
3829 (do_dispatch): Move to ix86-tune-sched-bd.c.
3830 (has_dispatch): Move to ix86-tune-sched-bd.c.
3831 * config/i386/t-i386: Add new object files.
3832 * config/i386/x86-tune-costs.h: New file.
3833 * config/i386/x86-tune-sched-atom.c: New file.
3834 * config/i386/x86-tune-sched-bd.c: New file.
3835 * config/i386/x86-tune-sched-core.c: New file.
3836 * config/i386/x86-tune-sched.c: New file.
3838 2017-10-11 Liu Hao <lh_mouse@126.com>
3840 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
3841 the generic version below instead.
3842 (should_colorize): Recognize Windows consoles as terminals
3844 * pretty-print.c [__MINGW32__] (write_all): New function.
3845 [__MINGW32__] (find_esc_head): Likewise.
3846 [__MINGW32__] (find_esc_terminator): Likewise.
3847 [__MINGW32__] (eat_esc_sequence): Likewise.
3848 [__MINGW32__] (mingw_ansi_fputs): New function that handles
3850 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
3853 2017-10-11 Richard Biener <rguenther@suse.de>
3855 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
3856 Properly call analyze_scalar_evolution with the loop of the stmt.
3858 2017-10-11 Richard Biener <rguenther@suse.de>
3860 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
3861 * tree-core.h (tree_base): Add chrec_var union member.
3862 * tree.h (CHREC_VAR): Remove.
3863 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
3864 * tree-chrec.h (build_polynomial_chrec): Adjust.
3865 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
3866 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
3868 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
3870 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
3871 * match.pd: ... here.
3872 ((T) X == (T) Y): Relax condition.
3874 2017-10-11 Bin Cheng <bin.cheng@arm.com>
3876 PR tree-optimization/82472
3877 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
3879 (break_alias_scc_partitions): Update postorder number.
3881 2017-10-11 Martin Liska <mliska@suse.cz>
3884 * opts.c (parse_no_sanitize_attribute): Do not use error_value
3886 * opts.h (parse_no_sanitize_attribute): Remove last argument.
3888 2017-10-11 Martin Liska <mliska@suse.cz>
3890 * print-rtl.c (print_insn): Move declaration of idbuf
3891 to same scope as name.
3893 2017-10-11 Martin Liska <mliska@suse.cz>
3898 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
3901 2017-10-11 Martin Liska <mliska@suse.cz>
3904 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
3907 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
3909 * config.gcc (default_gnu_indirect_function): Default to yes for
3910 arm*-*-linux* with glibc.
3912 2017-10-11 Richard Biener <rguenther@suse.de>
3914 * tree-scalar-evolution.c (get_scalar_evolution): Handle
3915 default-defs and types we do not want to analyze.
3916 (interpret_loop_phi): Replace unreachable code with an assert.
3917 (compute_scalar_evolution_in_loop): Remove and inline ...
3918 (analyze_scalar_evolution_1): ... here, replacing condition with
3919 what makes the intent clearer. Remove handling of cases
3920 get_scalar_evolution now handles.
3922 2017-10-10 Jim Wilson <wilson@tuliptree.org>
3924 PR rtl-optimization/81434
3925 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
3926 comment for main loop. In sched_group_found if, also add checks for
3927 pass and min_cost_group.
3929 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
3931 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
3932 (rs6000_insn_cost): New function.
3933 * config/rs6000/rs6000.md (cost): New attribute.
3935 2017-10-10 Jakub Jelinek <jakub@redhat.com>
3936 H.J. Lu <hongjiu.lu@intel.com>
3940 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
3941 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
3942 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
3943 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
3944 to OPTION_MASK_ISA_AVX512VL - builtins that have both
3945 OPTION_MASK_ISA_MMX and some other bit set require both
3946 mmx and the ISAs without the mmx bit.
3947 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
3948 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
3949 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
3950 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
3951 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
3952 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
3953 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
3954 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
3955 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
3956 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
3957 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
3958 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
3959 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
3960 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
3961 Add OPTION_MASK_ISA_MMX.
3963 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
3965 * config.gcc (armv7*-*-freebsd*): New target.
3966 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
3968 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
3970 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
3971 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
3974 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
3976 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
3977 a template parameter.
3978 (WIDE_INT_REF_FOR): Update accordingly.
3979 * tree.h (wi::int_traits <const_tree>): Delete.
3980 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
3981 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
3982 (wi::tree_to_wide_ref): New typedef.
3983 (wi::to_wide): New function.
3984 * calls.c (get_size_range): Use wi::to_wide when operating on
3986 * cgraph.c (cgraph_node::create_thunk): Likewise.
3987 * config/i386/i386.c (ix86_data_alignment): Likewise.
3988 (ix86_local_alignment): Likewise.
3989 * dbxout.c (stabstr_O): Likewise.
3990 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
3991 * expr.c (const_vector_from_tree): Likewise.
3992 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
3993 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
3994 (fold_negate_expr_1, int_const_binop_1, const_binop)
3995 (fold_convert_const_int_from_real, optimize_bit_field_compare)
3996 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
3997 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
3998 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
3999 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
4000 (fold_not_const, round_up_loc): Likewise.
4001 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
4002 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
4003 (alloca_call_type): Likewise.
4004 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
4005 * godump.c (go_output_typedef): Likewise.
4006 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
4007 * internal-fn.c (get_min_precision): Likewise.
4008 * ipa-cp.c (ipcp_store_vr_results): Likewise.
4009 * ipa-polymorphic-call.c
4010 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
4011 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
4012 (ipa_modify_call_arguments): Likewise.
4013 * match.pd: Likewise.
4014 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
4015 * print-tree.c (print_node_brief, print_node): Likewise.
4016 * stmt.c (expand_case): Likewise.
4017 * stor-layout.c (layout_type): Likewise.
4018 * tree-affine.c (tree_to_aff_combination): Likewise.
4019 * tree-cfg.c (group_case_labels_stmt): Likewise.
4020 * tree-data-ref.c (dr_analyze_indices): Likewise.
4021 (prune_runtime_alias_test_list): Likewise.
4022 * tree-dump.c (dequeue_and_dump): Likewise.
4023 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
4024 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
4025 * tree-pretty-print.c (dump_generic_node): Likewise.
4026 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
4027 (simple_iv_with_niters): Likewise.
4028 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
4029 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
4030 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
4031 * tree-ssa-loop-niter.c (split_to_var_and_offset)
4032 (refine_value_range_using_guard, number_of_iterations_ne_max)
4033 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
4034 (get_cst_init_from_scev, record_nonwrapping_iv)
4035 (scev_var_range_cant_overflow): Likewise.
4036 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
4037 * tree-ssa-pre.c (compute_avail): Likewise.
4038 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
4039 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
4040 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
4041 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
4042 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
4043 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
4044 (dump_case_nodes, try_switch_expansion): Likewise.
4045 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
4046 (vect_do_peeling): Likewise.
4047 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
4048 * tree-vect-stmts.c (vectorizable_load): Likewise.
4049 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
4050 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
4051 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
4052 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
4053 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
4054 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
4055 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
4056 (evrp_dom_walker::before_dom_children): Likewise.
4057 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
4058 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
4059 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
4060 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
4061 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
4062 (get_range_pos_neg): Likewise.
4063 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
4064 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
4065 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
4066 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
4067 * config/avr/avr.c (avr_fold_builtin): Likewise.
4068 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
4069 * config/msp430/msp430.c (msp430_attr): Likewise.
4070 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
4071 * config/powerpcspe/powerpcspe-c.c
4072 (altivec_resolve_overloaded_builtin): Likewise.
4073 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
4074 (rs6000_expand_ternop_builtin): Likewise.
4075 * config/rs6000/rs6000-c.c
4076 (altivec_resolve_overloaded_builtin): Likewise.
4077 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
4078 (rs6000_expand_ternop_builtin): Likewise.
4079 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
4081 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4083 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
4084 when copying loop nest with only one inner loop.
4086 2017-10-10 Richard Biener <rguenther@suse.de>
4088 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
4089 blocks if SCEV is active.
4090 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
4092 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
4093 (scev_initialize): Assert we are not yet initialized.
4095 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4097 * tree-loop-distribution.c (generate_loops_for_partition): Remove
4098 inner loop's exit stmt by making it always exit the loop, otherwise
4099 we would generate an infinite empty loop.
4101 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4103 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
4104 renaming variables in new preheader if it's deleted.
4106 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4108 * tree-loop-distribution.c (struct partition): Remove unused field
4109 loops of the structure.
4110 (partition_alloc, partition_free): Ditto.
4111 (build_rdg_partition_for_vertex): Ditto.
4113 2017-10-09 Jeff Law <law@redhat.com>
4115 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
4116 return type to match prototype and documentation.
4118 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4120 * config/rs6000/rs6000.c (processor_costs): Move to ...
4121 * config/rs6000/rs6000.h: ... here.
4122 (rs6000_cost): Declare.
4124 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
4126 * except.c (setjmp_fn): New global variable.
4127 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
4128 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
4129 if DONT_USE_BUILTIN_SETJMP is defined.
4131 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4133 * target.def (insn_cost): New hook.
4134 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
4135 * doc/tm.texi: Regenerate.
4136 * rtlanal.c (insn_cost): Use the new hook.
4138 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4140 * combine.c (combine_validate_cost): Compute the new insn_cost,
4141 not just pattern_cost.
4142 (try_combine): Adjust comment.
4144 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4146 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
4148 * combine.c (uid_insn_cost): Adjust comment.
4149 (combine_validate_cost): Adjust comment. Use pattern_cost instead
4151 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
4152 * dse.c (find_shift_sequence): Ditto.
4153 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
4154 (bb_valid_for_noce_process_p): Use pattern_cost.
4155 * rtl.h (insn_rtx_cost): Delete.
4156 (pattern_cost): New prototype.
4157 (insn_cost): New prototype.
4158 * rtlanal.c (insn_rtx_cost): Rename to...
4159 (pattern_cost): ... this.
4162 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
4164 * config/i386/i386.md (*jcc_2): Remove insn pattern.
4165 (*jcc<mode>_0_r_i387): Ditto.
4166 (*jccxf_r_i387): Ditto.
4167 (*jcc<mode>_r_i387): Ditto.
4168 (*jccu<mode>_r_i387): Ditto.
4169 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
4170 (*jcc): Rename from *jcc_1.
4172 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4174 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
4175 deferred rescans after the lvx/stvx recombination pre-pass.
4177 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
4179 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
4180 memory operation instruction support.
4181 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
4182 (rs6000-ibm-aix[789]*): Likewise.
4183 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
4184 Document new functions.
4186 2017-10-09 Richard Biener <rguenther@suse.de>
4188 PR tree-optimization/82397
4189 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
4190 equality only for semantically equal trees.
4192 2017-10-09 Richard Biener <rguenther@suse.de>
4194 PR tree-optimization/82449
4195 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
4196 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
4197 allow constant addresses.
4198 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
4201 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4203 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
4206 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4209 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
4212 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4215 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
4217 2017-10-09 Jakub Jelinek <jakub@redhat.com>
4220 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
4221 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
4223 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
4225 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
4226 (WI_BINARY_PREDICATE_RESULT): Likewise.
4227 (wi::binary_traits::operator_result): New type.
4228 (wi::binary_traits::predicate_result): Likewise.
4229 (generic_wide_int::operator~, unary generic_wide_int::operator-)
4230 (generic_wide_int::operator==, generic_wide_int::operator!=)
4231 (generic_wide_int::operator&, generic_wide_int::and_not)
4232 (generic_wide_int::operator|, generic_wide_int::or_not)
4233 (generic_wide_int::operator^, generic_wide_int::operator+
4234 (binary generic_wide_int::operator-, generic_wide_int::operator*):
4236 (operator~, unary operator-, operator==, operator!=, operator&)
4237 (operator|, operator^, operator+, binary operator-, operator*): New
4239 * expr.c (get_inner_reference): Use wi::bit_and_not.
4240 * fold-const.c (fold_binary_loc): Likewise.
4241 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
4242 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
4243 (bit_value_binop): Likewise.
4244 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
4245 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
4246 (extract_range_from_binary_expr_1): Likewise.
4247 (masked_increment): Likewise.
4248 (simplify_bit_ops_using_ranges): Likewise.
4250 2017-10-09 Martin Jambor <mjambor@suse.cz>
4253 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
4254 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
4255 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
4257 (hsa_fixup_mov_insn_type): New function.
4258 (hsa_op_with_type::get_in_type): Use it.
4259 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
4260 immediates in an assert.
4261 (hsa_op_with_type::extend_int_to_32bit): New method.
4262 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
4263 types. Convert to dest type if necessary.
4264 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
4265 (reg_for_gimple_ssa): Pass false as min32int to
4266 hsa_type_for_scalar_tree_type.
4267 (gen_hsa_addr): Fixup type when creating addresable temporary.
4268 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
4269 (gen_hsa_unary_operation): Extend operands and convert to dest type if
4270 necessary. Call hsa_fixup_mov_insn_type.
4271 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
4272 extend operands and convert to dest type if necessary.
4273 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
4274 to dest type if necessary.
4275 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
4276 if conversion nt necessary and size matches.
4277 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
4278 to dest type if necessary.
4279 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
4280 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
4282 (gen_hsa_clrsb): Likewise.
4283 (gen_hsa_ffs): Likewise.
4284 (gen_hsa_divmod): Extend operands and convert to dest type if
4286 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
4288 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
4290 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
4291 Remove empty default arguments. Use a brace block as output
4293 (conditional return): Ditto.
4295 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
4296 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
4298 (group_ending_nop): Ditto.
4299 (doloop_end): Ditto.
4300 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
4301 (splitters for those): Ditto.
4303 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
4305 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
4306 a conditional jump (and the compare for it) so that pc_rtx is the
4308 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
4309 for the deleted and renamed ctr<mode>_internal[234] patterns.
4310 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
4311 Delete second conditional return pattern.
4312 (ctr<mode>_internal2): Delete this second bdnz pattern.
4313 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
4314 (ctr<mode>_internal4): Delete this second bdz pattern.
4316 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
4318 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
4319 (always_initialized_rtx_for_ssa_name_p): New predicate.
4320 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
4321 (finish_out_of_ssa): Free new field of SA.
4322 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
4323 * tree-ssa-coalesce.c: Include tree-ssa.h.
4324 (get_parm_default_def_partitions): Remove extern keyword.
4325 (get_undefined_value_partitions): New function.
4326 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
4327 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
4329 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
4331 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
4333 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
4335 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
4337 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
4338 for targets that preffer 128bit.
4340 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
4342 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
4344 2017-10-08 Olivier Hainque <hainque@adacore.com>
4346 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
4347 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
4349 (thumb_set_return_address): Likewise.
4351 2017-10-08 Olivier Hainque <hainque@adacore.com>
4353 * common/config/arm/arm-common.c (arm_except_unwind_info):
4354 Handle DWARF2_UNWIND_INFO.
4356 2017-10-07 Michael Collison <michael.collison@arm.com>
4358 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
4361 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
4363 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
4364 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
4365 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
4366 defined, force the creation of a new block for a dispatch label.
4368 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
4370 * invoke.texi (Wsuggest-attribute=cold): Document.
4371 * common.opt (Wsuggest-attribute=cold): New
4372 * ipa-pure-const.c (warn_function_cold): New function.
4373 * predict.c (compute_function_frequency): Use it.
4374 * predict.h (warn_function_cold): Declare.
4376 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
4378 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
4381 2017-10-06 Martin Liska <mliska@suse.cz>
4383 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
4384 keyword for member functions.
4385 (struct sanopt_tree_couple): New struct.
4386 (struct sanopt_tree_couple_hash): New function.
4387 (struct sanopt_ctx): Add new hash_map.
4388 (has_dominating_ubsan_ptr_check): New function.
4389 (record_ubsan_ptr_check_stmt): Likewise.
4390 (maybe_optimize_ubsan_ptr_ifn): Likewise.
4391 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
4392 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
4394 2017-10-06 Sudakshina Das <sudi.das@arm.com>
4397 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
4398 aarch64_simd_valid_immediate on CONST_VECTORs.
4399 (aarch64_reg_or_bic_imm): Likewise.
4401 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
4403 PR rtl-optimization/82396
4404 * haifa-sched.c (ready_sort_real): Disable qsort checking.
4406 2017-10-06 Sebastian Pop <sebpop@gmail.com>
4408 * graphite-dependences.c (scop_get_reads): Move code to...
4409 (scop_get_must_writes): Move code to...
4410 (scop_get_may_writes): Move code to...
4411 (scop_get_reads_and_writes): ... here.
4412 (scop_get_dependences): Call scop_get_reads_and_writes.
4414 2017-10-06 Jakub Jelinek <jakub@redhat.com>
4416 PR tree-optimization/82434
4417 * fold-const.h (can_native_encode_type_p,
4418 can_native_encode_string_p): Remove.
4419 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
4420 don't encode anything, just return what would be otherwise returned.
4421 (native_encode_fixed, native_encode_complex, native_encode_vector):
4423 (native_encode_string): Likewise. Inline by hand
4424 can_native_encode_string_p.
4425 (can_native_encode_type_p): Remove.
4426 (can_native_encode_string_p): Remove.
4427 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
4428 STRING_CSTs using can_native_encode_string_p, test all
4429 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
4430 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
4431 argument from native_encode_expr.
4432 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
4433 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
4436 2017-10-06 Richard Biener <rguenther@suse.de>
4438 PR tree-optimization/82397
4439 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
4440 operand_equal_p but rely on data_ref_compare_tree for detecting
4442 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
4443 to match up with dr_group_sort_cmp.
4445 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4448 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
4450 * config/s390/s390-builtin-types.def: Regenerate.
4452 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4455 * config/s390/s390-builtin-types.def: Regenerate.
4456 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
4457 Change flag from B_VXE to B_VX.
4458 (s390_vec_min_dbl): Remove B_VXE flag.
4460 2017-10-06 Richard Biener <rguenther@suse.de>
4462 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
4463 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
4464 translate_isl_ast_to_gimple::is_valid_rename,
4465 translate_isl_ast_to_gimple::get_rename,
4466 translate_isl_ast_to_gimple::get_def_bb_for_const,
4467 translate_isl_ast_to_gimple::get_new_name,
4468 translate_isl_ast_to_gimple::collect_all_ssa_names,
4469 translate_isl_ast_to_gimple::copy_loop_phi_args,
4470 translate_isl_ast_to_gimple::collect_all_ssa_names,
4471 translate_isl_ast_to_gimple::copy_loop_phi_args,
4472 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
4473 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
4474 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
4475 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
4476 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
4477 translate_isl_ast_to_gimple::copy_cond_phi_args,
4478 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
4479 translate_isl_ast_to_gimple::edge_for_new_close_phis,
4480 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
4481 translate_isl_ast_to_gimple::rename_uses,
4482 translate_isl_ast_to_gimple::rename_all_uses): Remove.
4483 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
4484 (set_rename_for_each_def): Likewise.
4485 (graphite_copy_stmts_from_block): Handle debug stmt resetting
4486 here. Handle rewriting SCEV analyzable uses here.
4487 (copy_bb_and_scalar_dependences): Generate code for PHI
4489 (graphite_regenerate_ast_isl): Adjust.
4490 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
4491 (add_write, add_read): New functions.
4492 (build_cross_bb_scalars_def): Use it and simplify.
4493 (build_cross_bb_scalars_use): Likewise.
4494 (graphite_find_cross_bb_scalar_vars): Inline into...
4495 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
4496 simulating out-of-SSA. Compute liveout and add dependencies.
4497 (build_scops): Force an empty entry block.
4498 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
4500 (sese_build_liveouts): Declare.
4501 (sese_trivially_empty_bb_p): Likewise.
4502 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
4503 compute liveout and debug_liveout.
4504 (sese_bad_liveouts_use): Remove.
4505 (sese_reset_debug_liveouts_bb): Likewise.
4506 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
4507 (sese_build_liveouts): Build liveout and debug_liveout and store
4509 (new_sese_info): Adjust.
4510 (free_sese_info): Likewise.
4511 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
4512 do not build liveout here.
4513 (move_sese_in_condition): Adjust region entry.
4514 (scev_analyzable_p): Match up with chrec_apply requirements.
4515 (sese_trivially_empty_bb_p): New.
4516 * tree-into-ssa.c (get_reaching_def): Properly support generating
4517 default-defs for incremental rewrite of anonymous names.
4519 2017-10-06 Richard Biener <rguenther@suse.de>
4521 * graphite-sese-to-poly.c (extract_affine): For casts increasing
4522 precision do not perform modulo reduction.
4524 2017-10-06 Richard Biener <rguenther@suse.de>
4526 PR tree-optimization/82436
4527 * tree-vect-slp.c (vect_supported_load_permutation_p): More
4528 conservatively choose the vectorization factor when checking
4529 whether we can perform the required load permutation.
4530 (vect_transform_slp_perm_load): Assert when we may not fail.
4532 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
4534 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
4535 message for incompatible -msdata=* and -mcall-* options.
4537 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
4539 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
4540 rate for post-reload scheduling.
4542 2017-10-05 Tamar Christina <tamar.christina@arm.com>
4544 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
4546 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
4548 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
4549 to improve monte carlo in scimark.
4551 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
4553 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
4554 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
4555 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
4556 pentium4_cost, nocona_cost): Set reassociation width to 1.
4557 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
4558 width to 2 for fp operations and 1 otherwise.
4559 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
4561 (atom_cost): Set reassociation width to 2.
4562 (slm_cost, generic_cost): Set fp reassociation width
4563 to 2 and 1 otherwise.
4564 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
4565 (core_cost): Set fp reassociation width to 4 and vector to 2.
4566 (ix86_reassociation_width): Rewrite using cost table; special case
4567 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
4568 and TARGET_AVX128_OPTIMAL.
4569 * config/i386/i386.h (processor_costs): Add
4570 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
4571 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
4572 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
4573 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
4574 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
4575 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
4577 2017-10-05 Nathan Sidwell <nathan@acm.org>
4579 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
4581 2017-10-05 Tamar Christina <tamar.christina@arm.com>
4583 * config/arm/arm.c (arm_test_fpu_data): New.
4584 (arm_run_selftests): Call arm_test_fpu_data.
4586 2017-10-04 Nathan Sidwell <nathan@acm.org>
4588 * toplev.c (toplev::main): Remove excess parens on pretty_printer
4590 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
4592 2017-10-04 Sudakshina Das <sudi.das@arm.com>
4594 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
4595 check type for aarch64_simd_valid_immediate.
4596 (aarch64_output_simd_mov_immediate): Update prototype.
4597 (aarch64_simd_valid_immediate): Update prototype.
4598 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
4599 support for ORR-immediate.
4600 (and<mode>3): modified pattern to add support for BIC-immediate.
4601 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
4602 now checks for valid immediate for BIC and ORR based on new enum
4604 (aarch64_output_simd_mov_immediate): Function now used to output
4605 BIC/ORR imm as well based on new enum argument.
4606 * config/aarch64/constraints.md (Do): New vector immediate constraint.
4608 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
4609 (aarch64_reg_or_bic_imm): Likewise.
4611 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4613 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
4614 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
4616 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
4620 PR rtl-optimization/82396
4621 * haifa-sched.c (autopref_multipass_init): Simplify
4623 (autopref_rank_data): Simplify sort order.
4624 * sched-int.h (autopref_multipass_data_): Remove
4625 multi_mem_insn_p, min_offset and max_offset.
4627 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4629 * doc/sourcebuild.texi: Document vect_peeling_profitable.
4631 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4633 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
4636 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4638 * doc/sourcebuild.texi: Document vect_long_mult.
4640 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
4642 PR tree-optimization/82413
4643 * fold-const.c (build_range_check): Use widest_int when comparing
4644 the maximum ETYPE value with HIGH.
4646 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
4648 PR rtl-optimization/82396
4649 * haifa-sched.c (autopref_multipass_init): Simplify
4651 (autopref_rank_data): Simplify sort order.
4652 * sched-int.h (autopref_multipass_data_): Remove
4653 multi_mem_insn_p, min_offset and max_offset.
4655 2017-10-04 Jakub Jelinek <jakub@redhat.com>
4657 PR tree-optimization/82381
4658 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
4659 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
4662 PR tree-optimization/82374
4663 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
4664 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
4665 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
4666 current_function_decl to the new decl.
4668 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
4670 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
4671 helper macro for IEEE float128 hardware built-in functions.
4672 (SQRTF128_ODD): Add built-in functions with the round-to-odd
4674 (TRUNCF128_ODD): Likewise.
4675 (ADDF128_ODD): Likewise.
4676 (SUBF128_ODD): Likewise.
4677 (MULF128_ODD): Likewise.
4678 (DIVF128_ODD): Likewise.
4679 (FMAF128_ODD): Likewise.
4680 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
4681 UNSPEC_TRUNC_ROUND_TO_ODD.
4682 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
4683 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
4684 floating point round to odd instructions.
4685 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
4686 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
4687 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
4688 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
4689 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
4690 (trunc<mode>sf2_hw): Change the truncate with round to odd
4691 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
4692 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
4693 to odd hardware instructions.
4694 (sub<mode>3_odd): Likewise.
4695 (mul<mode>3_odd): Likewise.
4696 (div<mode>3_odd): Likewise.
4697 (sqrt<mode>2_odd): Likewise.
4698 (fma<mode>4_odd): Likewise.
4699 (fms<mode>4_odd): Likewise.
4700 (nfma<mode>4_odd): Likewise.
4701 (nfms<mode>4_odd): Likewise.
4702 (trunc<mode>df2_odd): Change the truncate with round to odd
4703 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
4705 * doc/extend.texi (PowerPC built-in functions): Update documentation
4706 for existing IEEE float128-bit built-in functions. Add built-in
4707 functions that generate the IEEE 128-bit floating point round to
4710 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
4712 PR rtl-optimization/77729
4713 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
4714 to (X&(C1&~C2))|C2 transformations.
4716 2017-10-03 Martin Jambor <mjambor@suse.cz>
4718 PR tree-optimization/82363
4719 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
4720 mismatch, mark lacc written regardless of racc.
4722 2017-10-03 Jakub Jelinek <jakub@redhat.com>
4724 PR tree-optimization/82381
4725 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
4726 stmt_to_insert nor wheather SSA_NAMEs are default defs.
4727 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
4728 fallthrough into reassoc_stmt_dominates_stmt_p.
4731 * combine.c (combine_instructions): Don't combine in unreachable
4734 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
4737 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
4738 function to not use the have_cpu variable. Do not set cpu_index,
4739 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
4741 (rs6000_valid_attribute_p): Remove duplicate initializations of
4742 old_optimize and func_optimize.
4743 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
4744 (rs6000_activate_target_options): Make global.
4745 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
4748 2017-10-02 Jakub Jelinek <jakub@redhat.com>
4750 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
4751 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
4752 Set *poffset to 0 and *psize and *pmax_size to -1 if
4753 *poffset + *psize overflows in HOST_WIDE_INT.
4755 PR tree-optimization/82387
4756 PR tree-optimization/82388
4757 PR tree-optimization/82389
4758 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
4759 instead of live_bytes non-NULL.
4761 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
4764 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
4767 2017-10-02 Richard Biener <rguenther@suse.de>
4769 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
4770 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
4771 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
4773 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
4775 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
4776 requested precision matches the type's.
4777 * calls.c (alloc_max_size): Calculate the new candidate size as
4778 a widest_int and use wi::to_widest when comparing it with the
4779 current candidate size.
4780 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
4781 zero rather than integer_zero_node.
4782 * match.pd: Check for a no-op conversion before using wi::add
4783 rather than after. Use tree_to_uhwi when summing small shift
4784 counts into an unsigned int.
4786 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
4787 Alan Hayward <alan.hayward@arm.com>
4788 David Sherwood <david.sherwood@arm.com>
4791 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
4792 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
4793 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
4794 POINTER_AND_FP_REGS.
4796 2017-10-02 Richard Biener <rguenther@suse.de>
4798 PR tree-optimization/82355
4799 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
4800 a mapping for the enclosing loop but avoid generating one for
4802 (copy_bb_and_scalar_dependences): Remove premature codegen
4803 error on PHIs in blocks duplicated into multiple places.
4804 * graphite-scop-detection.c
4805 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
4806 in the region use it as loop and nest to analyze the DR in.
4807 (try_generate_gimple_bb): Likewise.
4808 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
4809 (add_loop_constraints): For blocks in a loop not in the region
4810 create a dimension with a single iteration.
4811 * sese.h (gbb_loop_at_index): Remove assert.
4813 2017-10-01 Kevin Buettner <kevinb@redhat.com>
4815 * omp-expand.c (adjust_context_scope): New function.
4816 (expand_parallel_call): Call adjust_context_scope.
4818 2017-10-01 Jeff Law <law@redhat.com>
4820 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
4821 dom_opt_dom_walker class with direct access to private members.
4822 Add comments. Call test_for_singularity.
4823 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
4824 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
4825 m_dummy_cond anymore.
4826 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
4828 (pass_dominator:execute): Build the dummy_cond here and pass it
4829 to the dom_opt_dom_walker ctor.
4830 (test_for_singularity): New function.
4832 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
4833 Maya Rashish <coypu@sdf.org>
4835 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
4836 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
4837 (alpha*-*-netbsd*) Use nbsd_tm_file.
4838 (arm*-*-netbsdelf*) Likewise.
4839 (i[34567]86-*-netbsdelf*) Likewise.
4840 (x86_64-*-netbsd*) Likewise.
4841 (mips*-*-netbsd*) Likewise.
4842 (powerpc-*-netbsd*) Likewise.
4843 (sh*-*-netbsd*) Likewise.
4844 (sparc-*-netbsdelf*) Likewise.
4845 (sparc64-*-netbsd*) Likewise.
4846 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
4848 (vax-*-netbsdelf*) Likewise.
4849 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
4850 (UINT_FAST8_TYPE) Likewise.
4851 (INT_FAST16_TYPE) Check CHAR_FAST16.
4852 (UINT_FAST16_TYPE) Likewise.
4854 2017-09-30 Jakub Jelinek <jakub@redhat.com>
4857 * config/i386/i386.md
4858 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
4859 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
4860 *divmodsi4_zext_2): New define_insn_and_split.
4861 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
4862 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
4863 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
4864 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
4865 New define_insn_and_split.
4866 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
4867 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
4868 operands[1] having DImode when mode is SImode.
4870 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
4871 always SImode for DIV and MOD in REG_EQUAL notes.
4873 2017-09-29 Yury Gribov <tetra2005@gmail.com>
4876 * match.pd: Fix handling of NaNs in pattern.
4878 2017-09-29 Jeff Law <law@redhat.com>
4880 * sbitmap.c (bitmap_bit_in_range_p): New function.
4881 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
4882 * tree-ssa-dse.c (live_bytes_read): New function.
4883 (dse_classify_store): Ignore reads of dead bytes.
4885 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
4886 typos and whitespace errors.
4887 * config/i386/predicates.md (address_no_seg_operand): Likewise.
4888 * config/s390/s390.c (s390_emit_prologue): Likewise.
4890 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
4893 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
4894 with a symbol for LRA.
4896 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
4898 PR rtl-optimization/82338
4899 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
4901 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
4903 * genmodes.c (calc_wider_mode): Suppress qsort macro.
4904 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
4905 (qsort_chk): Declare.
4906 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
4907 (qsort_chk): New function.
4909 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4911 PR tree-optimization/82337
4912 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
4913 phi definition if the PHI result appears in an abnormal PHI.
4914 (find_basis_for_base_expr): Don't record a basis if the LHS of the
4915 basis appears in an abnormal PHI.
4917 2017-09-29 Richard Biener <rguenther@suse.de>
4919 * graphite-isl-ast-to-gimple.c
4920 (translate_isl_ast_to_gimple::set_codegen_error): New function.
4921 (binary_op_to_tree): Use it.
4922 (get_rename_from_scev): Likewise.
4923 (copy_loop_phi_nodes): Likewise.
4924 (copy_bb_and_scalar_dependences): Likewise.
4925 (translate_pending_phi_nodes): Likewise.
4927 2017-09-29 Jakub Jelinek <jakub@redhat.com>
4930 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
4931 for movabsq $(i32 << shift), r64.
4933 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
4935 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
4936 index when encoding %esp as %rsp to avoid 0x67 prefix.
4938 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
4940 * config/i386/i386.md (*movsf_internal, *movdf_internal):
4941 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
4943 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
4945 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
4946 Extensions with more than 16 double VFP registers.
4947 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
4948 to_clear_mask and all code related to it. Replace the remaining
4949 entry by a sbitmap and adapt code accordingly.
4951 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
4953 * brig-builtins.def: Change pure attributes to const.
4955 2017-09-28 Joseph Myers <joseph@codesourcery.com>
4957 * config.gcc (default_gnu_indirect_function): Default to yes for
4958 sparc*-*-linux* with glibc.
4960 2017-09-28 Joseph Myers <joseph@codesourcery.com>
4962 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
4963 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
4964 when creating .init_array and .fini_array sections with priority
4967 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
4970 * config/aarch64/aarch64.c
4971 (aarch64_builtin_support_vector_misalignment): Always return false
4972 when misalignment is unknown.
4974 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
4976 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
4977 this function to return false if the definition used by the swap
4978 instruction is artificial, or if the memory address from which the
4979 constant value is loaded is not represented by a base address held
4980 in a register or if the base address register is a frame or stack
4981 pointer. Additionally, return false if the base address of the
4982 loaded constant is a SYMBOL_REF but is not considered to be a
4984 (replace_swapped_load_constant): New function.
4985 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
4986 loaded constant vector with a load of a swapped constant vector.
4988 2017-09-27 Carl Love <cel@us.ibm.com>
4990 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
4991 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
4992 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
4995 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
4997 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
4998 first, always call autopref_rank_data otherwise.
5000 2017-09-27 Richard Biener <rguenther@suse.de>
5002 * graphite-scop-detection.c (find_scop_parameters): Move
5003 loop bound handling ...
5004 (gather_bbs::before_dom_children): ... here, avoiding the need
5005 to build scop_info->loop_nest.
5006 (record_loop_in_sese): Remove.
5007 * sese.h (sese_info_t::loop_nest): Remove.
5008 * sese.c (new_sese_info): Do not allocate loop_nest.
5009 (free_sese_info): Do not free loop_nest.
5011 2017-09-27 Jakub Jelinek <jakub@redhat.com>
5014 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
5015 lhs from calls if the lhs has addressable type.
5017 2017-09-27 Richard Biener <rguenther@suse.de>
5019 * graphite.h (scop::max_alias_set): New member.
5020 * graphite-scop-detection.c: Remove references to non-existing
5021 --param in comments.
5022 (build_alias_sets): Record the maximum alias set used for drs.
5023 (build_scops): Support zero as unlimited for
5024 --param graphite-max-arrays-per-scop.
5025 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
5027 (build_poly_sr_1): ... here. Compute alias set based on the
5028 maximum alias set used for drs rather than
5029 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
5031 2017-09-27 Richard Biener <rguenther@suse.de>
5033 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
5034 --param loop-block-tile-size=0 to disable tiling.
5036 2017-09-27 Richard Biener <rguenther@suse.de>
5038 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
5039 (graphite-max-nb-scop-params): Document special value zero.
5040 * domwalk.h (dom_walker::STOP): New symbolical constant.
5041 (dom_walker::dom_walker): Add optional parameter for bb to
5043 (dom_walker::~dom_walker): Declare.
5044 (dom_walker::before_dom_children): Document STOP return value.
5045 (dom_walker::m_user_bb_to_rpo): New member.
5046 (dom_walker::m_bb_to_rpo): Likewise.
5047 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
5048 mapping here if not provided by the user.
5049 (dom_walker::~dom_walker): Free bb to RPO mapping if not
5050 provided by the user.
5051 (dom_walker::STOP): Define.
5052 (dom_walker::walk): Do not compute bb to RPO mapping here.
5053 Support STOP return value from before_dom_children to stop
5055 * graphite-optimize-isl.c (optimize_isl): If the schedule
5056 is the same still generate code if -fgraphite-identity
5057 or -floop-parallelize-all are given.
5058 * graphite-scop-detection.c: Include cfganal.h.
5059 (gather_bbs::gather_bbs): Get and pass through bb to RPO
5061 (gather_bbs::before_dom_children): Return STOP for BBs
5063 (build_scops): Compute bb to RPO mapping and pass it to
5064 the domwalk. Treat --param graphite-max-nb-scop-params=0
5065 as not limiting the number of params.
5066 * graphite.c (graphite_initialize): Remove limit on the
5067 number of basic-blocks in a function.
5068 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
5069 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
5070 default value of 10.
5072 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5074 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
5075 Adjust code to eliminate needing to do the shift right 32-bits
5076 operation after XSCVDPSPN.
5078 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5080 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
5081 ((X / Y) != 0 -> X >= Y): Likewise.
5083 2017-09-26 Carl Love <cel@us.ibm.com>
5085 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
5086 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
5087 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
5088 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
5089 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
5090 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
5091 definitions and overloading.
5092 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
5093 statement for P9V_BUILTIN_XST_LEN_R.
5094 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
5095 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
5096 define_expand and define_insn for the instructions and builtins.
5097 * doc/extend.texi: Update the built-in documentation file for the new
5099 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
5100 define_insn for the instructions
5102 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
5105 * gcc/config/netbsd-protos.h: New file.
5106 * gcc/config/netbsd.c: New file.
5107 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
5108 * gcc/config/t-netbsd: New file.
5109 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
5110 (tmake_file) Add t-netbsd.
5111 (extra_objs) Add netbsd.o.
5113 2017-09-26 Janus Weil <janus@gcc.gnu.org>
5117 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
5119 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5121 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
5122 sign extension from a vector register to a GPR by doing a 32-bit
5123 direct move and then an EXTSW.
5124 (extendsi<mode>2 splitter): Likewise.
5125 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
5126 right or vector extract after doing XSCVDPSPN. Use
5127 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
5129 (movdi_from_sf_zero_ext): Likewise.
5130 (reload_gpr_from_vsxsf): Likewise.
5131 (p8_mfvsrd_4_disf): Delete, no longer used.
5132 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
5133 then needing to move the SFmode to a GPR to use the XSCVDPSP
5134 instruction instead of FRSP and XSCVDPSPN.
5135 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
5136 it is adjacent to the other XSCVSPDP insns.
5137 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
5138 SFmode to be in traditional Altivec registers.
5139 (vsx_xscvdpspn): Eliminate useless alternative constraint.
5140 (vsx_xscvspdpn): Likewise.
5141 (vsx_xscvspdpn_scalar): Likewise.
5143 2017-09-26 Martin Jambor <mjambor@suse.cz>
5145 * tree-sra.c (compare_access_positions): Put integral types first,
5146 stabilize sorting of integral types, remove conditions putting
5147 non-full-precision integers last.
5148 (sort_and_splice_var_accesses): Disable scalarization if a
5149 non-integert would be represented by a non-full-precision integer.
5151 2017-09-26 Joseph Myers <joseph@codesourcery.com>
5153 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
5154 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
5155 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
5156 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
5157 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
5158 conditionals inside the function instead of around it. Call
5159 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
5160 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
5162 2017-09-26 Richard Biener <rguenther@suse.de>
5164 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
5166 (scop_detection::build_scop_breadth): ... this. Removed.
5167 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
5168 (scop_detection::harmful_stmt_in_bb): Likewise.
5169 (scop_detection::graphite_can_represent_stmt): Likewise.
5170 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
5171 (scop_detection::can_represent_loop): Remove recursion, fold in ...
5172 (scop_detection::can_represent_loop_1): ... this. Removed.
5173 (scop_detection::harmful_loop_in_region): Simplify after inlining
5174 the above and remove more quadraticness.
5175 (build_scops): Adjust.
5176 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
5179 2017-09-26 Jakub Jelinek <jakub@redhat.com>
5182 * config/i386/i386.c (ix86_print_operand_address_as): Only test
5183 REGNO (base) == SP_REG if base is a REG.
5186 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
5187 if it is different SSA_NAME.
5188 (optimize_range_tests_cmp_bitwise): New function.
5189 (optimize_range_tests): Call it.
5191 2017-09-26 Richard Biener <rguenther@suse.de>
5193 PR tree-optimization/82321
5194 * graphite.c (canonicalize_loop_closed_ssa): Properly check
5195 for the def being inside the loop.
5197 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5199 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
5201 * config/s390/s390-builtins.def: Fix constraint on op4.
5203 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5205 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
5206 independent expanders.
5207 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
5208 ("vec_ordered", "vec_unordered"): New expanders.
5210 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5212 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
5215 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5217 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
5218 vec_unpacks_lo_v16qi.
5219 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
5221 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5223 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
5224 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
5225 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
5227 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5229 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
5231 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
5233 ("*vec_slb<mode>"): New insn pattern.
5234 ("vec_shr_<mode>"): New expander.
5235 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
5236 and force the shift count operand to V16QImode.
5237 ("vec_srb<mode>"): Set shift count mode to V16QI.
5239 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5241 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
5242 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
5243 ("vec_widen_smult_hi_<mode>"): New expander definitions.
5245 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
5248 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
5250 2017-09-26 Richard Biener <rguenther@suse.de>
5252 PR tree-optimization/82320
5253 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
5256 2017-09-25 Jeff Law <law@redhat.com>
5258 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
5259 prototype for new argument.
5260 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
5261 mostly extracted from rs6000_emit_allocate_stack.
5262 (rs6000_emit_probe_stack_range_stack_clash): New function.
5263 (rs6000_emit_allocate_stack): Call
5264 rs6000_emit_probe_stack_range_stack_clash as needed.
5265 (rs6000_emit_probe_stack_range): Add additional argument
5266 to call to gen_probe_stack_range{si,di}.
5267 (output_probe_stack_range): New.
5268 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
5269 (output_probe_stack_range_stack_clash): New.
5270 (rs6000_emit_prologue): Emit notes into dump file as requested.
5271 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
5272 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
5273 Add additional operand and pass it to output_probe_stack_range.
5275 2017-09-25 Bin Cheng <bin.cheng@arm.com>
5277 PR tree-optimization/82163
5278 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
5279 (checking_verify_loop_closed_ssa): New parameter.
5280 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
5281 (check_loop_closed_ssa_stmt): Delete.
5282 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
5283 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
5284 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
5287 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
5289 * brig-builtins.def: Treat HSAIL barrier builtins as
5290 setjmp/longjump style functions.
5292 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
5294 * target.def (constant_alignment): New hook.
5295 * defaults.h (CONSTANT_ALIGNMENT): Delete.
5296 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
5297 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
5298 * doc/tm.texi: Regenerate.
5299 * targhooks.h (default_constant_alignment): Declare.
5300 (constant_alignment_word_strings): Likewise.
5301 * targhooks.c (default_constant_alignment): New function.
5302 (constant_alignment_word_strings): Likewise.
5303 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
5304 instead of CONSTANT_ALIGNMENT.
5305 * varasm.c (align_variable, get_variable_align, build_constant_desc)
5306 (force_const_mem): Likewise.
5307 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
5308 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
5309 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
5310 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5311 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
5313 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
5314 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5315 constant_alignment_word_strings.
5316 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
5317 (CONSTANT_ALIGNMENT): Likewise.
5318 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5319 (arm_constant_alignment): New function.
5320 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
5321 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5322 constant_alignment_word_strings.
5323 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
5324 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5325 constant_alignment_word_strings.
5326 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
5327 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5328 (cris_constant_alignment): New function.
5329 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
5330 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5331 (epiphany_constant_alignment): New function.
5332 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
5333 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5334 constant_alignment_word_strings.
5335 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
5336 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5337 constant_alignment_word_strings.
5338 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
5339 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5340 constant_alignment_word_strings.
5341 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
5342 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
5343 * config/i386/i386.c (ix86_constant_alignment): Make static.
5344 Use the same interface as the target hook.
5345 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5346 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
5347 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5348 constant_alignment_word_strings.
5349 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
5350 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
5351 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5352 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
5353 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5354 constant_alignment_word_strings.
5355 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
5356 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5357 constant_alignment_word_strings.
5358 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
5359 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5360 constant_alignment_word_strings.
5361 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
5362 * config/microblaze/microblaze.c (microblaze_constant_alignment):
5364 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5365 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
5366 * config/mips/mips.c (mips_constant_alignment): New function.
5367 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5368 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
5369 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
5370 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5371 (mmix_constant_alignment): Make static. Use the same interface
5373 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
5374 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5375 constant_alignment_word_strings.
5376 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
5377 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5378 constant_alignment_word_strings.
5379 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
5380 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5381 constant_alignment_word_strings.
5382 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
5383 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5384 (rs6000_constant_alignment): New function.
5385 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
5386 * config/riscv/riscv.c (riscv_constant_alignment): New function.
5387 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5388 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
5389 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5390 (rs6000_constant_alignment): New function.
5391 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
5392 * config/s390/s390.c (s390_constant_alignment): New function.
5393 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5394 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
5395 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5396 constant_alignment_word_strings.
5397 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
5398 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5399 (sparc_constant_alignment): New function.
5400 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
5401 * config/spu/spu.c (spu_constant_alignment): New function.
5402 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5403 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
5404 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5405 constant_alignment_word_strings.
5406 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
5407 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5408 constant_alignment_word_strings.
5409 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
5410 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5411 constant_alignment_word_strings.
5412 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
5413 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5414 (visium_constant_alignment): New function.
5415 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
5416 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5417 (xtensa_constant_alignment): New function.
5418 * system.h (CONSTANT_ALIGNMENT): Poison.
5420 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
5422 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
5423 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
5424 (rs6000_builtin_valid_without_lhs): New helper function.
5425 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5426 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
5428 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
5430 * target.h (vec_perm_indices): Use unsigned short rather than
5432 (auto_vec_perm_indices): Likewise.
5433 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
5434 Use unsigned int rather than unsigned char.
5435 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
5437 2017-09-25 Richard Biener <rguenther@suse.de>
5439 * cfgloop.h (sort_sibling_loops): Declare.
5440 * cfgloop.c (sort_sibling_loops_cmp): New helper.
5441 (sort_sibling_loops): New function sorting the sibling loop list
5443 * graphite.c (graphite_transform_loops): Sort sibling loops.
5445 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
5447 * target.def (vec_perm_const_ok): Change sel parameter to
5449 * optabs-query.c (can_vec_perm_p): Update accordingly.
5450 * doc/tm.texi: Regenerate.
5451 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
5452 auto_vec_perm_indices and remove separate nelt field.
5453 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
5454 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
5455 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
5456 (aarch64_expand_vec_perm_const): Update accordingly.
5457 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
5458 to vec_perm_indices.
5459 * config/arm/arm.c (expand_vec_perm_d): Change perm to
5460 auto_vec_perm_indices and remove separate nelt field.
5461 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
5462 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
5463 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
5465 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
5466 to vec_perm_indices.
5467 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
5468 sel to vec_perm_indices.
5469 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
5470 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
5471 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
5473 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
5476 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
5479 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
5480 on the FUNCTION_DECL function context if it has a DIE that is a
5483 2017-09-25 Richard Biener <rguenther@suse.de>
5485 PR tree-optimization/82285
5486 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
5489 2017-09-25 Tom de Vries <tom@codesourcery.com>
5493 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
5496 2017-09-25 Richard Biener <rguenther@suse.de>
5498 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
5499 ISL errors other than isl_error_quota happen. Dump if the
5500 schedule is the same.
5501 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
5502 errors instead of aborting inside ISL.
5504 2017-09-25 Iain Sandoe <iain@codesourcery.com>
5507 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
5508 of libgcc_eh for m64.
5509 * config/i386/darwin64.h: Likewise.
5511 2017-09-25 Richard Biener <rguenther@suse.de>
5514 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
5515 attribute for incomplete types nor twice for complete ones.
5517 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
5520 * config/i386/i386.c (ix86_print_operand_address_as): Encode
5521 %esp as %rsp to avoid 0x67 prefix if there is no index or base
5524 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
5527 * config/i386/i386.opt (mprefer-avx256): Use
5528 ix86_target_flags variable.
5529 * config/i386/i386.c (ix86_target_string): Move
5530 -mprefer-avx256 to flag2_opts.
5532 2017-09-22 Jakub Jelinek <jakub@redhat.com>
5535 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
5536 and x != -1 | y != -1 into (x & y) != -1.
5538 2017-09-22 Steve Ellcey <sellcey@cavium.com>
5540 * config.gcc: Add new case statement to set
5541 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
5542 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
5543 s390x-*-linux* case statements. Added aarch64 to the list of
5544 supported architectures.
5546 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
5548 PR tree-optimization/82289
5549 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
5550 STMT_VINFO_RELEVANT_P.
5552 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
5553 Alan Hayward <alan.hayward@arm.com>
5554 David Sherwood <david.sherwood@arm.com>
5556 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
5557 for VR_RANGE only; don't allow VR_ANTI_RANGE.
5558 (extract_range_from_binary_expr_1): Don't call
5559 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
5561 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
5562 Alan Hayward <alan.hayward@arm.com>
5563 David Sherwood <david.sherwood@arm.com>
5565 * target.def (preferred_vector_alignment): New hook.
5566 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
5568 * doc/tm.texi: Regenerate.
5569 * targhooks.h (default_preferred_vector_alignment): Declare.
5570 * targhooks.c (default_preferred_vector_alignment): New function.
5571 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
5573 (DR_TARGET_ALIGNMENT): New macro.
5574 (aligned_access_p): Update commentary.
5575 (vect_known_alignment_in_bytes): New function.
5576 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
5578 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
5579 Calculate the misalignment based on the target alignment rather than
5581 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
5582 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
5583 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
5584 the target alignment, rather than masking the element misalignment
5585 with the number of elements in a vector. Also use the target
5586 alignment when calculating the maximum number of peels.
5587 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
5588 instead of TYPE_ALIGN_UNIT.
5589 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
5590 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
5591 (vect_create_addr_base_for_vector_ref): Update call accordingly.
5592 (vect_create_data_ref_ptr): Likewise.
5593 (vect_setup_realignment): Realign by ANDing with
5594 -DR_TARGET_MISALIGNMENT.
5595 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
5596 the number of peels based on DR_TARGET_ALIGNMENT.
5597 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
5598 with the guaranteed alignment boundary when deciding whether
5600 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
5601 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
5602 (ensure_base_align): Remove stmt_info parameter. Get the
5603 target base alignment from DR_TARGET_ALIGNMENT.
5604 (vectorizable_store): Update call accordingly. Interpret
5605 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
5607 (vectorizable_load): Likewise.
5609 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
5610 Alan Hayward <alan.hayward@arm.com>
5611 David Sherwood <david.sherwood@arm.com>
5613 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
5614 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
5615 (vect_enhance_data_refs_alignment): Likewise.
5617 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
5619 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
5620 error. Only quit immediately if parsing is complete.
5621 (BEGIN): Initialize fatal_err and parse_done.
5622 (begin fpu, end fpu): Check number of arguments.
5623 (begin arch, end arch): Likewise.
5624 (begin cpu, end cpu): Likewise.
5625 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
5626 (optalias): Likewise.
5628 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
5630 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
5631 * config/arm/arm-isa.h: Delete. Move definitions to ...
5632 * arm-cpus.in: ... here. Use new feature and fgroup values.
5633 * config/arm/arm.c (arm_option_override): Use lower case for feature
5635 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
5636 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
5637 * config/arm/parsecpu.awk (END): Add new command 'isa'.
5639 (print_isa_bits_for): New function.
5640 (gen_isa): New function.
5641 (gen_comm_data): Use print_isa_bits_for.
5642 (define feature): New keyword.
5643 (define fgroup): New keyword.
5644 * config/arm/t-arm (TM_H): Remove.
5645 (GTM_H): Add arm-isa.h.
5646 (arm-isa.h): Add rule to generate file.
5647 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
5648 case for feature bit names.
5650 2017-09-22 Richard Biener <rguenther@suse.de>
5652 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
5654 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
5655 print of no dependency loops ...
5656 * graphite.c (graphite_transform_loops): ... here.
5657 (canonicalize_loop_closed_ssa_form): Work from inner to outer
5659 (same_close_phi_node, remove_duplicate_close_phi,
5660 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
5661 (canonicalize_loop_closed_ssa): ... here and simplify.
5662 * graphite-optimize-isl.c: Include tree-vectorizer.h.
5663 (optimize_isl): Use dump_printf_loc to tell when we stopped
5664 optimizing because of an ISL timeout.
5666 2017-09-22 Richard Biener <rguenther@suse.de>
5668 PR tree-optimization/82291
5669 * tree-if-conv.c (predicate_mem_writes): Make sure to
5670 remove writes in blocks predicated with false.
5672 2017-09-22 Richard Biener <rguenther@suse.de>
5674 * sese.c: Include cfganal.h.
5675 (if_region_set_false_region): Remove.
5676 (create_if_region_on_edge): Likewise.
5677 (move_sese_in_condition): Re-implement without destroying
5680 2017-09-22 Richard Biener <rguenther@suse.de>
5682 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
5683 Verify both BBs contain loop PHI nodes before dispatching to
5685 (graphite_regenerate_ast_isl): Do not recompute dominators,
5686 do not verify three times. Restructure for clarity.
5687 * graphite-scop-detection.c (same_close_phi_node,
5688 remove_duplicate_close_phi, make_close_phi_nodes_unique,
5689 defined_in_loop_p, canonicalize_loop_closed_ssa,
5690 canonicalize_loop_closed_ssa_form): Simplify, remove excess
5691 checking and SSA rewrite, move to ...
5692 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
5693 (graphite_initialize): Do not pass in ctx, do not reset the
5694 SCEV cache, compute only dominators.
5695 (graphite_transform_loops): Allocate ISL ctx after
5696 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
5697 Maintain post-dominators only around build_scops.
5698 * sese.c (if_region_set_false_region): Make static. Free
5699 and recompute dominators.
5700 (move_sese_in_condition): Assert we don't get called with
5701 post-dominators computed.
5702 * sese.h (if_region_set_false_region): Remove.
5704 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
5706 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
5707 mode attribute for TARGET_AVX512VL.
5709 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
5711 * config/i386/i386.opt (mprefer-avx256): New option.
5712 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
5714 (ix86_preferred_simd_mode): Return 256-bit AVX modes
5715 for TARGET_PREFER_AVX256.
5716 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
5718 2017-09-21 Jeff Law <law@redhat.com>
5720 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
5721 Fix dump output if the only stack space is for pushed registers.
5723 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5725 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
5728 2017-09-21 Martin Sebor <msebor@redhat.com>
5731 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
5732 code (in C++) or code that triggers warnings.
5734 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
5736 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
5738 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
5740 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
5741 * sched-rgn.c: Ditto.
5742 * sel-sched-ir.c: Ditto.
5744 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
5746 * toplev.h (set_random_seed): Adjust return type.
5747 * toplev.c (init_local_tick): Move eager initialization of random_seed
5748 to get_random_seed. Adjust comment.
5749 (init_random_seed): Inline to get_random_seed, delete.
5750 (get_random_seed): Initialize random_seed lazily.
5751 (set_random_seed): Do not return previous value.
5752 (print_switch_value): Do not call get_random_seed.
5754 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
5756 * cgraph.c (delete_function_version): New, broken out from...
5757 (cgraph_node::delete_function_version): ...here. Rename to
5758 cgraph_node::delete_function_version_by_decl. Update all uses.
5759 (cgraph_node::remove): Call delete_function_version.
5761 2017-09-21 Jakub Jelinek <jakub@redhat.com>
5764 * tree-inline.c (expand_call_inline): Emit clobber stmts for
5765 VAR_DECLs to which addressable non-volatile parameters are mapped
5766 and for id->retvar after the return value assignment. Clear
5767 id->retval and id->retbnd after inlining.
5769 2017-09-21 Richard Biener <rguenther@suse.de>
5771 PR tree-optimization/82276
5772 PR tree-optimization/82244
5773 * tree-vrp.c (build_assert_expr_for): Set
5774 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
5776 (remove_range_assertions): Revert earlier change.
5778 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
5781 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
5783 2017-09-21 Richard Biener <rguenther@suse.de>
5785 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5786 Restore valid IL after code generation errors.
5787 * graphite.c (graphite_transform_loops): Diagnose code
5788 generation issues as MSG_MISSED_OPTIMIZATION and continue
5789 with processing SCOPs.
5791 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5792 Alan Hayward <alan.hayward@arm.com>
5793 David Sherwood <david.sherwood@arm.com>
5795 * calls.c (compute_argument_addresses): Use simplify_gen_binary
5796 rather than choosing between plus_constant and gen_rtx_<CODE>.
5797 * expr.c (emit_push_insn): Likewise.
5798 (expand_expr_real_2): Likewise.
5800 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5801 Alan Hayward <alan.hayward@arm.com>
5802 David Sherwood <david.sherwood@arm.com>
5804 * loop-unroll.c (split_iv): Call copy_rtx on the step.
5806 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5807 Alan Hayward <alan.hayward@arm.com>
5808 David Sherwood <david.sherwood@arm.com>
5810 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
5811 calling tree_to_uhwi.
5813 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5814 Alan Hayward <alan.hayward@arm.com>
5815 David Sherwood <david.sherwood@arm.com>
5817 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
5818 INTEGER_CST rather than a negative test for ADDR_EXPR.
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-vrp.c (extract_range_from_binary_expr_1): Check
5825 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
5827 2017-09-21 Richard Biener <rguenther@suse.de>
5829 PR tree-optimization/71351
5830 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
5831 graphite_create_new_loop_guard): Remove, fold remaining parts
5833 (translate_isl_ast_node_for): ... here and simplify.
5835 2017-09-21 Jakub Jelinek <jakub@redhat.com>
5838 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
5839 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
5840 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
5841 alternative always use QI mode, for -Os imov (=R,R) alternative
5842 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
5845 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5846 Jeff Law <law@redhat.com>
5848 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
5849 (allocate_stack_space): New function, partially extracted from
5851 (s390_emit_prologue): Track offset to most recent stack probe.
5852 Code to allocate space moved into allocate_stack_space.
5853 Dump actions when no stack is allocated.
5854 (s390_prologue_plus_offset): New function.
5855 (s390_emit_stack_probe): Likewise.
5857 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
5859 * common.opt (Wa, Wl, Wp, g, gz=): Add
5861 (gno-column-info): Remove.
5862 (gcolumn-info): Drop RejectNegative.
5864 (gno-record-gcc-switches): Remove.
5865 (grecord-gcc-switches): Drop RejectNegative.
5866 (gno-split-dwarf): Remove.
5867 (gsplit-dwarf): Drop RejectNegative.
5868 (gno-strict-dwarf): Remove.
5869 (gstrict-dwarf): Drop RejectNegative.
5870 * config/darwin.opt (gfull, gused): Add RejectNegative.
5871 * dwarf2out.c (gen_producer_string): Drop
5872 gno-record-gcc-switches handler.
5873 * optc-gen.awk: Add g to prefixes with negative forms.
5874 * opts-common.c (remapping_prefix_p): New.
5875 (find_opt): Check it.
5876 (generate_canonical_option): Test g prefix.
5877 (option_map): Add -gno- mapping.
5878 (add_misspelling_candidates): Check remapping_prefix_p.
5880 2017-09-20 Jeff Law <law@redhat.com>
5882 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
5883 thinko in stack clash protection support.
5885 * explow.c (compute_stack_clash_protection_loop_data): Use
5886 CONST_INT_P instead of explicit test. Verify object is a
5887 CONST_INT_P before looking at INTVAL.
5888 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
5889 instead of explicit test.
5891 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
5894 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
5895 address instead of to r1 and r11.
5897 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
5899 * config.gcc: Support "knm".
5900 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
5901 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
5903 * config/i386/i386.c (m_KNM): Define.
5904 (processor_target_table): Add "knm".
5906 (ix86_option_override_internal): Add "knm".
5907 (ix86_issue_rate): Add PROCESSOR_KNM.
5908 (ix86_adjust_cost): Ditto.
5909 (ia32_multipass_dfa_lookahead): Ditto.
5910 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
5911 (fold_builtin_cpu): Add M_INTEL_KNM.
5912 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
5913 (processor_type): Add PROCESSOR_KNM.
5914 * config/i386/x86-tune.def: Add m_KNM.
5915 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
5917 2017-09-20 Richard Biener <rguenther@suse.de>
5919 PR tree-optimization/80213
5920 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
5921 are allowed in empty BBs as well.
5922 (canonicalize_loop_closed_ssa): Also look for other complex
5924 (scop_detection::get_sese): Include the loop-closed PHI block
5926 (scop_detection::merge_sese): Remove code adding extra blocks.
5927 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
5928 (build_scops): Assert the final returned scop is invalid.
5930 2017-09-20 Richard Biener <rguenther@suse.de>
5932 PR tree-optimization/82264
5933 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
5935 (vn_phi_lookup): Likewise.
5936 (vn_phi_insert): Likewise.
5938 2017-09-20 Jakub Jelinek <jakub@redhat.com>
5940 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
5941 that fits into uhwi or shwi, add DW_AT_const_value regardless
5942 of early_dwarf without going through RTL, using add_AT_unsigned
5945 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
5946 (DEBUG_LTO_ABBREV_SECTION): Likewise.
5947 (DEBUG_LTO_MACINFO_SECTION): Likewise.
5948 (DEBUG_MACRO_SECTION): Likewise.
5949 (DEBUG_LTO_MACRO_SECTION): Likewise.
5950 (DEBUG_STR_DWO_SECTION): Likewise.
5951 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
5952 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
5953 (DEBUG_LTO_DWO_LINE_SECTION): Define.
5954 (DEBUG_LTO_LINE_STR_SECTION): Define.
5955 (init_sections_and_labels): Initialize debug_line_str_section
5956 variable. Initialize debug_loc_section for -gdwarf-5 to
5957 DEBUG_LOCLISTS_SECTION. Formatting fixes.
5959 2017-09-20 Richard Biener <rguenther@suse.de>
5961 * graphite-sese-to-poly.c (extract_affine): Properly handle
5962 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
5964 2017-09-20 Richard Biener <rguenther@suse.de>
5966 PR tree-optimization/81373
5967 * graphite-scop-detection.c (build_cross_bb_scalars_def):
5968 Force SESE live-out defs to be handled even if they are
5971 2017-09-19 Jeff Law <law@redhat.com>
5973 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
5974 nothing for stack adjustments with REG_STACK_CHECK.
5975 * sched-deps.c (parse_add_or_inc): Reject insns with
5976 REG_STACK_CHECK from dependency breaking.
5977 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
5978 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
5979 * reg-notes.def (STACK_CHECK): New note.
5981 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
5982 (ix86_expand_prologue): Dump stack clash info as needed.
5983 Call ix86_adjust_stack_and_probe_stack_clash as needed.
5985 * function.c (dump_stack_clash_frame_info): New function.
5986 * function.h (dump_stack_clash_frame_info): Prototype.
5987 (enum stack_clash_probes): New enum.
5989 * config/alpha/alpha.c (alpha_expand_prologue): Also check
5990 flag_stack_clash_protection.
5991 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
5992 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
5993 (arm_frame_pointer_required): Likewise.
5994 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
5995 (ia64_expand_prologue): Likewise.
5996 * config/mips/mips.c (mips_expand_prologue): Likewise.
5997 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
5998 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
5999 (sparc_flat_expand_prologue): Likewise.
6000 * config/spu/spu.c (spu_expand_prologue): Likewise.
6002 * explow.c: Include "params.h".
6003 (anti_adjust_stack_and_probe_stack_clash): New function.
6004 (get_stack_check_protect): Likewise.
6005 (compute_stack_clash_protection_loop_data): Likewise.
6006 (emit_stack_clash_protection_loop_start): Likewise.
6007 (emit_stack_clash_protection_loop_end): Likewise.
6008 (allocate_dynamic_stack_space): Use get_stack_check_protect.
6009 Use anti_adjust_stack_and_probe_stack_clash.
6010 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
6011 (emit_stack_clash_protection_loop_start): Likewise.
6012 (emit_stack_clash_protection_loop_end): Likewise.
6013 * rtl.h (get_stack_check_protect): Prototype.
6014 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
6015 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
6016 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
6018 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
6020 * doc/tm.texi: Rebuilt.
6021 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
6022 get_stack_check_protect.
6023 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
6024 * config/arm/arm.c (arm_expand_prologue): Likewise.
6025 (arm_frame_pointer_required): Likewise.
6026 * config/i386/i386.c (ix86_expand_prologue): Likewise.
6027 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
6028 * config/mips/mips.c (mips_expand_prologue): Likewise.
6029 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
6030 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
6031 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6032 (sparc_flat_expand_prologue): Likewise.
6034 * common.opt (-fstack-clash-protection): New option.
6035 * flag-types.h (enum stack_check_type): Note difference between
6036 -fstack-check= and -fstack-clash-protection.
6037 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
6038 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
6039 * toplev.c (process_options): Issue warnings/errors for cases
6040 not handled with -fstack-clash-protection.
6041 * doc/invoke.texi (-fstack-clash-protection): Document new option.
6042 (-fstack-check): Note additional problem with -fstack-check=generic.
6043 Note that -fstack-check is primarily for Ada and refer users
6044 to -fstack-clash-protection for stack-clash-protection.
6045 Document new params for stack clash protection.
6047 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
6049 * config/i386/i386.c (ix86_split_long_move): Do not handle
6050 address used for LEA in a special way.
6052 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
6054 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
6056 2017-09-19 Martin Sebor <msebor@redhat.com>
6059 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
6060 of incompatible types.
6062 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
6064 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6065 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
6066 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6067 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
6069 2017-09-19 Richard Biener <rguenther@suse.de>
6071 PR tree-optimization/82244
6072 * tree-vrp.c (remove_range_assertions): Do not propagate
6073 a constant to abnormals but replace the assert with a copy.
6075 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
6077 PR rtl-optimization/57878
6078 PR rtl-optimization/68988
6079 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
6080 avoidance test involving non_reload_pseudos. Move frequency test
6081 below the general fragmentation avoidance test.
6083 2017-09-19 Richard Biener <rguenther@suse.de>
6085 PR tree-optimization/69728
6086 * graphite-sese-to-poly.c (schedule_error): New global.
6087 (add_loop_schedule): Handle empty domain by failing the
6089 (build_original_schedule): Handle schedule_error.
6091 2017-09-19 Richard Biener <rguenther@suse.de>
6093 * graphite-scop-detection.c (scop_detection::can_represent_loop):
6094 Do not iterate to sibling loops but only to siblings of inner
6097 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
6100 * config/m68k/m68k.md (moveq feeding equality comparison): Check
6101 that the registers are different.
6103 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
6105 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
6106 to processor_model and "amdfam17h" to arch_names_table.
6107 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
6109 2017-09-18 Jakub Jelinek <jakub@redhat.com>
6112 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
6114 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6115 Alan Hayward <alan.hayward@arm.com>
6116 David Sherwood <david.sherwood@arm.com>
6118 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
6120 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
6122 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
6123 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
6124 vect_schedule_slp_instance.
6125 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
6126 Update call to vect_slp_analyze_node_operations. Simplify return
6128 (vect_slp_analyze_bb_1): Update call accordingly.
6129 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
6130 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
6131 (vect_schedule_slp): Update call accordingly.
6133 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6134 Alan Hayward <alan.hayward@arm.com>
6135 David Sherwood <david.sherwood@arm.com>
6137 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
6138 with types that aren't in fact scalar.
6140 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6142 * tree-vect-slp.c (vect_record_max_nunits): New function,
6144 (vect_build_slp_tree_1): ...here.
6145 (vect_build_slp_tree_2): Call it for phis too.
6147 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6149 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
6150 to vect_get_vec_def_for_operand when getting the mask operand.
6152 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6153 Alan Hayward <alan.hayward@arm.com>
6154 David Sherwood <david.sherwood@arm.com>
6156 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
6159 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6160 Alan Hayward <alan.hayward@arm.com>
6161 David Sherwood <david.sherwood@arm.com>
6163 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
6164 calculation for vector booleans.
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-stmts.c (can_vectorize_live_stmts): New function,
6172 (vect_transform_stmt): ...here.
6173 (vect_analyze_stmt): Use it instead of calling
6174 vectorizable_live_operation directly.
6176 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
6178 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
6179 non-SIMT targets in acc vector loops.
6181 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
6183 * configure.ac: Add arc and check if assembler supports gdwarf2.
6184 * configure: Regenerate.
6186 2017-09-18 Richard Biener <rguenther@suse.de>
6188 PR tree-optimization/82220
6189 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
6190 epilogue niters from the min_profitable_iters compute.
6192 2017-09-18 Jakub Jelinek <jakub@redhat.com>
6195 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
6196 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
6197 (ix86_init_pic_reg): Revert 2017-09-01 changes.
6199 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
6202 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
6203 switching to a new text section.
6205 2017-09-18 Richard Biener <rguenther@suse.de>
6207 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
6209 (build_alias_set): Reject aliases with no access function.
6211 2017-09-18 Richard Biener <rguenther@suse.de>
6213 PR tree-optimization/79622
6214 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
6216 (build_cross_bb_scalars_use): Likewise.
6218 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
6220 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
6222 2017-09-18 Alan Modra <amodra@gmail.com>
6225 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
6226 stack_pointer_rtx for count 0. Update comments. Break up
6227 large rtl expression.
6229 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
6231 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
6232 Increase to 20 bytes.
6233 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
6234 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
6235 or avx version of the stub.
6237 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
6240 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
6241 compute the minimum stack alignment. Also update preferred stack
6242 boundary for leaf functions.
6244 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
6246 PR tree-optimization/82228
6247 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
6250 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
6252 * common/config/nds32/nds32-common.c
6253 (nds32_option_optimization_table): Refine formatting.
6254 (nds32_option_optimization_table): Use -fsched-pressure and
6255 -fomit-frame-pointer for specific optimization level.
6257 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
6259 * config/nds32/nds32.c: Refine formatting and comments.
6260 * config/nds32/nds32.h: Likewise.
6261 * config/nds32/nds32.md: Likewise.
6262 * config/nds32/nds32-cost.c: Likewise.
6263 * config/nds32/nds32-isr.c: Likewise.
6264 * config/nds32/nds32-md-auxiliary.c: Likewise.
6265 * config/nds32/nds32-multiple.md: Likewise.
6266 * config/nds32/nds32-predicates.c: Likewise.
6268 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
6269 Jakub Jelinek <jakub@redhat.com>
6271 Add support for -std=c++2a.
6272 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
6274 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
6276 2017-09-15 Steve Ellcey <sellcey@cavium.com>
6279 * doc/extend.texi (Common Function Attributes): Add
6280 references to ARM, AArch64, and S/390 specific attributes.
6281 (Function Specific Option Pragmas): Add AArch64 and S/390
6282 to list of back ends that support the target pragma.
6284 2017-09-15 Nathan Sidwell <nathan@acm.org>
6286 * doc/standards.texi: Fix C++17 description. Update URLs for
6289 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
6291 * common.opt (Wcast-align=strict): New warning option.
6292 * doc/invoke.texi: Document -Wcast-align=strict.
6294 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
6296 * cgraph.h (cgraph_thunk_info): Add comments.
6297 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
6298 assert for VIRTUAL_* arguments stricter.
6300 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
6302 PR tree-optimization/71026
6303 * match.pd: Move RDIV patterns from fold-const.c
6304 * fold-const.c (distribute_real_division): Removed.
6305 (fold_binary_loc): Remove calls to distribute_real_divison.
6307 2017-09-15 Jakub Jelinek <jakub@redhat.com>
6309 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
6310 c++1z and gnu++1z as deprecated. Change other references to
6311 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
6312 Change -Wc++1z-compat to -Wc++17-compat.
6313 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
6314 * dwarf2out.c (highest_c_language): Handle C++17.
6315 (gen_compile_unit_die): Likewise.
6317 2017-09-15 Jakub Jelinek <jakub@redhat.com>
6319 PR rtl-optimization/82192
6320 * combine.c (make_extraction): Don't look through non-paradoxical
6321 SUBREGs or TRUNCATE if pos + len is or might be bigger than
6324 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6325 Alan Hayward <alan.hayward@arm.com>
6326 David Sherwood <david.sherwood@arm.com>
6328 * target.def (function_arg_offset): New hook.
6329 * targhooks.h (default_function_arg_offset): Declare.
6330 * targhooks.c (default_function_arg_offset): New function.
6331 * function.c (locate_and_pad_parm): Use
6332 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
6333 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
6334 (TARGET_FUNCTION_ARG_OFFSET): ...this.
6335 * doc/tm.texi: Regenerate.
6336 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
6337 * config/spu/spu.c (spu_function_arg_offset): New function.
6338 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
6339 * system.h (FUNCTION_ARG_OFFSET): Poison.
6341 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6342 Alan Hayard <alan.hayward@arm.com>
6343 David Sherwood <david.sherwood@arm.com>
6345 * target.def (truly_noop_truncation): New hook.
6346 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
6347 than TRULY_NOOP_TRUNCATION.
6348 * hooks.h (hook_bool_uint_uint_true): Declare.
6349 * hooks.c (hook_bool_uint_uint_true): New function.
6350 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
6351 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
6352 * doc/tm.texi: Regenerate.
6353 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
6354 rather than TRULY_NOOP_TRUNCATION in comments.
6355 (simplify_comparison): Likewise.
6356 (record_truncated_value): Likewise.
6357 * expmed.c (extract_bit_field_1): Likewise.
6358 (extract_split_bit_field): Likewise.
6359 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
6360 instead of TRULY_NOOP_TRUNCATION.
6361 * function.c (assign_parm_setup_block): Likewise.
6362 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
6363 * rtlhooks.c: Include target.h.
6364 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
6365 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
6366 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
6367 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
6368 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
6369 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
6370 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
6371 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
6372 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
6373 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
6374 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
6375 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
6376 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
6377 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
6378 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
6379 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
6380 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
6381 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
6382 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
6383 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
6384 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
6385 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
6386 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
6387 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
6388 * config/mips/mips.c (mips_truly_noop_truncation): New function.
6389 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
6390 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
6391 rather than TRULY_NOOP_TRUNCATION in comments.
6392 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
6393 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
6394 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
6395 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
6396 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
6397 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
6398 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
6399 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
6400 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
6401 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
6402 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
6403 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
6404 rather than TRULY_NOOP_TRUNCATION in comments.
6405 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
6406 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
6407 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
6408 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
6409 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
6410 TRULY_NOOP_TRUNCATION condition.
6411 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
6412 (TRULY_NOOP_TRUNCATION): Delete.
6413 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
6414 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
6415 * config/spu/spu.c (spu_truly_noop_truncation): New function.
6416 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
6417 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
6418 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
6419 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
6420 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
6421 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
6422 rather than TRULY_NOOP_TRUNCATION in comments.
6423 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
6424 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
6425 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
6426 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
6427 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
6428 * system.h (TRULY_NOOP_TRUNCATION): Poison.
6430 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
6433 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
6434 (*cmp_ior): Likewise.
6435 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
6436 (*ior_scc_scc_cmp): Likewise.
6437 (*and_scc_scc): Likewise.
6438 (*and_scc_scc_cmp): Likewise.
6440 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6441 Alan Hayard <alan.hayward@arm.com>
6442 David Sherwood <david.sherwood@arm.com>
6444 * target.def (can_change_mode_class): New hook.
6445 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
6446 (hard_regno_nregs): Likewise.
6447 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
6448 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
6449 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
6450 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
6451 (LOAD_EXTEND_OP): Update accordingly.
6452 * doc/tm.texi: Regenerate.
6453 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
6454 CANNOT_CHANGE_MODE_CLASS.
6455 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
6456 (REG_CAN_CHANGE_MODE_P): ...this new macro.
6457 * combine.c (simplify_set): Update accordingly.
6458 * emit-rtl.c (validate_subreg): Likewise.
6459 * recog.c (general_operand): Likewise.
6460 * regcprop.c (mode_change_ok): Likewise.
6461 * reload1.c (choose_reload_regs): Likewise.
6462 (inherit_piecemeal_p): Likewise.
6463 * rtlanal.c (simplify_subreg_regno): Likewise.
6464 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
6465 instead of CANNOT_CHANGE_MODE_CLASS.
6466 (reload_cse_simplify_operands): Likewise.
6467 * reload.c (push_reload): Use targetm.can_change_mode_class
6468 instead of CANNOT_CHANGE_MODE_CLASS.
6469 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
6470 REG_CANNOT_CHANGE_MODE_P.
6471 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6472 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
6473 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6474 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6475 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6476 (arm_can_change_mode_class): New function.
6477 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
6478 than CANNOT_CHANGE_MODE_CLASS in comments.
6479 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6480 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
6481 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
6482 (ix86_can_change_mode_class): ...this new function, inverting the
6483 sense of the return value.
6484 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6485 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6486 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6487 (ia64_can_change_mode_class): New function.
6488 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6489 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
6490 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
6491 (m32c_can_change_mode_class): ...this new function, inverting the
6492 sense of the return value.
6493 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6494 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6495 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
6496 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
6497 (mips_can_change_mode_class): ...this new function, inverting the
6498 sense of the return value.
6499 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6500 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6501 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6502 (msp430_can_change_mode_class): New function.
6503 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6504 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
6505 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6506 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6507 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6508 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
6509 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6510 (pa_cannot_change_mode_class): Replace with...
6511 (pa_can_change_mode_class): ...this new function, inverting the
6512 sense of the return value.
6513 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
6514 than CANNOT_CHANGE_MODE_CLASS in comments.
6515 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6516 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
6517 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6518 (pdp11_cannot_change_mode_class): Replace with...
6519 (pdp11_can_change_mode_class): ...this new function, inverting the
6520 sense of the return value.
6521 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6522 * config/powerpcspe/powerpcspe-protos.h
6523 (rs6000_cannot_change_mode_class_ptr): Delete.
6524 * config/powerpcspe/powerpcspe.c
6525 (rs6000_cannot_change_mode_class_ptr): Delete.
6526 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6527 (rs6000_option_override_internal): Assign to
6528 targetm.can_change_mode_class instead of
6529 rs6000_cannot_change_mode_class_ptr.
6530 (rs6000_cannot_change_mode_class): Replace with...
6531 (rs6000_can_change_mode_class): ...this new function, inverting the
6532 sense of the return value.
6533 (rs6000_debug_cannot_change_mode_class): Replace with...
6534 (rs6000_debug_can_change_mode_class): ...this new function.
6535 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6536 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
6537 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6538 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6539 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
6541 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
6542 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6543 (rs6000_option_override_internal): Assign to
6544 targetm.can_change_mode_class instead of
6545 rs6000_cannot_change_mode_class_ptr.
6546 (rs6000_cannot_change_mode_class): Replace with...
6547 (rs6000_can_change_mode_class): ...this new function, inverting the
6548 sense of the return value.
6549 (rs6000_debug_cannot_change_mode_class): Replace with...
6550 (rs6000_debug_can_change_mode_class): ...this new function.
6551 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6552 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
6553 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
6554 (s390_can_change_mode_class): ...this new function, inverting the
6555 sense of the return value.
6556 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6557 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6558 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
6559 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6560 (sh_cannot_change_mode_class): Replace with...
6561 (sh_can_change_mode_class): ...this new function, inverting the
6562 sense of the return value.
6563 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6564 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6565 (sparc_can_change_mode_class): New function.
6566 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6567 * config/spu/spu.c (spu_can_change_mode_class): New function.
6568 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6569 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6570 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6571 (visium_can_change_mode_class): New function.
6572 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
6574 2017-09-15 Richard Biener <rguenther@suse.de>
6576 PR tree-optimization/82217
6577 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
6578 but not undefined case.
6580 2017-09-15 Jakub Jelinek <jakub@redhat.com>
6583 * postreload.c (reload_cse_simplify_operands): Skip
6584 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
6586 2017-09-15 Richard Biener <rguenther@suse.de>
6588 PR tree-optimization/68823
6589 * graphite-scop-detection.c (build_alias_set): If we have a
6590 possible dependence check whether we can handle them by just
6591 looking at the DRs DR_ACCESS_FNs.
6592 (build_scops): If build_alias_set fails, fail the SCOP.
6594 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
6596 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
6597 to support float128 built-in functions that require the ISA 3.0
6599 (BU_FLOAT128_3_HW): Likewise.
6600 (SQRTF128): Add support for the IEEE 128-bit square root and fma
6602 (FMAF128): Likewise.
6604 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
6605 support for built-in functions that need the ISA 3.0 IEEE 128-bit
6606 floating point instructions.
6607 (rs6000_invalid_builtin): Likewise.
6608 (rs6000_builtin_mask_names): Likewise.
6609 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
6610 (RS6000_BTM_FLOAT128_HW): Likewise.
6611 (RS6000_BTM_COMMON): Likewise.
6612 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
6614 * doc/extend.texi (RS/6000 built-in functions): Document the
6615 IEEE 128-bit floating point square root and fused multiply-add
6618 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
6620 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
6621 reg (r2) isn't in the set of registers defined in the prologue.
6623 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6624 Alan Hayward <alan.hayward@arm.com>
6625 David Sherwood <david.sherwood@arm.com>
6627 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
6628 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
6629 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
6630 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
6631 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
6633 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
6634 max_vectorization_factor.
6635 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
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 (vect_min_worthwhile_factor): Delete.
6642 (vect_worthwhile_without_simd_p): Declare.
6643 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
6644 (vectorizable_reduction): Use it.
6645 * tree-vect-stmts.c (vectorizable_shift): Likewise.
6646 (vectorizable_operation): Likewise.
6648 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6649 Alan Hayward <alan.hayward@arm.com>
6650 David Sherwood <david.sherwood@arm.com>
6652 * tree-vectorizer.h (vect_get_num_copies): New function.
6653 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
6654 * tree-vect-loop.c (vectorizable_reduction): Likewise.
6655 (vectorizable_induction): Likewise.
6656 (vectorizable_live_operation): Likewise.
6657 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
6658 (vectorizable_bswap): Likewise.
6659 (vectorizable_call): Likewise.
6660 (vectorizable_conversion): Likewise.
6661 (vectorizable_assignment): Likewise.
6662 (vectorizable_shift): Likewise.
6663 (vectorizable_operation): Likewise.
6664 (vectorizable_store): Likewise.
6665 (vectorizable_load): Likewise.
6666 (vectorizable_condition): Likewise.
6667 (vectorizable_comparison): Likewise.
6668 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
6670 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6671 Alan Hayward <alan.hayward@arm.com>
6672 David Sherwood <david.sherwood@arm.com>
6674 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
6675 of vect_init_vector.
6677 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6678 Alan Hayward <alan.hayward@arm.com>
6679 David Sherwood <david.sherwood@arm.com>
6681 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
6682 an inline wrapper that provides a location.
6683 (gimple_build_vector): Likewise.
6684 * gimple-fold.c (gimple_build_vector_from_val): New function.
6685 (gimple_build_vector): Likewise.
6686 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
6687 functions to build the initial value. Always return a gimple value.
6688 (get_initial_defs_for_reduction): Likewise. Only compute
6690 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
6691 vect_init_vector on the results from get_initial_def(s)_for_reduction.
6692 (vectorizable_induction): Use gimple_build_vector rather than
6695 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6696 Alan Hayward <alan.hayward@arm.com>
6697 David Sherwood <david.sherwood@arm.com>
6699 * target.h (vec_perm_indices): New typedef.
6700 (auto_vec_perm_indices): Likewise.
6701 * optabs-query.h: Include target.h
6702 (can_vec_perm_p): Take a vec_perm_indices *.
6703 * optabs-query.c (can_vec_perm_p): Likewise.
6704 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
6705 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
6706 * tree-vect-generic.c (lower_vec_perm): Likewise.
6707 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
6708 (vect_grouped_load_supported): Likewise.
6709 (vect_shift_permute_load_chain): Likewise.
6710 (vect_permute_store_chain): Use auto_vec_perm_indices.
6711 (vect_permute_load_chain): Likewise.
6712 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
6713 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
6714 Update uses of can_vec_perm_p.
6715 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
6716 mode with a number of elements. Take a vec_perm_indices *.
6717 (vect_create_epilog_for_reduction): Update accordingly.
6718 Use auto_vec_perm_indices.
6719 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
6720 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
6721 (vect_transform_slp_perm_load): Likewise.
6722 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
6723 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
6724 (vect_gen_perm_mask_checked): Likewise.
6725 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
6726 (vect_gen_perm_mask_checked): Likewise.
6727 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
6728 (vectorizable_store): Likewise.
6729 (vectorizable_load): Likewise.
6730 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
6731 (vectorizable_bswap): Likewise.
6733 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6734 Alan Hayward <alan.hayward@arm.com>
6735 David Sherwood <david.sherwood@arm.com>
6737 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
6738 * tree.c (build_vector): Likewise.
6739 (build_vector_from_ctor): Update accordingly.
6740 (build_vector_from_val): Likewise.
6741 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
6742 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
6743 * tree-vect-generic.c (add_rshift): Likewise.
6744 (expand_vector_divmod): Likewise.
6745 (optimize_vector_constructor): Likewise.
6746 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
6747 (vect_transform_slp_perm_load): Likewise.
6748 (vect_schedule_slp_instance): Likewise.
6749 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
6750 (vectorizable_call): Likewise.
6751 (vect_gen_perm_mask_any): Likewise. Add elements in order.
6752 * expmed.c (make_tree): Likewise.
6753 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
6754 a vector passed to build_vector.
6755 (fold_convert_const): Likewise.
6756 (exact_inverse): Likewise.
6757 (fold_ternary_loc): Likewise.
6758 (fold_relational_const): Likewise.
6759 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
6760 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
6761 (const_unop): Likewise. Store the reduction accumulator in a
6762 variable rather than an array.
6763 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
6764 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
6765 the new vector, rather than constructing it after the input arrays.
6766 (native_interpret_vector): Use auto_vec<tree> when building
6767 a vector passed to build_vector. Add elements in order.
6768 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
6769 auto_vec<tree> when building a vector passed to build_vector.
6770 (vect_create_epilog_for_reduction): Likewise.
6771 (vectorizable_induction): Likewise.
6772 (get_initial_def_for_reduction): Likewise. Fix indentation of
6774 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
6776 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
6777 passed to build_vector.
6779 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6780 Alan Hayward <alan.hayward@arm.com>
6781 David Sherwood <david.sherwood@arm.com>
6783 * tree-core.h (tree_base::u): Add an "nelts" field.
6784 (tree_vector): Use VECTOR_CST_NELTS as the length.
6785 * tree.c (tree_size): Likewise.
6786 (make_vector): Initialize VECTOR_CST_NELTS.
6787 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
6788 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
6789 TYPE_VECTOR_SUBPARTS.
6790 * expr.c (const_vector_mask_from_tree): Consistently use "units"
6791 as the number of units, setting it from VECTOR_CST_NELTS.
6792 (const_vector_from_tree): Likewise.
6793 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
6794 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
6795 (fold_negate_expr_1): Likewise.
6796 (fold_convert_const): Likewise.
6797 (const_binop): Likewise. Differentiate the number of output and
6799 (const_unop): Likewise.
6800 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
6801 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
6802 in cases that did the opposite.
6804 2017-09-14 Richard Biener <rguenther@suse.de>
6806 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
6809 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
6811 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
6813 2017-09-14 Jakub Jelinek <jakub@redhat.com>
6816 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
6817 if and where to split a bb, except for splitting before debug insn
6818 sequences followed by non-label real insn. Delete debug insns
6819 in between basic blocks.
6821 * combine.c (make_compound_operation_int): Formatting fixes.
6823 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
6824 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6825 * config/netbsd.h (LINK_EH_SPEC): Likewise.
6826 * config/sol2.h (LINK_EH_SPEC): Likewise.
6827 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6828 * config/s390/linux.h (LINK_SPEC): Likewise.
6829 * config/freebsd.h (LINK_EH_SPEC): Likewise.
6830 * config/openbsd.h (LINK_EH_SPEC): Likewise.
6831 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6832 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
6833 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
6834 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6835 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
6836 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
6838 2017-09-13 Jakub Jelinek <jakub@redhat.com>
6840 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
6842 (ENDFILE_LINUX_SPEC): Likewise.
6843 (LINK_EH_SPEC): Likewise.
6844 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
6845 (LINK_OS_LINUX_SPEC32): Likewise.
6846 (LINK_OS_LINUX_SPEC64): Likewise.
6847 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
6848 (LINK_OS_LINUX_SPEC): Likewise.
6850 2017-09-13 Martin Liska <mliska@suse.cz>
6853 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
6854 CASE_HIGH is NULL_TREE.
6856 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
6857 Alan Hayward <alan.hayward@arm.com>
6858 David Sherwood <david.sherwood@arm.com>
6860 * target.def (secondary_memory_needed): New hook.
6861 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
6862 instead of SECONDARY_MEMORY_NEEDED.
6863 (secondary_memory_needed_mode): Likewise.
6864 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
6865 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
6866 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
6867 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
6868 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
6869 * doc/tm.texi: Regenerate.
6870 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
6871 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
6872 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6873 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
6874 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
6875 * config/i386/i386.c (inline_secondary_memory_needed): Put the
6876 mode argument first and change the reg_class arguments to reg_class_t.
6877 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
6878 Make static. Update the call to inline_secondary_memory_needed.
6879 (ix86_register_move_cost): Update the call to
6880 inline_secondary_memory_needed.
6881 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6882 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
6884 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
6885 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
6887 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
6888 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
6889 * config/mips/mips.c (mips_secondary_memory_needed): Make static
6890 and match hook interface. Add comment from mips.h.
6891 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6892 * config/mmix/mmix.md (truncdfsf2): Refer to
6893 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
6895 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
6896 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
6897 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6898 (pa_secondary_memory_needed): New function.
6899 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
6900 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
6901 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6902 (pdp11_secondary_memory_needed): Make static and match hook interface.
6903 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
6904 * config/powerpcspe/powerpcspe-protos.h
6905 (rs6000_secondary_memory_needed_ptr): Delete.
6906 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
6908 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6909 (rs6000_option_override_internal): Assign to
6910 targetm.secondary_memory_needed rather than
6911 rs6000_secondary_memory_needed_ptr.
6912 (rs6000_secondary_memory_needed): Match hook interface.
6913 (rs6000_debug_secondary_memory_needed): Likewise.
6914 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
6915 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
6916 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
6917 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6918 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
6919 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
6921 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
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/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
6929 * config/s390/s390.c (s390_secondary_memory_needed): New function.
6930 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6931 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
6932 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6933 (sparc_secondary_memory_needed): New function.
6934 * lra-constraints.c (check_and_process_move): Refer to
6935 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
6937 (curr_insn_transform): Likewise.
6938 (process_alt_operands): Use targetm.secondary_memory_needed
6939 instead of TARGET_SECONDARY_MEMORY_NEEDED.
6940 (check_secondary_memory_needed_p): Likewise.
6941 (choose_split_class): Likewise.
6942 * reload.c: Unconditionally include code that was previously
6943 conditional on SECONDARY_MEMORY_NEEDED.
6944 (push_secondary_reload): Use targetm.secondary_memory_needed
6945 instead of TARGET_SECONDARY_MEMORY_NEEDED.
6946 (push_reload): Likewise.
6947 * reload1.c: Unconditionally include code that was previously
6948 conditional on SECONDARY_MEMORY_NEEDED.
6949 (choose_reload_regs): Use targetm.secondary_memory_needed
6950 instead of TARGET_SECONDARY_MEMORY_NEEDED.
6951 (gen_reload): Likewise.
6952 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
6954 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
6955 Alan Hayward <alan.hayward@arm.com>
6956 David Sherwood <david.sherwood@arm.com>
6958 * target.def (secondary_memory_needed_mode): New hook:
6959 * targhooks.c (default_secondary_memory_needed_mode): Declare.
6960 * targhooks.h (default_secondary_memory_needed_mode): New function.
6961 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
6962 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
6963 * doc/tm.texi: Regenerate.
6964 * lra-constraints.c (check_and_process_move): Use
6965 targetm.secondary_memory_needed_mode instead of
6966 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
6967 (curr_insn_transform): Likewise.
6968 * reload.c (get_secondary_mem): Likewise.
6969 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
6970 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
6972 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
6973 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
6974 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
6975 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
6976 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
6978 * config/powerpcspe/powerpcspe-protos.h
6979 (rs6000_secondary_memory_needed_mode): Delete.
6980 * config/powerpcspe/powerpcspe.c
6981 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
6982 (rs6000_secondary_memory_needed_mode): Make static.
6983 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
6984 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
6986 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
6988 (rs6000_secondary_memory_needed_mode): Make static.
6989 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
6990 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
6991 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
6992 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
6993 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
6995 (sparc_secondary_memory_needed_mode): New function.
6996 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
6998 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
7000 * config/aarch64/constraints.md (Umq): New constraint.
7001 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
7003 (mov<mode>): Update condition.
7005 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7007 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
7008 when bitposition is the same.
7010 2017-09-13 Richard Biener <rguenther@suse.de>
7012 * dwarf2out.c (output_die_symbol): Remove.
7013 (output_die): Do not output a DIEs symbol.
7015 2017-09-13 Richard Biener <rguenther@suse.de>
7018 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
7019 default-def to avoid breaking iterator update with the weird
7020 interaction with cgraph_update_edges_for_call_stmt_node.
7022 2017-09-13 Richard Biener <rguenther@suse.de>
7024 * tree-cfg.c (verify_gimple_assign_binary): Add verification
7025 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
7026 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
7027 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
7029 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
7031 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
7032 Disable pc relative literal load irrespective of
7033 TARGET_FIX_ERR_A53_84341 for default.
7035 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
7037 * config/sparc/sparc.c (output_return): Output the source location of
7038 the insn in the delay slot, if any.
7039 (output_sibcall): Likewise.
7041 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
7044 * common.opt (-static-pie): New alias.
7045 (shared): Negate static-pie.
7046 (-no-pie): Update help text.
7048 (static-pie): New option.
7049 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
7050 -static-pie support.
7051 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
7052 (LINK_EH_SPEC): Likewise.
7053 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7054 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7055 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7056 * gcc.c (LINK_COMMAND_SPEC): Likewise.
7057 (init_gcc_specs): Likewise.
7058 (init_spec): Likewise.
7059 (display_help): Update help message for -pie.
7060 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
7063 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
7065 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
7066 (movdi_aarch64): Likewise.
7067 (movti_aarch64): Likewise.
7069 2017-09-12 Simon Wright <simon@pushface.org>
7072 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
7073 calculation of the minor version, always output as 0.
7075 2017-09-12 Jakub Jelinek <jakub@redhat.com>
7078 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
7079 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
7080 on it early, rather than manual conversion late. For
7081 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
7082 instead of performing manual conversion.
7084 2017-09-12 Carl Love <cel@us.ibm.com>
7086 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
7087 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
7089 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7090 VMULOSW): Add definitions.
7091 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7092 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7093 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7094 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7095 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7097 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7099 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
7101 (movti_aarch64): Likewise.
7102 (movdf_aarch64): Likewise.
7103 (movtf_aarch64): Likewise.
7104 (load_pairdi): Likewise.
7105 (store_pairdi): Likewise.
7106 (load_pairdf): Likewise.
7107 (store_pairdf): Likewise.
7108 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
7109 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
7110 (ldr_got_small_<mode>): Likewise.
7111 (ldr_got_small_28k_<mode>): Likewise.
7112 (ldr_got_tiny): Likewise.
7113 * config/aarch64/iterators.md (ldst_sz): New.
7114 (ldpstp_sz): Likewise.
7115 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
7117 (thunderx_load): Split load_8 to load_16.
7118 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
7120 (thunderx2t99_storepair_basic): Split store_8 to store_16.
7121 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
7122 (xgene1_store_pair): Split store_8 to store_16.
7123 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
7124 (falkor_st_0_st_sd): Split store_8 to store_16.
7126 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7128 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
7129 and store1/2/3/4 to store_4/8/12/16.
7130 * config/aarch64/aarch64.md: Update for rename.
7131 * config/arm/arm.md: Likewise.: Likewise.
7132 * config/arm/arm.c: Likewise.
7133 * config/arm/thumb1.md: Likewise.
7134 * config/arm/thumb2.md: Likewise.
7135 * config/arm/vfp.md: Likewise.
7136 * config/arm/arm-generic.md: Likewise.
7137 * config/arm/arm1020e.md: Likewise.
7138 * config/arm/arm1026ejs.md: Likewise.
7139 * config/arm/arm1136jfs.md: Likewise.
7140 * config/arm/arm926ejs.md: Likewise.
7141 * config/arm/cortex-a15.md: Likewise.
7142 * config/arm/cortex-a17.md: Likewise.
7143 * config/arm/cortex-a5.md: Likewise.
7144 * config/arm/cortex-a53.md: Likewise.
7145 * config/arm/cortex-a57.md: Likewise.
7146 * config/arm/cortex-a7.md: Likewise.
7147 * config/arm/cortex-a8.md: Likewise.
7148 * config/arm/cortex-a9.md: Likewise.
7149 * config/arm/cortex-m4.md: Likewise.
7150 * config/arm/cortex-m7.md: Likewise.
7151 * config/arm/cortex-r4.md: Likewise.
7152 * config/arm/exynos-m1.md: Likewise.
7153 * config/arm/fa526.md: Likewise.
7154 * config/arm/fa606te.md: Likewise.
7155 * config/arm/fa626te.md: Likewise.
7156 * config/arm/fa726te.md: Likewise.
7157 * config/arm/fmp626.md: Likewise.
7158 * config/arm/iwmmxt.md: Likewise.
7159 * config/arm/ldmstm.md: Likewise.
7160 * config/arm/marvell-pj4.md: Likewise.
7161 * config/arm/xgene1.md: Likewise.
7162 * config/aarch64/thunderx.md: Likewise.
7163 * config/aarch64/thunderx2t99.md: Likewise.
7164 * config/aarch64/falkor.md: Likewise.
7166 2017-09-12 Martin Liska <mliska@suse.cz>
7168 * attribs.c (private_lookup_attribute): New function.
7169 * attribs.h (private_lookup_attribute): Declared here.
7170 (lookup_attribute): Called from this place.
7172 2017-09-12 Richard Biener <rguenther@suse.de>
7174 PR tree-optimization/82157
7175 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
7176 stmts with side-effects.
7178 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7179 Alan Hayward <alan.hayward@arm.com>
7180 David Sherwood <david.sherwood@arm.com>
7182 * target.def (hard_regno_nregs): New hook.
7183 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
7184 * targhooks.h (default_hard_regno_nregs): Declare.
7185 * targhooks.c (default_hard_regno_nregs): New function.
7186 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
7187 (TARGET_HARD_REGNO_NREGS): ...this hook.
7188 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
7189 (CLASS_MAX_NREGS): Likewise.
7190 * doc/tm.texi: Regenerate.
7191 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
7192 instead of HARD_REGNO_NREGS.
7193 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
7194 HARD_REGNO_NREGS in the comment.
7195 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
7196 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
7197 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
7198 Return an unsigned int.
7199 (TARGET_HARD_REGNO_NREGS): Redefine.
7200 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
7201 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
7202 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
7203 (arc_hard_regno_nregs): New function.
7204 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
7205 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
7206 (arm_hard_regno_nregs): New function.
7207 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
7208 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
7209 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
7210 (TARGET_HARD_REGNO_NREGS): Redefine.
7211 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
7212 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
7213 (HARD_REGNO_NREGS): Delete.
7214 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
7215 (cr16_hard_regno_nregs): New function.
7216 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
7217 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
7218 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
7219 (cris_hard_regno_nregs): New function.
7220 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
7221 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
7222 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
7223 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
7224 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
7225 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
7226 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
7227 (frv_hard_regno_nregs): Make static. Take and return an
7229 (frv_class_max_nregs): Remove outdated copy of documentation.
7230 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
7231 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
7232 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
7233 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
7234 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
7235 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
7236 (TARGET_HARD_REGNO_NREGS): Redefine.
7237 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
7238 (CLASS_MAX_NREGS): Update comment.
7239 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
7240 (ia64_hard_regno_nregs): New function.
7241 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
7242 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
7243 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
7244 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
7245 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
7247 (m32c_hard_regno_nregs): Likewise. Make static.
7248 (TARGET_HARD_REGNO_NREGS): Redefine.
7249 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
7250 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
7251 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
7252 (m68k_hard_regno_nregs): New function.
7253 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
7254 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
7255 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
7256 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
7257 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
7258 Take and return an unsigned int.
7259 (TARGET_HARD_REGNO_NREGS): Redefine.
7260 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
7261 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
7262 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
7263 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
7264 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
7265 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
7266 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
7267 (msp430_hard_regno_nregs): Make static. Take and return an
7269 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
7270 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
7271 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
7272 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
7273 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
7274 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
7275 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
7276 (TARGET_HARD_REGNO_NREGS): Redefine.
7277 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
7278 (PA_HARD_REGNO_NREGS): ...this.
7279 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
7280 (PA_HARD_REGNO_NREGS): ...this.
7281 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
7282 (pa_hard_regno_nregs): New function.
7283 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
7284 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
7285 (pdp11_hard_regno_nregs): New function.
7286 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
7287 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
7288 (rs6000_hard_regno_nregs_hook): New function.
7289 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
7290 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
7291 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
7292 Take and return an unsigned int. Move earlier in file.
7293 (TARGET_HARD_REGNO_NREGS): Redefine.
7294 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
7295 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
7296 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
7297 (rl78_hard_regno_nregs): Make static. Take and return an
7299 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
7300 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
7301 (rs6000_hard_regno_nregs_hook): New function.
7302 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
7303 * config/rx/rx.c (rx_hard_regno_nregs): New function.
7304 (TARGET_HARD_REGNO_NREGS): Redefine.
7305 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
7306 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
7307 instead of HARD_REGNO_NREGS.
7308 (s390_hard_regno_nregs): New function.
7309 (s390_hard_regno_mode_ok): Add comment from s390.h.
7310 (TARGET_HARD_REGNO_NREGS): Redefine.
7311 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
7312 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
7313 (sh_hard_regno_nregs): New function.
7314 (sh_pass_in_reg_p): Use it.
7315 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
7316 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
7317 (sparc_hard_regno_nregs): New function.
7318 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
7319 * config/spu/spu.c (spu_hard_regno_nregs): New function.
7320 (spu_function_arg_advance): Use it, supplying a valid register number.
7321 (TARGET_HARD_REGNO_NREGS): Redefine.
7322 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
7323 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
7324 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
7325 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
7326 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
7327 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
7328 (CLASS_MAX_NREGS): Remove copy of old documentation.
7329 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
7330 (visium_hard_regno_nregs): New function.
7331 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
7332 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
7333 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
7334 xtensa_hard_regno_nregs): New function.
7335 * system.h (HARD_REGNO_NREGS): Poison.
7337 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7339 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
7340 hard_regno_nregs instead of HARD_REGNO_NREGS.
7341 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
7342 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
7343 (c6x_expand_epilogue): Likewise.
7344 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
7345 (frv_read_iacc_argument): Likewise.
7346 * config/sh/sh.c: Include regs.h.
7347 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
7348 (regs_used): Likewise.
7349 (output_stack_adjust): Likewise.
7350 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
7351 * expmed.c: Include regs.h.
7352 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
7353 * ree.c: Include regs.h.
7354 (combine_reaching_defs): Use hard_regno_nregs instead of
7356 (add_removable_extension): Likewise.
7358 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7360 * regs.h (hard_regno_nregs): Turn into a function.
7361 (end_hard_regno): Update accordingly.
7362 * caller-save.c (setup_save_areas): Likewise.
7363 (save_call_clobbered_regs): Likewise.
7364 (replace_reg_with_saved_mem): Likewise.
7365 (insert_restore): Likewise.
7366 (insert_save): Likewise.
7367 * combine.c (can_change_dest_mode): Likewise.
7368 (move_deaths): Likewise.
7369 (distribute_notes): Likewise.
7370 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
7371 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
7372 (rs6000_split_multireg_move): Likewise.
7373 (rs6000_register_move_cost): Likewise.
7374 (rs6000_memory_move_cost): Likewise.
7375 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
7376 (rs6000_split_multireg_move): Likewise.
7377 (rs6000_register_move_cost): Likewise.
7378 (rs6000_memory_move_cost): Likewise.
7379 * cselib.c (cselib_reset_table): Likewise.
7380 (cselib_lookup_1): Likewise.
7381 * emit-rtl.c (set_mode_and_regno): Likewise.
7382 * function.c (aggregate_value_p): Likewise.
7383 * ira-color.c (setup_profitable_hard_regs): Likewise.
7384 (check_hard_reg_p): Likewise.
7385 (calculate_saved_nregs): Likewise.
7386 (assign_hard_reg): Likewise.
7387 (improve_allocation): Likewise.
7388 (calculate_spill_cost): Likewise.
7389 * ira-emit.c (modify_move_list): Likewise.
7390 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
7391 (ira_hard_reg_in_set_p): Likewise.
7392 * ira.c (setup_reg_mode_hard_regset): Likewise.
7393 (clarify_prohibited_class_mode_regs): Likewise.
7394 (check_allocation): Likewise.
7395 * lra-assigns.c (find_hard_regno_for_1): Likewise.
7396 (lra_setup_reg_renumber): Likewise.
7397 (setup_try_hard_regno_pseudos): Likewise.
7398 (spill_for): Likewise.
7399 (assign_hard_regno): Likewise.
7400 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
7401 * lra-constraints.c (in_class_p): Likewise.
7402 (lra_constraint_offset): Likewise.
7403 (simplify_operand_subreg): Likewise.
7404 (lra_constraints): Likewise.
7405 (split_reg): Likewise.
7406 (split_if_necessary): Likewise.
7407 (invariant_p): Likewise.
7408 (inherit_in_ebb): Likewise.
7409 * lra-lives.c (process_bb_lives): Likewise.
7410 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
7411 (get_hard_regs): Likewise.
7412 (do_remat): Likewise.
7413 * lra-spills.c (assign_spill_hard_regs): Likewise.
7414 * mode-switching.c (create_pre_exit): Likewise.
7415 * postreload.c (reload_combine_recognize_pattern): Likewise.
7416 * recog.c (peep2_find_free_register): Likewise.
7417 * regcprop.c (kill_value_regno): Likewise.
7418 (set_value_regno): Likewise.
7419 (copy_value): Likewise.
7420 (maybe_mode_change): Likewise.
7421 (find_oldest_value_reg): Likewise.
7422 (copyprop_hardreg_forward_1): Likewise.
7423 * regrename.c (check_new_reg_p): Likewise.
7424 (regrename_do_replace): Likewise.
7425 * reload.c (push_reload): Likewise.
7426 (combine_reloads): Likewise.
7427 (find_dummy_reload): Likewise.
7428 (operands_match_p): Likewise.
7429 (find_reloads): Likewise.
7430 (find_equiv_reg): Likewise.
7431 (reload_adjust_reg_for_mode): Likewise.
7432 * reload1.c (count_pseudo): Likewise.
7433 (count_spilled_pseudo): Likewise.
7434 (find_reg): Likewise.
7435 (clear_reload_reg_in_use): Likewise.
7436 (free_for_value_p): Likewise.
7437 (allocate_reload_reg): Likewise.
7438 (choose_reload_regs): Likewise.
7439 (reload_adjust_reg_for_temp): Likewise.
7440 (emit_reload_insns): Likewise.
7441 (delete_output_reload): Likewise.
7442 * rtlanal.c (subreg_get_info): Likewise.
7443 * sched-deps.c (sched_analyze_reg): Likewise.
7444 * sel-sched.c (init_regs_for_mode): Likewise.
7445 (mark_unavailable_hard_regs): Likewise.
7446 (choose_best_reg_1): Likewise.
7447 (verify_target_availability): Likewise.
7448 * valtrack.c (dead_debug_insert_temp): Likewise.
7449 * var-tracking.c (track_loc_p): Likewise.
7450 (emit_note_insn_var_location): Likewise.
7451 * varasm.c (make_decl_rtl): Likewise.
7452 * reginfo.c (choose_hard_reg_mode): Likewise.
7453 (init_reg_modes_target): Refer directly to
7454 this_target_regs->x_hard_regno_nregs.
7456 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7458 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
7459 instead of hard_regno_nregs.
7461 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7463 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
7464 end_hard_regno instead of hard_regno_nregs.
7465 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
7466 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
7467 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
7468 * ira-color.c (improve_allocation): Likewise.
7469 * lra-assigns.c (find_hard_regno_for_1): Likewise.
7470 * lra-lives.c (mark_regno_live): Likewise.
7471 (mark_regno_dead): Likewise.
7472 * lra-remat.c (operand_to_remat): Likewise.
7473 * lra.c (collect_non_operand_hard_regs): Likewise.
7474 * postreload.c (reload_combine_note_store): Likewise.
7475 (move2add_valid_value_p): Likewise.
7476 * reload.c (regno_clobbered_p): Likewise.
7478 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7480 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
7482 * config/v850/v850.c (v850_reorg): Likewise.
7483 * reload.c (refers_to_regno_for_reload_p): Likewise.
7484 (find_equiv_reg): Likewise.
7485 * reload1.c (reload_reg_reaches_end_p): Likewise.
7487 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7489 * caller-save.c (add_used_regs): Use REG_NREGS instead of
7491 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
7492 * config/arm/arm.c (output_move_neon): Likewise.
7493 (arm_attr_length_move_neon): Likewise.
7494 (neon_split_vcombine): Likewise.
7495 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
7496 (c6x_mark_reg_written): Likewise.
7497 (c6x_dwarf_register_span): Likewise.
7498 * config/i386/i386.c (ix86_save_reg): Likewise.
7499 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
7500 (rws_access_reg): Likewise.
7501 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
7502 * mode-switching.c (create_pre_exit): Likewise.
7503 * ree.c (combine_reaching_defs): Likewise.
7504 (add_removable_extension): Likewise.
7505 * regcprop.c (find_oldest_value_reg): Likewise.
7506 (copyprop_hardreg_forward_1): Likewise.
7507 * reload.c (reload_inner_reg_of_subreg): Likewise.
7508 (push_reload): Likewise.
7509 (combine_reloads): Likewise.
7510 (find_dummy_reload): Likewise.
7511 (reload_adjust_reg_for_mode): Likewise.
7512 * reload1.c (find_reload_regs): Likewise.
7513 (forget_old_reloads_1): Likewise.
7514 (reload_reg_free_for_value_p): Likewise.
7515 (reload_adjust_reg_for_temp): Likewise.
7516 (emit_reload_insns): Likewise.
7517 (delete_output_reload): Likewise.
7518 * sel-sched.c (choose_best_reg_1): Likewise.
7519 (choose_best_pseudo_reg): Likewise.
7521 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7522 Alan Hayward <alan.hayward@arm.com>
7523 David Sherwood <david.sherwood@arm.com>
7525 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
7526 * target.def (slow_unaligned_access): New hook.
7527 * targhooks.h (default_slow_unaligned_access): Declare.
7528 * targhooks.c (default_slow_unaligned_access): New function.
7529 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
7530 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
7531 * doc/tm.texi: Regenerate.
7532 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
7533 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
7534 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
7536 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
7537 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
7539 (rs6000_slow_unaligned_access): New function.
7540 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
7541 (expand_block_compare): Likewise.
7542 (expand_strn_compare): Likewise.
7543 (rs6000_rtx_costs): Likewise.
7544 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
7545 (riscv_slow_unaligned_access): Likewise.
7546 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
7547 (riscv_slow_unaligned_access_p): ...this and make static.
7548 (riscv_option_override): Update accordingly.
7549 (riscv_slow_unaligned_access): New function.
7550 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
7551 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
7552 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
7553 (rs6000_slow_unaligned_access): New function.
7554 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
7555 (rs6000_rtx_costs): Likewise.
7556 * config/rs6000/rs6000-string.c (expand_block_compare)
7557 (expand_strn_compare): Use targetm.slow_unaligned_access instead
7558 of SLOW_UNALIGNED_ACCESS.
7559 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
7560 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
7561 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
7562 of SLOW_UNALIGNED_ACCESS.
7563 * expmed.c (simple_mem_bitfield_p): Likewise.
7564 * expr.c (alignment_for_piecewise_move): Likewise.
7565 (emit_group_load_1): Likewise.
7566 (emit_group_store): Likewise.
7567 (copy_blkmode_from_reg): Likewise.
7568 (emit_push_insn): Likewise.
7569 (expand_assignment): Likewise.
7570 (store_field): Likewise.
7571 (expand_expr_real_1): Likewise.
7572 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
7573 * lra-constraints.c (simplify_operand_subreg): Likewise.
7574 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
7575 * gimple-ssa-store-merging.c: Likewise in block comment at start
7577 * tree-ssa-strlen.c: Include target.h.
7578 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
7579 of SLOW_UNALIGNED_ACCESS.
7580 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
7582 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7584 PR rtl-optimization/82185
7585 * expmed.c (emit_store_flag_int): Only test tem if it has been
7588 2017-09-12 Richard Biener <rguenther@suse.de>
7591 * match.pd ((FTYPE) N CMP CST): Fix typo.
7593 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
7595 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
7597 (mips_near_type_p): Add 'short_call' attribute as a synonym
7599 * doc/extend.texi (short_call): Document new function attribute.
7601 2017-09-12 Jakub Jelinek <jakub@redhat.com>
7604 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
7605 assertion check that in the condition.
7606 (get_atomic_generic_size): Likewise. Before testing if parameter
7607 has pointer type, if it has array type, call for C++
7608 default_conversion to perform array-to-pointer conversion.
7610 2017-09-12 Richard Biener <rguenther@suse.de>
7612 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
7613 for operations we cannot scalarize.
7615 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
7617 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
7618 vectors heap vectors. Clean up comments.
7619 Make visited_bbs a reference.
7620 (profitable_jump_thread_path): Make GC
7621 vectors heap vectors. Clean up comments.
7623 (convert_and_register_jump_thread_path): Make GC vectors heap
7625 (check_subpath_and_update_thread_path): Same. Clean up comments.
7626 Make visited_bbs a reference.
7627 (handle_phi): Abstract common code to to
7628 register_jump_thread_path_if_profitable.
7629 Rename VAR_BB to DEF_BB.
7631 Make GC vectors heap vectors.
7632 Make visited_bbs a reference.
7633 (handle_assignment): Same.
7634 (register_jump_thread_path_if_profitable): New.
7635 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
7637 Make GC vectors heap vectors. Clean up comments.
7638 Make visited_bbs a reference.
7639 (find_jump_threads_backwards): Make visited_bbs live in the stack.
7640 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
7643 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
7646 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
7647 words of E_DImode object are reachable by xtensa_uimm8x4 access.
7649 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
7651 Revert r251800 and r251799.
7653 2017-09-11 Martin Jambor <mjambor@suse.cz>
7656 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
7657 arguments in advance.
7658 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
7659 use it to find predecessor edges.
7660 (naive_outof_ssa): Collect vector of predecessors.
7662 2017-09-08 Jason Merrill <jason@redhat.com>
7664 PR c++/70029 - ICE with ref-qualifier and -flto
7665 * langhooks.h (struct lang_hooks_for_types): Add
7666 copy_lang_qualifiers.
7667 * attribs.c (build_type_attribute_qual_variant): Use it.
7668 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
7670 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
7671 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
7673 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
7676 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
7677 (*mulsi3_sp64): New instruction.
7678 (mulsi3): New expander.
7680 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
7682 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
7684 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7686 * sancov.c: Include memmodel.h.
7688 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
7691 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
7694 2017-09-07 Carl Love <cel@us.ibm.com>
7696 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
7697 the sldi instruction.
7699 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
7701 * sancov.c: Include tm_p.h.
7703 2017-09-07 Jakub Jelinek <jakub@redhat.com>
7706 * output.h (switch_to_other_text_partition): New declaration.
7707 * varasm.c (switch_to_other_text_partition): New function.
7708 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
7709 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
7710 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
7711 to the other text partition before emitting LCL label and switch back
7712 after emitting the word after it.
7714 2017-09-07 Richard Biener <rguenther@suse.de>
7716 * passes.def (pass_split_crit_edges): Remove instance before PRE.
7717 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
7718 critical edges here, after loop init.
7719 (pass_data_pre): Remove PROP_no_crit_edges flags.
7720 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
7721 for valueization of call args to avoid leaking VN_TOP.
7722 (visit_use): Assert we do not visit default defs.
7723 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
7724 Use error_mark_node to more easily detect leaking VN_TOP.
7725 All default-defs are varying, not VN_TOP. Mark them visited.
7726 (run_scc_vn): Make code match comment.
7728 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
7730 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
7731 OPTION_MASK_FLOAT128_KEYWORD.
7732 (POWERPC_MASKS): Likewise.
7733 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
7734 support for the -mfloat128-type option, and make -mfloat128
7735 default on PowerPC Linux systems. Define or undefine
7736 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
7737 Define __float128 to be __ieee128 if IEEE 128-bit support is
7738 enabled, or undefine it.
7739 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
7740 Delete defining __FLOAT128_TYPE__.
7741 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
7742 -mfloat128-type option and make -mfloat128 default on PowerPC
7744 (TARGET_FLOAT128_TYPE): Likewise.
7745 (-mfloat128-type): Likewise.
7746 * config/rs6000/rs6000.c (rs6000_option_override_internal):
7747 Delete the -mfloat128-type option and make -mfloat128 default on
7748 PowerPC Linux systems. Always use __ieee128 to be the keyword for
7749 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
7750 128-bit floating point is enabled. Change tests from using
7751 -mfloat128-type to -mfloat128.
7752 (rs6000_mangle_type): Use the correct mangling for the __float128
7753 type even if normal long double is restricted to 64-bits.
7754 (floatn_mode): Enable the _Float128 type by default on VSX Linux
7756 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
7757 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
7758 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
7760 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
7761 documentation for -mfloat128.
7763 2017-09-06 Olivier Hainque <hainque@adacore.com>
7765 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
7767 2017-09-06 Wish Wu <wishwu007@gmail.com>
7768 Jakub Jelinek <jakub@redhat.com>
7770 * asan.c (initialize_sanitizer_builtins): Add
7771 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
7772 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
7773 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
7774 BT_FN_VOID_UINT64_PTR variables.
7775 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
7776 (BT_FN_VOID_UINT16_UINT16): Likewise.
7777 (BT_FN_VOID_UINT32_UINT32): Likewise.
7778 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
7779 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
7780 (BT_FN_VOID_UINT64_PTR): Likewise.
7781 * common.opt (flag_sanitize_coverage): New variable.
7782 (fsanitize-coverage=trace-pc): Remove.
7783 (fsanitize-coverage=): Add.
7784 * flag-types.h (enum sanitize_coverage_code): New enum.
7785 * fold-const.c (fold_range_test): Disable non-short-circuit
7786 optimization if flag_sanitize_coverage.
7787 (fold_truth_andor): Likewise.
7788 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
7789 * opts.c (COVERAGE_SANITIZER_OPT): Define.
7790 (coverage_sanitizer_opts): New array.
7791 (get_closest_sanitizer_option): Add OPTS argument, handle also
7792 OPT_fsanitize_coverage_.
7793 (parse_sanitizer_options): Adjusted to also handle
7794 OPT_fsanitize_coverage_.
7795 (common_handle_option): Add OPT_fsanitize_coverage_.
7796 * sancov.c (instrument_comparison, instrument_switch): New function.
7797 (sancov_pass): Add trace-cmp support.
7798 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
7799 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
7800 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
7801 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
7802 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
7803 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
7804 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
7805 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
7806 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
7807 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
7809 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
7811 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
7812 error. Only quit immediately if parsing is complete.
7813 (BEGIN): Initialize fatal_err and parse_done.
7814 (begin fpu, end fpu): Check number of arguments.
7815 (begin arch, end arch): Likewise.
7816 (begin cpu, end cpu): Likewise.
7817 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
7818 (optalias): Likewise.
7820 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
7822 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
7823 * config/arm/arm-isa.h: Delete. Move definitions to ...
7824 * arm-cpus.in: ... here. Use new feature and fgroup values.
7825 * config/arm/arm.c (arm_option_override): Use lower case for feature
7827 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
7828 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
7829 * config/arm/parsecpu.awk (END): Add new command 'isa'.
7831 (print_isa_bits_for): New function.
7832 (gen_isa): New function.
7833 (gen_comm_data): Use print_isa_bits_for.
7834 (define feature): New keyword.
7835 (define fgroup): New keyword.
7836 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
7837 (arm-isa.h): Add rule to generate file.
7838 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
7839 case for feature bit names.
7841 2017-09-06 Richard Biener <rguenther@suse.de>
7843 * tree-ssa-pre.c (NECESSARY): Remove.
7844 (create_expression_by_pieces): Do not touch pass-local flags.
7845 (insert_into_preds_of_block): Likewise.
7846 (do_pre_regular_insertion): Likewise.
7847 (eliminate_insert): Likewise.
7848 (eliminate_dom_walker::before_dom_children): Likewise.
7849 (fini_eliminate): Do not look at inserted_exprs.
7850 (mark_operand_necessary): Remove.
7851 (remove_dead_inserted_code): Replace with simple work-list
7852 algorithm based on inserted_exprs and SSA uses.
7853 (pass_pre::execute): Re-order fini_eliminate and
7854 remove_dead_inserted_code.
7856 2017-09-06 Olivier Hainque <hainque@adacore.com>
7858 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
7859 for VxWorks 7. Adjust surrounding comments.
7861 2017-09-06 Richard Biener <rguenther@suse.de>
7863 * gimple-ssa-strength-reduction.c
7864 (find_candidates_dom_walker::before_dom_children): Also allow
7867 2017-09-06 Richard Biener <rguenther@suse.de>
7869 PR tree-optimization/82108
7870 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
7871 for gap in the non-permutation SLP case.
7873 2017-09-06 Martin Jambor <mjambor@suse.cz>
7875 PR tree-optimization/82078
7876 * tree-sra.c (sort_and_splice_var_accesses): Move call to
7877 add_access_to_work_queue...
7878 (build_accesses_from_assign): ...here.
7879 (propagate_all_subaccesses): Make sure racc is the group
7880 representative, if there is one.
7882 2017-09-06 Jakub Jelinek <jakub@redhat.com>
7885 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
7888 2017-09-06 Richard Biener <rguenther@suse.de>
7890 * gimple-ssa-strength-reduction.c
7891 (find_candidates_dom_walker::before_doom_children): Use a
7892 type and not a mode check.
7894 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
7897 * config/arm/predicates.md (arm_general_adddi_operand): Create new
7899 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
7901 2017-09-05 Jeff Law <law@redhat.com>
7903 PR tree-optimization/64910
7904 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
7905 cases where we have 3 or more operands.
7907 2017-09-05 Jakub Jelinek <jakub@redhat.com>
7910 * omp-low.c (lower_omp_for): Recompute tree invariant if
7911 gimple_omp_for_initial/final is ADDR_EXPR.
7914 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
7915 into gimple val before gimplification fo the COND_EXPR.
7917 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
7919 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
7920 REGION_COPY argument.
7921 (thread_through_all_blocks): Remove unused argument to
7922 duplicate_thread_path.
7924 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7925 Alan Hayward <alan.hayward@arm.com>
7926 David Sherwood <david.sherwood@arm.com>
7928 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
7929 Take a scalar_mode rather than a machine_mode.
7930 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
7931 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
7932 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
7933 (aarch64_gen_adjusted_ldpstp): Likewise.
7934 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
7936 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7937 Alan Hayward <alan.hayward@arm.com>
7938 David Sherwood <david.sherwood@arm.com>
7940 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
7941 Take a scalar_int_mode instead of a machine_mode.
7942 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
7943 (aarch64_output_scalar_simd_mov_immediate): Likewise.
7944 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
7945 (aarch64_simd_attr_length_rglist): Delete.
7946 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
7947 a scalar_int_mode instead of a machine_mode.
7948 (aarch64_add_offset): Likewise.
7949 (aarch64_internal_mov_immediate): Likewise
7950 (aarch64_add_constant_internal): Likewise.
7951 (aarch64_add_constant): Likewise.
7952 (aarch64_movw_imm): Likewise.
7953 (aarch64_rtx_arith_op_extract_p): Likewise.
7954 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
7955 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
7956 Remove assert that the mode isn't a vector.
7957 (aarch64_output_scalar_simd_mov_immediate): Likewise.
7958 (aarch64_expand_mov_immediate): Update calls after above changes.
7959 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
7960 (aarch64_and_bitmask_imm): Check for scalar integer modes.
7961 (aarch64_move_imm): Likewise.
7962 (aarch64_can_const_movi_rtx_p): Likewise.
7963 (aarch64_strip_extend): Likewise.
7964 (aarch64_extr_rtx_p): Likewise.
7965 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
7966 a CONST_INT when the mode parameter is VOIDmode.
7967 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
7969 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7971 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
7972 * stor-layout.c (bitwise_mode_for_mode): Likewise.
7973 (bitwise_type_for_mode): Update accordingly.
7975 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7977 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
7978 * stor-layout.c (mode_for_size_tree): Likewise.
7979 (mode_for_array): Update accordingly.
7980 (layout_decl): Likewise.
7981 (compute_record_mode): Likewise. Only set the mode once.
7983 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7985 * target.def (get_mask_mode): Change return type to opt_mode.
7987 * doc/tm.texi: Regenerate.
7988 * targhooks.h (default_get_mask_mode): Return an opt_mode.
7989 * targhooks.c (default_get_mask_mode): Likewise.
7990 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
7991 * optabs-query.c (can_vec_mask_load_store_p): Update use of
7992 targetm.get_mask_mode.
7993 * tree.c (build_truth_vector_type): Likewise.
7995 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7997 * machmode.h (mode_for_vector): Return an opt_mode.
7998 * stor-layout.c (mode_for_vector): Likewise.
7999 (mode_for_int_vector): Update accordingly.
8000 (layout_type): Likewise.
8001 * config/i386/i386.c (emit_memmov): Likewise.
8002 (ix86_expand_set_or_movmem): Likewise.
8003 (ix86_expand_vector_init): Likewise.
8004 (ix86_get_mask_mode): Likewise.
8005 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
8007 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
8008 * expmed.c (extract_bit_field_1): Likewise.
8009 * expr.c (expand_expr_real_2): Likewise.
8010 * optabs-query.c (can_vec_perm_p): Likewise.
8011 (can_vec_mask_load_store_p): Likewise.
8012 * optabs.c (expand_vec_perm): Likewise.
8013 * targhooks.c (default_get_mask_mode): Likewise.
8014 * tree-vect-stmts.c (vectorizable_store): Likewise.
8015 (vectorizable_load): Likewise.
8016 (get_vectype_for_scalar_type_and_size): Likewise.
8018 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8020 * machmode.h (mode_for_int_vector): New function.
8021 * stor-layout.c (mode_for_int_vector): Likewise.
8022 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
8023 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
8024 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
8025 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
8026 (s390_expand_vcond): Likewise.
8028 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8030 * machmode.h (opt_machine_mode): New type.
8031 (opt_mode<T>): Allow construction from anything that can be
8033 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
8034 (mode_for_size): Return an opt_machine_mode.
8035 * stor-layout.c (mode_for_size): Likewise.
8036 (mode_for_size_tree): Update call accordingly.
8037 (bitwise_mode_for_mode): Likewise.
8038 (make_fract_type): Likewise.
8039 (make_accum_type): Likewise.
8040 * caller-save.c (replace_reg_with_saved_mem): Update call
8042 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8043 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8044 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8045 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8046 * expmed.c (extract_bit_field_1): Likewise.
8047 * reload.c (get_secondary_mem): Likewise.
8048 * varasm.c (assemble_integer): Likewise.
8049 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
8052 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8054 * machmode.h (decimal_float_mode_for_size): New function.
8055 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
8056 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
8057 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
8058 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
8059 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
8060 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
8062 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8064 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
8065 (get_builtin_sync_mode): Likewise.
8066 (expand_ifn_atomic_compare_exchange): Likewise.
8067 (expand_builtin_atomic_clear): Likewise.
8068 (expand_builtin_atomic_test_and_set): Likewise.
8069 (fold_builtin_atomic_always_lock_free): Likewise.
8070 * calls.c (compute_argument_addresses): Likewise.
8071 (emit_library_call_value_1): Likewise.
8072 (store_one_arg): Likewise.
8073 * combine.c (combine_instructions): Likewise.
8074 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
8075 * config/arm/arm.c (arm_function_value): Likewise.
8076 (aapcs_allocate_return_reg): Likewise.
8077 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
8078 * config/i386/i386.c (construct_container): Likewise.
8079 (ix86_gimplify_va_arg): Likewise.
8080 (ix86_expand_sse_cmp): Likewise.
8081 (emit_memmov): Likewise.
8082 (emit_memset): Likewise.
8083 (expand_small_movmem_or_setmem): Likewise.
8084 (ix86_expand_pextr): Likewise.
8085 (ix86_expand_pinsr): Likewise.
8086 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
8087 * config/microblaze/microblaze.c (microblaze_block_move_straight):
8089 * config/mips/mips.c (mips_function_value_1) Likewise.
8090 (mips_block_move_straight): Likewise.
8091 (mips_expand_ins_as_unaligned_store): Likewise.
8092 * config/powerpcspe/powerpcspe.c
8093 (rs6000_darwin64_record_arg_advance_flush): Likewise.
8094 (rs6000_darwin64_record_arg_flush): Likewise.
8095 * config/rs6000/rs6000.c
8096 (rs6000_darwin64_record_arg_advance_flush): Likewise.
8097 (rs6000_darwin64_record_arg_flush): Likewise.
8098 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
8099 (sparc_function_value_1): Likewise.
8100 * config/spu/spu.c (adjust_operand): Likewise.
8101 (spu_emit_branch_or_set): Likewise.
8102 (arith_immediate_p): Likewise.
8103 * emit-rtl.c (gen_lowpart_common): Likewise.
8104 * expr.c (expand_expr_real_1): Likewise.
8105 * function.c (assign_parm_setup_block): Likewise.
8106 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8107 * reload1.c (alter_reg): Likewise.
8108 * stor-layout.c (mode_for_vector): Likewise.
8109 (layout_type): Likewise.
8111 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8113 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
8114 (spu_convert_move): Likewise.
8115 * lower-subreg.c (resolve_simple_move): Likewise.
8117 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8120 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
8121 define_insn to a define_expand.
8122 (altivec_vsum2sws_direct): New define_insn.
8123 (altivec_vsumsws): Convert from a define_insn to a define_expand.
8125 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8127 * config/arm/arm.c (arm_option_params_internal): Improve setting of
8130 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
8135 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
8136 realign stack if stack alignment needed is less than incoming
8139 2017-09-05 Marek Polacek <polacek@redhat.com>
8142 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
8145 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8147 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
8149 2017-09-05 Richard Biener <rguenther@suse.de>
8151 PR tree-optimization/82084
8152 * fold-const.c (can_native_encode_string_p): Handle wide characters.
8154 2017-09-05 Richard Biener <rguenther@suse.de>
8156 PR tree-optimization/82102
8157 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
8159 2017-09-05 Martin Liska <mliska@suse.cz>
8161 PR tree-optimization/82032
8162 * tree-cfg.c (generate_range_test): New function.
8163 * tree-cfg.h (generate_range_test): Declared here.
8164 * tree-cfgcleanup.c (convert_single_case_switch): New function.
8165 (cleanup_control_expr_graph): Use it.
8166 * tree-switch-conversion.c (try_switch_expansion): Remove
8168 (emit_case_nodes): Use generate_range_test.
8170 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
8173 * config/i386/i386.md (*<btsc><mode>_mask): Add
8174 TARGET_USE_BT to insn constraint.
8175 (*btr<mode>_mask): Ditto.
8177 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
8179 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
8180 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
8182 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
8185 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
8186 TARGET_NEON and TARGET_IWMMXT.
8187 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
8188 TARGET_NEON and TARGET_IWMMXT.
8189 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
8191 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
8193 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
8194 (ix86_rewrite_tls_address): Ditto.
8195 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
8196 (ix86_rewrite_tls_address_1): Ditto.
8197 (ix86_rewrite_tls_address): Ditto.
8198 * config/i386/predicates.md (tls_address_pattern): New predicate.
8199 * config/i386/i386.md (TLS address splitter): New splitter.
8201 2017-09-04 Richard Biener <rguenther@suse.de>
8203 PR tree-optimization/82084
8204 * fold-const.h (can_native_encode_string_p): Declare.
8205 * fold-const.c (can_native_encode_string_p): Factor out from ...
8206 (native_encode_string): ... here.
8207 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
8208 vectorizing stores from constants we later cannot handle.
8210 2017-09-04 Marek Polacek <polacek@redhat.com>
8213 * doc/invoke.texi: Update -Wtautological-compare documentation.
8215 2017-09-04 Jeff Law <law@redhat.com>
8217 PR tree-optimization/64910
8218 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
8219 swap the first and last operand if the last is a constant.
8221 2017-09-04 Marek Polacek <polacek@redhat.com>
8224 * convert.c (do_narrow): When sanitizing signed integer overflows,
8225 bail out for signed types.
8226 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
8228 2017-09-04 Richard Biener <rguenther@suse.de>
8230 PR tree-optimization/82060
8231 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
8232 Move devirtualization after stmt folding and before EH/AB/noreturn
8233 cleanup to get the stmt refs canonicalized. Use a bool instead
8234 of gimple_modified_p since that doesn't work for NOPs. Schedule
8235 NOPs generated by folding for removal.
8237 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8238 Alan Hayward <alan.hayward@arm.com>
8239 David Sherwood <david.sherwood@arm.com>
8241 * coretypes.h (pad_direction): New enum.
8242 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
8243 (FUNCTION_ARG_PADDING): Likewise.
8244 * target.def (function_arg_padding): New hook.
8245 * targhooks.h (default_function_arg_padding): Declare.
8246 * targhooks.c (default_function_arg_padding): New function.
8247 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
8248 (TARGET_FUNCTION_ARG_PADDING): ...this.
8249 * doc/tm.texi: Regenerate.
8250 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
8251 instead of direction.
8252 (compute_argument_addresses): Likewise.
8253 (load_register_parameters): Likewise.
8254 (emit_library_call_value_1): Likewise.
8255 (store_one_arg): Use targetm.calls.function_arg_padding instead
8256 of FUNCTION_ARG_PADDING.
8257 (must_pass_in_stack_var_size_or_pad): Likewise.
8258 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
8259 (emit_group_store): Likewise.
8260 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
8261 instead of FUNCTION_ARG_PADDING.
8262 (emit_push_insn): Likewise, and propagate enum change throughout
8264 * function.h (direction): Delete.
8265 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
8267 * function.c (assign_parm_find_stack_rtl): Likewise.
8268 (assign_parm_setup_block_p): Likewise.
8269 (assign_parm_setup_block): Likewise.
8270 (gimplify_parameters): Likewise.
8271 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
8272 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
8274 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
8275 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8276 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
8277 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
8278 (aarch64_function_arg_padding): ...this new function.
8279 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
8280 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8281 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
8282 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8283 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
8284 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8285 (arm_pad_arg_upward): Replace with...
8286 (arm_function_arg_padding): ...this new function.
8287 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
8289 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
8290 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
8291 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8292 (ia64_hpux_function_arg_padding): Replace with...
8293 (ia64_function_arg_padding): ...this new function. Use pad_direction
8294 instead of direction. Check for TARGET_HPUX.
8295 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
8296 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8297 (iq2000_function_arg_padding): New function.
8298 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
8299 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
8300 (mips_function_arg_padding): ...this new function.
8301 (mips_pad_reg_upward): Update accordingly.
8302 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8303 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
8304 targetm.calls.function_arg_padding.
8305 (FUNCTION_ARG_PADDING): Delete.
8306 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8307 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
8308 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8309 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
8310 (nios2_block_reg_padding): Return pad_direction instead of direction.
8311 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
8312 instead of direction.
8313 (nios2_function_arg_padding): Likewise. Make static.
8314 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8315 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
8316 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
8317 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
8318 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8319 (pa_function_arg_padding): Make static. Return pad_direction instead
8321 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
8322 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8323 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
8324 instead of direction. Use targetm.calls.function_arg_padding.
8325 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
8326 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
8327 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
8328 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
8329 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
8331 (function_arg_padding): Rename to...
8332 (rs6000_function_arg_padding): ...this. Make static. Return
8333 pad_direction instead of direction.
8334 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
8335 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
8336 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8337 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
8338 instead of direction. Use targetm.calls.function_arg_padding.
8339 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
8340 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
8341 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
8342 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
8343 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8344 (function_arg_padding): Rename to...
8345 (rs6000_function_arg_padding): ...this. Make static. Return
8346 pad_direction instead of direction.
8347 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
8348 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
8349 * config/s390/s390.c (s390_function_arg_padding): New function.
8350 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8351 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
8352 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
8353 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8354 (function_arg_padding): Rename to...
8355 (sparc_function_arg_padding): ...this. Make static. Return
8356 pad_direction instead of direction.
8357 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
8358 * config/spu/spu.c (spu_function_arg_padding): New function.
8359 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8360 * system.h (FUNCTION_ARG_PADDING): Poison.
8362 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8363 Alan Hayward <alan.hayward@arm.com>
8364 David Sherwood <david.sherwood@arm.com>
8366 * target.def (modes_tieable_p): New hook.
8367 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
8368 (TARGET_MODES_TIEABLE_P): ...this.
8369 * doc/tm.texi.in: Regenerate.
8370 * hooks.h (hook_bool_mode_mode_true): Declare.
8371 * hooks.c (hook_bool_mode_mode_true): New function.
8372 * combine.c (subst): Use targetm.modes_tieable_p instead of
8374 * dse.c (find_shift_sequence): Likewise.
8375 * expmed.c (extract_low_bits): Likewise.
8376 * lower-subreg.c: Include target.h.
8377 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
8379 * rtlanal.c (rtx_cost): Likewise.
8380 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
8381 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
8382 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
8383 (TARGET_MODES_TIEABLE_P): Redefine.
8384 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
8385 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
8386 (TARGET_MODES_TIEABLE_P): Redefine.
8387 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
8388 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
8389 (arc_modes_tieable_p): New function.
8390 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
8391 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
8392 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
8393 (arm_modes_tieable_p): Make static.
8394 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
8395 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
8396 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
8397 (TARGET_MODES_TIEABLE_P): Redefine.
8398 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
8399 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
8400 (TARGET_MODES_TIEABLE_P): Redefine.
8401 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
8402 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
8403 (cr16_modes_tieable_p): New function.
8404 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
8405 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
8406 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
8407 (TRULY_NOOP_TRUNCATION): Update comment.
8408 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
8409 (TRULY_NOOP_TRUNCATION): Update comment.
8410 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
8411 (frv_modes_tieable_p): New function.
8412 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
8413 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
8414 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
8415 (TARGET_MODES_TIEABLE_P): Redefine.
8416 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
8417 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
8418 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
8419 (TARGET_MODES_TIEABLE_P): Redefine.
8420 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
8421 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
8422 (ia64_modes_tieable_p): New function.
8423 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
8424 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
8425 (iq2000_modes_tieable_p): New function.
8426 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
8427 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
8428 (lm32_modes_tieable_p): New function.
8429 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
8430 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
8431 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
8432 (TARGET_MODES_TIEABLE_P): Redefine.
8433 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
8434 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
8435 (m32r_modes_tieable_p): New function.
8436 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
8437 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
8438 (m68k_modes_tieable_p): New function.
8439 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
8440 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
8441 (mcore_modes_tieable_p): New function.
8442 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
8443 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
8445 (TARGET_MODES_TIEABLE_P): Redefine.
8446 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
8447 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
8448 * config/mips/mips.c (mips_modes_tieable_p): Make static.
8449 (TARGET_MODES_TIEABLE_P): Redefine.
8450 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
8451 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
8452 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
8453 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
8454 (mn10300_modes_tieable_p): ...this and make static.
8455 (TARGET_MODES_TIEABLE_P): Redefine.
8456 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
8457 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
8458 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
8459 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
8460 (msp430_modes_tieable_p): Make static.
8461 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
8462 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
8463 (TARGET_MODES_TIEABLE_P): Redefine.
8464 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
8465 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
8466 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
8467 (TARGET_MODES_TIEABLE_P): Redefine.
8468 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
8469 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
8470 * config/pa/pa.c (pa_modes_tieable_p): Make static.
8471 (TARGET_MODES_TIEABLE_P): Redefine.
8472 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
8473 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
8474 (pdp11_modes_tieable_p): New function.
8475 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
8476 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
8477 (rs6000_modes_tieable_p): New function.
8478 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
8479 * config/powerpcspe/powerpcspe.md: Update comment.
8480 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
8481 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
8482 (TARGET_MODES_TIEABLE_P): Redefine.
8483 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
8484 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
8485 (rl78_modes_tieable_p): New function.
8486 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
8487 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
8488 (rs6000_modes_tieable_p): New function.
8489 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
8490 * config/rs6000/rs6000.md: Update comment.
8491 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
8492 * config/rx/rx.c (rx_modes_tieable_p): New function.
8493 (TARGET_MODES_TIEABLE_P): Redefine.
8494 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
8495 * config/s390/s390.c (s390_modes_tieable_p): New function.
8496 (TARGET_MODES_TIEABLE_P): Redefine.
8497 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
8498 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
8499 (sh_modes_tieable_p): New function.
8500 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
8501 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
8502 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
8503 (sparc_modes_tieable_p): Make static.
8504 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
8505 * config/spu/spu.c (spu_modes_tieable_p): New function.
8506 (TARGET_MODES_TIEABLE_P): Redefine.
8507 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
8508 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
8509 (TARGET_MODES_TIEABLE_P): Redefine.
8510 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
8511 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
8512 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
8513 * config/v850/v850.c (v850_modes_tieable_p): New function.
8514 (TARGET_MODES_TIEABLE_P): Redefine.
8515 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
8516 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
8517 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
8518 (visium_modes_tieable_p): New function.
8519 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
8520 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
8521 (xtensa_modes_tieable_p): New function.
8522 * system.h (MODES_TIEABLE_P): Poison.
8524 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8525 Alan Hayward <alan.hayward@arm.com>
8526 David Sherwood <david.sherwood@arm.com>
8528 * target.def (hard_regno_mode_ok): New hook.
8529 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
8530 (TARGET_HARD_REGNO_MODE_OK): ...this.
8531 * doc/tm.texi.in: Regenerate.
8532 * hooks.h (hook_bool_uint_mode_true): Declare.
8533 * hooks.c (hook_bool_uint_mode_true): New function.
8534 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
8536 * genpreds.c (write_insn_preds_c): Add an include of target.h.
8537 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
8538 instead of HARD_REGNO_MODE_OK.
8539 * caller-save.c: Include target.h.
8540 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
8542 * combine.c (can_combine_p): Likewise.
8543 (combinable_i3pat): Likewise.
8544 (can_change_dest_mode): Likewise.
8545 * expr.c (init_expr_target): Likewise.
8546 (convert_move): Likewise.
8547 (convert_modes): Likewise.
8548 * ira.c (setup_prohibited_class_mode_regs): Likewise.
8549 (setup_prohibited_mode_move_regs): Likewise.
8550 * ira.h (target_ira): Likewise.
8551 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8552 * lra-constraints.c (process_alt_operands): Likewise.
8553 (split_reg): Likewise.
8554 * recog.c (peep2_find_free_register): Likewise.
8555 * ree.c (combine_reaching_defs): Likewise.
8556 * regcprop.c (maybe_mode_change): Likewise.
8557 * reginfo.c (init_reg_sets_1): Likewise.
8558 (choose_hard_reg_mode): Likewise.
8559 (simplifiable_subregs): Likewise.
8560 * regrename.c (check_new_reg_p): Likewise.
8561 * reload.c (find_valid_class): Likewise.
8562 (find_valid_class_1): Likewise.
8563 (reload_inner_reg_of_subreg): Likewise.
8564 (push_reload): Likewise.
8565 (combine_reloads): Likewise.
8566 (find_dummy_reload): Likewise.
8567 (find_reloads): Likewise.
8568 * reload1.c (find_reg): Likewise.
8569 (set_reload_reg): Likewise.
8570 (allocate_reload_reg): Likewise.
8571 (choose_reload_regs): Likewise.
8572 (reload_adjust_reg_for_temp): Likewise.
8573 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
8574 (simplify_subreg_regno): Likewise.
8575 * sel-sched.c (init_regs_for_mode): Likewise.
8576 * varasm.c (make_decl_rtl): Likewise.
8577 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
8578 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
8580 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
8581 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
8582 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8583 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
8584 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
8585 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8586 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
8587 (arc_mode_class): Delete.
8588 (HARD_REGNO_MODE_OK): Delete.
8589 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8590 (arc_hard_regno_mode_ok): Rename old array to...
8591 (arc_hard_regno_mode_ok_modes): ...this.
8592 (arc_conditional_register_usage): Update accordingly.
8593 (arc_mode_class): Make static.
8594 (arc_hard_regno_mode_ok): New function.
8595 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
8596 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
8597 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8598 (arm_hard_regno_mode_ok): Make static.
8599 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
8601 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
8602 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
8603 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
8605 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8606 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
8607 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
8608 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
8609 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
8610 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8611 * config/bfin/predicates.md (valid_reg_operand): Use
8612 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
8613 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
8614 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
8615 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8616 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
8617 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
8618 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8619 (cr16_hard_regno_mode_ok): Make static and return a bool.
8620 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
8621 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8622 (cris_hard_regno_mode_ok): New function.
8623 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
8624 (epiphany_mode_class): Delete.
8625 (HARD_REGNO_MODE_OK): Delete.
8626 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
8627 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8628 (hard_regno_mode_ok): Rename to...
8629 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
8630 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
8631 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
8633 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
8634 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
8635 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8636 (frv_hard_regno_mode_ok): Make static and return a bool.
8637 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
8639 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
8640 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
8641 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
8642 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
8644 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8645 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
8646 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
8647 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
8649 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8650 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
8651 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8652 (ia64_hard_regno_mode_ok): New function.
8653 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
8654 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8655 (iq2000_hard_regno_mode_ok): New function.
8656 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
8657 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8658 (lm32_hard_regno_mode_ok): New function.
8659 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
8660 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
8661 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
8662 instead of HARD_REGNO_MODE_OK.
8663 (m32c_hard_regno_ok): Rename to...
8664 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
8665 (m32c_cannot_change_mode_class): Update accordingly.
8666 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8667 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
8668 (m32r_mode_class): Delete.
8669 (HARD_REGNO_MODE_OK): Delete.
8670 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8671 (m32r_hard_regno_mode_ok): Rename to...
8672 (m32r_hard_regno_modes): ...this.
8673 (m32r_mode_class): Make static.
8674 (m32r_hard_regno_mode_ok): New function.
8675 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
8676 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
8677 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8678 (m68k_hard_regno_mode_ok): Make static.
8679 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
8680 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8681 (mcore_hard_regno_mode_ok): New function.
8682 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
8683 (HARD_REGNO_MODE_OK): Delete.
8684 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
8686 (microblaze_hard_regno_mode_ok_p): ...this and make static.
8687 (microblaze_hard_regno_mode_ok): New function.
8688 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8689 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
8690 (mips_hard_regno_mode_ok): Delete.
8691 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
8692 (mips_hard_regno_mode_ok_p): ...this and make static.
8693 (mips_hard_regno_mode_ok_p): Rename to...
8694 (mips_hard_regno_mode_ok_uncached): ...this.
8695 (mips_hard_regno_mode_ok): New function.
8696 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
8697 of HARD_REGNO_MODE_OK.
8698 (mips_option_override): Update after above name changes.
8699 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8700 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
8701 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
8702 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
8703 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
8704 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8705 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
8706 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
8707 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
8708 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8709 (msp430_hard_regno_mode_ok): Make static and return a bool.
8710 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
8711 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
8712 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
8714 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8715 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
8716 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
8717 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
8718 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
8719 (PA_HARD_REGNO_MODE_OK): ...this
8720 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
8721 (PA_HARD_REGNO_MODE_OK): ...this.
8722 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8723 (pa_hard_regno_mode_ok): New function.
8724 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
8725 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8726 (pdp11_hard_regno_mode_ok): New function.
8727 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
8728 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
8730 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
8732 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8733 (rs6000_hard_regno_mode_ok): Rename to...
8734 (rs6000_hard_regno_mode_ok_uncached): ...this.
8735 (rs6000_init_hard_regno_mode_ok): Update accordingly.
8736 (rs6000_hard_regno_mode_ok): New function.
8737 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
8738 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
8739 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
8740 (riscv_hard_regno_mode_ok): ...this and make static.
8741 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8742 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
8743 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
8744 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8745 (rl78_hard_regno_mode_ok): Make static and return bool.
8746 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
8747 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
8749 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
8750 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8751 (rs6000_hard_regno_mode_ok): Rename to...
8752 (rs6000_hard_regno_mode_ok_uncached): ...this.
8753 (rs6000_init_hard_regno_mode_ok): Update accordingly.
8754 (rs6000_hard_regno_mode_ok): New function.
8755 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
8756 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
8757 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8758 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
8759 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
8760 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
8761 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8762 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
8763 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
8764 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8765 (sh_hard_regno_mode_ok): Make static.
8766 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
8767 instead of HARD_REGNO_MODE_OK.
8768 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
8769 (sparc_mode_class): Delete.
8770 (HARD_REGNO_MODE_OK): Delete.
8771 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8772 (hard_regno_mode_classes): Make static.
8773 (sparc_mode_class): Likewise.
8774 (sparc_hard_regno_mode_ok): New function.
8775 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
8776 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
8777 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
8779 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8780 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
8781 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
8782 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
8783 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
8784 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8785 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
8786 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
8787 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8788 (visium_hard_regno_mode_ok): New function.
8789 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
8790 instead of HARD_REGNO_MODE_OK.
8791 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
8792 (HARD_REGNO_MODE_OK): Delete.
8793 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
8794 (xtensa_hard_regno_mode_ok_p): ...this and make static.
8795 (xtensa_option_override): Update accordingly.
8796 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8797 (xtensa_hard_regno_mode_ok): New function.
8798 * system.h (HARD_REGNO_MODE_OK): Poison.
8800 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8801 Alan Hayward <alan.hayward@arm.com>
8802 David Sherwood <david.sherwood@arm.com>
8804 * target.def (hard_regno_call_part_clobbered): New hook.
8805 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
8806 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
8807 * doc/tm.texi: Regenerate.
8808 * hooks.h (hook_bool_uint_mode_false): Declare.
8809 * hooks.c (hook_bool_uint_mode_false): New function.
8810 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8811 * cselib.c (cselib_process_insn): Use
8812 targetm.hard_regno_call_part_clobbered instead of
8813 HARD_REGNO_CALL_PART_CLOBBERED.
8814 * ira-conflicts.c (ira_build_conflicts): Likewise.
8815 * ira-costs.c (ira_tune_allocno_costs): Likewise.
8816 * lra-constraints.c (need_for_call_save_p): Likewise.
8817 * lra-lives.c: Include target.h.
8818 (check_pseudos_live_through_calls): Use
8819 targetm.hard_regno_call_part_clobbered instead of
8820 HARD_REGNO_CALL_PART_CLOBBERED.
8821 * regcprop.c: Include target.h.
8822 (copyprop_hardreg_forward_1): Use
8823 targetm.hard_regno_call_part_clobbered instead of
8824 HARD_REGNO_CALL_PART_CLOBBERED.
8825 * reginfo.c (choose_hard_reg_mode): Likewise.
8826 * regrename.c (check_new_reg_p): Likewise.
8827 * reload.c (find_equiv_reg): Likewise.
8828 * reload1.c (emit_reload_insns): Likewise.
8829 * sched-deps.c (deps_analyze_insn): Likewise.
8830 * sel-sched.c (init_regs_for_mode): Likewise.
8831 (mark_unavailable_hard_regs): Likewise.
8832 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
8833 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8834 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
8836 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8837 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8838 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
8840 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
8842 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8843 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8844 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
8846 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8847 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8848 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
8850 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8851 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
8853 * config/powerpcspe/powerpcspe.c
8854 (rs6000_hard_regno_call_part_clobbered): New function.
8855 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8856 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8857 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
8859 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8860 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8861 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
8863 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8864 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8865 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
8867 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8868 Alan Hayward <alan.hayward@arm.com>
8869 David Sherwood <david.sherwood@arm.com>
8871 * rtl.h (subreg_memory_offset): Declare.
8872 * emit-rtl.c (subreg_memory_offset): New function.
8873 * expmed.c (store_bit_field_1): Use it.
8874 * expr.c (undefined_operand_subword_p): Likewise.
8875 * simplify-rtx.c (simplify_subreg): Likewise.
8877 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
8879 PR rtl-optimization/57448
8882 * optabs.c (expand_atomic_load): Place compiler memory barriers if
8883 using atomic_load pattern.
8884 (expand_atomic_store): Likewise.
8886 2017-09-04 Jakub Jelinek <jakub@redhat.com>
8889 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
8890 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
8891 handling. Use replace_call_with_value with NULL instead of
8892 gsi_replace, unlink_stmt_vdef and release_defs.
8894 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
8897 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
8899 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8902 * rtl.h (emit_library_call_value_1): Declare.
8903 (emit_library_call): Replace declaration with a series of overloads.
8904 Remove the parameter count argument.
8905 (emit_library_call_value): Likewise.
8906 * calls.c (emit_library_call_value_1): Make global. Replace varargs
8907 with an "rtx_mode_t *".
8908 (emit_library_call_value): Delete.
8909 (emit_library_call): Likewise.
8910 * asan.c (asan_emit_stack_protection): Update calls accordingly.
8911 (asan_emit_allocas_unpoison): Likewise.
8912 * builtins.c (expand_builtin_powi): Likewise.
8913 (expand_asan_emit_allocas_unpoison): Likewise.
8914 * cfgexpand.c (expand_main_function): Likewise.
8915 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
8916 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
8917 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
8918 * config/arm/arm.c (arm_trampoline_init): Likewise.
8919 (arm_call_tls_get_addr): Likewise.
8920 (arm_expand_divmod_libfunc): Likewise.
8921 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
8922 (smulsi3_highpart): Likewise.
8923 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
8924 (c6x_expand_compare): Likewise.
8925 (c6x_expand_movmem): Likewise.
8926 * config/frv/frv.c (frv_trampoline_init): Likewise.
8927 * config/i386/i386.c (ix86_trampoline_init): Likewise.
8928 (ix86_expand_divmod_libfunc): Likewise.
8929 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
8930 (ia64_expand_compare): Likewise.
8931 (ia64_profile_hook): Likewise.
8932 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
8933 (nonlocal_goto): Likewise.
8934 (restore_stack_nonlocal): Likewise.
8935 * config/m32r/m32r.c (block_move_call): Likewise.
8936 (m32r_trampoline_init): Likewise.
8937 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
8938 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
8939 (m68k_call_m68k_read_tp): Likewise.
8940 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
8941 (microblaze_expand_divide): Likewise.
8942 * config/mips/mips.h (mips_args): Likewise.
8943 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
8944 (MIPS_ICACHE_SYNC): Likewise.
8945 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
8946 (nios2_trampoline_init): Likewise.
8947 * config/pa/pa.c (hppa_tls_call): Likewise.
8948 (pa_trampoline_init): Likewise.
8949 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
8950 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
8951 (expand_strn_compare): Likewise.
8952 (rs6000_generate_compare): Likewise.
8953 (rs6000_expand_float128_convert): Likewise.
8954 (output_profile_hook): Likewise.
8955 (rs6000_trampoline_init): Likewise.
8956 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
8957 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
8958 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
8959 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
8960 (rs6000_generate_compare): Likewise.
8961 (rs6000_expand_float128_convert): Likewise.
8962 (output_profile_hook): Likewise.
8963 (rs6000_trampoline_init): Likewise.
8964 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
8965 * config/sh/sh.c (sh_trampoline_init): Likewise.
8966 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
8967 (sparc_emit_float_lib_cmp): Likewise.
8968 (sparc32_initialize_trampoline): Likewise.
8969 (sparc64_initialize_trampoline): Likewise.
8970 (sparc_profile_hook): Likewise.
8971 * config/spu/spu.c (ea_load_store): Likewise.
8972 * config/spu/spu.md (floatunssidf2): Likewise.
8973 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
8974 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
8975 * config/visium/visium.c (expand_block_move_4): Likewise.
8976 (expand_block_move_2): Likewise.
8977 (expand_block_move_1): Likewise.
8978 (expand_block_set_4): Likewise.
8979 (expand_block_set_2): Likewise.
8980 (expand_block_set_1): Likewise.
8981 (visium_trampoline_init): Likewise.
8982 (visium_profile_hook): Likewise.
8983 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
8984 (xtensa_setup_frame_addresses): Likewise.
8985 (xtensa_trampoline_init): Likewise.
8986 * except.c (sjlj_emit_function_enter): Likewise.
8987 (sjlj_emit_function_exit): Likewise.
8988 * explow.c (allocate_dynamic_stack_space): Likewise.
8989 (probe_stack_range): Likewise.
8990 * expr.c (convert_mode_scalar): Likewise.
8991 * optabs.c (expand_binop): Likewise.
8992 (expand_twoval_binop_libfunc): Likewise.
8993 (expand_unop): Likewise.
8994 (prepare_cmp_insn): Likewise.
8995 (prepare_float_lib_cmp): Likewise.
8996 (expand_float): Likewise.
8997 (expand_fix): Likewise.
8998 (expand_fixed_convert): Likewise.
8999 (maybe_emit_sync_lock_test_and_set): Likewise.
9000 (expand_atomic_compare_and_swap): Likewise.
9001 (expand_mem_thread_fence): Likewise.
9002 (expand_atomic_fetch_op): Likewise.
9004 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
9006 * doc/generic.texi (OpenACC): Adjust URL.
9007 * doc/invoke.texi (C Dialect Options): Ditto.
9009 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
9011 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
9012 predicate for operand 1. Add (m,<S>) constraint.
9013 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
9014 Prevent memory operand 1 with register operand 2.
9016 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
9018 PR rtl-optimization/82024
9019 * combine.c (try_combine): If the combination result is a PARALLEL,
9020 and we only need to retain the SET in there that would be placed
9021 at I2, check that we can place that at I3 instead, before doing so.
9023 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9026 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
9028 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
9029 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
9032 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
9033 Jeff Law <law@redhat.com>
9035 * varasm.c (bss_initializer_p): Do not put constants into .bss
9036 (categorize_decl_for_section): Handle bss_initializer_p returning
9037 false when DECL_INITIAL is NULL.
9039 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9042 * config/s390/s390.c (s390_can_inline_p): New function.
9044 2017-09-01 Jeff Law <law@redhat.com>
9046 PR tree-optimization/82052
9047 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
9048 Always initialize the returned slot after a hash table miss
9049 when INSERT is true.
9051 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
9053 * config/s390/s390.md (mem_signal_fence): Remove.
9054 * doc/md.texi (mem_signal_fence): Remove.
9055 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
9057 * target-insns.def (mem_signal_fence): Remove.
9059 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9062 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
9065 * asan.c (create_odr_indicator): Strip name encoding from assembler
9066 name before appending it after __odr_asan_.
9068 2017-09-01 Martin Liska <mliska@suse.cz>
9070 PR tree-optimization/82059
9071 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
9072 frequency only when an edge is redirected.
9074 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9076 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
9077 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
9078 (arc_conditional_register_usage): Remove ARC600 lp_count
9080 (arc_file_start): Emit Tag_ARC_CPU_variation.
9081 (arc_can_use_doloop_p): New conditions to use ZOLs.
9082 (hwloop_fail): New function.
9083 (hwloop_optimize): Likewise.
9084 (hwloop_pattern_reg): Likewise.
9085 (arc_doloop_hooks): New struct, to be used with reorg_loops.
9086 (arc_reorg_loops): New function, calls reorg_loops.
9087 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
9088 (arc600_corereg_hazard): Remove ZOL checking, case handled by
9090 (arc_loop_hazard): Remove function, functionality moved into
9092 (arc_hazard): Remove arc_loop_hazard call.
9093 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
9094 into hwloop_optimize.
9095 (arc_label_align): Remove ZOL handling.
9096 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
9097 * config/arc/arc.md (doloop_begin): Remove pattern.
9098 (doloop_begin_i): Likewise.
9099 (doloop_end_i): Likewise.
9100 (doloop_fallback): Likewise.
9101 (doloop_fallback_m): Likewise.
9102 (doloop_end): Reimplement expand.
9103 (arc_lp): New pattern for LP instruction.
9104 (loop_end): New pattern.
9105 (loop_fail): Likewise.
9106 (decrement_and_branch_until_zero): Likewise.
9107 * config/arc/arc.opt (mlpc-width): New option.
9108 * doc/invoke.texi (mlpc-width): Document option.
9110 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9112 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
9113 (arc_ccfsm_advance): Fix checking for delay slots.
9114 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
9116 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9118 * config/arc/arc.md (movqi_insn): Add stores to save constant long
9120 (movhi_insn): Update store instruction constraint which are saving
9121 6-bit short immediates.
9122 (movsi_insn): Consider also short scaled load operations.
9123 (zero_extendhisi2_i): Use Usd constraint instead of T.
9124 (extendhisi2_i): Add q constraint.
9125 (arc_clzsi2): Add type and length attributes.
9126 (arc_ctzsi2): Likewise.
9127 * config/arc/constraints.md (Usc): Update constraint, the
9128 assembler can parse two relocations for a single instruction.
9130 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9132 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
9133 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
9135 2017-08-31 Olivier Hainque <hainque@adacore.com>
9137 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
9138 match as powerpc-wrs-vxworks*.
9140 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
9142 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
9143 register constraint for by-element operand.
9144 (aarch64_mls_elt_merge<mode>): Likewise.
9146 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9148 * config/arc/arc.c (arc_can_follow_jump): Check for short
9151 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9153 * config.gcc: Use g.opt for arc.
9154 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
9155 functionality moved to ...
9156 (legitimate_scaled_address_p): New function, ...here.
9157 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
9158 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
9159 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
9161 (arc_override_options): Handle G option.
9162 (arc_output_pic_addr_const): Correct function definition.
9163 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
9164 (arc_decl_anon_ns_mem_p): Delete.
9165 (arc_in_small_data_p): Overhaul this function to take into
9166 consideration the value given via G option.
9167 (arc_rewrite_small_data_1): Renamed and corrected old
9168 arc_rewrite_small_data function.
9169 (arc_rewrite_small_data): New function.
9170 (small_data_pattern): Don't use pic_offset_table_rtx.
9171 * config/arc/arc.h (CC1_SPEC): Recognize G option.
9172 * config/arc/simdext.md (movmisalignv2hi): Use
9173 prepare_move_operands function.
9175 (movmisalign*): Likewise.
9176 * doc/invoke.texi (ARC options): Document -G option.
9178 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9180 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
9182 * config/arc/arc.c (arc_print_operand): Output scalled address for
9183 sdata whenever is possible.
9184 (arc_in_small_data_p): Allow sdata for 64bit datum when double
9185 load/stores are available.
9186 (compact_sda_memory_operand): Check for the alignment required by
9187 code density instructions.
9188 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
9190 * config/arc/constraints.md (Usd): Update constraint.
9191 (Us0): New constraint.
9192 (Usc): Update constraint.
9194 2017-08-31 Richard Biener <rguenther@suse.de>
9197 * dwarf2out.c (dwarf2out_early_global_decl): Process each
9200 2017-08-31 Tamar Christina <tamar.christina@arm.com>
9202 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
9203 Resize type_signature.
9205 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9206 Alan Hayward <alan.hayward@arm.com>
9207 David Sherwood <david.sherwood@arm.com>
9209 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
9210 subregs whose inner modes can be stored in GPRs.
9211 (aarch64_classify_index): Likewise.
9213 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9214 Alan Hayward <alan.hayward@arm.com>
9215 David Sherwood <david.sherwood@arm.com>
9217 * config/aarch64/iterators.md (V_cmp_result): Rename to...
9218 (V_INT_EQUIV): ...this.
9219 (v_cmp_result): Rename to...
9220 (v_int_equiv): ...this.
9221 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
9222 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
9223 (copysign<mode>3): Likewise.
9224 (aarch64_simd_bsl<mode>_internal): Likewise.
9225 (aarch64_simd_bsl<mode>): Likewise.
9226 (vec_cmp<mode><mode>): Likewise.
9227 (vcond<mode><mode>): Likewise.
9228 (vcond<v_cmp_mixed><mode>): Likewise.
9229 (vcondu<mode><v_cmp_mixed>): Likewise.
9230 (aarch64_cm<optab><mode>): Likewise.
9231 (aarch64_cmtst<mode>): Likewise.
9232 (aarch64_fac<optab><mode>): Likewise.
9233 (vec_perm_const<mode>): Likewise.
9234 (vcond_mask_<mode><v_cmp_result>): Rename to...
9235 (vcond_mask_<mode><v_int_equiv>): ...this.
9236 (vec_cmp<mode><v_cmp_result>): Rename to...
9237 (vec_cmp<mode><v_int_equiv>): ...this.
9239 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9240 Alan Hayward <alan.hayward@arm.com>
9241 David Sherwood <david.sherwood@arm.com>
9243 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
9245 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
9246 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
9247 (UNSPEC_LD4_DREG): New unspecs.
9248 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
9249 (aarch64_ld2<mode>_dreg_be): Replace with...
9250 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
9252 (aarch64_ld3<mode>_dreg_le)
9253 (aarch64_ld3<mode>_dreg_be): Replace with...
9254 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
9256 (aarch64_ld4<mode>_dreg_le)
9257 (aarch64_ld4<mode>_dreg_be): Replace with...
9258 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
9261 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9263 PR tree-optimization/81987
9264 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
9265 insert an initializer in a location not dominated by the stride
9268 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
9270 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
9271 on the entire header of the finally block in the fallthru case.
9273 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
9275 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
9277 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
9279 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
9280 rs6000_emit_move_from_cr and call renamed function.
9281 (rs6000_emit_prologue): Call renamed functions.
9282 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
9283 movesi_from_cr, remove volatile CRs.
9285 2017-08-30 Jon Beniston <jon@beniston.com>
9286 Richard Biener <rguenther@suse.de>
9288 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
9289 of VECTOR_MODE_P check.
9290 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
9291 element vector types.
9293 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9295 * df.h (df_read_modify_subreg_p): Remove in favor of...
9296 * rtl.h (read_modify_subreg_p): ...this new function. Take a
9297 const_rtx instead of an rtx.
9298 * cprop.c (local_cprop_find_used_regs): Update accordingly.
9299 * df-problems.c (df_word_lr_mark_ref): Likewise.
9300 * ira-lives.c (mark_pseudo_reg_live): Likewise.
9301 (mark_pseudo_reg_dead): Likewise.
9302 (mark_ref_dead): Likewise.
9303 * reginfo.c (init_subregs_of_mode): Likewise.
9304 * sched-deps.c (sched_analyze_1): Likewise.
9305 * df-scan.c (df_def_record_1): Likewise.
9306 (df_uses_record): Likewise.
9307 (df_read_modify_subreg_p): Remove in favor of...
9308 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
9309 const_rtx instead of an rtx.
9311 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9312 Alan Hayward <alan.hayward@arm.com>
9313 David Sherwood <david.sherwood@arm.com>
9315 * rtl.h (partial_subreg_p): New function.
9316 * caller-save.c (save_call_clobbered_regs): Use it.
9317 * calls.c (expand_call): Likewise.
9318 * combine.c (combinable_i3pat): Likewise.
9319 (simplify_set): Likewise.
9320 (make_extraction): Likewise.
9321 (make_compound_operation_int): Likewise.
9322 (gen_lowpart_or_truncate): Likewise.
9323 (force_to_mode): Likewise.
9324 (make_field_assignment): Likewise.
9325 (reg_truncated_to_mode): Likewise.
9326 (record_truncated_value): Likewise.
9327 (move_deaths): Likewise.
9328 * cse.c (record_jump_cond): Likewise.
9329 (cse_insn): Likewise.
9330 * cselib.c (cselib_lookup_1): Likewise.
9331 * expmed.c (extract_bit_field_using_extv): Likewise.
9332 * function.c (assign_parm_setup_reg): Likewise.
9333 * ifcvt.c (noce_convert_multiple_sets): Likewise.
9334 * ira-build.c (create_insn_allocnos): Likewise.
9335 * lra-coalesce.c (merge_pseudos): Likewise.
9336 * lra-constraints.c (match_reload): Likewise.
9337 (simplify_operand_subreg): Likewise.
9338 (curr_insn_transform): Likewise.
9339 * lra-lives.c (process_bb_lives): Likewise.
9340 * lra.c (new_insn_reg): Likewise.
9341 (lra_substitute_pseudo): Likewise.
9342 * regcprop.c (mode_change_ok): Likewise.
9343 (maybe_mode_change): Likewise.
9344 (copyprop_hardreg_forward_1): Likewise.
9345 * reload.c (push_reload): Likewise.
9346 (find_reloads): Likewise.
9347 (find_reloads_subreg_address): Likewise.
9348 * reload1.c (alter_reg): Likewise.
9349 (eliminate_regs_1): Likewise.
9350 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9352 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
9354 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
9355 back to if statements, including unpack.
9357 2017-08-30 Martin Liska <mliska@suse.cz>
9360 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
9362 (func_checker::compare_asm_inputs_outputs): ... this function.
9363 (func_checker::compare_gimple_asm): Use the function to compare
9364 also ASM constrains.
9365 * ipa-icf-gimple.h: Rename the function.
9367 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9368 Alan Hayward <alan.hayward@arm.com>
9369 David Sherwood <david.sherwood@arm.com>
9371 * coretypes.h (complex_mode): New type.
9372 * gdbhooks.py (build_pretty_printer): Handle it.
9373 * machmode.h (complex_mode): New class.
9374 (complex_mode::includes_p): New function.
9375 (is_complex_int_mode): Likewise.
9376 (is_complex_float_mode): Likewise.
9377 * genmodes.c (get_mode_class): Handle complex mode classes.
9378 * function.c (expand_function_end): Use is_complex_int_mode.
9380 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9381 Alan Hayward <alan.hayward@arm.com>
9382 David Sherwood <david.sherwood@arm.com>
9384 * coretypes.h (scalar_mode_pod): New typedef.
9385 * gdbhooks.py (build_pretty_printer): Handle it.
9386 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
9387 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
9388 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
9389 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
9392 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9393 Alan Hayward <alan.hayward@arm.com>
9394 David Sherwood <david.sherwood@arm.com>
9396 * machmode.h (mode_for_vector): Take a scalar_mode instead
9398 * stor-layout.c (mode_for_vector): Likewise.
9399 * explow.c (promote_mode): Use as_a <scalar_mode>.
9400 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
9402 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9403 Alan Hayward <alan.hayward@arm.com>
9404 David Sherwood <david.sherwood@arm.com>
9406 * target.def (preferred_simd_mode): Take a scalar_mode
9407 instead of a machine_mode.
9408 * targhooks.h (default_preferred_simd_mode): Likewise.
9409 * targhooks.c (default_preferred_simd_mode): Likewise.
9410 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
9411 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
9412 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
9413 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
9414 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
9415 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
9416 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
9417 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
9419 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
9420 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
9421 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
9422 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
9423 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
9424 * doc/tm.texi: Regenerate.
9425 * optabs-query.c (can_vec_mask_load_store_p): Return false for
9428 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9429 Alan Hayward <alan.hayward@arm.com>
9430 David Sherwood <david.sherwood@arm.com>
9432 * target.def (scalar_mode_supported_p): Take a scalar_mode
9433 instead of a machine_mode.
9434 * targhooks.h (default_scalar_mode_supported_p): Likewise.
9435 * targhooks.c (default_scalar_mode_supported_p): Likewise.
9436 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
9437 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
9438 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
9439 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
9440 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
9441 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
9442 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
9443 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
9444 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
9445 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
9446 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
9447 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
9449 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
9450 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
9451 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
9452 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
9453 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
9455 * doc/tm.texi: Regenerate.
9457 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9458 Alan Hayward <alan.hayward@arm.com>
9459 David Sherwood <david.sherwood@arm.com>
9461 * coretypes.h (opt_scalar_mode): New typedef.
9462 * gdbhooks.py (build_pretty_printers): Handle it.
9463 * machmode.h (mode_iterator::get_2xwider): Add overload for
9465 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
9467 * expr.c (convert_mode_scalar): Likewise.
9468 * omp-low.c (omp_clause_aligned_alignment): Likewise.
9469 * optabs.c (expand_float): Likewise.
9470 (expand_fix): Likewise.
9471 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
9473 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9474 Alan Hayward <alan.hayward@arm.com>
9475 David Sherwood <david.sherwood@arm.com>
9477 * optabs.c (expand_float): Explicitly check for scalars before
9478 using a branching expansion.
9479 (expand_fix): Likewise.
9481 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9482 Alan Hayward <alan.hayward@arm.com>
9483 David Sherwood <david.sherwood@arm.com>
9485 * expr.c (convert_mode): Split scalar handling out into...
9486 (convert_mode_scalar): ...this new function. Treat the modes
9489 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9490 Alan Hayward <alan.hayward@arm.com>
9491 David Sherwood <david.sherwood@arm.com>
9493 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
9495 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
9497 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9498 Alan Hayward <alan.hayward@arm.com>
9499 David Sherwood <david.sherwood@arm.com>
9501 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
9502 rather than a machine_mode.
9503 (fixed_from_string): Likewise.
9504 (fixed_convert): Likewise.
9505 (fixed_convert_from_int): Likewise.
9506 (fixed_convert_from_real): Likewise.
9507 (real_convert_from_fixed): Likewise.
9508 * fixed-value.c (fixed_from_double_int): Likewise.
9509 (fixed_from_string): Likewise.
9510 (fixed_convert): Likewise.
9511 (fixed_convert_from_int): Likewise.
9512 (fixed_convert_from_real): Likewise.
9513 (real_convert_from_fixed): Likewise.
9514 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
9516 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9517 Alan Hayward <alan.hayward@arm.com>
9518 David Sherwood <david.sherwood@arm.com>
9520 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
9521 of separate mode class checks. Do not allow vector modes here.
9522 (immed_wide_int_const): Use as_a <scalar_mode>.
9523 * explow.c (trunc_int_for_mode): Likewise.
9524 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
9525 (wi::shwi): Likewise.
9526 (wi::min_value): Likewise.
9527 (wi::max_value): Likewise.
9528 * dwarf2out.c (loc_descriptor): Likewise.
9529 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
9532 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9533 Alan Hayward <alan.hayward@arm.com>
9534 David Sherwood <david.sherwood@arm.com>
9536 * tree.h (SCALAR_TYPE_MODE): New macro.
9537 * expr.c (expand_expr_addr_expr_1): Use it.
9538 (expand_expr_real_2): Likewise.
9539 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
9540 (fold_convert_const_fixed_from_int): Likewise.
9541 (fold_convert_const_fixed_from_real): Likewise.
9542 (native_encode_fixed): Likewise
9543 (native_encode_complex): Likewise
9544 (native_encode_vector): Likewise.
9545 (native_interpret_fixed): Likewise.
9546 (native_interpret_real): Likewise.
9547 (native_interpret_complex): Likewise.
9548 (native_interpret_vector): Likewise.
9549 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
9550 (simd_clone_adjust_argument_types): Likewise.
9551 (simd_clone_init_simd_arrays): Likewise.
9552 (simd_clone_adjust): Likewise.
9553 * stor-layout.c (layout_type): Likewise.
9554 * tree.c (build_minus_one_cst): Likewise.
9555 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
9556 * tree-inline.c (estimate_move_cost): Likewise.
9557 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
9558 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
9559 (vectorizable_reduction): Likewise.
9560 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
9561 (vect_recog_mixed_size_cond_pattern): Likewise.
9562 (check_bool_pattern): Likewise.
9563 (adjust_bool_pattern): Likewise.
9564 (search_type_for_mask_1): Likewise.
9565 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
9566 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
9567 (vectorizable_load): Likewise.
9568 (vectorizable_store): Likewise.
9569 * ubsan.c (ubsan_encode_value): Likewise.
9570 * varasm.c (output_constant): Likewise.
9572 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9573 Alan Hayward <alan.hayward@arm.com>
9574 David Sherwood <david.sherwood@arm.com>
9576 * coretypes.h (scalar_mode): New class.
9577 * machmode.h (scalar_mode): Likewise.
9578 (scalar_mode::includes_p): New function.
9579 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
9580 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
9581 * genmodes.c (get_mode_class): Handle remaining scalar modes.
9582 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
9583 * expmed.c (store_bit_field_1): Likewise.
9584 (extract_bit_field_1): Likewise.
9585 * expr.c (write_complex_part): Likewise.
9586 (read_complex_part): Likewise.
9587 (emit_move_complex_push): Likewise.
9588 (expand_expr_real_2): Likewise.
9589 * function.c (assign_parm_setup_reg): Likewise.
9590 (assign_parms_unsplit_complex): Likewise.
9591 * optabs.c (expand_binop): Likewise.
9592 * rtlanal.c (subreg_get_info): Likewise.
9593 * simplify-rtx.c (simplify_immed_subreg): Likewise.
9594 * varasm.c (output_constant_pool_2): Likewise.
9596 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9597 Alan Hayward <alan.hayward@arm.com>
9598 David Sherwood <david.sherwood@arm.com>
9600 * expmed.c (extract_high_half): Use scalar_int_mode and remove
9602 (expmed_mult_highpart_optab): Likewise.
9603 (expmed_mult_highpart): Likewise.
9605 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9606 Alan Hayward <alan.hayward@arm.com>
9607 David Sherwood <david.sherwood@arm.com>
9609 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
9610 instead of a machine_mode.
9611 (builtin_memset_read_str): Likewise.
9612 * builtins.c (c_readstr): Likewise.
9613 (builtin_memcpy_read_str): Likewise.
9614 (builtin_strncpy_read_str): Likewise.
9615 (builtin_memset_read_str): Likewise.
9616 (builtin_memset_gen_str): Likewise.
9617 (expand_builtin_signbit): Use scalar_int_mode for local variables.
9618 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
9619 instead of a machine_mode.
9620 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
9622 (make_extraction): Likewise.
9623 (try_widen_shift_mode): Take and return scalar_int_modes instead
9625 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
9626 a scalar_int_mode instead of a machine_mode.
9627 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
9628 (avr_addr_space_pointer_mode): Likewise.
9629 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
9630 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
9631 (msp430_unwind_word_mode): Likewise.
9632 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
9633 (spu_addr_space_pointer_mode): Likewise.
9634 (spu_addr_space_address_mode): Likewise.
9635 (spu_libgcc_cmp_return_mode): Likewise.
9636 (spu_libgcc_shift_count_mode): Likewise.
9637 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
9638 (rl78_addr_space_pointer_mode): Likewise.
9639 (fl78_unwind_word_mode): Likewise.
9640 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
9642 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
9643 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
9644 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
9645 (mips_valid_pointer_mode): Likewise.
9646 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
9647 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
9648 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
9650 (ft32_addr_space_address_mode): Likewise.
9651 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
9652 scalar_int_mode instead of a machine_mode.
9653 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
9655 (m32c_addr_space_address_mode): Likewise.
9656 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
9657 (rs6000_eh_return_filter_mode): Likewise.
9658 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
9659 (rs6000_eh_return_filter_mode): Likewise.
9660 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
9661 (s390_libgcc_shift_count_mode): Likewise.
9662 (s390_unwind_word_mode): Likewise.
9663 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
9665 * target.def (mode_rep_extended): Likewise.
9666 (valid_pointer_mode): Likewise.
9667 (addr_space.valid_pointer_mode): Likewise.
9668 (eh_return_filter_mode): Return a scalar_int_mode rather than
9670 (libgcc_cmp_return_mode): Likewise.
9671 (libgcc_shift_count_mode): Likewise.
9672 (unwind_word_mode): Likewise.
9673 (addr_space.pointer_mode): Likewise.
9674 (addr_space.address_mode): Likewise.
9675 * doc/tm.texi: Regenerate.
9676 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
9678 (do_jump): Use scalar_int_mode for local variables.
9679 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
9680 rather than a machine_mode.
9681 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
9682 (scompare_loc_descriptor_wide): Likewise.
9683 (scompare_loc_descriptor_narrow): Likewise.
9684 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
9686 * except.c (sjlj_emit_dispatch_table): Likewise.
9687 (expand_builtin_eh_copy_values): Likewise.
9688 * explow.c (convert_memory_address_addr_space_1): Likewise.
9689 Take a scalar_int_mode rather than a machine_mode.
9690 (convert_memory_address_addr_space): Take a scalar_int_mode rather
9691 than a machine_mode.
9692 (memory_address_addr_space): Use scalar_int_mode for local variables.
9693 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
9694 rather than a machine_mode.
9695 * expmed.c (mask_rtx): Likewise.
9696 (init_expmed_one_conv): Likewise.
9697 (expand_mult_highpart_adjust): Likewise.
9698 (extract_high_half): Likewise.
9699 (expmed_mult_highpart_optab): Likewise.
9700 (expmed_mult_highpart): Likewise.
9701 (expand_smod_pow2): Likewise.
9702 (expand_sdiv_pow2): Likewise.
9703 (emit_store_flag_int): Likewise.
9704 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
9706 (extract_low_bits): Likewise.
9707 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
9709 * expr.c (pieces_addr::adjust): Likewise.
9710 (can_store_by_pieces): Likewise.
9711 (store_by_pieces): Likewise.
9712 (clear_by_pieces_1): Likewise.
9713 (expand_expr_addr_expr_1): Likewise.
9714 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
9715 (expand_expr_real_1): Likewise.
9716 (try_casesi): Likewise.
9717 * final.c (shorten_branches): Likewise.
9718 * fold-const.c (fold_convert_const_int_from_fixed): Change the
9719 type of "mode" to machine_mode.
9720 * internal-fn.c (expand_arith_overflow_result_store): Take a
9721 scalar_int_mode rather than a machine_mode.
9722 (expand_mul_overflow): Use scalar_int_mode for local variables.
9723 * loop-doloop.c (doloop_modify): Likewise.
9724 (doloop_optimize): Likewise.
9725 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
9726 than a machine_mode.
9727 (expand_doubleword_shift_condmove): Likewise.
9728 (expand_doubleword_shift): Likewise.
9729 (expand_doubleword_clz): Likewise.
9730 (expand_doubleword_popcount): Likewise.
9731 (expand_doubleword_parity): Likewise.
9732 (expand_absneg_bit): Use scalar_int_mode for local variables.
9733 (prepare_float_lib_cmp): Likewise.
9734 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
9735 rather than a machine_mode.
9736 (convert_memory_address_addr_space): Likewise.
9737 (get_mode_bounds): Likewise.
9738 (get_address_mode): Return a scalar_int_mode rather than a
9740 * rtlanal.c (get_address_mode): Likewise.
9741 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
9742 than a machine_mode.
9743 * targhooks.c (default_mode_rep_extended): Likewise.
9744 (default_valid_pointer_mode): Likewise.
9745 (default_addr_space_valid_pointer_mode): Likewise.
9746 (default_eh_return_filter_mode): Return a scalar_int_mode rather
9747 than a machine_mode.
9748 (default_libgcc_cmp_return_mode): Likewise.
9749 (default_libgcc_shift_count_mode): Likewise.
9750 (default_unwind_word_mode): Likewise.
9751 (default_addr_space_pointer_mode): Likewise.
9752 (default_addr_space_address_mode): Likewise.
9753 * targhooks.h (default_eh_return_filter_mode): Likewise.
9754 (default_libgcc_cmp_return_mode): Likewise.
9755 (default_libgcc_shift_count_mode): Likewise.
9756 (default_unwind_word_mode): Likewise.
9757 (default_addr_space_pointer_mode): Likewise.
9758 (default_addr_space_address_mode): Likewise.
9759 (default_mode_rep_extended): Take a scalar_int_mode rather than
9761 (default_valid_pointer_mode): Likewise.
9762 (default_addr_space_valid_pointer_mode): Likewise.
9763 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
9765 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
9766 rather than a machine_mode.
9767 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
9768 for local variables.
9769 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
9770 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
9771 than a machine_mode.
9773 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9774 Alan Hayward <alan.hayward@arm.com>
9775 David Sherwood <david.sherwood@arm.com>
9777 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
9778 the mode argument to scalar_int_mode.
9779 (do_jump_by_parts_zero_rtx): Likewise.
9780 (do_jump_by_parts_equality_rtx): Likewise.
9781 (do_jump_by_parts_greater): Take a mode argument.
9782 (do_jump_by_parts_equality): Likewise.
9783 (do_jump_1): Update calls accordingly.
9785 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9786 Alan Hayward <alan.hayward@arm.com>
9787 David Sherwood <david.sherwood@arm.com>
9789 * is-a.h (safe_dyn_cast): New function.
9790 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
9791 (jump_table_for_label): Likewise.
9792 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
9793 instead of an rtx_insn *.
9794 (shorten_branches): Use dyn_cast instead of LABEL_P and
9795 JUMP_TABLE_DATA_P. Use jump_table_for_label and
9796 rtx_jump_table_data::get_data_mode.
9797 (final_scan_insn): Likewise.
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 * combine.c (try_combine): Use is_a <scalar_int_mode> when
9804 trying to combine a full-register integer set with a subreg
9807 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9808 Alan Hayward <alan.hayward@arm.com>
9809 David Sherwood <david.sherwood@arm.com>
9811 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
9812 that is always either address_mode or pointer_mode.
9814 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9815 Alan Hayward <alan.hayward@arm.com>
9816 David Sherwood <david.sherwood@arm.com>
9818 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
9819 when the two are known to be equal.
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 * simplify-rtx.c (simplify_const_unary_operation): Use
9826 is_a <scalar_int_mode> instead of checking for a nonzero
9827 precision. Forcibly convert op_mode to a scalar_int_mode
9828 in that case. More clearly differentiate the operand and
9829 result modes and use the former when deciding what the value
9830 of a count-bits operation should be. Use is_int_mode instead
9831 of checking for a MODE_INT. Remove redundant check for whether
9832 this mode has a zero precision.
9834 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9835 Alan Hayward <alan.hayward@arm.com>
9836 David Sherwood <david.sherwood@arm.com>
9838 * optabs.c (widen_leading): Change the type of the mode argument
9839 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
9840 (widen_bswap): Likewise.
9841 (expand_parity): Likewise.
9842 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
9843 (expand_ffs): Likewise.
9844 (epand_unop): Check for scalar integer modes before calling the
9847 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9848 Alan Hayward <alan.hayward@arm.com>
9849 David Sherwood <david.sherwood@arm.com>
9851 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
9853 (expand_expr_real_1): Update call accordingly.
9855 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9856 Alan Hayward <alan.hayward@arm.com>
9857 David Sherwood <david.sherwood@arm.com>
9859 * expmed.c (store_bit_field_using_insv): Add op0_mode and
9860 value_mode arguments. Use scalar_int_mode internally.
9861 (store_bit_field_1): Rename the new integer mode from imode
9862 to op0_mode and use it instead of GET_MODE (op0). Update calls
9863 to store_split_bit_field, store_bit_field_using_insv and
9864 store_fixed_bit_field.
9865 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
9866 Use scalar_int_mode internally. Use a bit count rather than a mode
9867 when calculating the largest bit size for get_best_mode.
9868 Update calls to store_split_bit_field and store_fixed_bit_field_1.
9869 (store_fixed_bit_field_1): Add mode and value_mode arguments.
9870 Remove assertion that OP0 has a scalar integer mode.
9871 (store_split_bit_field): Add op0_mode and value_mode arguments.
9872 Update calls to extract_fixed_bit_field.
9873 (extract_bit_field_using_extv): Add an op0_mode argument.
9874 Use scalar_int_mode internally.
9875 (extract_bit_field_1): Rename the new integer mode from imode to
9876 op0_mode and use it instead of GET_MODE (op0). Update calls to
9877 extract_split_bit_field, extract_bit_field_using_extv and
9878 extract_fixed_bit_field.
9879 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
9880 to extract_split_bit_field and extract_fixed_bit_field_1.
9881 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
9882 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
9884 (extract_split_bit_field): Add an op0_mode argument. Update call
9885 to extract_fixed_bit_field.
9887 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9888 Alan Hayward <alan.hayward@arm.com>
9889 David Sherwood <david.sherwood@arm.com>
9891 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
9892 * explow.c (hard_function_value): Likewise.
9893 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
9894 convert_to_mode call outside the loop.
9895 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
9896 for the mode iterator. Require the mode specified by max_pieces
9898 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
9900 (copy_blkmode_to_reg): Likewise.
9901 (set_storage_via_setmem): Likewise.
9902 * optabs.c (prepare_cmp_insn): Likewise.
9903 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
9904 * stor-layout.c (finish_bitfield_representative): Likewise.
9906 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9907 Alan Hayward <alan.hayward@arm.com>
9908 David Sherwood <david.sherwood@arm.com>
9910 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
9911 * expr.c (convert_move): Use them.
9912 (convert_modes): Likewise.
9913 (store_expr_with_bounds): Likewise.
9915 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9916 Alan Hayward <alan.hayward@arm.com>
9917 David Sherwood <david.sherwood@arm.com>
9919 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
9920 parameter for the mode of "x". Remove the "known_x", "known_mode"
9921 and "known_ret" arguments. Change the type of the mode argument
9923 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
9924 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
9925 (reg_num_sign_bit_copies_for_combine): Likewise.
9926 * rtlanal.c (nonzero_bits1): Likewise.
9927 (num_sign_bit_copies1): Likewise.
9928 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
9929 (reg_num_sign_bit_copies_general): Likewise.
9930 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
9931 (reg_nonzero_bits_general): Likewise.
9933 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9934 Alan Hayward <alan.hayward@arm.com>
9935 David Sherwood <david.sherwood@arm.com>
9937 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
9938 than in subroutines. Return 1 for non-integer modes.
9939 (cached_num_sign_bit_copies): Change the type of the mode parameter
9941 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
9942 classes. Handle CONST_INT_P first and then check whether X also
9943 has a scalar integer mode. Check the same thing for inner registers
9944 of a SUBREG and for values that are being extended or truncated.
9946 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9947 Alan Hayward <alan.hayward@arm.com>
9948 David Sherwood <david.sherwood@arm.com>
9950 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
9951 in subroutines. Return the mode mask for non-integer modes.
9952 (cached_nonzero_bits): Change the type of the mode parameter
9954 (nonzero_bits1): Likewise. Remove early exit for other mode
9955 classes. Handle CONST_INT_P first and then check whether X
9956 also has a scalar integer mode.
9958 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9959 Alan Hayward <alan.hayward@arm.com>
9960 David Sherwood <david.sherwood@arm.com>
9962 * expr.c (widest_int_mode_for_size): Make the comment match the code.
9963 Return a scalar_int_mode and assert that the size is greater than
9965 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
9966 (op_by_pieces_d::op_by_pieces_d): Likewise.
9967 (op_by_pieces_d::run): Likewise.
9968 (can_store_by_pieces): Likewise.
9970 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9971 Alan Hayward <alan.hayward@arm.com>
9972 David Sherwood <david.sherwood@arm.com>
9974 * combine.c (extract_left_shift): Add a mode argument and update
9976 (make_compound_operation_int): Change the type of the mode parameter
9977 to scalar_int_mode and update the call to extract_left_shift.
9979 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9980 Alan Hayward <alan.hayward@arm.com>
9981 David Sherwood <david.sherwood@arm.com>
9983 * combine.c (simplify_and_const_int): Change the type of the mode
9984 parameter to scalar_int_mode.
9985 (simplify_and_const_int_1): Likewise. Update recursive call.
9987 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9988 Alan Hayward <alan.hayward@arm.com>
9989 David Sherwood <david.sherwood@arm.com>
9991 * combine.c (simplify_compare_const): Check that the mode is a
9992 scalar_int_mode (rather than VOIDmode) before testing its
9994 (simplify_comparison): Move COMPARISON_P handling out of the
9995 loop and restrict the latter part of the loop to scalar_int_modes.
9996 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
9997 and when considering SUBREG_REGs. Use is_int_mode instead of
9998 checking GET_MODE_CLASS against MODE_INT.
10000 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10001 Alan Hayward <alan.hayward@arm.com>
10002 David Sherwood <david.sherwood@arm.com>
10004 * combine.c (try_widen_shift_mode): Move check for equal modes to...
10005 (simplify_shift_const_1): ...here. Use scalar_int_mode for
10006 shift_unit_mode and for modes involved in scalar shifts.
10008 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10009 Alan Hayward <alan.hayward@arm.com>
10010 David Sherwood <david.sherwood@arm.com>
10012 * combine.c (force_int_to_mode): New function, split out from...
10013 (force_to_mode): ...here. Keep xmode up-to-date and use it
10014 instead of GET_MODE (x).
10016 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10017 Alan Hayward <alan.hayward@arm.com>
10018 David Sherwood <david.sherwood@arm.com>
10020 * optabs-query.h (extraction_insn::struct_mode): Change type to
10021 opt_scalar_int_mode and update comment.
10022 (extraction_insn::field_mode): Change type to scalar_int_mode.
10023 (extraction_insn::pos_mode): Likewise.
10024 * combine.c (make_extraction): Update accordingly.
10025 * optabs-query.c (get_traditional_extraction_insn): Likewise.
10026 (get_optab_extraction_insn): Likewise.
10027 * recog.c (simplify_while_replacing): Likewise.
10028 * expmed.c (narrow_bit_field_mem): Change the type of the mode
10029 parameter to opt_scalar_int_mode.
10031 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10032 Alan Hayward <alan.hayward@arm.com>
10033 David Sherwood <david.sherwood@arm.com>
10035 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
10036 to a scalar_int_mode instead of a machine_mode.
10037 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
10038 (get_best_mode): Return a boolean and use a pointer argument to store
10039 the selected mode. Replace the limit mode parameter with a bit limit.
10040 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
10041 for the values returned by bit_field_mode_iterator::next_mode.
10042 (store_bit_field): Update call to get_best_mode.
10043 (store_fixed_bit_field): Likewise.
10044 (extract_fixed_bit_field): Likewise.
10045 * expr.c (optimize_bitfield_assignment_op): Likewise.
10046 * fold-const.c (optimize_bit_field_compare): Likewise.
10047 (fold_truth_andor_1): Likewise.
10048 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
10049 Update for new type of m_mode.
10050 (get_best_mode): As above.
10052 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10053 Alan Hayward <alan.hayward@arm.com>
10054 David Sherwood <david.sherwood@arm.com>
10056 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
10057 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
10058 (store_bit_field): Check is_a <scalar_int_mode> before calling
10059 strict_volatile_bitfield_p.
10060 (extract_bit_field): Likewise.
10062 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10063 Alan Hayward <alan.hayward@arm.com>
10064 David Sherwood <david.sherwood@arm.com>
10066 * target.def (cstore_mode): Return a scalar_int_mode.
10067 * doc/tm.texi: Regenerate.
10068 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
10069 * targhooks.h (default_cstore_mode): Likewise.
10070 * targhooks.c (default_cstore_mode): Likewise, using a forced
10072 * expmed.c (emit_cstore): Expect the target of the cstore to be
10075 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10076 Alan Hayward <alan.hayward@arm.com>
10077 David Sherwood <david.sherwood@arm.com>
10079 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
10081 (niter_desc): Likewise mode.
10082 (iv_analyze): Add a mode parameter.
10084 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
10085 and change its type to scalar_int_mode.
10086 * loop-iv.c: Update commentary at head of file.
10087 (iv_constant): Pass the mode paraeter before the rtx it describes
10088 and change its type to scalar_int_mode. Remove VOIDmode handling.
10089 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
10090 (iv_extend): Likewise.
10091 (shorten_into_mode): Likewise.
10092 (iv_add): Use scalar_int_mode.
10093 (iv_mult): Likewise.
10094 (iv_shift): Likewise.
10095 (canonicalize_iv_subregs): Likewise.
10096 (get_biv_step_1): Pass the outer_mode parameter before the rtx
10097 it describes and change its mode to scalar_int_mode. Also change
10098 the type of the returned inner_mode to scalar_int_mode.
10099 (get_biv_step): Likewise, turning outer_mode from a pointer
10100 into a direct parameter. Update call to get_biv_step_1.
10101 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
10102 iv_constant and get_biv_step.
10103 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
10104 and change its type to scalar_int_mode. Don't initialise iv->mode
10105 to VOIDmode and remove later checks for its still being VOIDmode.
10106 Update calls to iv_analyze_op and iv_analyze_expr. Check
10107 is_a <scalar_int_mode> when changing the mode under consideration.
10108 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
10109 Update call to iv_analyze_expr.
10110 (iv_analyze_op): Add a mode parameter. Reject subregs whose
10111 inner register is not also a scalar_int_mode. Update call to
10113 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
10114 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
10115 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
10116 separate mode class checks. Update calls to iv_analyze. Remove
10117 fix-up of VOIDmodes after iv_analyze_biv.
10118 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
10119 don't have a scalar_int_mode. Update call to biv_p.
10121 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10122 Alan Hayward <alan.hayward@arm.com>
10123 David Sherwood <david.sherwood@arm.com>
10125 * cfgexpand.c (convert_debug_memory_address): Use
10126 as_a <scalar_int_mode>.
10127 * combine.c (expand_compound_operation): Likewise.
10128 (make_extraction): Likewise.
10129 (change_zero_ext): Likewise.
10130 (simplify_comparison): Likewise.
10131 * cse.c (cse_insn): Likewise.
10132 * dwarf2out.c (minmax_loc_descriptor): Likewise.
10133 (mem_loc_descriptor): Likewise.
10134 (loc_descriptor): Likewise.
10135 * expmed.c (init_expmed_one_mode): Likewise.
10136 (synth_mult): Likewise.
10137 (emit_store_flag_1): Likewise.
10138 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
10139 of a comparison with size.
10140 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
10141 (reduce_to_bit_field_precision): Likewise.
10142 * function.c (expand_function_end): Likewise.
10143 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
10144 * loop-doloop.c (doloop_modify): Likewise.
10145 * optabs.c (expand_binop): Likewise.
10146 (expand_unop): Likewise.
10147 (expand_copysign_absneg): Likewise.
10148 (prepare_cmp_insn): Likewise.
10149 (maybe_legitimize_operand): Likewise.
10150 * recog.c (const_scalar_int_operand): Likewise.
10151 * rtlanal.c (get_address_mode): Likewise.
10152 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10153 (simplify_cond_clz_ctz): Likewise.
10154 * tree-nested.c (get_nl_goto_field): Likewise.
10155 * tree.c (build_vector_type_for_mode): Likewise.
10156 * var-tracking.c (use_narrower_mode): Likewise.
10158 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10159 Alan Hayward <alan.hayward@arm.com>
10160 David Sherwood <david.sherwood@arm.com>
10162 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
10163 * builtins.c (expand_builtin_signbit): Use it.
10164 * cfgexpand.c (expand_debug_expr): Likewise.
10165 * dojump.c (do_jump): Likewise.
10166 (do_compare_and_jump): Likewise.
10167 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
10168 * expmed.c (make_tree): Likewise.
10169 * expr.c (expand_expr_real_2): Likewise.
10170 (expand_expr_real_1): Likewise.
10171 (try_casesi): Likewise.
10172 * fold-const-call.c (fold_const_call_ss): Likewise.
10173 * fold-const.c (unextend): Likewise.
10174 (extract_muldiv_1): Likewise.
10175 (fold_single_bit_test): Likewise.
10176 (native_encode_int): Likewise.
10177 (native_encode_string): Likewise.
10178 (native_interpret_int): Likewise.
10179 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
10180 * internal-fn.c (expand_addsub_overflow): Likewise.
10181 (expand_neg_overflow): Likewise.
10182 (expand_mul_overflow): Likewise.
10183 (expand_arith_overflow): Likewise.
10184 * match.pd: Likewise.
10185 * stor-layout.c (layout_type): Likewise.
10186 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10187 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
10188 * tree-ssanames.c (get_range_info): Likewise.
10189 * tree-switch-conversion.c (array_value_type) Likewise.
10190 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
10191 (vect_recog_divmod_pattern): Likewise.
10192 (vect_recog_mixed_size_cond_pattern): Likewise.
10193 * tree-vrp.c (extract_range_basic): Likewise.
10194 (simplify_float_conversion_using_ranges): Likewise.
10195 * tree.c (int_fits_type_p): Likewise.
10196 * ubsan.c (instrument_bool_enum_load): Likewise.
10197 * varasm.c (mergeable_string_section): Likewise.
10198 (narrowing_initializer_constant_valid_p): Likewise.
10199 (output_constant): Likewise.
10201 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10202 Alan Hayward <alan.hayward@arm.com>
10203 David Sherwood <david.sherwood@arm.com>
10205 * machmode.h (NARROWEST_INT_MODE): New macro.
10206 * expr.c (alignment_for_piecewise_move): Use it instead of
10207 GET_CLASS_NARROWEST_MODE (MODE_INT).
10208 (push_block): Likewise.
10209 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
10211 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10213 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10214 Alan Hayward <alan.hayward@arm.com>
10215 David Sherwood <david.sherwood@arm.com>
10217 * postreload.c (move2add_valid_value_p): Change the type of the
10218 mode parameter to scalar_int_mode.
10219 (move2add_use_add2_insn): Add a mode parameter and use it instead
10221 (move2add_use_add3_insn): Likewise.
10222 (reload_cse_move2add): Update accordingly.
10224 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10225 Alan Hayward <alan.hayward@arm.com>
10226 David Sherwood <david.sherwood@arm.com>
10228 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
10230 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
10231 * optabs.c (expand_unop): Likewise.
10233 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10234 Alan Hayward <alan.hayward@arm.com>
10235 David Sherwood <david.sherwood@arm.com>
10237 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
10238 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
10239 (popcount_loc_descriptor): Likewise.
10240 (bswap_loc_descriptor): Likewise.
10241 (rotate_loc_descriptor): Likewise.
10242 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
10243 calling the functions above.
10245 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10246 Alan Hayward <alan.hayward@arm.com>
10247 David Sherwood <david.sherwood@arm.com>
10249 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
10251 (try_combine): Likewise.
10252 (simplify_if_then_else): Likewise.
10253 * cse.c (cse_insn): Likewise.
10254 * dwarf2out.c (mem_loc_descriptor): Likewise.
10255 * emit-rtl.c (gen_lowpart_common): Likewise.
10256 * simplify-rtx.c (simplify_truncation): Likewise.
10257 (simplify_binary_operation_1): Likewise.
10258 (simplify_const_relational_operation): Likewise.
10259 (simplify_ternary_operation): Likewise.
10260 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
10262 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10263 Alan Hayward <alan.hayward@arm.com>
10264 David Sherwood <david.sherwood@arm.com>
10266 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
10267 * reload.c (push_reload): Likewise.
10268 (find_reloads): Likewise.
10270 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10271 Alan Hayward <alan.hayward@arm.com>
10272 David Sherwood <david.sherwood@arm.com>
10274 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
10275 (make_compound_operation_int): Likewise.
10276 (change_zero_ext): Likewise.
10277 * expr.c (convert_move): Likewise.
10278 (convert_modes): Likewise.
10279 * fwprop.c (forward_propagate_subreg): Likewise.
10280 * loop-iv.c (get_biv_step_1): Likewise.
10281 * optabs.c (widen_operand): Likewise.
10282 * postreload.c (move2add_valid_value_p): Likewise.
10283 * recog.c (simplify_while_replacing): Likewise.
10284 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10285 (simplify_binary_operation_1): Likewise. Remove redundant
10286 mode equality check.
10288 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10289 Alan Hayward <alan.hayward@arm.com>
10290 David Sherwood <david.sherwood@arm.com>
10292 * combine.c (combine_simplify_rtx): Add checks for
10293 is_a <scalar_int_mode>.
10294 (simplify_if_then_else): Likewise.
10295 (make_field_assignment): Likewise.
10296 (simplify_comparison): Likewise.
10297 * ifcvt.c (noce_try_bitop): Likewise.
10298 * loop-invariant.c (canonicalize_address_mult): Likewise.
10299 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10301 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10302 Alan Hayward <alan.hayward@arm.com>
10303 David Sherwood <david.sherwood@arm.com>
10305 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
10306 is_a <scalar_int_mode> instead of != BLKmode.
10308 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10309 Alan Hayward <alan.hayward@arm.com>
10310 David Sherwood <david.sherwood@arm.com>
10312 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
10313 instead of != VOIDmode.
10314 * combine.c (if_then_else_cond): Likewise.
10315 (change_zero_ext): Likewise.
10316 * dwarf2out.c (mem_loc_descriptor): Likewise.
10317 (loc_descriptor): Likewise.
10318 * rtlanal.c (canonicalize_condition): Likewise.
10319 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
10321 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10322 Alan Hayward <alan.hayward@arm.com>
10323 David Sherwood <david.sherwood@arm.com>
10325 * simplify-rtx.c (simplify_binary_operation_1): Use
10326 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
10328 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10329 Alan Hayward <alan.hayward@arm.com>
10330 David Sherwood <david.sherwood@arm.com>
10332 * wide-int.h (int_traits<unsigned char>) New class.
10333 (int_traits<unsigned short>) Likewise.
10334 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
10335 Use GET_MODE_UNIT_PRECISION and remove redundant test for
10337 * combine.c (set_nonzero_bits_and_sign_copies): Use
10338 is_a <scalar_int_mode>.
10339 (find_split_point): Likewise.
10340 (combine_simplify_rtx): Likewise.
10341 (simplify_logical): Likewise.
10342 (expand_compound_operation): Likewise.
10343 (expand_field_assignment): Likewise.
10344 (make_compound_operation): Likewise.
10345 (extended_count): Likewise.
10346 (change_zero_ext): Likewise.
10347 (simplify_comparison): Likewise.
10348 * dwarf2out.c (scompare_loc_descriptor): Likewise.
10349 (ucompare_loc_descriptor): Likewise.
10350 (minmax_loc_descriptor): Likewise.
10351 (mem_loc_descriptor): Likewise.
10352 (loc_descriptor): Likewise.
10353 * expmed.c (init_expmed_one_mode): Likewise.
10354 * lra-constraints.c (lra_constraint_offset): Likewise.
10355 * optabs.c (prepare_libcall_arg): Likewise.
10356 * postreload.c (move2add_note_store): Likewise.
10357 * reload.c (operands_match_p): Likewise.
10358 * rtl.h (load_extend_op): Likewise.
10359 * rtlhooks.c (gen_lowpart_general): Likewise.
10360 * simplify-rtx.c (simplify_truncation): Likewise.
10361 (simplify_unary_operation_1): Likewise.
10362 (simplify_binary_operation_1): Likewise.
10363 (simplify_const_binary_operation): Likewise.
10364 (simplify_const_relational_operation): Likewise.
10365 (simplify_subreg): Likewise.
10366 * stor-layout.c (bitwise_mode_for_mode): Likewise.
10367 * var-tracking.c (adjust_mems): Likewise.
10368 (prepare_call_arguments): Likewise.
10370 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10371 Alan Hayward <alan.hayward@arm.com>
10372 David Sherwood <david.sherwood@arm.com>
10374 * machmode.h (is_int_mode): New fuction.
10375 * combine.c (find_split_point): Use it.
10376 (combine_simplify_rtx): Likewise.
10377 (simplify_if_then_else): Likewise.
10378 (simplify_set): Likewise.
10379 (simplify_shift_const_1): Likewise.
10380 (simplify_comparison): Likewise.
10381 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
10382 * cse.c (notreg_cost): Likewise.
10383 (cse_insn): Likewise.
10384 * cselib.c (cselib_lookup_1): Likewise.
10385 * dojump.c (do_jump_1): Likewise.
10386 (do_compare_rtx_and_jump): Likewise.
10387 * dse.c (get_call_args): Likewise.
10388 * dwarf2out.c (rtl_for_decl_init): Likewise.
10389 (native_encode_initializer): Likewise.
10390 * expmed.c (emit_store_flag_1): Likewise.
10391 (emit_store_flag): Likewise.
10392 * expr.c (convert_modes): Likewise.
10393 (store_field): Likewise.
10394 (expand_expr_real_1): Likewise.
10395 * fold-const.c (fold_read_from_constant_string): Likewise.
10396 * gimple-ssa-sprintf.c (get_format_string): Likewise.
10397 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
10398 * optabs.c (expand_binop): Likewise.
10399 (expand_unop): Likewise.
10400 (expand_abs_nojump): Likewise.
10401 (expand_one_cmpl_abs_nojump): Likewise.
10402 * simplify-rtx.c (mode_signbit_p): Likewise.
10403 (val_signbit_p): Likewise.
10404 (val_signbit_known_set_p): Likewise.
10405 (val_signbit_known_clear_p): Likewise.
10406 (simplify_relational_operation_1): Likewise.
10407 * tree.c (vector_type_mode): Likewise.
10409 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10410 Alan Hayward <alan.hayward@arm.com>
10411 David Sherwood <david.sherwood@arm.com>
10413 * machmode.h (smallest_mode_for_size): Fix formatting.
10414 (smallest_int_mode_for_size): New function.
10415 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
10416 instead of smallest_mode_for_size.
10417 * combine.c (make_extraction): Likewise.
10418 * config/arc/arc.c (arc_expand_movmem): Likewise.
10419 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
10420 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
10421 * config/s390/s390.c (s390_expand_insv): Likewise.
10422 * config/sparc/sparc.c (assign_int_registers): Likewise.
10423 * config/spu/spu.c (spu_function_value): Likewise.
10424 (spu_function_arg): Likewise.
10425 * coverage.c (get_gcov_type): Likewise.
10426 (get_gcov_unsigned_t): Likewise.
10427 * dse.c (find_shift_sequence): Likewise.
10428 * expmed.c (store_bit_field_1): Likewise.
10429 * expr.c (convert_move): Likewise.
10430 (store_field): Likewise.
10431 * internal-fn.c (expand_arith_overflow): Likewise.
10432 * optabs-query.c (get_best_extraction_insn): Likewise.
10433 * optabs.c (expand_twoval_binop_libfunc): Likewise.
10434 * stor-layout.c (layout_type): Likewise.
10435 (initialize_sizetypes): Likewise.
10436 * targhooks.c (default_get_mask_mode): Likewise.
10437 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
10439 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10440 Alan Hayward <alan.hayward@arm.com>
10441 David Sherwood <david.sherwood@arm.com>
10443 * machmode.h (opt_mode::else_blk): New function.
10444 (int_mode_for_mode): Declare.
10445 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
10446 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
10448 * cfgexpand.c (expand_debug_expr): Likewise.
10449 * combine.c (gen_lowpart_or_truncate): Likewise.
10450 (gen_lowpart_for_combine): Likewise.
10451 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
10452 * config/avr/avr.c (avr_to_int_mode): Likewise.
10453 (avr_out_plus_1): Likewise.
10454 (avr_out_plus): Likewise.
10455 (avr_out_round): Likewise.
10456 * config/i386/i386.c (ix86_split_to_parts): Likewise.
10457 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
10458 (s390_expand_vcond): Likewise.
10459 * config/spu/spu.c (spu_split_immediate): Likewise.
10460 (spu_expand_mov): Likewise.
10461 * dse.c (get_stored_val): Likewise.
10462 * expmed.c (store_bit_field_1): Likewise.
10463 (convert_extracted_bit_field): Use int_mode_for_mode instead of
10465 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
10466 (extract_low_bits): Likewise.
10467 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
10468 handling rather than repeating the check.
10469 (emit_group_store): Likewise.
10470 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
10471 * optabs.c (expand_absneg_bit): Likewise.
10472 (expand_copysign_absneg): Likewise.
10473 (expand_copysign_bit): Likewise.
10474 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
10475 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
10476 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
10477 * var-tracking.c (prepare_call_arguments): Likewise.
10478 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
10479 int_mode_for_mode instead of mode_for_size.
10480 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
10482 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10483 Alan Hayward <alan.hayward@arm.com>
10484 David Sherwood <david.sherwood@arm.com>
10486 * machmode.h (int_mode_for_size): New function.
10487 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
10488 instead of mode_for_size.
10489 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
10491 * combine.c (expand_field_assignment): Use int_mode_for_size
10492 instead of mode_for_size.
10493 (make_extraction): Likewise.
10494 (simplify_shift_const_1): Likewise.
10495 (simplify_comparison): Likewise.
10496 * dojump.c (do_jump): Likewise.
10497 * dwarf2out.c (mem_loc_descriptor): Likewise.
10498 * emit-rtl.c (init_derived_machine_modes): Likewise.
10499 * expmed.c (flip_storage_order): Likewise.
10500 (convert_extracted_bit_field): Likewise.
10501 * expr.c (copy_blkmode_from_reg): Likewise.
10502 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
10503 * internal-fn.c (expand_mul_overflow): Likewise.
10504 * lower-subreg.c (simple_move): Likewise.
10505 * optabs-libfuncs.c (init_optabs): Likewise.
10506 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10507 * tree.c (vector_type_mode): Likewise.
10508 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
10509 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
10510 * tree-vect-generic.c (expand_vector_parallel): Likewise.
10511 * tree-vect-stmts.c (vectorizable_load): Likewise.
10512 (vectorizable_store): Likewise.
10514 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10515 Alan Hayward <alan.hayward@arm.com>
10516 David Sherwood <david.sherwood@arm.com>
10518 * coretypes.h (pod_mode): New type.
10519 (scalar_int_mode_pod): New typedef.
10520 * machmode.h (pod_mode): New class.
10521 (int_n_data_t::m): Change type to scalar_int_mode_pod.
10522 * genmodes.c (emit_mode_int_n): Update accordingly.
10523 * lower-subreg.h (target_lower_subreg): Change type to
10524 scalar_int_mode_pod.
10525 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
10526 scalar_int_mode_pod.
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 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
10533 machine_mode to scalar_int_mode.
10534 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
10535 (rs6000_option_override_internal): Remove cast to int.
10536 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
10537 machine_mode to scalar_int_mode.
10538 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
10539 (rs6000_option_override_internal): Remove cast to int.
10540 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
10541 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
10543 * config/s390/s390.c (s390_expand_builtin): Likewise.
10544 * coretypes.h (scalar_int_mode): New type.
10545 (opt_scalar_int_mode): New typedef.
10546 * machmode.h (scalar_int_mode): New class.
10547 (scalar_int_mode::includes_p): New function.
10548 (byte_mode): Change type to scalar_int_mode.
10549 (word_mode): Likewise.
10550 (ptr_mode): Likewise.
10551 * emit-rtl.c (byte_mode): Likewise.
10552 (word_mode): Likewise.
10553 (ptr_mode): Likewise.
10554 (init_derived_machine_modes): Update accordingly.
10555 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
10556 and MODE_PARTIAL_INT.
10557 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
10558 opt_scalar_int_mode.
10560 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10561 Alan Hayward <alan.hayward@arm.com>
10562 David Sherwood <david.sherwood@arm.com>
10564 * target.def (libgcc_floating_mode_supported_p): Take a
10566 * doc/tm.texi: Regenerate.
10567 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
10569 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
10570 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
10573 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10574 Alan Hayward <alan.hayward@arm.com>
10575 David Sherwood <david.sherwood@arm.com>
10577 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
10578 * doc/tm.texi: Regenerate.
10579 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
10580 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
10581 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
10582 * targhooks.h (default_floatn_mode): Likewise.
10583 * targhooks.c (default_floatn_mode): Likewise.
10584 * tree.c (build_common_tree_nodes): Update accordingly.
10586 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10587 Alan Hayward <alan.hayward@arm.com>
10588 David Sherwood <david.sherwood@arm.com>
10590 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
10591 (mode_iterator::iterate_p): Likewise.
10592 (mode_iterator::get_wider): Likewise.
10593 * expr.c (init_expr_target): Use opt_scalar_float_mode.
10595 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10596 Alan Hayward <alan.hayward@arm.com>
10597 David Sherwood <david.sherwood@arm.com>
10599 * coretypes.h (opt_scalar_float_mode): New typedef.
10600 * machmode.h (float_mode_for_size): New function.
10601 * emit-rtl.c (double_mode): Delete.
10602 (init_emit_once): Use float_mode_for_size.
10603 * stor-layout.c (layout_type): Likewise.
10604 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
10606 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10607 Alan Hayward <alan.hayward@arm.com>
10608 David Sherwood <david.sherwood@arm.com>
10610 * output.h (assemble_real): Take a scalar_float_mode.
10611 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
10612 * config/arm/arm.md (consttable_4): Likewise.
10613 (consttable_8): Likewise.
10614 (consttable_16): Likewise.
10615 * config/mips/mips.md (consttable_float): Likewise.
10616 * config/s390/s390.c (s390_output_pool_entry): Likewise.
10617 * varasm.c (assemble_real): Take a scalar_float_mode.
10618 (output_constant_pool_2): Update accordingly.
10619 (output_constant): Likewise.
10621 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10622 Alan Hayward <alan.hayward@arm.com>
10623 David Sherwood <david.sherwood@arm.com>
10625 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
10626 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
10627 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
10628 (native_encode_real): Likewise.
10629 (native_interpret_real): Likewise.
10630 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
10631 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10633 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10634 Alan Hayward <alan.hayward@arm.com>
10635 David Sherwood <david.sherwood@arm.com>
10637 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
10638 <scalar_float_mode>. Simplify.
10639 (gen_extend_conv_libfunc): Likewise.
10641 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10642 Alan Hayward <alan.hayward@arm.com>
10643 David Sherwood <david.sherwood@arm.com>
10645 * coretypes.h (scalar_float_mode): New type.
10646 * machmode.h (mode_traits::from_int): Use machine_mode if
10647 USE_ENUM_MODES is defined.
10648 (is_a): New function.
10650 (dyn_cast): Likewise.
10651 (scalar_float_mode): New class.
10652 (scalar_float_mode::includes_p): New function.
10653 (is_float_mode): Likewise.
10654 * gdbhooks.py (MachineModePrinter): New class.
10655 (build_pretty_printer): Use it for scalar_float_mode.
10656 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
10657 (format_helper::format_helper): Turn into a template.
10658 * genmodes.c (get_mode_class): New function.
10659 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
10660 or machine_mode if none.
10661 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
10662 as_a <scalar_float_mode>.
10663 * dwarf2out.c (mem_loc_descriptor): Likewise.
10664 (insert_float): Likewise.
10665 (add_const_value_attribute): Likewise.
10666 * simplify-rtx.c (simplify_immed_subreg): Likewise.
10667 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
10668 (expand_unop): Update accordingly.
10669 (expand_abs_nojump): Likewise.
10670 (expand_copysign_absneg): Take a scalar_float_mode.
10671 (expand_copysign_bit): Likewise.
10672 (expand_copysign): Update accordingly.
10674 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10675 Alan Hayward <alan.hayward@arm.com>
10676 David Sherwood <david.sherwood@arm.com>
10678 * coretypes.h (opt_mode): New class.
10679 * machmode.h (opt_mode): Likewise.
10680 (opt_mode::else_void): New function.
10681 (opt_mode::require): Likewise.
10682 (opt_mode::exists): Likewise.
10683 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
10684 (GET_MODE_2XWIDER_MODE): Likewise.
10685 (mode_iterator::get_wider): Update accordingly.
10686 (mode_iterator::get_2xwider): Likewise.
10687 (mode_iterator::get_known_wider): Likewise, turning into a template.
10688 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
10689 forcing a wider mode to exist.
10690 * config/cr16/cr16.h (LONG_REG_P): Likewise.
10691 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
10692 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
10693 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
10694 * lower-subreg.c (init_lower_subreg): Likewise.
10695 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
10696 on the final iteration.
10697 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
10698 a wider mode exists before asking for a move pattern.
10699 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
10700 forcing a wider mode to exist.
10701 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
10702 returning false if no such mode exists.
10703 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
10704 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
10705 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
10706 Avoid checking for a MODE_INT if we already know the mode is not a
10708 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
10709 forcing a wider mode to exist.
10710 (expmed_mult_highpart_optab): Likewise.
10711 (expmed_mult_highpart): Likewise.
10712 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
10714 * lto-streamer-in.c (lto_input_mode_table): Likewise.
10715 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
10716 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
10717 * internal-fn.c (expand_mul_overflow): Update use of
10718 GET_MODE_2XWIDER_MODE.
10719 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10720 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
10721 GET_MODE_WIDER_MODE.
10722 (convert_plusminus_to_widen): Likewise.
10723 * tree-switch-conversion.c (array_value_type): Likewise.
10724 * var-tracking.c (emit_note_insn_var_location): Likewise.
10725 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10726 Return false inside rather than outside the loop if no wider mode
10728 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
10729 and GET_MODE_2XWIDER_MODE
10730 (can_compare_p): Use else_void.
10731 * gdbhooks.py (OptMachineModePrinter): New class.
10732 (build_pretty_printer): Use it for opt_mode.
10734 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10735 Alan Hayward <alan.hayward@arm.com>
10736 David Sherwood <david.sherwood@arm.com>
10738 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
10739 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
10741 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10742 Alan Hayward <alan.hayward@arm.com>
10743 David Sherwood <david.sherwood@arm.com>
10745 * machmode.h (mode_traits): New structure.
10746 (get_narrowest_mode): New function.
10747 (mode_iterator::start): Likewise.
10748 (mode_iterator::iterate_p): Likewise.
10749 (mode_iterator::get_wider): Likewise.
10750 (mode_iterator::get_known_wider): Likewise.
10751 (mode_iterator::get_2xwider): Likewise.
10752 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
10753 (FOR_EACH_MODE): Likewise.
10754 (FOR_EACH_MODE_FROM): Likewise.
10755 (FOR_EACH_MODE_UNTIL): Likewise.
10756 (FOR_EACH_WIDER_MODE): Likewise.
10757 (FOR_EACH_2XWIDER_MODE): Likewise.
10758 * builtins.c (expand_builtin_strlen): Use new mode iterators.
10759 * combine.c (simplify_comparison): Likewise
10760 * config/i386/i386.c (type_natural_mode): Likewise.
10761 * cse.c (cse_insn): Likewise.
10762 * dse.c (find_shift_sequence): Likewise.
10763 * emit-rtl.c (init_derived_machine_modes): Likewise.
10764 (init_emit_once): Likewise.
10765 * explow.c (hard_function_value): Likewise.
10766 * expmed.c (extract_fixed_bit_field_1): Likewise.
10767 (extract_bit_field_1): Likewise.
10768 (expand_divmod): Likewise.
10769 (emit_store_flag_1): Likewise.
10770 * expr.c (init_expr_target): Likewise.
10771 (convert_move): Likewise.
10772 (alignment_for_piecewise_move): Likewise.
10773 (widest_int_mode_for_size): Likewise.
10774 (emit_block_move_via_movmem): Likewise.
10775 (copy_blkmode_to_reg): Likewise.
10776 (set_storage_via_setmem): Likewise.
10777 (compress_float_constant): Likewise.
10778 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10779 * optabs-query.c (get_best_extraction_insn): Likewise.
10780 * optabs.c (expand_binop): Likewise.
10781 (expand_twoval_unop): Likewise.
10782 (expand_twoval_binop): Likewise.
10783 (widen_leading): Likewise.
10784 (widen_bswap): Likewise.
10785 (expand_parity): Likewise.
10786 (expand_unop): Likewise.
10787 (prepare_cmp_insn): Likewise.
10788 (prepare_float_lib_cmp): Likewise.
10789 (expand_float): Likewise.
10790 (expand_fix): Likewise.
10791 (expand_sfix_optab): Likewise.
10792 * postreload.c (move2add_use_add2_insn): Likewise.
10793 * reg-stack.c (reg_to_stack): Likewise.
10794 * reginfo.c (choose_hard_reg_mode): Likewise.
10795 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
10796 * stor-layout.c (mode_for_size): Likewise.
10797 (smallest_mode_for_size): Likewise.
10798 (mode_for_vector): Likewise.
10799 (finish_bitfield_representative): Likewise.
10800 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
10801 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
10802 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10803 * var-tracking.c (prepare_call_arguments): Likewise.
10805 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10806 Alan Hayward <alan.hayward@arm.com>
10807 David Sherwood <david.sherwood@arm.com>
10809 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
10810 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
10811 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
10812 * machmode.h (mode_size): Move earlier in file.
10813 (mode_precision): Likewise.
10814 (mode_inner): Likewise.
10815 (mode_nunits): Likewise.
10816 (mode_unit_size): Likewise.
10817 (unit_unit_precision): Likewise.
10818 (mode_wider): Likewise.
10819 (mode_2xwider): Likewise.
10820 (machine_mode): New class.
10821 (mode_to_bytes): New function.
10822 (mode_to_bits): Likewise.
10823 (mode_to_precision): Likewise.
10824 (mode_to_inner): Likewise.
10825 (mode_to_unit_size): Likewise.
10826 (mode_to_unit_precision): Likewise.
10827 (mode_to_nunits): Likewise.
10828 (GET_MODE_SIZE): Use mode_to_bytes.
10829 (GET_MODE_BITSIZE): Use mode_to_bits.
10830 (GET_MODE_PRECISION): Use mode_to_precision.
10831 (GET_MODE_INNER): Use mode_to_inner.
10832 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
10833 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
10834 (GET_MODE_NUNITS): Use mode_to_nunits.
10835 * system.h (ALWAYS_INLINE): New macro.
10836 * config/powerpcspe/powerpcspe-c.c
10837 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
10838 int for arg1_mode and arg2_mode.
10840 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10841 Alan Hayward <alan.hayward@arm.com>
10842 David Sherwood <david.sherwood@arm.com>
10844 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
10845 Prefix mode names with E_ in case statements.
10846 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10847 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
10848 (aarch64_split_simd_move): Likewise.
10849 (aarch64_gen_storewb_pair): Likewise.
10850 (aarch64_gen_loadwb_pair): Likewise.
10851 (aarch64_gen_store_pair): Likewise.
10852 (aarch64_gen_load_pair): Likewise.
10853 (aarch64_get_condition_code_1): Likewise.
10854 (aarch64_constant_pool_reload_icode): Likewise.
10855 (get_rsqrte_type): Likewise.
10856 (get_rsqrts_type): Likewise.
10857 (get_recpe_type): Likewise.
10858 (get_recps_type): Likewise.
10859 (aarch64_gimplify_va_arg_expr): Likewise.
10860 (aarch64_simd_container_mode): Likewise.
10861 (aarch64_emit_load_exclusive): Likewise.
10862 (aarch64_emit_store_exclusive): Likewise.
10863 (aarch64_expand_compare_and_swap): Likewise.
10864 (aarch64_gen_atomic_cas): Likewise.
10865 (aarch64_emit_bic): Likewise.
10866 (aarch64_emit_atomic_swap): Likewise.
10867 (aarch64_emit_atomic_load_op): Likewise.
10868 (aarch64_evpc_trn): Likewise.
10869 (aarch64_evpc_uzp): Likewise.
10870 (aarch64_evpc_zip): Likewise.
10871 (aarch64_evpc_ext): Likewise.
10872 (aarch64_evpc_rev): Likewise.
10873 (aarch64_evpc_dup): Likewise.
10874 (aarch64_gen_ccmp_first): Likewise.
10875 (aarch64_gen_ccmp_next): Likewise.
10876 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
10877 (alpha_emit_xfloating_libcall): Likewise.
10878 (emit_insxl): Likewise.
10879 (alpha_arg_type): Likewise.
10880 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
10881 (arc_preferred_simd_mode): Likewise.
10882 (arc_secondary_reload): Likewise.
10883 (get_arc_condition_code): Likewise.
10884 (arc_print_operand): Likewise.
10885 (arc_legitimate_constant_p): Likewise.
10886 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10887 * config/arc/arc.md (casesi_load): Likewise.
10888 (casesi_compact_jump): Likewise.
10889 * config/arc/predicates.md (proper_comparison_operator): Likewise.
10890 (cc_use_register): Likewise.
10891 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10892 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
10893 (arm_init_iwmmxt_builtins): Likewise.
10894 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
10895 (neon_expand_vector_init): Likewise.
10896 (arm_attr_length_move_neon): Likewise.
10897 (maybe_get_arm_condition_code): Likewise.
10898 (arm_emit_vector_const): Likewise.
10899 (arm_preferred_simd_mode): Likewise.
10900 (arm_output_iwmmxt_tinsr): Likewise.
10901 (thumb1_output_casesi): Likewise.
10902 (thumb2_output_casesi): Likewise.
10903 (arm_emit_load_exclusive): Likewise.
10904 (arm_emit_store_exclusive): Likewise.
10905 (arm_expand_compare_and_swap): Likewise.
10906 (arm_evpc_neon_vuzp): Likewise.
10907 (arm_evpc_neon_vzip): Likewise.
10908 (arm_evpc_neon_vrev): Likewise.
10909 (arm_evpc_neon_vtrn): Likewise.
10910 (arm_evpc_neon_vext): Likewise.
10911 (arm_validize_comparison): Likewise.
10912 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
10913 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
10914 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
10915 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
10916 (c6x_preferred_simd_mode): Likewise.
10917 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
10918 (epiphany_rtx_costs): Likewise.
10919 * config/epiphany/predicates.md (proper_comparison_operator):
10921 * config/frv/frv.c (condexec_memory_operand): Likewise.
10922 (frv_emit_move): Likewise.
10923 (output_move_single): Likewise.
10924 (output_condmove_single): Likewise.
10925 (frv_hard_regno_mode_ok): Likewise.
10926 (frv_matching_accg_mode): Likewise.
10927 * config/h8300/h8300.c (split_adds_subs): Likewise.
10928 (h8300_rtx_costs): Likewise.
10929 (h8300_print_operand): Likewise.
10930 (compute_mov_length): Likewise.
10931 (output_logical_op): Likewise.
10932 (compute_logical_op_length): Likewise.
10933 (compute_logical_op_cc): Likewise.
10934 (h8300_shift_needs_scratch_p): Likewise.
10935 (output_a_shift): Likewise.
10936 (compute_a_shift_length): Likewise.
10937 (compute_a_shift_cc): Likewise.
10938 (expand_a_rotate): Likewise.
10939 (output_a_rotate): Likewise.
10940 * config/i386/i386.c (classify_argument): Likewise.
10941 (function_arg_advance_32): Likewise.
10942 (function_arg_32): Likewise.
10943 (function_arg_64): Likewise.
10944 (function_value_64): Likewise.
10945 (ix86_gimplify_va_arg): Likewise.
10946 (ix86_legitimate_constant_p): Likewise.
10947 (put_condition_code): Likewise.
10948 (split_double_mode): Likewise.
10949 (ix86_avx256_split_vector_move_misalign): Likewise.
10950 (ix86_expand_vector_logical_operator): Likewise.
10951 (ix86_split_idivmod): Likewise.
10952 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
10953 (ix86_build_const_vector): Likewise.
10954 (ix86_build_signbit_mask): Likewise.
10955 (ix86_match_ccmode): Likewise.
10956 (ix86_cc_modes_compatible): Likewise.
10957 (ix86_expand_branch): Likewise.
10958 (ix86_expand_sse_cmp): Likewise.
10959 (ix86_expand_sse_movcc): Likewise.
10960 (ix86_expand_int_sse_cmp): Likewise.
10961 (ix86_expand_vec_perm_vpermi2): Likewise.
10962 (ix86_expand_vec_perm): Likewise.
10963 (ix86_expand_sse_unpack): Likewise.
10964 (ix86_expand_int_addcc): Likewise.
10965 (ix86_split_to_parts): Likewise.
10966 (ix86_vectorize_builtin_gather): Likewise.
10967 (ix86_vectorize_builtin_scatter): Likewise.
10968 (avx_vpermilp_parallel): Likewise.
10969 (inline_memory_move_cost): Likewise.
10970 (ix86_tieable_integer_mode_p): Likewise.
10971 (x86_maybe_negate_const_int): Likewise.
10972 (ix86_expand_vector_init_duplicate): Likewise.
10973 (ix86_expand_vector_init_one_nonzero): Likewise.
10974 (ix86_expand_vector_init_one_var): Likewise.
10975 (ix86_expand_vector_init_concat): Likewise.
10976 (ix86_expand_vector_init_interleave): Likewise.
10977 (ix86_expand_vector_init_general): Likewise.
10978 (ix86_expand_vector_set): Likewise.
10979 (ix86_expand_vector_extract): Likewise.
10980 (emit_reduc_half): Likewise.
10981 (ix86_emit_i387_round): Likewise.
10982 (ix86_mangle_type): Likewise.
10983 (ix86_expand_round_sse4): Likewise.
10984 (expand_vec_perm_blend): Likewise.
10985 (canonicalize_vector_int_perm): Likewise.
10986 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
10987 (expand_vec_perm_1): Likewise.
10988 (expand_vec_perm_interleave3): Likewise.
10989 (expand_vec_perm_even_odd_pack): Likewise.
10990 (expand_vec_perm_even_odd_1): Likewise.
10991 (expand_vec_perm_broadcast_1): Likewise.
10992 (ix86_vectorize_vec_perm_const_ok): Likewise.
10993 (ix86_expand_vecop_qihi): Likewise.
10994 (ix86_expand_mul_widen_hilo): Likewise.
10995 (ix86_expand_sse2_abs): Likewise.
10996 (ix86_expand_pextr): Likewise.
10997 (ix86_expand_pinsr): Likewise.
10998 (ix86_preferred_simd_mode): Likewise.
10999 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
11000 * config/i386/sse.md (*andnot<mode>3): Likewise.
11001 (<mask_codefor><code><mode>3<mask_name>): Likewise.
11002 (*<code><mode>3): Likewise.
11003 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
11004 (ia64_expand_atomic_op): Likewise.
11005 (ia64_arg_type): Likewise.
11006 (ia64_mode_to_int): Likewise.
11007 (ia64_scalar_mode_supported_p): Likewise.
11008 (ia64_vector_mode_supported_p): Likewise.
11009 (expand_vec_perm_broadcast): Likewise.
11010 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
11011 (iq2000_function_arg_advance): Likewise.
11012 (iq2000_function_arg): Likewise.
11013 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
11014 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
11015 (m68k_libcall_value): Likewise.
11016 (m68k_function_value): Likewise.
11017 (sched_attr_op_type): Likewise.
11018 * config/mcore/mcore.c (mcore_output_move): Likewise.
11019 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
11021 (microblaze_function_arg): Likewise.
11022 * config/mips/mips.c (mips16_build_call_stub): Likewise.
11023 (mips_print_operand): Likewise.
11024 (mips_mode_ok_for_mov_fmt_p): Likewise.
11025 (mips_vector_mode_supported_p): Likewise.
11026 (mips_preferred_simd_mode): Likewise.
11027 (mips_expand_vpc_loongson_even_odd): Likewise.
11028 (mips_expand_vec_unpack): Likewise.
11029 (mips_expand_vi_broadcast): Likewise.
11030 (mips_expand_vector_init): Likewise.
11031 (mips_expand_vec_reduc): Likewise.
11032 (mips_expand_msa_cmp): Likewise.
11033 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
11034 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
11035 (cc_flags_for_mode): Likewise.
11036 * config/msp430/msp430.c (msp430_print_operand): Likewise.
11037 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
11038 (nds32_output_casesi_pc_relative): Likewise.
11039 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11040 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
11041 (nvptx_gen_unpack): Likewise.
11042 (nvptx_gen_pack): Likewise.
11043 (nvptx_gen_shuffle): Likewise.
11044 (nvptx_gen_wcast): Likewise.
11045 (nvptx_preferred_simd_mode): Likewise.
11046 * config/pa/pa.c (pa_secondary_reload): Likewise.
11047 * config/pa/predicates.md (base14_operand): Likewise.
11048 * config/powerpcspe/powerpcspe-c.c
11049 (altivec_resolve_overloaded_builtin): Likewise.
11050 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
11052 (rs6000_preferred_simd_mode): Likewise.
11053 (output_vec_const_move): Likewise.
11054 (rs6000_expand_vector_extract): Likewise.
11055 (rs6000_split_vec_extract_var): Likewise.
11056 (reg_offset_addressing_ok_p): Likewise.
11057 (rs6000_legitimate_offset_address_p): Likewise.
11058 (rs6000_legitimize_address): Likewise.
11059 (rs6000_emit_set_const): Likewise.
11060 (rs6000_const_vec): Likewise.
11061 (rs6000_emit_move): Likewise.
11062 (spe_build_register_parallel): Likewise.
11063 (rs6000_darwin64_record_arg_recurse): Likewise.
11064 (swap_selector_for_mode): Likewise.
11065 (spe_init_builtins): Likewise.
11066 (paired_init_builtins): Likewise.
11067 (altivec_init_builtins): Likewise.
11068 (do_load_for_compare): Likewise.
11069 (rs6000_generate_compare): Likewise.
11070 (rs6000_expand_float128_convert): Likewise.
11071 (emit_load_locked): Likewise.
11072 (emit_store_conditional): Likewise.
11073 (rs6000_output_function_epilogue): Likewise.
11074 (rs6000_handle_altivec_attribute): Likewise.
11075 (rs6000_function_value): Likewise.
11076 (emit_fusion_gpr_load): Likewise.
11077 (emit_fusion_p9_load): Likewise.
11078 (emit_fusion_p9_store): Likewise.
11079 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
11080 (fusion_gpr_mem_load): Likewise.
11081 (fusion_addis_mem_combo_load): Likewise.
11082 (fusion_addis_mem_combo_store): Likewise.
11083 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
11084 (fusion_gpr_mem_load): Likewise.
11085 (fusion_addis_mem_combo_load): Likewise.
11086 (fusion_addis_mem_combo_store): Likewise.
11087 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11089 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
11090 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
11091 (rs6000_preferred_simd_mode): Likewise.
11092 (output_vec_const_move): Likewise.
11093 (rs6000_expand_vector_extract): Likewise.
11094 (rs6000_split_vec_extract_var): Likewise.
11095 (reg_offset_addressing_ok_p): Likewise.
11096 (rs6000_legitimate_offset_address_p): Likewise.
11097 (rs6000_legitimize_address): Likewise.
11098 (rs6000_emit_set_const): Likewise.
11099 (rs6000_const_vec): Likewise.
11100 (rs6000_emit_move): Likewise.
11101 (rs6000_darwin64_record_arg_recurse): Likewise.
11102 (swap_selector_for_mode): Likewise.
11103 (paired_init_builtins): Likewise.
11104 (altivec_init_builtins): Likewise.
11105 (rs6000_expand_float128_convert): Likewise.
11106 (emit_load_locked): Likewise.
11107 (emit_store_conditional): Likewise.
11108 (rs6000_output_function_epilogue): Likewise.
11109 (rs6000_handle_altivec_attribute): Likewise.
11110 (rs6000_function_value): Likewise.
11111 (emit_fusion_gpr_load): Likewise.
11112 (emit_fusion_p9_load): Likewise.
11113 (emit_fusion_p9_store): Likewise.
11114 * config/rx/rx.c (rx_gen_move_template): Likewise.
11115 (flags_from_mode): Likewise.
11116 * config/s390/predicates.md (s390_alc_comparison): Likewise.
11117 (s390_slb_comparison): Likewise.
11118 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
11119 (s390_vector_mode_supported_p): Likewise.
11120 (s390_cc_modes_compatible): Likewise.
11121 (s390_match_ccmode_set): Likewise.
11122 (s390_canonicalize_comparison): Likewise.
11123 (s390_emit_compare_and_swap): Likewise.
11124 (s390_branch_condition_mask): Likewise.
11125 (s390_rtx_costs): Likewise.
11126 (s390_secondary_reload): Likewise.
11127 (__SECONDARY_RELOAD_CASE): Likewise.
11128 (s390_expand_cs): Likewise.
11129 (s390_preferred_simd_mode): Likewise.
11130 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
11131 * config/sh/sh.c (sh_print_operand): Likewise.
11132 (dump_table): Likewise.
11133 (sh_secondary_reload): Likewise.
11134 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11135 * config/sh/sh.md (casesi_worker_1): Likewise.
11136 (casesi_worker_2): Likewise.
11137 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
11138 (fcc_comparison_operator): Likewise.
11139 * config/sparc/sparc.c (sparc_expand_move): Likewise.
11140 (emit_soft_tfmode_cvt): Likewise.
11141 (sparc_preferred_simd_mode): Likewise.
11142 (output_cbranch): Likewise.
11143 (sparc_print_operand): Likewise.
11144 (sparc_expand_vec_perm_bmask): Likewise.
11145 (vector_init_bshuffle): Likewise.
11146 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11147 (spu_vector_mode_supported_p): Likewise.
11148 (spu_expand_insv): Likewise.
11149 (spu_emit_branch_or_set): Likewise.
11150 (spu_handle_vector_attribute): Likewise.
11151 (spu_builtin_splats): Likewise.
11152 (spu_builtin_extract): Likewise.
11153 (spu_builtin_promote): Likewise.
11154 (spu_expand_sign_extend): Likewise.
11155 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11156 (tilegx_simd_int): Likewise.
11157 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
11158 (tilepro_simd_int): Likewise.
11159 * config/v850/v850.c (const_double_split): Likewise.
11160 (v850_print_operand): Likewise.
11161 (ep_memory_offset): Likewise.
11162 * config/vax/vax.c (vax_rtx_costs): Likewise.
11163 (vax_output_int_move): Likewise.
11164 (vax_output_int_add): Likewise.
11165 (vax_output_int_subtract): Likewise.
11166 * config/visium/predicates.md (visium_branch_operator): Likewise.
11167 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
11168 (visium_print_operand_address): Likewise.
11169 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11170 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
11171 (xtensa_expand_conditional_branch): Likewise.
11172 (xtensa_copy_incoming_a7): Likewise.
11173 (xtensa_output_literal): Likewise.
11174 * dfp.c (decimal_real_maxval): Likewise.
11175 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11177 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11178 Alan Hayward <alan.hayward@arm.com>
11179 David Sherwood <david.sherwood@arm.com>
11181 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
11182 (mode_nunits_inline): Likewise.
11183 (mode_inner_inline): Likewise.
11184 (mode_unit_size_inline): Likewise.
11185 (mode_unit_precision_inline): Likewise.
11186 (emit_insn_modes_h): Likewise. Also emit a #define of the
11188 (emit_mode_wider): Add an E_ prefix to mode names.
11189 (emit_mode_complex): Likewise.
11190 (emit_mode_inner): Likewise.
11191 (emit_mode_adjustments): Likewise.
11192 (emit_mode_int_n): Likewise.
11193 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
11194 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
11195 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
11196 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
11197 (CRC32_BUILTIN, ENTRY): Likewise.
11198 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
11199 (aarch64_pop_regs): Likewise.
11200 (aarch64_process_components): Likewise.
11201 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
11202 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
11203 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
11204 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
11205 * config/arm/arm.c (arm_init_libfuncs): Likewise.
11206 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
11208 * config/i386/i386-builtin.def (pcmpestr): Likewise.
11209 (pcmpistr): Likewise.
11210 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
11211 * config/mmix/mmix.c (mmix_output_condition): Likewise.
11212 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
11214 * config/rl78/rl78.c (mduc_regs): Likewise.
11215 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
11216 (htm_expand_builtin): Likewise.
11217 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
11218 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
11219 * config/xtensa/xtensa.c (print_operand): Likewise.
11220 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
11221 (NUM_MODE_VECTOR_INT): Likewise.
11222 * genoutput.c (null_operand): Likewise.
11223 (output_operand_data): Likewise.
11224 * genrecog.c (print_parameter_value): Likewise.
11225 * lra.c (debug_operand_data): Likewise.
11227 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11229 * dwarf2out.c (scompare_loc_descriptor_wide)
11230 (scompare_loc_descriptor_narrow): New functions, split out from...
11231 (scompare_loc_descriptor): ...here.
11232 * expmed.c (emit_store_flag_int): New function, split out from...
11233 (emit_store_flag): ...here.
11235 2017-08-30 Richard Biener <rguenther@suse.de>
11237 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
11238 (dwarf2out_early_finish): Move setting of AT_pubnames from
11239 early debug output to early finish.
11241 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11243 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
11244 and -mdata-region to the assembler.
11246 2017-08-30 Richard Biener <rguenther@suse.de>
11248 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
11250 (gen_subprogram_die): Add DW_AT_object_pointer only early.
11251 (dwarf2out_early_global_decl): Only generate a DIE for the
11252 abstract origin if it doesn't already exist or is a declaration DIE.
11253 (resolve_addr): Do not add the linkage name twice when
11254 generating a stub DIE for the DW_TAG_GNU_call_site target.
11256 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11258 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11259 Use machine_mode rather than int for arg1_mode.
11261 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
11264 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
11265 that the second argument of the built-in functions to unpack
11266 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
11267 switch statement instead a lot of if statements.
11268 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
11269 Allow 64-bit values to be in Altivec registers as well as
11270 traditional floating point registers.
11271 (pack<mode>, FMOVE128_VSX iterator): Likewise.
11273 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
11275 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
11276 MAX_REGS_PER_ADDRESS == 1.
11278 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
11280 * config/i386/i386.opt (flag_fentry): Do not init to -1.
11281 * config/i386/i386.c (ix86_option_override_internal): Simplify
11282 setting of opts->x_flag_entry.
11284 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11285 Jakub Jelinek <jakub@redhat.com>
11286 Richard Biener <rguenther@suse.de>
11288 PR tree-optimization/81503
11289 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
11290 folded constant fits in the target type; reorder tests for clarity.
11292 2017-08-29 Martin Liska <mliska@suse.cz>
11294 * passes.def: Include pass_lower_switch.
11295 * stmt.c (dump_case_nodes): Remove and move to
11296 tree-switch-conversion.
11297 (case_values_threshold): Likewise.
11298 (expand_switch_as_decision_tree_p): Likewise.
11299 (emit_case_decision_tree): Likewise.
11300 (expand_case): Likewise.
11301 (balance_case_nodes): Likewise.
11302 (node_has_low_bound): Likewise.
11303 (node_has_high_bound): Likewise.
11304 (node_is_bounded): Likewise.
11305 (emit_case_nodes): Likewise.
11306 (struct simple_case_node): New struct.
11307 (add_case_node): Remove.
11308 (emit_case_dispatch_table): Use vector instead of case_list.
11309 (reset_out_edges_aux): Remove.
11310 (compute_cases_per_edge): Likewise.
11311 (expand_case): Build list of simple_case_node.
11312 (expand_sjlj_dispatch_table): Use it.
11313 * tree-switch-conversion.c (struct case_node): Moved from
11314 stmt.c and adjusted.
11315 (emit_case_nodes): Likewise.
11316 (node_has_low_bound): Likewise.
11317 (node_has_high_bound): Likewise.
11318 (node_is_bounded): Likewise.
11319 (case_values_threshold): Likewise.
11320 (reset_out_edges_aux): Likewise.
11321 (compute_cases_per_edge): Likewise.
11322 (add_case_node): Likewise.
11323 (dump_case_nodes): Likewise.
11324 (balance_case_nodes): Likewise.
11325 (expand_switch_as_decision_tree_p): Likewise.
11326 (emit_jump): Likewise.
11327 (emit_case_decision_tree): Likewise.
11328 (try_switch_expansion): Likewise.
11329 (do_jump_if_equal): Likewise.
11330 (emit_cmp_and_jump_insns): Likewise.
11331 (fix_phi_operands_for_edge): New function.
11332 (record_phi_operand_mapping): Likewise.
11333 (class pass_lower_switch): New pass.
11334 (pass_lower_switch::execute): New function.
11335 (make_pass_lower_switch): Likewise.
11336 (conditional_probability):
11337 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
11338 * tree-pass.h: Add make_pass_lower_switch.
11340 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11343 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
11346 2017-08-29 Richard Biener <rguenther@suse.de>
11348 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
11349 we do not add a DW_AT_inline attribute twice.
11350 (gen_subprogram_die): Remove code setting DW_AT_inline on
11351 DECL_ABSTRACT_P nodes.
11353 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
11355 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
11356 calls to internal functions.
11357 (gimplify_modify_expr): Likewise.
11358 * tree-call-cdce.c (use_internal_fn): Likewise.
11359 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
11360 (convert_to_divmod): Set the nothrow flag.
11361 * tree-if-conv.c (predicate_mem_writes): Likewise.
11362 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
11363 (vectorizable_call): Likewise.
11364 (vectorizable_store): Likewise.
11365 (vectorizable_load): Likewise.
11366 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
11367 (vect_recog_mask_conversion_pattern): Likewise.
11369 2017-08-29 Martin Liska <mliska@suse.cz>
11372 * gcc.c (driver_handle_option): Add new argument.
11373 * opts-common.c (handle_option): Pass
11374 target_option_override_hook.
11375 * opts-global.c (lang_handle_option): Add new option.
11376 (set_default_handlers): Add new argument.
11377 (decode_options): Likewise.
11378 * opts.c (target_handle_option): Likewise.
11379 (common_handle_option): Call target_option_override_hook.
11380 * opts.h (struct cl_option_handler_func): Add hook for
11381 target option override.
11382 (struct cl_option_handlers): Likewise.
11383 (set_default_handlers): Add new argument.
11384 (decode_options): Likewise.
11385 (common_handle_option): Likewise.
11386 (target_handle_option): Likewise.
11387 * toplev.c (toplev::main): Pass targetm.target_option.override
11390 2017-08-29 Richard Biener <rguenther@suse.de>
11391 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
11393 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
11394 life to the active subtree.
11396 2017-08-28 Jeff Law <law@redhat.com>
11398 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
11399 derive_equivalences.
11400 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
11401 Code moved into....
11402 (edge_info::derive_equivalences): New private member function
11404 * tree-ssa-dom.c (class edge_info): Changed from a struct
11405 to a class. Add ctor/dtor, methods and data members.
11406 (edge_info::edge_info): Renamed from allocate_edge_info.
11407 Initialize additional members.
11408 (edge_info::~edge_info): New.
11409 (free_dom_edge_info): Delete the edge info.
11410 (record_edge_info): Use new class & associated member functions.
11411 Tighten forms for testing for edge equivalences.
11412 (record_temporary_equivalences): Iterate over the simple
11413 equivalences rather than assuming there's only one per edge.
11414 (cprop_into_successor_phis): Iterate over the simple
11415 equivalences rather than assuming there's only one per edge.
11416 (optimize_stmt): Use operand_equal_p rather than pointer
11417 equality for mini-DSE code.
11419 2017-08-28 Nathan Sidwell <nathan@acm.org>
11421 * gcc.c (execute): Fold SIGPIPE handling into switch
11422 statement. Adjust internal error message.
11424 2017-08-28 Richard Biener <rguenther@suse.de>
11427 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
11428 Do nothing for removed DIEs.
11430 2017-08-28 Richard Biener <rguenther@suse.de>
11432 PR tree-optimization/81977
11433 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
11436 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
11439 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
11440 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
11441 using targetm.gen_mem_thread_fence.
11443 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
11446 * config/i386/i386.md (*<btsc><mode>): Change operand 2
11447 predicate to register_operand. Reorder operands.
11448 (*btr<mode>): Ditto.
11449 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
11450 (*btr<mode>_mask): Ditto.
11452 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
11454 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
11455 * config/rs6000/xmmintrin.h: New file.
11456 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
11458 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11461 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
11462 parameter and_insn and return it.
11463 (recombine_lvx_pattern): Insert a copy to ensure availability of
11464 the base register of the copied masking operation at the point of
11465 the instruction replacement.
11466 (recombine_stvx_pattern): Likewise.
11468 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
11470 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
11471 undocumented switches.
11472 (-mpower9-dform-vector): Likewise.
11473 (-mpower9-dform): Likewise.
11474 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
11475 comments to delete references to -mpower9-dform* switches.
11476 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
11477 Delete reference to -mpower9-dform* switches, test for
11478 -mpower9-vector instead.
11479 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
11480 (OTHER_P9_VECTOR_MASKS): Likewise.
11481 (POWERPC_MASKS): Likewise.
11482 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
11483 tests against -mpower9-dform* to -mpower9-vector. Delete code
11484 that checked for -mpower9-dform* consistancy with other options.
11485 Add test for -mpower9-misc to enable other power9 switches.
11486 (rs6000_init_hard_regno_mode_ok): Likewise.
11487 (rs6000_option_override_internal): Likewise.
11488 (rs6000_emit_prologue): Likewise.
11489 (rs6000_emit_epilogue): Likewise.
11490 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
11491 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
11492 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
11494 (emit_fusion_p9_store): Likewise.
11495 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
11496 resetting these macros if the assembler does not support ISA 3.0
11498 (TARGET_P9_DFORM_VECTOR): Likewise.
11499 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
11500 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
11502 2017-08-25 Alan Modra <amodra@gmail.com>
11504 PR rtl-optimization/81747
11505 * cse.c (cse_extended_basic_block): Don't attempt to record
11506 equivalences for degenerate conditional branches that branch
11507 to their fall-through.
11509 2017-08-24 Martin Sebor <msebor@redhat.com>
11511 PR middle-end/81908
11512 * gimple-fold.c (size_must_be_zero_p): New function.
11513 (gimple_fold_builtin_memory_op): Call it.
11515 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
11517 * config/rs6000/mm_malloc.h: New file.
11519 2017-08-24 Bin Cheng <bin.cheng@arm.com>
11521 PR tree-optimization/81913
11522 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
11523 analysis when either IVs in condition can wrap.
11525 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
11527 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
11528 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
11530 2017-08-24 Richard Biener <rguenther@suse.de>
11533 * targhooks.c (default_target_can_inline_p): Properly
11534 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
11535 is present and always compare.
11536 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
11537 infer -mfpmath=sse from TARGET_SSE_P.
11538 (ix86_can_inline_p): Properly use target_option_default_node when
11539 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
11541 2017-08-24 Richard Biener <rguenther@suse.de>
11544 * dwarf2out.c (output_die): Handle flag_generate_offload like
11546 (output_comp_unit): Likewise.
11547 (gen_array_type_die): Likewise.
11548 (dwarf2out_early_finish): Likewise.
11549 (note_variable_value_in_expr): Likewise.
11550 (dwarf2out_finish): Likewise. Adjust assert.
11551 * cgraphunit.c (symbol_table::compile): Move setting of
11552 flag_generate_offload earlier ...
11553 (symbol_table::finalize_compilation_unit): ... here, before
11554 early debug finalization.
11556 2017-08-24 Richard Biener <rguenther@suse.de>
11558 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
11559 and ipa-fnsummary.h.
11560 (ix86_can_inline_p): When ix86_fpmath flags do not match
11561 check whether the callee uses FP math at all.
11563 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
11565 PR middle-end/81931
11566 * tree-ssanames.c (get_nonzero_bits): Use element_precision
11567 instead of TYPE_PRECISION.
11569 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
11570 Alan Hayward <alan.hayward@arm.com>
11571 David Sherwood <david.sherwood@arm.com>
11573 * combine.c (make_extraction): Use subreg_offset_from_lsb.
11575 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
11577 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
11579 (ix86_frame::stack_realign_allocate): New field.
11580 (struct machine_frame_state): Modify comments.
11581 (machine_frame_state::sp_realigned_fp_end): New field.
11582 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
11583 layout calculation.
11584 (sp_valid_at): Add assertion to assure no attempt to access invalid
11585 offset of a realigned stack.
11586 (fp_valid_at): Likewise.
11587 (choose_baseaddr): Modify comments.
11588 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
11589 ix86_expand_prologue.
11590 (ix86_expand_prologue): Modify stack realignment and allocation.
11591 (ix86_expand_epilogue): Modify comments.
11592 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
11593 avx2_runtime, avx512f, and avx512f_runtime.
11595 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
11597 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
11598 (mstackrealign): Do not init to -1.
11599 * config/i386/i386.c (ix86_option_override_internal):
11600 Check opts_set, not opts when setting default value of
11601 opts->x_ix86_force_align_arg_pointer.
11603 2017-08-23 Richard Biener <rguenther@suse.de>
11605 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
11606 lang_hooks.decl_printable_name.
11607 * print-rtl-function.c (print_rtx_function): Likewise.
11608 * tree-pretty-print.c (dump_function_header): Likewise.
11610 2017-08-23 Richard Biener <rguenther@suse.de>
11613 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
11614 -g0 at compile-time.
11616 2017-08-23 Tamar Christina <tamar.christina@arm.com>
11618 PR middle-end/19706
11619 * doc/sourcebuild.texi (Other hardware attributes):
11622 2017-08-23 Tamar Christina <tamar.christina@arm.com>
11624 PR middle-end/19706
11625 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
11626 Fix single-use check.
11628 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
11630 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
11632 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
11634 * doc/install.texi: Modify to add more details on running selected
11637 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
11639 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
11640 is combined with -mabi=ms.
11641 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
11644 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11646 PR tree-optimization/81488
11647 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
11648 and cached_basis fields.
11649 (MAX_SPREAD): New constant.
11650 (alloc_cand_and_find_basis): Initialize new fields.
11651 (clear_visited): New function.
11652 (create_phi_basis_1): Rename from create_phi_basis, set visited
11653 and cached_basis fields.
11654 (create_phi_basis): New wrapper function.
11655 (phi_add_costs_1): Rename from phi_add_costs, add spread
11656 parameter, set visited field, short-circuit when limits reached.
11657 (phi_add_costs): New wrapper function.
11658 (record_phi_increments_1): Rename from record_phi_increments, set
11660 (record_phi_increments): New wrapper function.
11661 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
11662 (phi_incr_cost): New wrapper function.
11663 (all_phi_incrs_profitable_1): Rename from
11664 all_phi_incrs_profitable, set visited field.
11665 (all_phi_incrs_profitable): New wrapper function.
11667 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
11668 Alan Hayward <alan.hayward@arm.com>
11669 David Sherwood <david.sherwood@arm.com>
11671 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
11672 that takes the outer and inner modes.
11673 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
11674 comparison as the canonical test for a paradoxical subreg.
11675 * combine.c (simplify_set): Use paradoxical_subreg_p.
11676 (make_extraction): Likewise.
11677 (force_to_mode): Likewise.
11678 (rtx_equal_for_field_assignment_p): Likewise.
11679 (gen_lowpart_for_combine): Likewise.
11680 (simplify_comparison): Likewise.
11681 * cse.c (equiv_constant): Likewise.
11682 * expmed.c (store_bit_field_1): Likewise.
11683 * final.c (alter_subreg): Likewise.
11684 * fwprop.c (propagate_rtx): Likewise.
11685 (forward_propagate_subreg): Likewise.
11686 * ira-conflicts.c (ira_build_conflicts): Likewise.
11687 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
11688 * lra-constraints.c (curr_insn_transform): Likewise.
11689 (split_reg): Likewise.
11690 * lra-eliminations.c (move_plus_up): Likewise.
11691 (lra_eliminate_regs_1): Likewise.
11692 * recog.c (general_operand): Likewise.
11693 * ree.c (combine_reaching_defs): Likewise.
11694 * reload.c (push_reload): Likewise.
11695 (find_reloads): Likewise.
11696 * reload1.c (elimination_effects): Likewise.
11697 (compute_reload_subreg_offset): Likewise.
11698 (choose_reload_regs): Likewise.
11699 * rtlanal.c (subreg_lsb_1): Likewise.
11700 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11701 (simplify_subreg): Likewise.
11702 * var-tracking.c (track_loc_p): Likewise.
11703 * emit-rtl.c (byte_lowpart_offset): Likewise.
11704 (paradoxical_subreg_p): Delete out-of-line definition.
11706 2017-08-22 Jeff Law <law@redhat.com>
11708 PR tree-optimization/81741
11709 PR tree-optimization/71947
11710 * tree-ssa-dom.c: Include tree-inline.h.
11711 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
11712 equivalences if one is more expensive to compute than the other.
11713 * tree-ssa-scopedtables.h (class const_or_copies): Make
11714 record_const_or_copy_raw method private.
11715 (class avail_exprs_stack): New method simplify_binary_operation.
11716 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
11717 avail_exprs_stack::simplify_binary_operation as needed.
11718 (avail_exprs_stack::simplify_binary_operation): New function.
11720 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
11722 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
11723 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
11724 (DOT_SYMBOLS): Likewise.
11725 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
11726 (RELOCATABLE_NEEDS_FIXUP): Likewise.
11727 (RS6000_ABI_NAME): Likewise.
11728 (TARGET_CMODEL): Likewise.
11729 (TOC_SECTION_ASM_OP): Likewise.
11730 (SET_CMODEL): New macro.
11731 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
11733 2017-08-22 Richard Biener <rguenther@suse.de>
11735 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
11736 to allow for free-lang-data replacements similar to verify_type_variant.
11738 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
11739 Alan Hayward <alan.hayward@arm.com>
11740 David Sherwood <david.sherwood@arm.com>
11742 * config/aarch64/aarch64.md (casesi): Use DImode rather than
11743 VOIDmode for the LABEL_REF.
11745 2017-08-22 Richard Biener <rguenther@suse.de>
11747 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
11749 2017-08-22 Richard Biener <rguenther@suse.de>
11751 * common.opt (feliminate-dwarf2-dups): Ignore.
11752 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
11753 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
11754 same_die_p_wrap, compute_section_prefix,
11755 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
11756 (comdat_symbol_id, comdat_symbol_number): Likewise.
11757 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
11759 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
11760 (output_die): Mark unreachable path unreachable.
11761 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
11762 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
11763 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
11764 (dwarf2out_early_finish): Likewise.
11766 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
11768 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
11770 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
11773 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
11774 not VAR_P. Filter attribute warnings with OPT_Wattributes.
11775 (avr_attribute_table) <io, io_low, address>: Initialize
11776 .decl_required with true.
11778 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
11780 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
11781 undocumented debugging options.
11782 (-mvsx-scalar-double): Likewise.
11783 (-mallow-df-permute): Likewise.
11784 (-mvectorize-builtins): Likewise.
11785 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
11786 (rs6000_builtin_vectorized_function): Likewise.
11787 (rs6000_builtin_md_vectorized_function): Likewise.
11788 (rs6000_opt_vars): Likewise.
11790 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
11793 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
11794 (*btrq_imm): Rename from *btrq.
11795 (*btcq_imm): Rename from *btcq.
11796 (btsc): New code attribute.
11797 (*<btsc><mode>): New insn pattern.
11798 (*btr<mode>): Ditto.
11799 (*<btsc><mode>_mask): New insn_and_split pattern.
11800 (*btr<mode>_mask): Ditto.
11802 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11803 Alan Hayward <alan.hayward@arm.com>
11804 David Sherwood <david.sherwood@arm.com>
11806 * function.c (pad_below): Simplify padding calculation.
11808 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11809 Alan Hayward <alan.hayward@arm.com>
11810 David Sherwood <david.sherwood@arm.com>
11812 * target.def (function_prologue): Remove frame size argument.
11813 (function_epilogue): Likewise.
11814 * doc/tm.texi: Regenerate.
11815 * final.c (final_start_function): Update call to function_prologue.
11816 (final_end_function): Update call to function_epilogue.
11817 (default_function_pro_epilogue): Remove frame size argument.
11818 * output.h (default_function_pro_epilogue): Likewise.
11819 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
11820 (arm_output_function_prologue): Likewise.
11821 * config/frv/frv.c (frv_function_prologue): Likewise.
11822 (frv_function_epilogue): Likewise.
11823 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
11824 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
11825 (ia64_output_function_epilogue): Likewise.
11826 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
11827 (m32r_output_function_epilogue): Likewise.
11828 * config/microblaze/microblaze.c (microblaze_function_prologue)
11829 (microblaze_function_epilogue): Likewise.
11830 * config/mips/mips.c (mips_output_function_prologue): Likewise.
11831 (mips_output_function_epilogue): Likewise.
11832 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
11833 (mmix_target_asm_function_epilogue): Likewise.
11834 * config/msp430/msp430.c (msp430_start_function): Likewise.
11835 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
11836 (nds32_asm_function_epilogue): Likewise.
11837 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
11838 * config/pa/pa.c (pa_output_function_prologue): Likewise.
11839 (pa_output_function_epilogue): Likewise.
11840 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
11841 (rs6000_output_function_epilogue): Likewise.
11842 * config/rl78/rl78.c (rl78_start_function): Likewise.
11843 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
11844 (rs6000_output_function_epilogue): Likewise.
11845 * config/rx/rx.c (rx_output_function_prologue): Likewise.
11846 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
11847 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
11848 (sparc_asm_function_epilogue): Likewise.
11850 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11852 * tree.h (type_has_mode_precision_p): New function.
11853 * convert.c (convert_to_integer_1): Use it.
11854 * expr.c (expand_expr_real_2): Likewise.
11855 (expand_expr_real_1): Likewise.
11856 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
11857 * match.pd: Likewise.
11858 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
11859 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
11860 * tree-tailcall.c (process_assignment): Likewise.
11861 * tree-vect-loop.c (vectorizable_reduction): Likewise.
11862 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
11863 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
11864 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11865 (vectorizable_assignment): Likewise.
11866 (vectorizable_shift): Likewise.
11867 (vectorizable_operation): Likewise.
11868 * tree-vrp.c (register_edge_assert_for_2): Likewise.
11870 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
11872 * match.pd: Add pow (C, x) simplification.
11874 2017-08-21 Richard Biener <rguenther@suse.de>
11876 PR tree-optimization/81900
11877 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
11878 for blocks with abnormal predecessors.
11879 (compute_antic): Do not set visited flag prematurely.
11881 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
11884 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
11886 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11888 * stor-layout.h (vector_type_mode): Move to...
11889 * tree.h (vector_type_mode): ...here.
11890 * stor-layout.c (vector_type_mode): Move to...
11891 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
11893 2017-08-21 Richard Biener <rguenther@suse.de>
11895 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
11896 register_external_die hooks.
11897 (debug_false_tree_charstarstar_uhwistar): Declare.
11898 (debug_nothing_tree_charstar_uhwi): Likewise.
11899 * debug.c (do_nothing_debug_hooks): Adjust.
11900 (debug_false_tree_charstarstar_uhwistar): New do nothing.
11901 (debug_nothing_tree_charstar_uhwi): Likewise.
11902 * dbxout.c (dbx_debug_hooks): Adjust.
11903 (xcoff_debug_hooks): Likewise.
11904 * sdbout.c (sdb_debug_hooks): Likewise.
11905 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
11906 * dwarf2out.c (macinfo_label_base): New global.
11907 (dwarf2out_register_external_die): New function for the
11908 register_external_die hook.
11909 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
11910 (dwarf2_debug_hooks): Use them.
11911 (dwarf2_lineno_debug_hooks): Adjust.
11912 (struct die_struct): Add with_offset flag.
11913 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
11914 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
11915 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
11916 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
11917 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
11918 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
11919 defining section names for the early LTO debug variants.
11920 (reset_indirect_string): New helper.
11921 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
11922 (print_dw_val): Add support for offsetted symbol references.
11923 (get_ultimate_context): Split out from is_cxx.
11924 (is_cxx): Use get_ultimate_context.
11925 (is_fortran): Add decl overload.
11926 (compute_comp_unit_symbol): Split out worker from
11927 compute_section_prefix.
11928 (compute_section_prefix): Call compute_comp_unit_symbol and
11929 set comdat_type_p here.
11930 (output_die): Skip DIE symbol output for the LTO added one.
11931 Handle DIE symbol references with offset.
11932 (output_comp_unit): Guard section name mangling properly.
11933 For LTO debug sections emit a symbol at the section beginning
11934 which we use to refer to its DIEs.
11935 (add_abstract_origin_attribute): For DIEs registered via
11936 dwarf2out_register_external_die directly refer to the early
11937 DIE rather than indirectly through the shadow one we created.
11938 Remove obsolete call to dwarf2out_abstract_function for
11939 non-function/block origins.
11940 (gen_array_type_die): When generating early LTO debug do
11941 not emit DW_AT_string_length.
11942 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
11943 late when in LTO. As suggested place a gcc_unreachable for
11944 the DECL_ABSTRACT_P case.
11945 (gen_subprogram_die): Avoid another specification DIE
11946 for early built declarations/definitions for the late LTO case.
11947 (gen_variable_die): Add type references for late duplicated VLA dies
11949 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
11950 we have the abstract instance already.
11951 (process_scope_var): Adjust decl DIE contexts in LTO which
11952 first puts them in limbo.
11953 (gen_decl_die): Do not generate type DIEs late apart from
11954 types for VLAs or for decls we do not yet have a DIE. Do not
11955 call dwarf2out_abstract_function late.
11956 (dwarf2out_early_global_decl): Make sure to create DIEs
11957 for abstract instances of a decl first.
11958 (dwarf2out_late_global_decl): Adjust comment.
11959 (output_macinfo_op): With multiple macro sections use
11960 macinfo_label_base to distinguish labels.
11961 (output_macinfo): Likewise. Update macinfo_label_base.
11962 Pass in the line info label.
11963 (note_variable_value_in_expr): When generating LTO resolve
11964 all variable values here by generating DIEs as needed.
11965 (init_sections_and_labels): Add early LTO debug flag parameter
11966 and generate different sections and names if set. Add generation
11967 counter for the labels so we can have multiple of them.
11968 (reset_dies): Helper to allow DIEs to be output multiple times.
11969 (dwarf2out_finish): When outputting DIEs to the fat part of an
11970 LTO object first reset DIEs.
11971 (dwarf2out_early_finish): Output early DIEs when generating LTO.
11972 (modified_type_die): Check for decl_ultimate_origin being self
11974 (gen_type_die_with_usage): Likewise.
11975 (gen_typedef_die): Allow decl_ultimate_origin being self.
11976 (set_decl_abstract_flags): Remove.
11977 (set_block_abstract_flags): Likewise.
11978 (dwarf2out_abstract_function): Treat the early generated DIEs
11979 as the abstract copy and only add DW_AT_inline and
11980 DW_AT_artificial here and call set_decl_origin_self.
11981 If the DIE has an abstract origin don't do anything.
11982 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
11983 if we have none yet (Go fails to build one, PR78628).
11984 (variably_modified_type_p): Prevent endless recursion for Ada
11985 cyclic pointer types.
11986 * lto-streamer-in.c: Include debug.h.
11987 (dref_queue): New global.
11988 (lto_read_tree_1): Stream in DIE references.
11989 (lto_input_tree): Register DIE references.
11990 (input_function): Stream DECL_DEBUG_ARGS.
11991 * lto-streamer-out.c: Include debug.h.
11992 (lto_write_tree_1): Output DIE references.
11993 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
11994 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
11995 (output_function): Stream DECL_DEBUG_ARGS.
11996 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
11997 Stream DECL_ABSTRACT_ORIGIN.
11998 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
11999 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
12000 DECL_CONTEXT for file-scope decls.
12001 * lto-streamer.h (struct dref_entry): Declare.
12002 (dref_queue): Likewise.
12003 * cfgexpand.c (pass_expand::execute): Do not call the
12004 outlining_inline_function hook here.
12005 * lto-wrapper.c (debug_obj): New global.
12006 (tool_cleanup): Unlink it if required.
12007 (debug_objcopy): New function.
12008 (run_gcc): Handle early debug sections in the IL files by
12009 extracting them to separate files, partially linkin them and
12010 feeding the result back as result to the linker.
12011 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
12012 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
12013 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
12014 sections into a separate segment.
12015 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
12017 (darwin_asm_dwarf_section): Likewise.
12018 (darwin_asm_output_dwarf_offset): Likewise.
12019 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
12021 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12022 Alan Hayward <alan.hayward@arm.com>
12023 David Sherwood <david.sherwood@arm.com>
12025 * read-md.h (md_reader::record_potential_iterator_use): Replace
12026 pointer argument with an rtx and an index.
12027 * read-rtl.c (iterator_group::apply_iterator): Likewise.
12028 (apply_mode_iterator): Likewise.
12029 (apply_code_iterator): Likewise.
12030 (apply_int_iterator): Likewise.
12031 (apply_subst_iterator): Likewise.
12032 (record_iterator_use): Likewise.
12033 (record_attribute_use): Likewise.
12034 (md_reader::record_potential_iterator_use): Likewise. Update calls
12035 to record_iterator_use and apply_iterator.
12036 (iterator_use): Replace ptr with x and index.
12037 (attribute_use): Likewise.
12038 (apply_attribute_uses): Update calls to apply_iterator.
12039 (apply_iterators): Likewise. Update initialization of iterator_use.
12040 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
12041 and record_potential_iterator_use.
12042 (rtx_reader::read_rtx_operand): Likewise.
12044 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12045 Alan Hayward <alan.hayward@arm.com>
12046 David Sherwood <david.sherwood@arm.com>
12048 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
12051 2017-08-21 Richard Biener <rguenther@suse.de>
12053 PR middle-end/81884
12054 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
12055 at struct end conservatively when comparing common bases.
12057 2017-08-21 Richard Biener <rguenther@suse.de>
12059 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
12060 (mem_ref_in_stmt): Remove.
12061 (determine_max_movement): Use ref index to get at the reference.
12062 (invariantness_dom_walker::before_dom_children): Deal with
12063 lim data already initialized.
12064 (gather_mem_refs_stmt): Initialize lim data and record ref index.
12066 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12068 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
12069 (TARGET_ISA_ROUND): Ditto.
12070 (TARGET_ROUND): Ditto.
12071 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
12072 * config/i386/i386.md: Ditto.
12073 * config/i386/sse.md: Ditto.
12074 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
12075 with OPTION_MASK_ISA_SSE4_1.
12077 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12080 * doc/extend.texi (x86 Built-in Functions): Correct the name of
12081 __builtin_ia32_lzcnt_u16.
12083 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
12086 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
12087 (rs6000_set_current_function): Rewrite function to use it.
12089 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
12092 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
12093 and TYPE_WARN_IF_NOT_ALIGN.
12094 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
12095 (handle_warn_if_not_align): New.
12096 (place_union_field): Call handle_warn_if_not_align.
12097 (place_field): Call handle_warn_if_not_align.
12098 Copy TYPE_WARN_IF_NOT_ALIGN.
12099 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
12100 (layout_type): Likewise.
12101 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
12103 (tree_decl_common): Add warn_if_not_align.
12104 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
12105 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
12106 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
12107 (DECL_WARN_IF_NOT_ALIGN): Likewise.
12108 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
12109 * doc/extend.texi: Document warn_if_not_aligned attribute.
12110 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
12112 2017-08-17 Martin Liska <mliska@suse.cz>
12115 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
12116 (get_data_dependence): Use it as pointer type.
12117 (distribute_loop): Likewise.
12119 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12121 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
12122 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
12123 (p8_vmrgow_<mode>_direct): New define_insn.
12124 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
12125 handle endianness for vmrgew and vmrgow permute patterns.
12127 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
12129 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
12130 * config/rs6000/rs6000-cpus.def: Remove comment.
12131 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
12132 (POWERPC_MASKS): Likewise.
12133 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
12134 use of TARGET_VSX_TIMODE.
12135 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
12136 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
12137 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
12138 (rs6000_option_override_internal): Remove dead code.
12139 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
12140 (rs6000_legitimize_reload_address): Likewise.
12141 (rs6000_legitimate_address_p): Likewise.
12142 (rs6000_opt_masks): Delete "vsx-timode".
12143 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
12144 from function comment.
12145 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
12146 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
12147 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
12149 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
12150 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
12151 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
12152 (VSX_TI): Likewise.
12154 (define_peephole2): Likewise.
12156 2017-08-17 Martin Sebor <msebor@redhat.com>
12159 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
12160 past the end of an array.
12161 (test_pp_format): Add test cases.
12163 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
12165 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
12166 missing ECF_NOTHROW flags.
12168 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
12171 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
12172 operands residing in integer registers.
12173 (*vsx_le_perm_load_<mode>): Likewise.
12174 (*vsx_le_perm_store_<mode>): Likewise.
12175 (define_peephole2): Add peepholes to optimize the above.
12177 2017-08-17 Marek Polacek <polacek@redhat.com>
12179 PR middle-end/81814
12180 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
12181 to mimic what shorten_compare did. Change the return type to bool.
12182 (fold_cond_expr_with_comparison): Update call to
12183 operand_equal_for_comparison_p.
12184 (fold_ternary_loc): Likewise.
12186 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
12188 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
12190 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
12192 2017-08-17 Richard Biener <rguenther@suse.de>
12194 * tree-ssa-structalias.c (solve_graph): When propagating
12195 to successors update the graphs succ edges and avoid duplicate work.
12197 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
12200 * config/i386/i386.c (ix86_option_override_internal): Save target
12201 specific options after ix86_stack_protector_guard_reg was changed.
12203 2017-08-17 Richard Biener <rguenther@suse.de>
12205 PR tree-optimization/81827
12206 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
12207 (new_var_info): Initialize it conservatively.
12208 (get_call_vi): Mark register vars.
12209 (new_scalar_tmp_constraint_exp): Likewise.
12210 (handle_rhs_call): Likewise.
12211 (handle_const_call): Likewise.
12212 (create_function_info_for): Likewise.
12213 (solve_constraints): Sort varinfos to separate register from
12214 non-register vars to pack points-to solution bitmaps during
12217 2017-08-17 Marek Polacek <polacek@redhat.com>
12219 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
12221 2017-08-17 Richard Biener <rguenther@suse.de>
12223 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
12224 to true when overflow is undefined and we saturated the result.
12226 2017-08-17 Alan Modra <amodra@gmail.com>
12229 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
12230 Don't use store multiple if only one reg needs saving.
12231 (interesting_frame_related_regno): New function.
12232 (rs6000_frame_related): Don't emit frame info for regs that
12234 (rs6000_emit_epilogue): Likewise.
12236 2017-08-16 Nathan Sidwell <nathan@acm.org>
12238 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
12239 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
12240 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
12241 * tree.c (free_lang_data_in_type): Use else-if chain. Always
12242 clear TYPE_LANG_1. Remove obsolete member-function stripping.
12243 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
12244 (verify_type): Adjust for TYPE_BINFO move.
12245 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
12246 process TYPE_BINFO directly.
12247 (hash_tree): Likewise.
12248 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
12250 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
12253 2017-08-16 David Malcolm <dmalcolm@redhat.com>
12255 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
12257 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
12260 * config/i386/i386.md (*anddi_1_btr): Change predicates of
12261 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
12262 Add ix86_binary_operator_ok to insn constraint.
12263 (*iordi_1_bts): Ditto.
12264 (*xordi_1_btc): Ditto.
12265 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
12266 Update corresponding peephole2 pattern.
12270 2017-08-16 Bin Cheng <bin.cheng@arm.com>
12272 PR tree-optimization/81832
12273 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
12274 copy loop header which has IFN_LOOP_DIST_ALIAS call.
12276 2017-08-16 Marek Polacek <polacek@redhat.com>
12279 * fold-const.c (fold_indirect_ref_1): Restore original behavior
12280 regarding size_zero_node.
12282 2017-08-16 Martin Liska <mliska@suse.cz>
12285 * config.gcc: Respect previously set extra_objs in case
12288 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
12290 PR tree-optimization/81835
12291 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
12292 the phi SSA_NAME. Check that the condition in a COND_EXPR does
12293 not depend on the phi.
12295 2017-08-16 Alan Modra <amodra@gmail.com>
12297 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
12300 2017-08-16 Alan Modra <amodra@gmail.com>
12302 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
12303 (save_reg_p): ..into this. Update all callers.
12304 (first_reg_to_save): Simplify.
12306 2017-08-16 Alan Modra <amodra@gmail.com>
12308 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
12311 2017-08-15 Joseph Myers <joseph@codesourcery.com>
12315 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
12316 constant count if that count is less than 32.
12318 2017-08-15 Nathan Sidwell <nathan@acm.org>
12320 * gcc.c (execute): Emit friendlier message if inferior is killed
12321 by an external cause.
12323 2017-08-15 Richard Biener <rguenther@suse.de>
12325 PR tree-optimization/81790
12326 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
12327 CONSTRUCTORs from simplifying and VN.
12329 2017-08-14 Martin Sebor <msebor@redhat.com>
12331 * builtin-attrs.def: Add comments.
12333 2017-08-14 Martin Sebor <msebor@redhat.com>
12336 * doc/extend.texi (attribute nonstring): Document new attribute.
12338 2017-08-14 Martin Sebor <msebor@redhat.com>
12341 * tree-diagnostic.c (default_tree_printer): Handle %G.
12342 * gimple-pretty-print.h (percent_G_format): Declare new function.
12343 * gimple-pretty-print.c (percent_G_format): Define.
12344 * tree-pretty-print.c (percent_K_format): Add argument.
12346 2017-08-14 Martin Sebor <msebor@redhat.com>
12348 PR translation/79998
12349 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
12350 Remove a stray space.
12352 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
12355 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
12356 (*iordi_1_bts): Ditto.
12357 (*xordi_1_btc): Ditto.
12359 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12362 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
12363 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12365 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
12366 quoted strings, and make more translator-friendly.
12367 (darwin_rs6000_override_options): Likewise.
12368 (rs6000_option_override_internal): Likewise.
12369 (rs6000_return_in_memory): Fix overlong line.
12370 (init_cmulative_args): Use quoted strings, and make more
12371 translator-friendly.
12372 (rs6000_pass_by_reference): Fix overlong line.
12373 (def_builtin): Use quoted strings.
12374 (altivec_expand_predicate_builtin): Use quoted strings, and make
12375 more translator-friendly.
12376 (htm_expand_builtin): Use quoted strings.
12377 (cpu_expand_builtin): Use quoted strings, and make more
12378 translator-friendly.
12379 (altivec_expand_builtin): Likewise.
12380 (paired_expand_predicate_builtin): Likewise.
12381 (rs6000_invalid_builtin): Likewise.
12382 (builtin_function_type): Use quoted strings.
12383 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
12384 more translator-friendly.
12385 (rs6000_trampoline_init): Likewise.
12386 (rs6000_handle_altivec_attribute): Likewise.
12387 (rs6000_inner_target_options): Use quoted strings.
12388 (rs6000_disable_incompatible_switches): Likewise.
12389 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
12390 strings, and make more translator-friendly.
12391 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
12393 2017-08-14 Bin Cheng <bin.cheng@arm.com>
12395 PR tree-optimization/81799
12396 * tree-loop-distribution.c (version_loop_by_alias_check): Force
12397 cond_expr to simple gimple operand.
12399 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
12401 PR middle-end/46932
12402 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
12404 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
12408 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
12409 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
12410 TARGET_GASISR_PROLOGUES.
12411 * config/avr/avr.c (avr_option_override): Same.
12412 (avr_pass_pre_proep::execute): Same.
12414 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
12417 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
12418 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
12420 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
12422 * config/i386/i386.md (*load_tp_<mode>): Redefine as
12423 define_insn_and_split. Split to a memory load from 0 in
12424 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
12425 using PTR mode iterator.
12426 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
12427 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
12428 (*add_tp_<mode>): Redefine as define_insn_and_split.
12429 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
12430 address space. Merge with *add_tp_x32 using PTR mode iterator.
12431 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
12432 Split to an add with a memory load from 0 in
12433 DEFAULT_TLS_SEG_REG address space.
12435 2017-08-12 Andrew Pinski <apinski@cavium.com>
12437 * config/aarch64/aarch64-option-extensions.def (rdma):
12438 Fix feature string to what Linux prints out in /proc/cpuinfo.
12440 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
12443 * dwarf2out.c (modified_type_die): For C typedef types that have
12444 an ultimate origin, process the ultimate origin instead of the
12446 (gen_typedef_die): Assert that input DECLs have no ultimate
12448 (gen_type_die_with_usage): For typedef variants that have an
12449 ultimate origin, just call gen_decl_die on the original DECL.
12450 (process_scope_var): Avoid creating DIEs for local typedefs and
12451 concrete static variables.
12453 2017-08-12 Alan Modra <amodra@gmail.com>
12457 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
12458 match gnu-user.h startfile.
12459 (ENDFILE_LINUX_SPEC): Similarly.
12461 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
12464 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
12466 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
12468 2017-08-11 Tamar Christina <tamar.christina@arm.com>
12469 * config/aarch64/aarch64.md (mov<mode>): Change.
12470 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
12471 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
12472 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
12474 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
12476 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
12477 for storage order barriers.
12479 2017-08-11 Martin Liska <mliska@suse.cz>
12482 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
12483 variables of void type.
12485 2017-08-11 Martin Liska <mliska@suse.cz>
12487 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
12488 TARGET_SUPPORTS_ALIASES.
12489 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
12490 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
12491 (optimize_weakref): Likewise.
12492 * symtab.c (symtab_node::noninterposable_alias): Likewise.
12493 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
12494 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
12496 2017-08-11 Martin Liska <mliska@suse.cz>
12499 * config/i386/i386.c (make_resolver_func): Do complete
12500 refactoring of the function.
12502 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
12505 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
12506 * config/i386/i386.c (ix86_stack_protect_guard): Use
12507 ix86_stack_protect_guard_symbol_str to generate varible declaration.
12508 * doc/invoke.texi (x86 Options): Document
12509 -mstack-protector-guard-symbol= option.
12511 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
12513 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
12514 * config/i386/i386.c (ix86_split_stack_guard): New function.
12515 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
12516 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
12517 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
12518 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
12519 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
12520 (split_stack_space_check): Call ix86_split_stack_guard.
12522 2017-08-10 Martin Sebor <msebor@redhat.com>
12524 * print-tree.c (print_node): Print location using the established
12526 Replace spaces with colons.
12527 (debug_raw, debug): Ditto.
12529 2017-08-10 Martin Sebor <msebor@redhat.com>
12532 * pretty-print.c (pp_format): Correct the handling of %s precision.
12534 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
12537 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
12539 (ix86_finalize_stack_frame_flags): This. Also clear
12540 frame_pointer_needed if -fno-omit-frame-pointer is used without
12542 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
12543 with ix86_finalize_stack_frame_flags.
12544 (ix86_expand_epilogue): Likewise.
12545 (ix86_expand_split_stack_prologue): Likewise.
12546 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
12548 2017-08-10 Martin Liska <mliska@suse.cz>
12551 * c-attribs.c (handle_target_attribute):
12552 Report warning for an empty string argument of target attribute.
12554 2017-08-09 Jakub Jelinek <jakub@redhat.com>
12557 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
12559 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
12560 or DECL_NONLOCAL labels.
12561 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
12562 or DECL_NONLOCAL labels here.
12564 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
12566 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
12567 to indicate when early gimple folding has been disabled.
12568 (rs6000_gimple_fold_builtin): Add debug content.
12569 (rs6000_invalid_builtin): Fix whitespace.
12570 (rs6000_expand_builtin): Fix whitespace.
12571 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
12573 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
12576 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
12577 SAVE_MULTIPLE if not all the registers that saves, should be saved.
12579 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
12581 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
12582 (qdf24xx): Likewise.
12583 * config/aarch64/aarch64.md: Include falkor.md.
12584 * config/aarch64/falkor.md: New.
12586 2017-08-09 Marek Polacek <polacek@redhat.com>
12589 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
12590 * diagnostic.c (emit_diagnostic): Add a comment.
12591 (emit_diagnostic_valist): New function.
12593 2017-08-09 Marek Polacek <polacek@redhat.com>
12596 * input.c (make_location): New overload.
12597 * input.h (make_location): Declare.
12599 2017-08-08 Alan Modra <amodra@gmail.com>
12600 H.J. Lu <hongjiu.lu@intel.com>
12603 * gcc.c (NO_PIE_SPEC): Delete.
12604 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
12606 (LINK_PIE_SPEC): ..to here.
12607 (LINK_COMMAND_SPEC): Support -no-pie.
12608 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
12609 chain of crtbegin*.o selection, update for PIE_SPEC changes and
12611 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
12612 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
12613 (ENDFILE_CRTEND_SPEC): Similarly.
12615 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
12618 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
12619 (mstack-protector-guard-offset=): Ditto.
12620 * config/i386/i386.c (ix86_option_override): Handle
12621 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
12623 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
12624 ix86_stack_protect_guard_offset variables.
12625 (TARGET_STACK_PROTECT_GUARD): Always define.
12626 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
12627 and -mstack-protector-guard-offset= options.
12629 2017-08-08 Bin Cheng <bin.cheng@arm.com>
12631 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
12632 boundary case for the last candidate.
12634 2017-08-08 Bin Cheng <bin.cheng@arm.com>
12636 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
12637 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
12639 2017-08-08 Tamar Christina <tamar.christina@arm.com>
12641 PR middle-end/19706
12642 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
12643 * config/aarch64/aarch64-builtins.c
12644 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
12645 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
12646 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
12648 2017-08-08 Tamar Christina <tamar.christina@arm.com>
12649 Andrew Pinski <pinskia@gmail.com>
12651 PR middle-end/19706
12652 * internal-fn.def (XORSIGN): New.
12653 * optabs.def (xorsign_optab): New.
12654 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
12655 (convert_expand_mult_copysign): New.
12656 (pass_optimize_widening_mul::execute): Call
12657 convert_expand_mult_copysign.
12659 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12661 PR tree-optimization/81354
12662 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
12663 Insert on edges rather than explicitly creating landing pads.
12664 (analyze_candidates_and_replace): Commit edge inserts.
12666 2017-08-08 Richard Biener <rguenther@suse.de>
12668 PR middle-end/81719
12669 * tree-ssa-loop-niter.c: Include tree-dfa.h.
12670 (expand_simple_operations): Also look through ADDR_EXPRs with
12671 MEM_REF bases treating them as POINTER_PLUS_EXPR.
12673 2017-08-08 Richard Biener <rguenther@suse.de>
12675 PR tree-optimization/81723
12676 * tree-vect-slp.c (struct bst_traits): New hash traits.
12677 (bst_fail): New global.
12678 (vect_build_slp_tree_2): New worker, split out from ...
12679 (vect_build_slp_tree): ... this now wrapping it with using
12680 bst_fail set to cache SLP tree build fails. Properly handle
12682 (vect_analyze_slp_instance): Allocate and free bst_fail.
12684 2017-08-08 Martin Liska <mliska@suse.cz>
12687 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
12688 LABEL_DECLs that can be from a different function.
12690 2017-08-08 Bin Cheng <bin.cheng@arm.com>
12692 PR tree-optimization/81744
12693 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
12694 loop's number of iterations.
12696 2017-08-08 Martin Liska <mliska@suse.cz>
12698 * asan.c: Include header files.
12699 * attribs.c (build_decl_attribute_variant): New function moved
12701 (build_type_attribute_qual_variant): Likewise.
12702 (cmp_attrib_identifiers): Likewise.
12703 (simple_cst_list_equal): Likewise.
12704 (omp_declare_simd_clauses_equal): Likewise.
12705 (attribute_value_equal): Likewise.
12706 (comp_type_attributes): Likewise.
12707 (build_type_attribute_variant): Likewise.
12708 (lookup_ident_attribute): Likewise.
12709 (remove_attribute): Likewise.
12710 (merge_attributes): Likewise.
12711 (merge_type_attributes): Likewise.
12712 (merge_decl_attributes): Likewise.
12713 (merge_dllimport_decl_attributes): Likewise.
12714 (handle_dll_attribute): Likewise.
12715 (attribute_list_equal): Likewise.
12716 (attribute_list_contained): Likewise.
12717 * attribs.h (lookup_attribute): New function moved from tree.[ch].
12718 (lookup_attribute_by_prefix): Likewise.
12719 * bb-reorder.c: Include header files.
12720 * builtins.c: Likewise.
12721 * calls.c: Likewise.
12722 * cfgexpand.c: Likewise.
12723 * cgraph.c: Likewise.
12724 * cgraphunit.c: Likewise.
12725 * convert.c: Likewise.
12726 * dwarf2out.c: Likewise.
12727 * final.c: Likewise.
12728 * fold-const.c: Likewise.
12729 * function.c: Likewise.
12730 * gimple-expr.c: Likewise.
12731 * gimple-fold.c: Likewise.
12732 * gimple-pretty-print.c: Likewise.
12733 * gimple.c: Likewise.
12734 * gimplify.c: Likewise.
12735 * hsa-common.c: Likewise.
12736 * hsa-gen.c: Likewise.
12737 * internal-fn.c: Likewise.
12738 * ipa-chkp.c: Likewise.
12739 * ipa-cp.c: Likewise.
12740 * ipa-devirt.c: Likewise.
12741 * ipa-fnsummary.c: Likewise.
12742 * ipa-inline.c: Likewise.
12743 * ipa-visibility.c: Likewise.
12745 * lto-cgraph.c: Likewise.
12746 * omp-expand.c: Likewise.
12747 * omp-general.c: Likewise.
12748 * omp-low.c: Likewise.
12749 * omp-offload.c: Likewise.
12750 * omp-simd-clone.c: Likewise.
12751 * opts-global.c: Likewise.
12752 * passes.c: Likewise.
12753 * predict.c: Likewise.
12754 * sancov.c: Likewise.
12755 * sanopt.c: Likewise.
12756 * symtab.c: Likewise.
12757 * toplev.c: Likewise.
12758 * trans-mem.c: Likewise.
12759 * tree-chkp.c: Likewise.
12760 * tree-eh.c: Likewise.
12761 * tree-into-ssa.c: Likewise.
12762 * tree-object-size.c: Likewise.
12763 * tree-parloops.c: Likewise.
12764 * tree-profile.c: Likewise.
12765 * tree-ssa-ccp.c: Likewise.
12766 * tree-ssa-live.c: Likewise.
12767 * tree-ssa-loop.c: Likewise.
12768 * tree-ssa-sccvn.c: Likewise.
12769 * tree-ssa-structalias.c: Likewise.
12770 * tree-ssa.c: Likewise.
12771 * tree-streamer-in.c: Likewise.
12772 * tree-vectorizer.c: Likewise.
12773 * tree-vrp.c: Likewise.
12774 * tsan.c: Likewise.
12775 * ubsan.c: Likewise.
12776 * varasm.c: Likewise.
12777 * varpool.c: Likewise.
12778 * tree.c: Remove functions moved to attribs.[ch].
12779 * tree.h: Likewise.
12780 * config/aarch64/aarch64.c: Add attrs.h header file.
12781 * config/alpha/alpha.c: Likewise.
12782 * config/arc/arc.c: Likewise.
12783 * config/arm/arm.c: Likewise.
12784 * config/avr/avr.c: Likewise.
12785 * config/bfin/bfin.c: Likewise.
12786 * config/c6x/c6x.c: Likewise.
12787 * config/cr16/cr16.c: Likewise.
12788 * config/cris/cris.c: Likewise.
12789 * config/darwin.c: Likewise.
12790 * config/epiphany/epiphany.c: Likewise.
12791 * config/fr30/fr30.c: Likewise.
12792 * config/frv/frv.c: Likewise.
12793 * config/ft32/ft32.c: Likewise.
12794 * config/h8300/h8300.c: Likewise.
12795 * config/i386/winnt.c: Likewise.
12796 * config/ia64/ia64.c: Likewise.
12797 * config/iq2000/iq2000.c: Likewise.
12798 * config/lm32/lm32.c: Likewise.
12799 * config/m32c/m32c.c: Likewise.
12800 * config/m32r/m32r.c: Likewise.
12801 * config/m68k/m68k.c: Likewise.
12802 * config/mcore/mcore.c: Likewise.
12803 * config/microblaze/microblaze.c: Likewise.
12804 * config/mips/mips.c: Likewise.
12805 * config/mmix/mmix.c: Likewise.
12806 * config/mn10300/mn10300.c: Likewise.
12807 * config/moxie/moxie.c: Likewise.
12808 * config/msp430/msp430.c: Likewise.
12809 * config/nds32/nds32-isr.c: Likewise.
12810 * config/nds32/nds32.c: Likewise.
12811 * config/nios2/nios2.c: Likewise.
12812 * config/nvptx/nvptx.c: Likewise.
12813 * config/pa/pa.c: Likewise.
12814 * config/pdp11/pdp11.c: Likewise.
12815 * config/powerpcspe/powerpcspe.c: Likewise.
12816 * config/riscv/riscv.c: Likewise.
12817 * config/rl78/rl78.c: Likewise.
12818 * config/rx/rx.c: Likewise.
12819 * config/s390/s390.c: Likewise.
12820 * config/sh/sh.c: Likewise.
12821 * config/sol2.c: Likewise.
12822 * config/sparc/sparc.c: Likewise.
12823 * config/spu/spu.c: Likewise.
12824 * config/stormy16/stormy16.c: Likewise.
12825 * config/tilegx/tilegx.c: Likewise.
12826 * config/tilepro/tilepro.c: Likewise.
12827 * config/v850/v850.c: Likewise.
12828 * config/vax/vax.c: Likewise.
12829 * config/visium/visium.c: Likewise.
12830 * config/xtensa/xtensa.c: Likewise.
12832 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
12835 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
12836 constraints since the -mupper-regs-* switches have been
12838 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
12839 into a vector from a double word element that was extracted from
12840 another vector, and eliminate extra XXPERMDI instructions.
12841 (vsx_concat_<mode>_2): Likewise.
12842 (vsx_concat_<mode>_3): Likewise.
12843 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
12844 concat to allow optimizing inserts from previous extracts.
12846 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
12848 * config/i386/i386.c (ix86_stack_protect_guard): Generate
12849 memory reference to a SSP offset in TLS address space.
12850 (ix86_print_operand) <case '@'>: Remove.
12851 (ix86_print_operand_punct_valid_p): Remove '@' code.
12852 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
12853 UNSPEC_SP_TLS_TEST.
12854 (stack_tls_protect_set_<mode>): Remove.
12855 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
12856 (stack_tls_protect_test_<mode>): Remove.
12857 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
12859 2017-08-07 Olivier Hainque <hainque@adacore.com>
12862 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
12864 2017-08-07 Douglas Rupp <rupp@adacore.com>
12866 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
12867 variable was referenced as multidir in command.
12869 2017-08-07 Jakub Jelinek <jakub@redhat.com>
12872 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
12875 2017-08-07 Martin Liska <mliska@suse.cz>
12877 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
12878 * config/rl78/rl78.c: Add include of attribs.h.
12879 * config/sh/sh.c: Likewise.
12880 * config/v850/v850.c: Likewise.
12882 2017-08-07 Tom de Vries <tom@codesourcery.com>
12884 PR middle-end/78266
12885 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
12887 2017-08-07 Martin Liska <mliska@suse.cz>
12889 * config/mips/mips.c: Include attribs.h.
12891 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
12894 * doc/invoke.texi: Document change in behvaior for -Ofast for
12897 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
12899 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
12901 (aarch64_pop_regs): Likewise.
12902 (aarch64_gen_load_pair): Likewise.
12903 (aarch64_save_callee_saves): Likewise.
12904 (aarch64_restore_callee_saves): Likewise.
12906 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
12908 * config/i386/i386.c: Revert the last change.
12910 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
12913 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
12915 (ix86_finalize_stack_frame_flags): This. Also clear
12916 frame_pointer_needed if -fno-omit-frame-pointer is used without
12918 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
12919 with ix86_finalize_stack_frame_flags.
12920 (ix86_expand_epilogue): Likewise.
12921 (ix86_expand_split_stack_prologue): Likewise.
12923 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
12926 * config/i386/i386.c (get_builtin_code_for_version): Set priority
12927 to P_AES for Westmere.
12929 2017-08-07 Jonathan Yong <10walls@gmail.com>
12931 * config/i386/mingw.opt (fset-stack-executable): Removed.
12932 * config/i386/cygming.opt (fset-stack-executable): Moved
12934 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
12936 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
12938 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
12940 2017-08-07 Marek Polacek <polacek@redhat.com>
12942 PR middle-end/81737
12943 * fold-const.c (fold_indirect_ref_1): Check type_domain.
12945 2017-08-07 Martin Liska <mliska@suse.cz>
12947 * attribs.h (canonicalize_attr_name): New function.
12948 (cmp_attribs): Move from c-format.c and adjusted.
12949 (is_attribute_p): Moved from tree.h.
12950 * tree-inline.c: Add new includes.
12951 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
12952 (private_is_attribute_p): Remove.
12953 (private_lookup_attribute): Likewise.
12954 (private_lookup_attribute_by_prefix): Simplify.
12955 (remove_attribute): Use is_attribute_p.
12956 * tree.h: Remove removed declarations.
12958 2017-08-07 Jakub Jelinek <jakub@redhat.com>
12960 PR middle-end/81698
12961 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
12962 instead of computing it in the function. Formatting fix.
12963 (expand_case): Don't rely on default_edge being the first edge,
12964 clear it if removing it, pass default_edge to
12965 emit_case_dispatch_table.
12966 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
12969 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
12971 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
12972 insn in the function, emit NOP after the insn.
12974 2017-08-06 Tom de Vries <tom@codesourcery.com>
12976 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
12979 2017-08-06 Tom de Vries <tom@codesourcery.com>
12981 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
12984 2017-08-04 Yury Gribov <tetra2005@gmail.com>
12986 PR tree-optimization/57371
12987 * match.pd: New pattern.
12989 2017-08-04 Marek Polacek <polacek@redhat.com>
12991 PR middle-end/81695
12992 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
12993 perform the computation in offset_int.
12995 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
12997 PR tree-optimization/81136
12998 * tree-vectorizer.h: Include tree-hash-traits.h.
12999 (vec_base_alignments): New typedef.
13000 (vec_info): Add a base_alignments field.
13001 (vect_record_base_alignments): Declare.
13002 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
13004 (DR_IS_CONDITIONAL_IN_STMT): New macro.
13005 (create_data_ref): Add an is_conditional_in_stmt argument.
13006 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
13007 the is_conditional_in_stmt field.
13008 (data_ref_loc): Add an is_conditional_in_stmt field.
13009 (get_references_in_stmt): Set the is_conditional_in_stmt field.
13010 (find_data_references_in_stmt): Update call to create_data_ref.
13011 (graphite_find_data_references_in_stmt): Likewise.
13012 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
13013 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
13014 (vect_record_base_alignment): New function.
13015 (vect_record_base_alignments): Likewise.
13016 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
13017 for nested statements even if we fail to compute a misalignment.
13018 Use pooled base alignments for unconditional references.
13019 (vect_find_same_alignment_drs): Compare base addresses instead
13021 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
13022 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
13024 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13026 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
13027 Add an explicit name for the enum. Use auto_vec for slp_instances
13028 and grouped_stores.
13029 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
13031 (_bb_vec_info): Add a constructor and destructor.
13032 (vinfo_for_stmt): Return NULL for uids of -1 as well.
13033 (destroy_loop_vec_info): Delete.
13034 (vect_destroy_datarefs): Likewise.
13035 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
13036 (vec_info::vec_info): New function.
13037 (vec_info::~vec_info): Likewise.
13038 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
13039 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
13040 destroy_loop_vec_info.
13041 * tree-vect-loop.c (new_loop_vec_info): Replace with...
13042 (_loop_vec_info::_loop_vec_info): ...this.
13043 (destroy_loop_vec_info): Replace with...
13044 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
13045 the stmt_vec_infos. Leave handling of vec_info information to its
13046 destructor. Remove explicit vector releases.
13047 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
13048 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
13049 * tree-vect-slp.c (new_bb_vec_info): Replace with...
13050 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
13051 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
13052 (destroy_bb_vec_info): Replace with...
13053 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
13054 information to its destructor.
13055 (vect_slp_analyze_bb_1): Use new and delete instead of
13056 new_bb_vec_info and destroy_bb_vec_info.
13057 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
13060 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13062 * tree-data-ref.h (subscript): Add access_fn field.
13063 (data_dependence_relation): Add could_be_independent_p.
13064 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
13065 (same_access_functions): Move to tree-data-ref.c.
13066 * tree-data-ref.c (ref_contains_union_access_p): New function.
13067 (access_fn_component_p): Likewise.
13068 (access_fn_components_comparable_p): Likewise.
13069 (dr_analyze_indices): Add a reference to access_fn_component_p.
13070 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
13072 (constant_access_functions): Likewise.
13073 (add_other_self_distances): Likewise.
13074 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
13075 (initialize_data_dependence_relation): Use XCNEW and remove
13076 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
13077 of access functions that have the same type. Allow the
13078 subsequence to end with different bases in some circumstances.
13079 Record the chosen access functions in SUB_ACCESS_FN.
13080 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
13081 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
13082 (subscript_dependence_tester_1): Likewise dra and drb.
13083 (build_classic_dist_vector): Update calls accordingly.
13084 (subscript_dependence_tester): Likewise.
13085 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
13086 DDR_COULD_BE_INDEPENDENT_P.
13087 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
13088 comp_alias_ddrs instead of may_alias_ddrs.
13089 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
13091 (vect_analyze_data_ref_dependence): Use it if
13092 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
13093 distance vectors if that fails.
13094 (dependence_distance_ge_vf): New function.
13095 (vect_prune_runtime_alias_test_list): Use it. Don't clear
13096 LOOP_VINFO_MAY_ALIAS_DDRS.
13098 2017-08-04 Richard Biener <rguenther@suse.de>
13100 PR middle-end/81705
13101 * fold-const.c (fold_binary_loc): Properly restrict
13102 minus_var0 && minus_var1 case when associating undefined overflow
13105 2017-08-04 Tom de Vries <tom@codesourcery.com>
13107 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
13109 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13111 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13112 Don't start diagnostic messages with a capital letter.
13113 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13115 (rs6000_invalid_builtin): Likewise.
13116 (rs6000_trampoline_init): Likewise.
13118 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13121 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
13122 after setting changeable df flags.
13124 2017-08-03 Richard Biener <rguenther@suse.de>
13126 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
13127 up if the use is in USE - X.
13129 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
13131 * toplev.c (dumpfile.h): New include.
13132 (internal_error_reentered): New static function. Use it...
13133 (internal_error_function): ...here to handle reentered internal_error.
13135 2017-08-03 Richard Biener <rguenther@suse.de>
13137 PR middle-end/81148
13138 * fold-const.c (split_tree): Add minus_var and minus_con
13139 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
13140 here but always use minus_*.
13141 (associate_trees): Assert we never associate with MINUS_EXPR
13142 and NULL first operand. Do not recurse for PLUS_EXPR operands
13143 when associating as MINUS_EXPR either.
13144 (fold_binary_loc): Track minus_var and minus_con.
13146 2017-08-03 Tom de Vries <tom@codesourcery.com>
13149 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
13150 ACCEL_COMPILER, apply finish_options on
13151 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
13153 2017-08-03 Tom de Vries <tom@codesourcery.com>
13156 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
13157 function_entry_patch_area_size > 0.
13159 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13162 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
13163 instead of 10??LU, perform unit multiplication in wide_int,
13164 don't change alloc_object_size_limit if the limit is larger
13167 PR tree-optimization/81655
13168 PR tree-optimization/81588
13169 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
13170 the case when ranges[i].low and high are 1 for unsigned type with
13173 PR middle-end/81052
13174 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
13175 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
13177 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13179 * tree-vrp.h: Add include guard.
13181 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
13184 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
13185 (ud2): New insn pattern.
13186 * config/i386/i386.c (ix86_expand_epilogue):
13187 For naked functions, generate ud2 instead of trap insn.
13189 2017-08-02 Marek Polacek <polacek@redhat.com>
13192 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
13194 2017-08-02 Tom de Vries <tom@codesourcery.com>
13195 Cesar Philippidis <cesar@codesourcery.com>
13197 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
13198 Add missing edge probabilities.
13200 2017-08-02 Tamar Christina <tamar.christina@arm.com>
13202 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
13203 Correct endianness.
13205 2017-08-02 Jakub Jelinek <jakub@redhat.com>
13207 PR middle-end/79499
13208 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
13209 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
13210 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
13211 prologue_seq sequences - if any.
13213 2017-08-02 Richard Biener <rguenther@suse.de>
13215 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
13216 via vectors if supported, integer extracts via punning if supported
13217 or otherwise vector extracts.
13219 2017-08-02 Richard Biener <rguenther@suse.de>
13221 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
13223 (bitmap_insert_into_set): ... this.
13225 2017-08-02 Richard Biener <rguenther@suse.de>
13227 PR tree-optimization/81633
13229 2015-08-17 Alan Hayward <alan.hayward@arm.com>
13231 PR tree-optimization/71752
13232 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
13234 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
13236 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
13237 (machine_function::call_ms2sysv_pad_out): Remove field.
13238 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
13239 (ix86_compute_frame_layout): Likewise.
13241 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
13244 * config/i386/i386.c (ix86_set_func_type): Disallow naked
13245 attribute with interrupt attribute.
13247 2017-08-01 Andrew Pinski <apinski@cavium.com>
13249 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
13250 BIT_INSERT_EXPR's operand 1
13251 to see if the types precision matches.
13253 2017-08-01 Martin Liska <mliska@suse.cz>
13255 PR middle-end/70140
13256 * builtins.c (expand_builtin_memcpy_args): Remove.
13257 (expand_builtin_memcpy): Call newly added function
13258 expand_builtin_memory_copy_args.
13259 (expand_builtin_memcpy_with_bounds): Likewise.
13260 (expand_builtin_mempcpy): Remove last argument.
13261 (expand_builtin_mempcpy_with_bounds): Likewise.
13262 (expand_builtin_memory_copy_args): New function created from
13263 expand_builtin_mempcpy_args with small modifications.
13264 (expand_builtin_mempcpy_args): Remove.
13265 (expand_builtin_stpcpy): Remove unused argument.
13266 (expand_builtin): Likewise.
13267 (expand_builtin_with_bounds): Likewise.
13269 2017-08-01 Martin Liska <mliska@suse.cz>
13272 Make mempcpy more optimal (PR middle-end/70140).
13274 2017-08-01 Jakub Jelinek <jakub@redhat.com>
13277 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
13278 __builtin_vec_cmpne verify both arguments are compatible vectors
13279 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
13280 verify arg1_type is a pointer or array type. For __builtin_vec_st,
13281 move computation of aligned to after checking the argument types.
13285 * config/rs6000/vsx.md (vextract_fp_from_shorth,
13286 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
13289 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
13290 Doug Rupp <rupp@adacore.com>
13291 Olivier Hainque <hainque@adacore.com>
13293 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
13294 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
13296 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
13297 for TARGET_OS_CPP_BUILTIN.
13298 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
13299 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
13301 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
13302 passing required abi options to the assembler for EABI configurations.
13303 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
13304 of .text.hot and .text.unlikely sections for kernel modules when
13305 using ARM style exceptions.
13306 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
13307 options. Add EXTRA_CC1_SPEC.
13308 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
13310 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
13312 (ARM_TARGET2_DWARF_FORMAT): Define.
13313 * config/arm/t-vxworks: Adjust multilib control to removal of the
13314 Diab command line options.
13316 2017-08-01 Martin Liska <mliska@suse.cz>
13318 PR gcov-profile/81561
13319 * gcov.c (unblock): Make unblocking safe as we need to preserve
13320 index correspondence of blocks and block_lists.
13322 2017-08-01 Richard Biener <rguenther@suse.de>
13324 PR tree-optimization/81181
13325 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
13326 (compute_antic): ... end of iteration here.
13328 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
13330 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
13331 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
13332 (ftree-slp-vectorize): Likewise.
13333 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
13334 can no longer be set independent of flag_tree_loop_vectorize.
13335 * omp-general.c (emp_max_vf): Likewise.
13336 * opts.c (enable_fdo_optimizations): Remove references to
13337 flag_tree_vectorize, these are now implicit.
13338 (common_handle_option): Remove handling for OPT_ftree_vectorize,
13339 and leave it for the options machinery.
13341 2017-08-01 Martin Liska <mliska@suse.cz>
13343 PR middle-end/70140
13344 * builtins.c (expand_builtin_memcpy_args): Remove.
13345 (expand_builtin_memcpy): Call newly added function
13346 expand_builtin_memory_copy_args.
13347 (expand_builtin_memcpy_with_bounds): Likewise.
13348 (expand_builtin_mempcpy): Remove last argument.
13349 (expand_builtin_mempcpy_with_bounds): Likewise.
13350 (expand_builtin_memory_copy_args): New function created from
13351 expand_builtin_mempcpy_args with small modifications.
13352 (expand_builtin_mempcpy_args): Remove.
13353 (expand_builtin_stpcpy): Remove unused argument.
13354 (expand_builtin): Likewise.
13355 (expand_builtin_with_bounds): Likewise.
13357 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
13360 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
13361 print "ds:" only for immediates in generic address space.
13363 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
13366 * config/i386/i386.c (ix86_funciton_naked): New prototype.
13367 (ix86_function_ok_for_sibcall): Return false for naked functions.
13369 2017-08-01 Richard Biener <rguenther@suse.de>
13371 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
13372 (compute_antic): Seed worklist with exit block predecessors.
13373 * cfganal.c (dfs_find_deadend): For a cycle return the source
13374 of the edge closing it.
13376 2017-08-01 Tamar Christina <tamar.christina@arm.com>
13378 * config/aarch64/aarch64.c
13379 (aarch64_can_const_movi_rtx_p): Move 0 check.
13381 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13383 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
13384 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
13386 * match.pd: Ditto in address comparison pattern.
13388 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13390 PR tree-optimization/81627
13391 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
13392 closed ssa form for store-store chain.
13394 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13396 PR tree-optimization/81620
13397 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
13398 for store-store chain.
13400 2017-08-01 Jakub Jelinek <jakub@redhat.com>
13402 PR tree-optimization/81588
13403 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
13404 ranges[i].in_p, invert comparison code ccode. For >/>=,
13405 swap rhs1 and rhs2 and comparison code unconditionally,
13406 for </<= don't do that. Don't swap rhs1/rhs2 again if
13407 ranges[i].in_p, instead invert comparison code ccode if
13408 opcode or oe->rank is BIT_IOR_EXPR.
13411 * optabs.def (vec_extract_optab, vec_init_optab): Change from
13412 a direct optab to conversion optab.
13413 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
13414 with GET_MODE_INNER as last argument instead of optab_handler.
13415 * expmed.c (extract_bit_field_1): Likewise. Use vector from
13416 vector extraction if possible and optab is available.
13417 * expr.c (store_constructor): Use convert_optab_handler instead
13418 of optab_handler. Use vector initialization from smaller
13419 vectors if possible and optab is available.
13420 * tree-vect-stmts.c (vectorizable_load): Likewise.
13421 * doc/md.texi (vec_extract, vec_init): Document that the optabs
13422 now have two modes.
13423 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
13424 of vec_init from half-sized vectors with the same element mode.
13425 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
13426 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
13427 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
13428 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
13429 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
13430 after mode in gen_vec_extract* calls.
13431 (vec_extract<mode>): Renamed to ...
13432 (vec_extract<mode><ssescalarmodelower>): ... this.
13433 (vec_extract<mode><ssehalfvecmodelower>): New expander.
13434 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
13435 element mode after mode in gen_vec_init* calls.
13436 (VEC_INIT_HALF_MODE): New mode iterator.
13437 (vec_init<mode>): Renamed to ...
13438 (vec_init<mode><ssescalarmodelower>): ... this.
13439 (vec_init<mode><ssehalfvecmodelower>): New expander.
13440 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
13441 (vec_extractv2sfsf): ... this.
13442 (vec_initv2sf): Renamed to ...
13443 (vec_initv2sfsf): ... this.
13444 (vec_extractv2si): Renamed to ...
13445 (vec_extractv2sisi): ... this.
13446 (vec_initv2si): Renamed to ...
13447 (vec_initv2sisi): ... this.
13448 (vec_extractv4hi): Renamed to ...
13449 (vec_extractv4hihi): ... this.
13450 (vec_initv4hi): Renamed to ...
13451 (vec_initv4hihi): ... this.
13452 (vec_extractv8qi): Renamed to ...
13453 (vec_extractv8qiqi): ... this.
13454 (vec_initv8qi): Renamed to ...
13455 (vec_initv8qiqi): ... this.
13456 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
13457 (vec_init<mode>): Renamed to ...
13458 (vec_init<mode><VEC_base_l>): ... this.
13459 (vec_extract<mode>): Renamed to ...
13460 (vec_extract<mode><VEC_base_l>): ... this.
13461 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
13462 (vec_initv2sfsf): ... this.
13463 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
13464 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
13465 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
13466 element mode after mode in gen_vec_init* calls.
13467 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
13468 (vec_init<mode><Vel>): ... this.
13469 (vec_extract<mode>): Renamed to ...
13470 (vec_extract<mode><Vel>): ... this.
13471 * config/aarch64/iterators.md (Vel): New mode attribute.
13472 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
13473 Add element mode after mode in gen_vec_extract* calls.
13474 * config/s390/vector.md (non_vec_l): New mode attribute.
13475 (vec_extract<mode>): Renamed to ...
13476 (vec_extract<mode><non_vec_l>): ... this.
13477 (vec_init<mode>): Renamed to ...
13478 (vec_init<mode><non_vec_l>): ... this.
13479 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
13480 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
13482 * config/arm/iterators.md (V_elem_l): New mode attribute.
13483 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
13484 (vec_extract<mode><V_elem_l>): ... this.
13485 (vec_extractv2di): Renamed to ...
13486 (vec_extractv2didi): ... this.
13487 (vec_init<mode>): Renamed to ...
13488 (vec_init<mode><V_elem_l>): ... this.
13489 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
13490 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
13491 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
13492 Add element mode after gen_vec_extract* calls.
13493 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
13494 (vec_init<mode><unitmode>): ... this.
13495 (vec_extract<mode>): Renamed to ...
13496 (vec_extract<mode><unitmode>): ... this.
13497 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
13498 (vec_init<mode><unitmode>): ... this.
13499 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
13500 (vec_initv2sfsf): ... this.
13501 (vec_extractv2sf): Renamed to ...
13502 (vec_extractv2sfsf): ... this.
13503 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
13504 Add element mode after gen_vec_extract* calls.
13505 * config/mips/mips.md (unitmode): New mode iterator.
13506 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
13507 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
13508 * config/spu/spu.md (inner_l): New mode attribute.
13509 (vec_init<mode>): Renamed to ...
13510 (vec_init<mode><inner_l>): ... this.
13511 (vec_extract<mode>): Renamed to ...
13512 (vec_extract<mode><inner_l>): ... this.
13513 * config/sparc/sparc.md (veltmode): New mode iterator.
13514 (vec_init<VMALL:mode>): Renamed to ...
13515 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
13516 * config/ia64/vect.md (vec_initv2si): Renamed to ...
13517 (vec_initv2sisi): ... this.
13518 (vec_initv2sf): Renamed to ...
13519 (vec_initv2sfsf): ... this.
13520 (vec_extractv2sf): Renamed to ...
13521 (vec_extractv2sfsf): ... this.
13522 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
13523 (vec_init<mode>): Renamed to ...
13524 (vec_init<mode><VEC_base_l>): ... this.
13525 (vec_extract<mode>): Renamed to ...
13526 (vec_extract<mode><VEC_base_l>): ... this.
13527 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
13528 (vec_initv2sfsf): ... this.
13529 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
13530 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
13531 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
13532 gen_vec_init* calls.
13534 2017-08-01 Richard Biener <rguenther@suse.de>
13536 PR tree-optimization/81297
13537 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
13538 TREE_OVERFLOW from INTEGER_CSTs.
13540 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
13542 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
13544 2017-07-31 Carl Love <cel@us.ibm.com>
13546 * config/rs6000/rs6000-c: Add support for built-in functions
13547 vector signed char vec_xl_be (signed long long, signed char *);
13548 vector unsigned char vec_xl_be (signed long long, unsigned char *);
13549 vector signed int vec_xl_be (signed long long, signed int *);
13550 vector unsigned int vec_xl_be (signed long long, unsigned int *);
13551 vector signed long long vec_xl_be (signed long long, signed long long *);
13552 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
13553 vector signed short vec_xl_be (signed long long, signed short *);
13554 vector unsigned short vec_xl_be (signed long long, unsigned short *);
13555 vector double vec_xl_be (signed long long, double *);
13556 vector float vec_xl_be (signed long long, float *);
13557 * config/rs6000/altivec.h (vec_xl_be): Add #define.
13558 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
13559 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
13561 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
13562 (altivec_expand_builtin): Add switch statement to call
13563 altivec_expand_xl_be for each builtin.
13564 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
13565 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
13566 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
13567 __builtin_vsx_le_be_v16qi.
13568 * doc/extend.texi: Update the built-in documentation file for the
13569 new built-in functions.
13571 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
13574 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
13576 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
13578 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13580 * config.gcc: Add z14.
13581 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
13582 CPU model numbers for z13s and z14.
13583 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
13585 * config/s390/s390-opts.h (enum processor_type): Rename
13586 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
13587 * config/s390/s390.c (processor_table): Add field for CPU name to
13588 be passed to Binutils.
13589 (s390_asm_output_machine_for_arch): Use the new field in
13590 processor_table for Binutils.
13591 (s390_expand_builtin): Replace arch12 with z14.
13592 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
13593 (s390_get_sched_attrmask): Likewise.
13594 (s390_get_unit_mask): Likewise.
13595 * config/s390/s390.opt: Add z14 to processor_type enum.
13597 2017-07-31 Martin Jambor <mjambor@suse.cz>
13600 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
13601 regardless of optimization level.
13603 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
13604 Martin Liska <mliska@suse.cz>
13606 * predict.def: Remove old comment and adjust probability.
13607 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
13608 PREDICT statements.
13610 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
13613 * config/i386/i386.c (ix86_function_naked): New function.
13614 (ix86_can_use_return_insn_p): Return false for naked functions.
13615 (ix86_expand_prologue): Skip prologue for naked functions.
13616 (ix86_expand_epilogue): Skip epilogue for naked functions
13617 and emit trap instruction.
13618 (ix86_warn_func_return): New function.
13619 (ix86_attribute_table): Add "naked" attribute specification.
13620 (TARGET_WARN_FUNC_RETURN): Define.
13621 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
13623 2017-07-31 Martin Liska <mliska@suse.cz>
13625 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
13626 (dump_gimple_bb_header): Always dump BB info.
13627 (pp_cfg_jump): Do not append info about BB when dumping a jump.
13629 2017-07-31 Martin Liska <mliska@suse.cz>
13632 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
13633 also with current_function_decl non-null equality.
13635 2017-07-31 Jakub Jelinek <jakub@redhat.com>
13638 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
13639 change type to the element type, instead add eltype variable and
13640 use it where we are interested in the element type.
13642 PR tree-optimization/81603
13643 * ipa-polymorphic-call.c
13644 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
13645 offset arithmetic in offset_int, bail out if the resulting bit offset
13646 doesn't fit into shwi.
13648 2017-07-31 Martin Liska <mliska@suse.cz>
13650 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
13651 (gimplify_save_expr): Fix comment.
13653 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
13656 * config/i386/i386.c (ix86_function_arg): Update arguments for
13658 (ix86_compute_frame_layout): Set the initial stack offset to
13659 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
13660 INCOMING_FRAME_SP_OFFSET.
13661 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
13662 stack before exception handler returns.
13663 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
13664 the 'ERROR_CODE' for exception handler.
13666 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
13668 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
13669 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
13670 (ASM_OUTPUT_REG_POP): Ditto.
13671 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
13672 instead of asm_fprintf to output pure string.
13674 2017-07-29 Jakub Jelinek <jakub@redhat.com>
13676 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
13677 to imported_module_or_decl hook.
13678 (debug_nothing_tree_tree_tree_bool): Remove.
13679 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
13680 * debug.c (do_nothing_debug_hooks): Use
13681 debug_nothing_tree_tree_tree_bool_bool instead of
13682 debug_nothing_tree_tree_tree_bool.
13683 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
13684 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
13685 * sdbout.c (sdb_debug_hooks): Likewise.
13686 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
13687 (gen_namespace_die): Add DW_AT_export_symbols attribute if
13689 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
13690 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
13691 attribute, don't add anything.
13693 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13695 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
13696 (fold_build2_stat_loc): Likewise.
13697 (fold_build3_stat_loc): Likewise.
13698 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
13699 (fold_build1_loc): Remove macro.
13700 (fold_build2_loc): Likewise.
13701 (fold_build3_loc): Likewise.
13703 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13705 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
13706 (gimple_build_debug_bind_source_stat): Likewise.
13707 * gimple.h (gimple_build_debug_bind): Remove macro.
13708 (gimple_build_debug_bind_source): Likewise.
13710 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13712 * bitmap.c (bitmap_alloc): Adjust.
13713 (bitmap_gc_alloc): Likewise.
13714 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
13716 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13718 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
13719 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
13720 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
13721 (bitmap_gc_alloc_stat): Likewise.
13722 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
13724 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13726 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
13727 * rtl.h (shallow_copy_rtx): Remove macro.
13729 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13731 * emit-rtl.c (gen_raw_REG): Adjust.
13732 * gengenrtl.c (gendef): Likewise.
13733 * rtl.c (rtx_alloc_stat): Remove _stat from name.
13734 * rtl.h (rtx_alloc): Remove macro.
13736 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13738 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
13739 (build_tree_list_stat): Likewise.
13740 * tree.h (build_tree_list): Remove macro.
13741 (build_tree_list_vec): Likewise.
13743 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13745 * tree.c (make_vector_stat): Remove _stat from name.
13746 (build_vector_stat): Likewise.
13747 * tree.h (make_vector_stat): Remove macro.
13748 (build_vector_stat): Likewise.
13750 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13752 * tree.h (build_var_debug_value): Remove prototype.
13754 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13756 * tree.c (tree_cons_stat): Remove _stat from name.
13757 * tree.h (tree_cons): Remove macro.
13759 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13761 * tree.c (build_vl_exp_stat): Remove _stat from name.
13762 * tree.h (build_vl_exp): Remove macro.
13764 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13766 * tree.c (build_decl_stat): Remove _stat from name.
13767 * tree.h (build_decl): Remove macro.
13769 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13771 * gimple.c (gimple_build_with_ops_stat): Adjust.
13772 (gimple_alloc_stat): Remove _stat from name.
13773 * gimple.h (gimple_alloc): Remove macro.
13775 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13777 * tree.c (make_tree_vec_stat): Remove _stat from name.
13778 (grow_tree_vec_stat): Likewise.
13779 * tree.h (make_tree_vec_stat): Adjust prototype.
13780 (grow_tree_vec_stat): Likewise.
13781 (make_tree_vec): Remove macro.
13782 (grow_tree_vec): Likewise.
13784 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13786 * fold-const.c (fold_build1_stat_loc): Adjust.
13787 (fold_build2_stat_loc): Likewise.
13788 (fold_build3_stat_loc): Likewise.
13789 * tree.c (build0_stat): Remove _stat from name.
13790 (build1_stat): Likewise.
13791 (build2_stat): Likewise.
13792 (build3_stat): Likewise.
13793 (build4_stat): Likewise.
13794 (build5_stat): Likewise.
13795 * tree.h (build1_loc): Remove macro, and rename _stat function
13797 (build2_loc): Likewise.
13798 (build3_loc): Likewise.
13799 (build4_loc): Likewise.
13800 (build5_loc): Likewise.
13802 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13804 * tree.c (make_int_cst_stat): Remove _stat from name.
13805 * tree.h (make_int_cst_stat): Adjust prototype.
13806 (make_int_cst): Remove macro.
13808 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13810 * tree.c (make_tre_binfo_stat): Remove _stat from name.
13811 * tree.h (make_tree_binfo_stat): Adjust prototype.
13812 (make_tree_binfo): Remove.
13814 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13816 * tree.c (copy_node_stat): Rename to copy_node.
13817 (build_distinct_type_copy): Adjust.
13818 * tree.h (copy_node_stat): Adjust prototype.
13819 (copy_node): Remove macro.
13821 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13823 * tree.c (make_node_stat): rename to make_node.
13824 (build_tree_list_stat): Adjust.
13825 (build0_stat): Likewise.
13826 (build2_stat): Likewise.
13827 (build3_stat): Likewise.
13828 (build4_stat): Likewise.
13829 (build5_stat): Likewise.
13830 (build_decl_stat): Likewise.
13831 * tree.h (make_node_stat): Adjust prototype.
13832 (make_node): remove macro.
13834 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
13836 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
13837 (PPC_FEATURE2_SCV): Likewise.
13838 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
13840 2017-07-28 Tamar Christina <tamar.christina@arm.com>
13842 * config/aarch64/aarch64.c
13843 (aarch64_internal_mov_immediate): Add new special pattern.
13844 * config/aarch64/aarch64.md (*movdi_aarch64):
13845 Add reg/32bit const mov case.
13847 2017-07-28 Tamar Christina <tamar.christina@arm.com>
13848 Richard Sandiford <richard.sandiford@linaro.org>
13850 * config/aarch64/aarch64.md (mov<mode>): Generalize.
13851 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
13852 Add integer and movi cases.
13853 (movi-split-hf-df-sf split, fp16): New.
13854 (enabled): Added TARGET_FP_F16INST.
13855 * config/aarch64/iterators.md (GPF_HF): New.
13856 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
13858 2017-07-28 Tamar Christina <tamar.christina@arm.com>
13860 * config/aarch64/aarch64.c
13861 (aarch64_simd_container_mode): Add prototype.
13862 (aarch64_expand_mov_immediate): Add HI support.
13863 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
13864 (aarch64_can_const_movi_rtx_p): New.
13865 (aarch64_preferred_reload_class):
13866 Remove restrictions of using FP registers for certain SIMD operations.
13867 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
13868 (aarch64_valid_floating_const): Add integer move validation.
13869 (aarch64_simd_imm_scalar_p): Remove.
13870 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
13871 (aarch64_legitimate_constant_p): Expand list of supported cases.
13872 * config/aarch64/aarch64-protos.h
13873 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
13874 (aarch64_reinterpret_float_as_int): New.
13875 (aarch64_simd_imm_scalar_p): Remove.
13876 * config/aarch64/constraints.md (Uvi): New.
13877 (Dd): Split into Ds and new Dd.
13878 * config/aarch64/aarch64.md (*movsi_aarch64):
13880 (*movdi_aarch64): Add SIMD mov case.
13882 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13884 * tree-predcom.c: (struct chain): Handle store-store chain in which
13885 stores for elimination only store loop invariant values.
13886 (execute_pred_commoning_chain): Ditto.
13887 (prepare_initializers_chain_store_elim): Ditto.
13888 (prepare_finalizers): Ditto.
13889 (is_inv_store_elimination_chain): New function.
13890 (initialize_root_vars_store_elim_1): New function.
13892 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13894 * tree-predcom.c: Revise general description of the pass.
13895 (enum chain_type): New enum type for store elimination.
13896 (struct chain): New field supporting store elimination.
13897 (struct component): Ditto.
13898 (dump_chain): Dump store-stores chain.
13899 (release_chain): Release resources.
13900 (split_data_refs_to_components): Compute and create component
13901 contains only stores for elimination.
13902 (get_chain_last_ref_at): New function.
13903 (make_invariant_chain): Initialization.
13904 (make_rooted_chain): Specify chain type in parameter and record it.
13905 (add_looparound_copies): Skip for store-stores chain.
13906 (determine_roots_comp): Compute type of chain and pass it to
13908 (initialize_root_vars_store_elim_2): New function.
13909 (finalize_eliminated_stores): New function.
13910 (remove_stmt): Handle store for elimination.
13911 (execute_pred_commoning_chain): Execute predictive commoning on
13912 store-store chains.
13913 (determine_unroll_factor): Skip unroll for store-stores chain.
13914 (prepare_initializers_chain_store_elim): New function.
13915 (prepare_initializers_chain): Hanlde store-store chain.
13916 (prepare_finalizers_chain, prepare_finalizers): New function.
13917 (tree_predictive_commoning_loop): Return integer value indicating
13918 if loop is unrolled or lcssa form is corrupted.
13919 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
13921 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13923 * tree-predcom.c (initialize_root): Delete.
13924 (execute_pred_commoning_chain): Initialize root vars and replace
13925 reference of non-combined chain directly, rather than call above
13928 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13930 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
13931 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
13933 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13935 * tree-predcom.c (struct chain): New field init_seq.
13936 (release_chain): Release init_seq.
13937 (prepare_initializers_chain): Record intialization stmts in above
13939 (insert_init_seqs): New function.
13940 (tree_predictive_commoning_loop): Call insert_init_seqs.
13942 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13944 * tree-predcom.c (determine_roots_comp): Skip trivial components.
13946 2017-07-28 Richard Biener <rguenther@suse.de>
13948 * match.pd: Remove superfluous :c.
13949 * genmatch.c (simplify::id): Add member.
13950 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
13952 (current_id): New global.
13953 (dt_node::parent): Move from ...
13954 (dt_operand::parent): ... here. Add for_id member.
13955 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
13956 (decision_tree::find_node): Relax order requirement when
13957 merging DT_TRUE nodes to ones inbetween the current simplify
13958 and the one we try to merge with. Add diagnostic whenever
13959 we need to enforce pattern order by not merging.
13960 (decision_tree::insert): Set current_id.
13961 (decision_tree::print_node): Dump parent node and for_id.
13962 (parser::last_id): Add member.
13963 (parser::push_simplify): Assign unique id.
13964 (parser::parser): Initialize last_id.
13966 2017-07-28 Martin Liska <mliska@suse.cz>
13969 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
13970 gimple_build_debug_bind.
13972 2017-07-28 Richard Biener <rguenther@suse.de>
13974 PR tree-optimization/81502
13975 * match.pd: Add pattern combining BIT_INSERT_EXPR with
13977 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
13979 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
13980 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
13981 for BIT_FIELD_REF args.
13982 * fold-const.c (make_bit_field_ref): Likewise.
13983 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
13985 2017-07-28 Jakub Jelinek <jakub@redhat.com>
13988 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
13989 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
13990 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
13991 Or it into SANITIZER_UNDEFINED.
13992 * ubsan.c: Include gimple-fold.h and varasm.h.
13993 (ubsan_expand_ptr_ifn): New function.
13994 (instrument_pointer_overflow): New function.
13995 (maybe_instrument_pointer_overflow): New function.
13996 (instrument_object_size): Formatting fix.
13997 (pass_ubsan::execute): Call instrument_pointer_overflow
13998 and maybe_instrument_pointer_overflow.
13999 * internal-fn.c (expand_UBSAN_PTR): New function.
14000 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
14001 * sanitizer.def (__ubsan_handle_pointer_overflow,
14002 __ubsan_handle_pointer_overflow_abort): New builtins.
14003 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
14004 * internal-fn.def (UBSAN_PTR): New internal function.
14005 * opts.c (sanitizer_opts): Add pointer-overflow.
14006 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
14007 * fold-const.c (build_range_check): Compute pointer range check in
14008 integral type if pointer arithmetics would be needed. Formatting
14011 2017-07-28 Martin Liska <mliska@suse.cz>
14014 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
14015 parameters that are of a variable-length.
14017 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
14019 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
14021 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
14022 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
14023 (CRT_CALL_STATIC_FUNCTION): Likewise.
14024 (ASM_DEFAULT_SPEC): New define.
14025 (ASM_SPEC32): Likewise.
14026 (ASM_SPEC64): Likewise.
14027 (ASM_SPEC_COMMON): Likewise.
14028 (ASM_SPEC): Likewise.
14029 (INVALID_64BIT): Likewise.
14030 (LINK_OS_DEFAULT_SPEC): Likewise.
14031 (LINK_OS_SPEC32): Likewise.
14032 (LINK_OS_SPEC64): Likewise.
14033 (POWERPC_LINUX): Likewise.
14034 (PTRDIFF_TYPE): Likewise.
14035 (RESTORE_FP_PREFIX): Likewise.
14036 (RESTORE_FP_SUFFIX): Likewise.
14037 (SAVE_FP_PREFIX): Likewise.
14038 (SAVE_FP_SUFFIX): Likewise.
14039 (SIZE_TYPE): Likewise.
14040 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
14041 (TARGET_64BIT): Likewise.
14042 (TARGET_64BIT): Likewise.
14043 (TARGET_AIX): Likewise.
14044 (WCHAR_TYPE_SIZE): Likewise.
14045 (WCHAR_TYPE): Undefine.
14046 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
14047 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
14048 (CPP_OS_RTEMS_SPEC): Delete.
14049 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
14050 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
14052 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
14054 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14056 PR tree-optimization/81578
14057 * tree-parloops.c (build_new_reduction): Bail out if
14058 reduction_code isn't one of the standard OpenMP reductions.
14059 Move the details printing after that decision.
14061 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14063 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
14064 related to reload_in_progress.
14065 (splat_input_operand): Likewise.
14066 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
14068 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
14070 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
14071 (TARGET_INSTANTIATE_DECLS): Likewise.
14072 (legitimate_indexed_address_p): Delete reload_in_progress code.
14073 (rs6000_debug_legitimate_address_p): Likewise.
14074 (rs6000_eliminate_indexed_memrefs): Likewise.
14075 (rs6000_emit_le_vsx_store): Likewise.
14076 (rs6000_emit_move_si_sf_subreg): Likewise.
14077 (rs6000_emit_move): Likewise.
14078 (register_to_reg_type): Likewise.
14079 (rs6000_pre_atomic_barrier): Likewise.
14080 (rs6000_machopic_legitimize_pic_address): Likewise.
14081 (rs6000_allocate_stack_temp): Likewise.
14082 (rs6000_address_for_fpconvert): Likewise.
14083 (rs6000_address_for_altivec): Likewise.
14084 (rs6000_secondary_memory_needed_rtx): Delete function.
14085 (rs6000_check_sdmode): Likewise.
14086 (rs6000_alloc_sdmode_stack_slot): Likewise.
14087 (rs6000_instantiate_decls): Likewise.
14088 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
14089 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
14090 Delete reload_in_progress.
14091 (*vec_reload_and_plus_<mptrsize>): Likewise.
14092 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
14093 (vsx_div_v2di): Likewise.
14094 (vsx_udiv_v2di): Likewise.
14096 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14098 * config/rs6000/rs6000.opt (mlra): Replace with stub.
14099 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
14100 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
14101 (rs6000_debug_reg_global): Delete print of LRA status.
14102 (rs6000_option_override_internal): Delete dead LRA related code.
14103 (rs6000_lra_p): Delete function.
14104 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
14106 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
14108 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
14109 * config/riscv/rtems.h: New file.
14111 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14112 Sudakshina Das <sudi.das@arm.com>
14114 * config/aarch64/aarch64.md
14115 (define_split for and<mode>3nr_compare): Move
14116 non aarch64_logical_operand to a register.
14117 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
14118 register immediate operand to a register.
14119 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
14121 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14123 PR middle-end/81564
14124 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
14126 2017-07-27 Richard Biener <rguenther@suse.de>
14128 PR tree-optimization/81573
14129 PR tree-optimization/81494
14130 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
14131 multi defuse cycle case.
14133 2017-07-27 Richard Biener <rguenther@suse.de>
14135 PR tree-optimization/81571
14136 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
14139 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
14141 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
14142 earlier and only if MASK_FPU is set. Adjust formatting.
14144 2017-07-27 Martin Liska <mliska@suse.cz>
14146 * opt-functions.awk: Add validation of value of Init.
14147 * optc-gen.awk: Pass new argument.
14149 2017-07-27 Martin Liska <mliska@suse.cz>
14151 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
14152 Fix wrong condition.
14154 2017-07-27 Martin Liska <mliska@suse.cz>
14156 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
14157 BBs and edges seen by autoFDO.
14159 2017-07-27 Richard Biener <rguenther@suse.de>
14161 PR tree-optimization/81502
14162 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
14163 with incompatible but same sized type.
14164 (execute_update_addresses_taken): Likewise.
14166 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
14168 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
14169 flag_tree_loop_vectorize rather than flag_tree_vectorize.
14171 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14174 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
14175 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
14176 Change s_operand to memory_operand.
14178 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
14180 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
14181 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
14182 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
14183 Emit instructions rather than returning an expression. Handle TFmode
14184 and KFmode by casting to TImode.
14185 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
14186 (rs6000_emit_le_vsx_store): Likewise.
14187 * config/rs6000/vsx.md (VSX_TI): New iterator.
14188 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
14189 (*vsx_le_undo_permute_<mode>): Likewise.
14190 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
14191 emit the split sequence.
14192 (*vsx_le_perm_store_<mode>): Likewise.
14194 2017-07-27 Jakub Jelinek <jakub@redhat.com>
14196 PR tree-optimization/81555
14197 PR tree-optimization/81556
14198 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
14199 if true, force CHANGED for the recursive invocation.
14200 (reassociate_bb): Remember original length of ops array, pass
14201 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
14203 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
14204 attributes for noipa attribute. For naked attribute use
14205 lookup_attribute first before lookup_attribute_spec.
14206 * final.c (rest_of_handle_final): Disable IPA RA for functions with
14208 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
14209 for functions with noipa attribute.
14210 (cgraph_externally_visible_p): Return true for functions with noipa
14212 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
14213 for functions with noipa attribute.
14214 * doc/extend.texi: Document noipa function attribute.
14215 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
14216 also for functions with noipa attribute.
14217 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
14219 2017-07-26 Andrew Pinski <apinski@cavium.com>
14221 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
14222 vec_unalign_load_cost and vec_unalign_store_cost.
14224 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14226 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
14227 -mvsx-small-integer option.
14228 (ISA_3_0_MASKS_IEEE): Likewise.
14229 (OTHER_VSX_VECTOR_MASKS): Likewise.
14230 (POWERPC_MASKS): Likewise.
14231 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
14232 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
14233 code, only testing for DImode being allowed in non-VSX floating
14235 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
14236 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
14238 (rs6000_option_override_internal): Delete -mvsx-small-integer.
14239 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
14240 TARGET_P8_VECTOR test.
14241 (rs6000_secondary_reload_simple_move): Likewise.
14242 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
14243 since TARGET_P9_VECTOR was already tested.
14244 (rs6000_opt_masks): Remove -mvsx-small-integer.
14245 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
14246 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14248 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
14249 test for TARGET_VEXTRACTUB was used, and that uses
14251 (p9 extract splitter): Likewise.
14252 (vsx_extract_<mode>_di_p9): Likewise.
14253 (vsx_extract_<mode>_store_p9): Likewise.
14254 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
14255 for TARGET_P9_VECTOR was used. Delete code that is now dead with
14256 the elimination of TARGET_VSX_SMALL_INTEGER.
14257 (vsx_extract_<mode>_p8): Likewise.
14258 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
14259 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
14260 (vsx_set_<mode>_p9): Likewise.
14261 (vsx_set_v4sf_p9): Likewise.
14262 (vsx_set_v4sf_p9_zero): Likewise.
14263 (vsx_insert_extract_v4sf_p9): Likewise.
14264 (vsx_insert_extract_v4sf_p9_2): Likewise.
14265 * config/rs6000/rs6000.md (sign extend splitter): Change
14266 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
14267 (floatsi<mode>2_lfiwax_mem): Likewise.
14268 (floatunssi<mode>2_lfiwzx_mem): Likewise.
14269 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
14270 since a test for TARGET_P9_VECTOR was used.
14271 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14272 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
14273 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14274 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
14275 TARGET_P8_VECTOR test.
14276 (fix_trunc<mode>si2_stfiwx): Likewise.
14277 (fix_trunc<mode>si2_internal): Likewise.
14278 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
14279 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14281 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
14282 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
14283 TARGET_P8_VECTOR test.
14284 (fixuns_trunc<mode>si2_stfiwx): Likewise.
14285 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
14286 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14288 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
14289 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
14290 since a test for TARGET_P9_VECTOR was used.
14291 (splitter for loading small constants): Likewise.
14293 2017-07-26 Andrew Pinski <apinski@cavium.com>
14295 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
14298 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
14301 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
14302 (fp_valid_at): Likewise.
14304 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
14306 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
14307 (qdf24xx_addrcost_table): Likewise.
14308 (cortexa57_tunings): Update to use generic_branch_cost.
14309 (cortexa72_tunings): Likewise.
14310 (cortexa73_tunings): Likewise.
14311 (qdf24xx_tunings): Likewise.
14313 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
14315 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
14316 (thunderx2t99_branch_cost): Likewise.
14317 (cortexa35_tunings): Update to use generic_branch_cost.
14318 (cortexa53_tunings): Likewise.
14319 (cortexa57_tunings): Likewise.
14320 (cortexa72_tunings): Likewise.
14321 (cortexa73_tunings): Likewise.
14322 (thunderx2t99_tunings): Likewise.
14324 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
14326 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
14327 (sparc_option_override): Honour MASK_FSMULD.
14328 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
14329 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
14330 * config/sparc/sparc.opt (mfsmuld): New option.
14331 * doc/invoke.texi (mfsmuld): Document option.
14333 2017-07-26 Marek Polacek <polacek@redhat.com>
14335 PR middle-end/70992
14336 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
14338 2017-07-26 Richard Biener <rguenther@suse.de>
14340 * gimple-match-head.c (do_valueize): Return OP if valueize
14342 (get_def): New helper to get at the def stmt of a SSA name
14343 if valueize allows.
14344 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
14345 do_valueize to get at the def stmt.
14346 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
14348 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
14350 PR middle-end/46932
14351 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
14353 2017-07-26 Martin Liska <mliska@suse.cz>
14356 * function.c (expand_function_start): Make expansion of
14357 nonlocal_goto_save_area after parm_birth_insn.
14359 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
14361 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
14362 from all CPU target flags enable members.
14364 2017-07-26 Richard Biener <rguenther@suse.de>
14366 * genmatch.c (dt_simplify::gen): Make iterator vars const.
14367 (decision_tree::gen): Make 'type' const.
14368 (write_predicate): Likewise.
14370 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
14372 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
14373 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
14374 (rs6000_option_override_internal): Likewise.
14375 (rs6000_expand_vector_set): Likewise.
14376 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
14377 (TARGET_UPPER_REGS_SF): Likewise.
14378 (TARGET_UPPER_REGS_DI): Likewise.
14379 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
14380 (TARGET_DIRECT_MOVE_64BIT): Likewise.
14381 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
14382 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14383 (Splitters for DI constants in Altivec registers): Likewise.
14384 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
14385 (vsx_set_v4sf_p9): Likewise.
14386 (vsx_set_v4sf_p9_zero): Likewise.
14387 (vsx_insert_extract_v4sf_p9): Likewise.
14388 (vsx_insert_extract_v4sf_p9_2): Likewise.
14390 2017-07-25 Carl Love <cel@us.ibm.com>
14392 * doc/extend.texi: Update the built-in documentation file for the
14393 existing built-in functions
14394 vector signed char vec_cnttz (vector signed char);
14395 vector unsigned char vec_cnttz (vector unsigned char);
14396 vector signed short vec_cnttz (vector signed short);
14397 vector unsigned short vec_cnttz (vector unsigned short);
14398 vector signed int vec_cnttz (vector signed int);
14399 vector unsigned int vec_cnttz (vector unsigned int);
14400 vector signed long long vec_cnttz (vector signed long long);
14401 vector unsigned long long vec_cnttz (vector unsigned long long);
14403 2017-07-25 Andrew Pinski <apinski@cavium.com>
14405 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
14406 accesses where the use is for the first operand of a BIT_INSERT.
14408 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
14411 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
14412 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
14414 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
14416 * config/i386/gstabs.h: Delete.
14417 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
14419 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
14421 * config/i386/i386.c (ix86_decompose_address): Do not check for
14422 register RTX when looking at index_reg or base_reg.
14423 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
14425 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
14427 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
14428 to update EH info here.
14430 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
14432 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
14434 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
14436 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
14438 2017-07-25 Torsten Duwe <duwe@suse.de>
14440 * common.opt: Introduce -fpatchable-function-entry
14441 command line option, and its variables function_entry_patch_area_size
14442 and function_entry_patch_area_start.
14443 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
14444 including a two-value parser.
14445 * target.def (print_patchable_function_entry): New target hook.
14446 * targhooks.h (default_print_patchable_function_entry): New function.
14447 * targhooks.c (default_print_patchable_function_entry): Likewise.
14448 * toplev.c (process_options): Switch off IPA-RA if
14449 patchable function entries are being generated.
14450 * varasm.c (assemble_start_function): Look at the
14451 patchable-function-entry command line switch and current
14452 function attributes and maybe generate NOP instructions by
14453 calling the print_patchable_function_entry hook.
14454 * doc/extend.texi: Document patchable_function_entry attribute.
14455 * doc/invoke.texi: Document -fpatchable_function_entry
14456 command line option.
14457 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
14459 * doc/tm.texi: Re-generate.
14461 2017-07-25 Jakub Jelinek <jakub@redhat.com>
14464 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
14465 TARGET_AVX512DQ rather than TARGET_AVX512BW.
14467 2017-07-25 Tamar Christina <tamar.christina@arm.com>
14469 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
14471 2017-07-25 Richard Biener <rguenther@suse.de>
14473 PR tree-optimization/81455
14474 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
14475 not walk in cycles when looking for guards.
14477 2017-07-25 Richard Biener <rguenther@suse.de>
14479 PR tree-optimization/81529
14480 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
14481 when optimizing backedge uses.
14483 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
14485 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
14487 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
14488 to dl_section_ref. On AIX, append an expression to subtract
14489 the size of the section length to dl_section_ref.
14491 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
14493 * configure.ac: If any of the config.* scripts fail, exit 1.
14494 * configure: Regenerate.
14496 2017-07-25 Richard Biener <rguenther@suse.de>
14498 PR middle-end/81546
14499 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
14500 of immediate uses, be more verbose on errors.
14502 2017-07-25 Richard Biener <rguenther@suse.de>
14504 PR tree-optimization/81510
14505 * tree-vect-loop.c (vect_is_simple_reduction): When the
14506 reduction stmt is not inside the loop bail out.
14508 2017-07-25 Richard Biener <rguenther@suse.de>
14510 PR tree-optimization/81303
14511 * tree-vect-loop-manip.c (vect_loop_versioning): Build
14512 profitability check against LOOP_VINFO_NITERSM1.
14514 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
14516 * domwalk.c (cmp_bb_postorder): Simplify.
14517 (sort_bbs_postorder): New function. Use it...
14518 (dom_walker::walk): ...here to optimize common cases.
14520 2017-07-25 Martin Liska <mliska@suse.cz>
14523 * ipa-visibility.c (function_and_variable_visibility): Make the
14524 redirection just on target that supports aliasing.
14525 Fix GNU coding style.
14527 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
14530 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
14532 * config/arm/rtems.h: Likewise.
14533 * config/bfin/rtems.h: Likewise.
14534 * config/i386/rtemself.h: Likewise.
14535 * config/lm32/rtems.h: Likewise.
14536 * config/m32c/rtems.h: Likewise.
14537 * config/m68k/rtemself.h: Likewise.
14538 * config/microblaze/rtems.h: Likewise.
14539 * config/mips/rtems.h: Likewise.
14540 * config/moxie/rtems.h: Likewise.
14541 * config/nios2/rtems.h: Likewise.
14542 * config/powerpcspe/rtems.h: Likewise.
14543 * config/rs6000/rtems.h: Likewise.
14544 * config/rtems.h: Likewise.
14545 * config/sh/rtems.h: Likewise.
14546 * config/sh/rtemself.h: Likewise.
14547 * config/sparc/rtemself.h: Likewise.
14549 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
14552 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
14553 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
14554 * tree-ssa-structalias.c (alias_get_name): Same.
14556 2017-07-25 Bin Cheng <bin.cheng@arm.com>
14559 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
14560 instructions if no du chain is found.
14562 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
14564 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
14566 2017-07-25 Richard Biener <rguenther@suse.de>
14568 PR middle-end/81505
14569 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
14572 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
14574 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
14575 upper-regs options.
14576 (ISA_2_7_MASKS_SERVER): Likewise.
14577 (ISA_3_0_MASKS_IEEE): Likewise.
14578 (OTHER_P8_VECTOR_MASKS): Likewise.
14579 (OTHER_VSX_VECTOR_MASKS): Likewise.
14580 (POWERPC_MASKS): Likewise.
14581 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
14582 duplicate list of options.
14583 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
14584 explicit -mupper-regs options.
14585 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
14586 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
14587 alias for -mupper-regs-df.
14588 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
14589 (rs6000_init_hard_regno_mode_ok): Likewise.
14590 (rs6000_option_override_internal): Likewise.
14591 (rs6000_opt_masks): Likewise.
14592 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
14593 options in terms of whether -mvsx or -mpower8-vector was used.
14594 (TARGET_UPPER_REGS_DI): Likewise.
14595 (TARGET_UPPER_REGS_SF): Likewise.
14596 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
14597 -mupper-regs-* options.
14599 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
14601 * passes.c (emergency_dump_function): Print some empty lines and a
14602 header before the RTL dump.
14604 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
14606 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
14608 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
14611 * config/aarch64/aarch64.c (aarch64_classify_symbol):
14612 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
14614 2017-07-24 Carl Love <cel@us.ibm.com>
14616 * config/rs6000/rs6000-c.c: Add support for built-in functions
14617 vector float vec_extract_fp32_from_shorth (vector unsigned short);
14618 vector float vec_extract_fp32_from_shortl (vector unsigned short);
14619 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
14620 vec_extract_fp_from_shortl): Add defines for the two builtins.
14621 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
14622 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
14624 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
14625 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
14626 * doc/extend.texi: Update the built-in documentation file for the
14627 new built-in function.
14629 2017-07-24 Jakub Jelinek <jakub@redhat.com>
14632 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
14634 * doc/generic.texi: Likewise.
14635 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
14636 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
14638 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
14640 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
14641 (aarch64_mls_elt_merge<mode>): Likewise.
14643 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
14645 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
14646 having __cxa_atexit.
14648 2017-07-23 Michael Collison <michael.collison@arm.com>
14650 * config/arm/arm.c (arm_option_override): Deprecate
14651 use of -mstructure-size-boundary.
14652 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
14653 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
14655 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14658 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
14659 Reduce cost estimate for direct moves.
14661 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
14664 * config/i386/i386.c (ix86_option_override_internal): Disable
14665 BMI, BMI2 and TBM instructions for -m16.
14667 2017-07-21 Carl Love <cel@us.ibm.com>
14669 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14670 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14671 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14672 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14673 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14674 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
14675 VMULOSW): New enum "unspec" values.
14676 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
14677 altivec_vmulosw): New patterns.
14678 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14679 VMULOSW): Add definitions.
14681 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
14683 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
14684 (qdf24xx): Likewise.
14685 * config/aarch64/aarch64-options-extensions.def (rdma); New.
14686 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
14687 (AARCH64_FL_V8_1): Renumber.
14688 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
14689 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
14690 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
14691 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
14692 rdma to feature modifiers list.
14694 2017-07-21 Yury Gribov <tetra2005@gmail.com>
14696 PR middle-end/56727
14697 * ipa-visibility (function_and_variable_visibility): Convert
14698 recursive PLT call to direct call if appropriate.
14700 2017-07-21 Andrew Pinski <apinski@cavium.com>
14702 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
14703 operand 1 to see if the types precision matches.
14704 * fold-const.c (operand_equal_p): Likewise.
14706 2017-07-21 Richard Biener <rguenther@suse.de>
14708 PR tree-optimization/81303
14709 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
14710 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
14711 (vect_peeling_hash_get_lowest_cost): Adjust.
14712 (vect_enhance_data_refs_alignment): Likewise. Use
14713 vect_get_peeling_costs_all_drs to compute the penalty for no
14714 peeling to match up costs.
14716 2017-07-21 Richard Biener <rguenther@suse.de>
14718 PR tree-optimization/81500
14719 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
14720 we didn't identify a reduction path.
14722 2017-07-21 Tom de Vries <tom@codesourcery.com>
14723 Cesar Philippidis <cesar@codesourcery.com>
14725 PR gcov-profile/81442
14726 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
14729 2017-07-21 Tom de Vries <tom@codesourcery.com>
14732 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
14734 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
14735 nvptx_override_options_after_change.
14737 2017-07-21 Ulrich Drepper <drepper@redhat.com>
14739 * dwarf2out.c (output_file_names): Avoid double testing for
14740 dwarf_version >= 5.
14742 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
14744 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
14746 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
14748 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
14750 (try_crossjump_to_edge): Do not punt on partitioned functions.
14752 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
14754 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
14755 Put all BBs reachable only via paths crossing cold region to cold
14757 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
14759 2016-07-21 Richard Biener <rguenther@suse.de>
14761 PR tree-optimization/81303
14762 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
14763 into account prologue and epilogue iterations when raising
14764 min_profitable_iters to sth at least covering one vector iteration.
14766 2017-07-21 Tamar Christina <tamar.christina@arm.com>
14768 * config/arm/arm.c (arm_test_cpu_arch_dat):
14771 2017-07-20 Nathan Sidwell <nathan@acm.org>
14773 Remove TYPE_METHODS.
14774 * tree.h (TYPE_METHODS): Delete.
14775 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
14776 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
14777 (dbxout_type_methods): Scan TYPE_FIELDS.
14778 (dbxout_type): Don't check TYPE_METHODS here.
14779 * function.c (use_register_for_decl): Always ignore register for
14780 class types when not optimizing.
14781 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
14782 * tree.c (free_lang_data_in_type): Stitch out member functions and
14783 templates from TYPE_FIELDS.
14784 (build_distinct_type_copy, verify_type_variant,
14785 verify_type): Member fns are on TYPE_FIELDS.
14786 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
14787 * tree-pretty-print.c (dump_generic_node): Likewise.
14789 2017-07-20 Jakub Jelinek <jakub@redhat.com>
14792 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
14793 V2TImode and V4TImode.
14794 (ix86_expand_vector_extract): Likewise.
14795 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
14796 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
14797 (ssescalarmode): Handle V4TImode and V2TImode.
14798 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
14799 (*vec_extractv2ti, *vec_extractv4ti): New insns.
14800 (VEXTRACTI128_MODE): New mode iterator.
14801 (splitter for *vec_extractv?ti first element): New.
14802 (VEC_INIT_MODE): New mode iterator.
14803 (vec_init<mode>): Consolidate 3 expanders into one using
14804 VEC_INIT_MODE mode iterator.
14806 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
14808 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
14809 non_spilled_static_chain_regno_p.
14811 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
14813 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
14815 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
14817 * bb-reorder.c (connect_traces): Allow copying of blocks within
14820 2017-07-20 Richard Biener <rguenther@suse.de>
14822 * gimple.h (gimple_phi_result): Add gphi * overload.
14823 (gimple_phi_result_ptr): Likewise.
14824 (gimple_phi_arg): Likewise. Adjust index assert to only
14825 allow actual argument accesses rather than all slots available
14827 (gimple_phi_arg_def): Add gphi * overload.
14828 * tree-phinodes.c (make_phi_node): Initialize only actual
14830 (resize_phi_node): Clear memory not covered by old node,
14831 do not initialize excess argument slots.
14832 (reserve_phi_args_for_new_edge): Initialize new argument slot
14835 2017-07-20 Bin Cheng <bin.cheng@arm.com>
14837 PR tree-optimization/81388
14839 2016-07-21 Bin Cheng <bin.cheng@arm.com>
14841 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
14842 by removing computation of may_be_zero.
14844 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
14845 Tom de Vries <tom@codesourcery.com>
14847 PR middle-end/81030
14848 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
14849 when gimple level profile disagrees with what RTL expander did.
14851 2017-07-20 Richard Biener <rguenther@suse.de>
14853 PR tree-optimization/61171
14854 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
14855 (vect_analyze_stmt): Add slp instance parameter.
14856 (vectorizable_reduction): Likewise.
14857 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
14858 (vect_is_simple_reduction): Deal with chains not detected
14859 as SLP reduction chain, specifically not properly associated
14860 chains containing a mix of plus/minus.
14861 (get_reduction_op): Remove.
14862 (get_initial_defs_for_reduction): Simplify, pass in whether
14863 this is a reduction chain, pass in the SLP node for the PHIs.
14864 (vect_create_epilog_for_reduction): Get the SLP instance as
14866 (vectorizable_reduction): Get the SLP instance as arg.
14867 During analysis remember the SLP node with the PHIs in the
14868 instance. Simplify getting at the vectorized reduction PHIs.
14869 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
14870 through SLP instance.
14871 (vect_slp_analyze_operations): Likewise.
14872 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
14873 (vect_transform_stmt): Likewise.
14875 2017-07-20 Tom de Vries <tom@codesourcery.com>
14877 PR tree-optimization/81489
14878 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
14879 read of phi arg location to before loop that modifies phi.
14881 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14883 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
14886 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
14888 PR middle-end/81331
14889 * except.c (execute): Fix ordering issue.
14891 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
14893 PR rtl-optimization/81423
14894 * combine.c (make_compound_operation_int): Don't try to optimize
14895 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
14897 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
14899 PR rtl-optimization/81423
14900 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
14901 with a constant that is -1 in the truncated to mode.
14903 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
14905 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
14906 (determine_unlikely_bbs): ... here.
14907 * predict.h (propagate_unlikely_bbs_forward): Declare.
14908 * cfgexpand.c (pass_expand::execute): Use it.
14909 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
14911 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
14912 propagate_unlikely_bbs_forward.
14914 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
14916 PR middle-end/81331
14917 * except.c (maybe_add_nop_after_section_switch): New function.
14920 2017-07-19 Tom de Vries <tom@codesourcery.com>
14922 * gimple.h (gimple_phi_set_arg): Make assert more strict.
14924 2017-07-19 Tom de Vries <tom@codesourcery.com>
14926 * gimple.h (gimple_phi_arg): Make assert more strict.
14928 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
14930 * config.gcc (powerpc*-*-*): Add mmintrin.h.
14931 * config/rs6000/mmintrin.h: New file.
14932 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
14934 2017-07-19 Jakub Jelinek <jakub@redhat.com>
14936 PR tree-optimization/81346
14937 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
14939 2017-07-19 Tom de Vries <tom@codesourcery.com>
14941 * config/nvptx/nvptx.md (VECIM): Add V2DI.
14943 2017-07-19 Tom de Vries <tom@codesourcery.com>
14945 * config/nvptx/nvptx-modes.def: Add V2DImode.
14946 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
14947 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
14948 (nvptx_output_mov_insn): Handle lack of mov.b128.
14949 (nvptx_print_operand): Handle 'H' and 'L' codes.
14950 (nvptx_vector_mode_supported): Allow V2DImode.
14951 (nvptx_preferred_simd_mode): New function.
14952 (nvptx_data_alignment): New function.
14953 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
14954 nvptx_preferred_simd_mode.
14955 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
14957 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
14959 2017-07-19 Tom de Vries <tom@codesourcery.com>
14961 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
14962 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
14963 (nvptx_vector_mode_supported): New function. Allow V2SImode.
14964 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
14965 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
14966 (mov<VECIM>_insn): New define_insn.
14967 (define_expand "mov<VECIM>): New define_expand.
14969 2017-07-19 Tom de Vries <tom@codesourcery.com>
14971 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
14973 2017-07-19 Jakub Jelinek <jakub@redhat.com>
14975 PR tree-optimization/81346
14976 * fold-const.h (fold_div_compare, range_check_type): Declare.
14977 * fold-const.c (range_check_type): New function.
14978 (build_range_check): Use range_check_type.
14979 (fold_div_compare): No longer static, rewritten into
14980 a match.pd helper function.
14981 (fold_comparison): Don't call fold_div_compare here.
14982 * match.pd (X / C1 op C2): New optimization using fold_div_compare
14983 as helper function.
14985 2017-07-19 Nathan Sidwell <nathan@acm.org>
14987 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
14988 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
14989 * tree.c (find_decls_types_r, verify_type): Use
14990 TYPE_{MIN,MAX}_VALUE_RAW.
14991 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
14992 (hash_tree): Likewise.
14993 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
14995 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
14998 2017-07-18 Tom de Vries <tom@codesourcery.com>
15000 PR middle-end/81464
15001 * omp-expand.c (expand_omp_for_static_chunk): Handle
15002 equal-argument loop exit phi.
15004 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
15007 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
15008 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
15009 operand 2 predicate.
15010 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
15011 operand 2 predicate.
15012 (ror,rol -> rorx splitters): Use const_int_operand as
15013 operand 2 predicate.
15015 2017-06-18 Richard Biener <rguenther@suse.de>
15017 PR tree-optimization/81410
15018 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
15019 the gap in the ! slp_perm SLP case after each group.
15021 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15023 PR middle-end/81463
15024 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
15027 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15029 PR middle-end/81462
15030 * predict.c (set_even_probabilities): Cleanup; do not affect
15031 probabilities that are already known.
15032 (combine_predictions_for_bb): Call even when count is set.
15034 2017-07-18 Nathan Sidwell <nathan@acm.org>
15036 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
15039 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15042 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
15043 optimization for loop niter analysis.
15045 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
15048 * config/avr/avr.c (avr_optimize_casesi): Don't use
15049 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
15051 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
15053 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
15054 body_cost_vec from _vect_peel_extended_info.
15055 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
15056 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
15059 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15061 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
15063 2017-07-18 Richard Biener <rguenther@suse.de>
15065 PR tree-optimization/80620
15066 PR tree-optimization/81403
15067 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
15068 info when re-using a VN table entry.
15070 2017-07-18 Richard Biener <rguenther@suse.de>
15072 PR tree-optimization/81418
15073 * tree-vect-loop.c (vectorizable_reduction): Properly compute
15074 vectype_in. Verify that with lane-reducing reduction operations
15075 we have a single def-use cycle.
15077 2017-07-17 Carl Love <cel@us.ibm.com>
15079 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
15081 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15082 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15083 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15084 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15085 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15086 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15087 VMULOSW): New enum "unspec" values.
15088 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
15089 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
15090 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15091 altivec_vmulosw): New patterns.
15092 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15093 VMULOSW): Add definitions.
15094 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
15096 * config/alpha/alpha.c: Include predict.h.
15098 2017-07-17 Yury Gribov <tetra2005@gmail.com>
15100 * tree-vrp.c (compare_assert_loc): Fix comparison function
15101 to return predictable results.
15103 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15105 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
15107 (subdi3): Likewise.
15108 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
15109 * doc/invoke.texi (mexpand-adddi): Update text.
15111 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15113 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
15114 that also clobbers the CC register. The old expand code is moved
15116 (*arc_clzsi2): ... here.
15117 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
15118 the CC register. The old expand code is moved to ...
15119 (arc_ctzsi2): ... here.
15121 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15123 * config/arc/arc.opt (mindexed-loads): Use initial value
15124 TARGET_INDEXED_LOADS_DEFAULT.
15125 (mauto-modify-reg): Use initial value
15126 TARGET_AUTO_MODIFY_REG_DEFAULT.
15127 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15128 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15129 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15130 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15132 2017-07-17 Martin Liska <mliska@suse.cz>
15135 * opts.c (finish_options): Do not allow -fgnu-tm
15136 w/ -fsanitize={kernel-,}address. Say sorry.
15138 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15141 * tree-loop-distribution.c (classify_partition): Only assert on
15142 numer of iterations.
15143 (merge_dep_scc_partitions): Delete prameter. Update function call.
15144 (distribute_loop): Remove code handling loop with unknown niters.
15145 (pass_loop_distribution::execute): Skip loop with unknown niters.
15147 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15150 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
15151 function sort_partitions_by_post_order.
15153 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15155 PR tree-optimization/81374
15156 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
15157 the max index of basic blocks, rather than number of basic blocks.
15159 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15161 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
15163 (arc_legitimate_pic_operand_p): Likewise.
15164 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
15166 (arc_needs_pcl_p): Likewise.
15167 (arc_legitimate_pc_offset_p): Likewise.
15168 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
15169 function is also used in constrains.md.
15170 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
15171 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
15172 PLUS. Only return true/false in known cases, otherwise assert.
15173 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
15174 is already called in arc_legitimate_constant_p.
15175 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
15177 (LEGITIMATE_PIC_OPERAND_P): Use
15178 arc_raw_symbolic_reference_mentioned_p function.
15179 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
15184 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15185 Andrew Burgess <andrew.burgess@embecosm.com>
15187 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
15188 (arc_return_address_register): New function.
15189 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
15190 (arc_handle_fndecl_attribute): Add naked attribute.
15191 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15192 (TARGET_WARN_FUNC_RETURN): Likewise.
15193 (arc_allocate_stack_slots_for_args): New function.
15194 (arc_warn_func_return): Likewise.
15195 (machine_function): Change type fn_type.
15196 (arc_compute_function_type): Consider new naked function type,
15197 change function return type.
15198 (arc_must_save_register): Adapt to handle new
15199 arc_compute_function_type's return type.
15200 (arc_expand_prologue): Likewise.
15201 (arc_expand_epilogue): Likewise.
15202 (arc_return_address_regs): Delete.
15203 (arc_return_address_register): New function.
15204 (arc_epilogue_uses): Use above function.
15205 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
15206 (arc_function_type): Change encoding, add naked type.
15207 (ARC_INTERRUPT_P): Change to handle the new encoding.
15208 (ARC_FAST_INTERRUPT_P): Likewise.
15209 (ARC_NORMAL_P): Define.
15210 (ARC_NAKED_P): Likewise.
15211 (arc_compute_function_type): Delete prototype.
15212 * config/arc/arc.md (in_ret_delay_slot): Use
15213 arc_return_address_register function.
15214 (simple_return): Likewise.
15215 (p_return_i): Likewise.
15217 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15219 PR tree-optimization/81428
15220 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
15221 can't be built for those types.
15223 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
15225 Remove stuff dead since r239246.
15227 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
15228 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
15229 (avr_inform_devices): Remove dead stuff.
15231 2017-07-17 Tamar Christina <tamar.christina@arm.com>
15233 * config/arm/arm_neon.h: Fix softp typo.
15235 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15237 PR tree-optimization/81365
15238 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
15239 aggregate moves onto bb predecessor edges, make sure there are no
15240 loads that could alias the lhs in between the start of bb and the
15243 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
15246 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
15247 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
15248 [LSHIFTRT, outer_code = TRUNCATE]: Same.
15250 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15252 PR tree-optimization/81396
15253 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
15254 (init_symbolic_number): Initialize it to 1.
15255 (perform_symbolic_merge): Add n_ops from both operands into the new
15257 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
15258 without base_addr as useless if they need more than one operation.
15259 (bswap_replace): Handle !bswap case for NULL base_addr.
15261 2017-07-17 Tom de Vries <tom@codesourcery.com>
15264 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
15267 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
15269 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
15270 conditional builtin define __FIX_LEON3FT_B2BST.
15272 2017-07-17 Daniel Cederman <cederman@gaisler.com>
15274 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
15275 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
15278 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
15280 PR rtl-optimization/81424
15281 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
15282 to remove potential trapping from operands if -fnon-call-exceptions.
15284 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15286 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
15287 profile_proability for scalling.
15288 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
15290 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15292 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
15294 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15296 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
15297 fixpoint arithmetics.
15299 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15301 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
15302 fixpoint arithmetics.
15304 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15306 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
15307 fixpoint arithmetics.
15309 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15311 * profile-count.h (profile_probability::from_reg_br_prob_note,
15312 profile_probability::to_reg_br_prob_note): New functions.
15313 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
15314 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
15315 * predict.c (probability_reliable_p): Update.
15316 (edge_probability_reliable_p): Update.
15317 (br_prob_note_reliable_p): Update.
15318 (invert_br_probabilities): Update.
15319 (add_reg_br_prob_note): New function.
15320 (combine_predictions_for_insn): Update.
15321 * asan.c (asan_clear_shadow): Update.
15322 * cfgbuild.c (compute_outgoing_frequencies): Update.
15323 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
15324 (update_br_prob_note): Update.
15325 (rtl_verify_edges): Update.
15326 (purge_dead_edges): Update.
15327 (fixup_reorder_chain): Update.
15328 * emit-rtl.c (try_split): Update.
15329 * ifcvt.c (cond_exec_process_insns): Update.
15330 (cond_exec_process_if_block): Update.
15331 (dead_or_predicable): Update.
15332 * internal-fn.c (expand_addsub_overflow): Update.
15333 (expand_neg_overflow): Update.
15334 (expand_mul_overflow): Update.
15335 * loop-doloop.c (doloop_modify): Update.
15336 * loop-unroll.c (compare_and_jump_seq): Update.
15337 * optabs.c (emit_cmp_and_jump_insn_1): Update.
15338 * predict.h: Update.
15339 * reorg.c (mostly_true_jump): Update.
15341 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
15342 * config/alpha/alpha.c (emit_unlikely_jump): Update.
15343 * config/arc/arc.c: (emit_unlikely_jump): Update.
15344 * config/arm/arm.c: (emit_unlikely_jump): Update.
15345 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
15346 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
15347 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
15348 (ix86_print_operand): Update.
15349 (ix86_split_fp_branch): Update.
15350 (predict_jump): Update.
15351 * config/ia64/ia64.c (ia64_print_operand): Update.
15352 * config/mmix/mmix.c (mmix_print_operand): Update.
15353 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
15354 (rs6000_expand_split_stack_prologue): Update.
15355 * config/rs6000/rs6000.c: Update.
15356 * config/s390/s390.c (s390_expand_vec_strlen): Update.
15357 (s390_expand_vec_movstr): Update.
15358 (s390_expand_cs_tdsi): Update.
15359 (s390_expand_split_stack_prologue): Update.
15360 * config/sh/sh.c (sh_print_operand): Update.
15361 (expand_cbranchsi4): Update.
15362 (expand_cbranchdi4): Update.
15363 * config/sparc/sparc.c (output_v9branch): Update.
15364 * config/spu/spu.c (get_branch_target): Update.
15365 (ea_load_store_inline): Update.
15366 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
15367 * config/tilepro/tilepro.c: Update.
15369 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
15371 * gimplify.c (mostly_copy_tree_r): Revert latest change.
15372 (gimplify_save_expr): Likewise.
15374 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15376 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
15378 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15380 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
15382 * timevar.def (TV_IPA_FNSUMMARY): Define.
15384 2017-07-16 Daniel Cederman <cederman@gaisler.com>
15386 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
15387 to back store errata sensitive sequence from being generated.
15388 (sqrtdf2_fix): Likewise.
15390 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15392 * tree-ssa-threadupdate.c (compute_path_counts,
15393 update_joiner_offpath_counts): Use profile_probability.
15395 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
15398 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15400 * config/arm/arm-c.c (arm_cpu_builtins): Define
15401 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
15403 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
15405 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15406 array entries to represent __ieee128 versions of the
15407 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
15408 scalar_extract_sig, and scalar_insert_exp built-in functions.
15409 (altivec_resolve_overloaded_builtin): Add special case handling
15410 for the __builtin_scalar_insert_exp function, as represented by
15411 the P9V_BUILTIN_VEC_VSIEDP constant.
15412 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
15413 exponent support for __ieee128 argument.
15414 (VSESQP): Add scalar extract signature support for __ieee128
15416 (VSTDCNQP): Add scalar test negative support for __ieee128
15418 (VSIEQP): Add scalar insert exponent support for __int128 argument
15419 with __ieee128 result.
15420 (VSIEQPF): Add scalar insert exponent support for __ieee128
15421 argument with __ieee128 result.
15422 (VSTDCQP): Add scalar test data class support for __ieee128
15424 (VSTDCNQP): Add overload support for scalar test negative with
15425 __ieee128 argument.
15426 (VSTDCQP): Add overload support for scalar test data class
15427 __ieee128 argument.
15428 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
15429 UNSPEC_VSX_SXSIGDP.
15430 (UNSPEC_VSX_SIEXPQP): New constant.
15431 (xsxexpqp): New insn for VSX scalar extract exponent quad
15433 (xsxsigqp): New insn for VSX scalar extract significand quad
15435 (xsiexpqpf): New insn for VSX scalar insert exponent quad
15436 precision with floating point argument.
15437 (xststdcqp): New expand for VSX scalar test data class quad
15439 (xststdcnegqp): New expand for VSX scalar test negative quad
15441 (xststdcqp): New insn to match expansions for VSX scalar test data
15442 class quad precision and VSX scalar test negative quad precision.
15443 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
15444 special case operand checking to enforce that second operand of
15445 VSX scalar test data class with quad precision argument is a 7-bit
15447 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
15448 prototypes and descriptions of __ieee128 versions of
15449 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
15450 scalar_test_data_class, and scalar_test_neg built-in functions.
15452 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15454 PR tree-optimization/81162
15455 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
15456 replace a negate with an add.
15458 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
15460 * doc/invoke.texi (arm/-mcpu): Document +crypto.
15462 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15464 * config/arm/arm-c.c (arm_cpu_builtins): Define
15465 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
15467 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15469 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
15470 (armv8-r): Set ARM Cortex-R52 as default CPU.
15471 * config/arm/arm-tables.opt: Regenerate.
15472 * config/arm/arm-tune.md: Regenerate.
15473 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
15475 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
15476 extension for -mcpu=cortex-r52.
15478 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15480 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
15481 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
15482 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
15483 (fp-armv8): Define it as FP_ARMv8 only.
15484 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
15485 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
15487 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
15488 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
15489 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
15490 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
15491 than TARGET_FPU_ARMV8.
15492 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
15493 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
15494 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
15496 * config/arm/neon.md (neon_vrint): Likewise.
15497 (neon_vcvt): Likewise.
15498 (neon_<fmaxmin_op><mode>): Likewise.
15499 (<fmaxmin><mode>3): Likewise.
15500 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
15501 * config/arm/predicates.md (arm_cond_move_operator): Check against
15502 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
15504 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
15506 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
15507 to top of function.
15509 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15511 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
15512 loop in comment with memset.
15514 2017-07-14 Martin Liska <mliska@suse.cz>
15516 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
15517 * dwarf2out.c (is_java): Remove the function.
15518 (output_pubname): Remove usage of the function.
15519 (lower_bound_default): Remove usage of DW_LANG_Java.
15520 (gen_compile_unit_die): Likewise.
15521 * gcc.c: Remove compiler defaults for .java and .zip files.
15522 * gimple-expr.c (remove_suffix): Change as there's no longer
15523 extension than 4-letter one.
15524 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
15525 (gimplify_save_expr): Likewise.
15526 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
15527 as it's possible even for other languages than Java.
15528 * langhooks.h (struct lang_hooks): Remove Java from a comment.
15529 * lto-opts.c (lto_write_options): Remove reference to Java.
15530 * opts.c (strip_off_ending): Update file extension handling.
15531 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
15532 * tree-eh.c (lower_resx): Likewise.
15533 * tree.c (free_lang_data_in_type): Remove dead code.
15534 (find_decls_types_r): Likewise.
15535 (build_common_builtin_nodes): Remove Java from a comment.
15536 (verify_type): Remove dead code.
15537 * varasm.c (assemble_external): Remove Java from a comment.
15539 2017-07-14 Martin Liska <mliska@suse.cz>
15541 * opts.c (finish_options): Add quotes.
15542 (common_handle_option): Likewise.
15544 2017-07-14 Martin Liska <mliska@suse.cz>
15546 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
15547 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
15548 Remove N_SO_PASCAL.
15549 * dwarf2out.c (lower_bound_default): Do not handle
15551 (gen_compile_unit_die): Likewise.
15552 * gcc.c: Remove default extension binding for GNU Pascal.
15553 * stmt.c: Remove Pascal language from a comment.
15554 * xcoffout.c: Likewise.
15556 2017-07-13 David Malcolm <dmalcolm@redhat.com>
15559 * diagnostic-show-locus.c (fixit_cmp): New function.
15560 (layout::layout): Sort m_fixit_hints.
15561 (column_range::column_range): Assert that the values are valid.
15562 (struct char_span): New struct.
15563 (correction::overwrite): New method.
15564 (struct source_line): New struct.
15565 (line_corrections::add_hint): Add assertions. Reimplement memcpy
15566 calls in terms of classes source_line and char_span, and
15567 correction::overwrite.
15568 (selftest::test_overlapped_fixit_printing_2): New function.
15569 (selftest::diagnostic_show_locus_c_tests): Call it.
15571 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
15573 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
15574 early if there is no lhs.
15576 2017-07-13 Martin Liska <mliska@suse.cz>
15578 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
15579 (gen_reference_type_die): Likewise.
15580 * stor-layout.c: Remove Pascal-related comment.
15582 2017-07-13 Martin Liska <mliska@suse.cz>
15584 * opts.c (finish_options): Add quotes to error messages.
15585 (parse_sanitizer_options): Likewise.
15587 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
15589 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
15591 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
15593 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
15595 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
15597 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
15599 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
15601 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
15604 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
15605 provides the hardware capability bits, define the macro
15606 __BUILTIN_CPU_SUPPORTS__.
15607 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
15608 if GLIBC does not provide the hardware capability bits. Add a
15609 gcc_unreachable call if the built-in cpu function is neither
15610 __builtin_cpu_is nor __builtin_cpu_supports.
15611 (rs6000_get_function_versions_dispatcher): Change the warning
15612 that an old GLIBC is used which does not export the capability
15613 bits to be an error.
15614 * doc/extend.texi (target_clones attribute): Document the
15615 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
15616 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
15617 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
15618 the macros defined by GCC if the newer GLIBC is available.
15620 2017-07-12 Jeff Law <law@redhat.com>
15622 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
15623 remaining includes slightly.
15624 * config/riscv/riscv-builtins.c: Include profile-count.h.
15626 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
15629 * config/avr/avr.c (avr_set_current_function): In diagnostic
15630 messages: Quote keywords and (parts of) identifiers.
15631 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
15634 2017-07-12 Carl Love <cel@us.ibm.com>
15636 * config/rs6000/rs6000-c.c: Add support for built-in functions
15637 vector bool char vec_revb (vector bool char);
15638 vector bool short vec_revb (vector short char);
15639 vector bool int vec_revb (vector bool int);
15640 vector bool long long vec_revb (vector bool long long);
15641 * doc/extend.texi: Update the built-in documentation file for the
15642 new built-in functions.
15644 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15646 * config/s390/s390.md: Remove movcc splitter.
15648 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15650 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
15651 load/store on condition.
15653 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
15656 * config/avr/avr.c (avr_encode_section_info)
15657 [progmem && !TREE_READONLY]: Error if progmem object needs
15660 2017-07-11 Michael Collison <michael.collison@arm.com>
15662 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
15665 2017-07-11 Carl Love <cel@us.ibm.com>
15667 * config/rs6000/rs6000-c.c: Add support for builtins
15668 vector unsigned int vec_parity_lsbb (vector signed int);
15669 vector unsigned int vec_parity_lsbb (vector unsigned int);
15670 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
15671 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
15672 vector unsigned long long vec_parity_lsbb (vector signed long long);
15673 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
15674 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
15675 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
15676 * doc/extend.texi: Update the built-in documentation file for the
15677 new built-in functions.
15679 2017-07-11 David Malcolm <dmalcolm@redhat.com>
15681 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
15682 (layout::m_primary_loc): New field.
15683 (layout::layout): Initialize new field. Move location filtering
15684 logic from here to...
15685 (layout::maybe_add_location_range): ...this new method. Add
15686 support for filtering to just the lines already specified by other
15688 (layout::will_show_line_p): New method.
15689 (gcc_rich_location::add_location_if_nearby): New method.
15690 (selftest::test_add_location_if_nearby): New test function.
15691 (selftest::diagnostic_show_locus_c_tests): Call it.
15692 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
15695 2017-07-11 Tom de Vries <tom@codesourcery.com>
15697 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
15698 (bb_first_real_insn): New function.
15699 (nvptx_single): Add extra initialization of broadcasted condition
15702 2017-07-11 Nathan Sidwell <nathan@acm.org>
15704 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
15706 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
15708 * doc/extend.texi (AVR Function Attributes): Remove weblink to
15709 Binutils doc as TEXI will mess them up.
15710 * doc/invoke.texi (AVR Options): Same here.
15712 2017-07-11 Daniel Cederman <cederman@gaisler.com>
15714 * config/sparc/sparc.opt (mfix-ut700): New option.
15715 (mfix-gr712rc): Likewise.
15716 (sparc_fix_b2bst): New variable.
15717 * doc/invoke.texi (SPARC options): Document them.
15718 (ARM options): Fix warnings.
15719 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
15720 instructions to prevent sequences that can trigger the store-store
15721 errata for certain LEON3FT processors.
15722 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
15723 (sparc_option_override): Set sparc_fix_b2bst appropriately.
15724 * config/sparc/sparc.md (fix_b2bst): New attribute.
15725 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
15727 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
15730 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
15732 (*rsqrtsf2_sse): Ditto.
15734 (div<mode>3): Macroize insn from divdf3 and divsf3
15735 using MODEF mode iterator.
15737 2017-07-10 Martin Sebor <msebor@redhat.com>
15739 PR tree-optimization/80397
15740 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
15741 instead of testing for equality to INTEGER_TYPE.
15743 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
15745 * config.gcc: Remove uclibc from arc target spec.
15747 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
15749 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
15751 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15754 * lto-wrapper.c (remove_option): New function.
15755 (merge_and_complain): Merge PIC/PIE options more realistically.
15757 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
15759 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
15763 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
15764 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
15765 * config.in: Regenerate.
15766 * configure: Regenerate.
15767 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
15768 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
15769 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
15770 (TARGET_GASISR_PROLOGUES): ...target mask.
15771 * common/config/avr/avr-common.c
15772 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
15773 Set -mgas-isr-prologues.
15774 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
15775 INSERT_PASS_BEFORE for it.
15776 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
15777 * config/avr/avr.c (avr_option_override)
15778 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
15779 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
15780 (avr_attribute_table) <no_gccisr>: Add new function attribute.
15781 (avr_set_current_function) <is_no_gccisr>: Init machine field.
15782 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
15784 (make_avr_pass_pre_proep): New function.
15785 (emit_push_sfr) <treg>: Add argument to function and use it
15786 instead of TMP_REG.
15787 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
15788 and set machine->gasisr.yes.
15789 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
15790 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
15791 __gcc_isr.n_pushed to .L__stack_usage.
15792 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
15793 (avr_asm_final_postscan_insn): ...this new static function.
15794 * config/avr/avr.h (machine_function)
15795 <is_no_gccisr, use_L__stack_usage>: New fields.
15796 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
15797 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
15798 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
15799 (gasisr, *gasisr): New expander and insn.
15800 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
15801 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
15802 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
15804 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
15806 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
15809 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
15811 Move jump-tables out of .text again.
15814 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
15815 (ASM_OUTPUT_ADDR_VEC): New function.
15816 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
15817 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
15818 INSN_ADDRESSes as asm comment.
15819 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
15820 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
15821 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
15822 * config/avr/avr.md (*tablejump): Adjust comment.
15823 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
15824 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
15826 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
15827 (avr_output_addr_vec): New proto.
15828 (avr_log_t) <insn_addresses>: New field.
15830 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
15833 * config/i386/i386.c (ix86_function_arg_advance): Set
15834 outgoing_args_on_stack to true if there are outgoing arguments
15836 (ix86_function_arg): Likewise.
15837 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
15838 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
15839 * config/i386/i386.h (machine_function): Add
15840 outgoing_args_on_stack.
15842 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
15844 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
15845 supporting pthreds.
15846 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
15848 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
15850 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
15851 (REAL_H): Remove $(MACHMODE_H).
15852 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
15854 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
15855 $(MACHMODE_H) and double-int.h.
15856 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
15858 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
15861 2017-07-07 Andrew Pinski <apinski@cavium.com>
15863 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
15864 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
15866 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15868 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
15869 Add warning if GCC was not configured to link against a GLIBC that
15870 exports the hardware capability bits.
15871 (make_resolver_func): Make resolver function private and not a
15872 COMDAT function. Create the name with clone_function_name instead
15873 of make_unique_name.
15876 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
15877 correct operand in doing the split.
15879 2017-07-07 Carl Love <cel@us.ibm.com>
15881 * config/rs6000/rs6000-c: Add support for built-in function
15882 vector unsigned short vec_pack_to_short_fp32 (vector float,
15884 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
15885 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
15886 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
15887 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
15888 (convert_4f32_8i16): Add define_expand.
15889 * doc/extend.texi: Update the built-in documentation file for the
15890 new built-in function.
15892 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15894 * config/sparc/m8.md: New file.
15895 * config/sparc/sparc.md: Include m8.md.
15897 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15899 * config/sparc/sparc.opt: New option -mvis4b.
15900 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
15901 (sparc_option_override): Handle VIS4B.
15902 (enum sparc_builtins): Define
15903 SPARC_BUILTIN_DICTUNPACK{8,16,32},
15904 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
15905 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
15906 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
15907 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
15908 (check_constant_argument): New function.
15909 (sparc_vis_init_builtins): Define builtins
15910 __builtin_vis_dictunpack{8,16,32},
15911 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
15912 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
15913 __builtin_vis_fpcmpde{8,16,32}shl and
15914 __builtin_vis_fpcmpur{8,16,32}shl.
15915 (sparc_expand_builtin): Check that the constant operands to
15916 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
15917 constant and in range.
15918 * config/sparc/sparc-c.c (sparc_target_macros): Handle
15920 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
15921 (SPARC_IMM5_P): Likewise.
15922 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
15923 (enabled): Handle vis4b.
15924 (UNSPEC_DICTUNPACK): New unspec.
15925 (UNSPEC_FPCMPSHL): Likewise.
15926 (UNSPEC_FPUCMPSHL): Likewise.
15927 (UNSPEC_FPCMPDESHL): Likewise.
15928 (UNSPEC_FPCMPURSHL): Likewise.
15929 (cpu_feature): New CPU feature `vis4b'.
15930 (dictunpack{8,16,32}): New insns.
15931 (FPCSMODE): New mode iterator.
15932 (fpcscond): New code iterator.
15933 (fpcsucond): Likewise.
15934 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
15935 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
15936 (fpcmpde{8,16,32}{si,di}shl): Likewise.
15937 (fpcmpur{8,16,32}{si,di}shl): Likewise.
15938 * config/sparc/constraints.md: Define constraints `q' for unsigned
15939 2-bit integer constants and `t' for unsigned 5-bit integer
15941 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
15943 (imm5_operand_dictunpack16): Likewise.
15944 (imm5_operand_dictunpack32): Likewise.
15945 (imm2_operand): Likewise.
15946 * doc/invoke.texi (SPARC Options): Document -mvis4b.
15947 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
15948 ditunpack* and fpcmp*shl builtins.
15950 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15952 * config.gcc: Handle m8 in --with-{cpu,tune} options.
15953 * config.in: Add HAVE_AS_SPARC6 define.
15954 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
15956 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
15958 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
15959 (CPP_CPU_SPEC): Handle m8.
15960 (ASM_CPU_SPEC): Likewise.
15961 * config/sparc/sparc-opts.h (enum processor_type): Add
15963 * config/sparc/sparc.c (m8_costs): New struct.
15964 (sparc_option_override): Handle TARGET_CPU_m8.
15965 (sparc32_initialize_trampoline): Likewise.
15966 (sparc64_initialize_trampoline): Likewise.
15967 (sparc_issue_rate): Likewise.
15968 (sparc_register_move_cost): Likewise.
15969 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
15970 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
15971 (ASM_CPU64_DEFAULT_SPEC): Likewise.
15972 (CPP_CPU_SPEC): Handle M8.
15973 (ASM_CPU_SPEC): Likewise.
15974 (AS_M8_FLAG): Define.
15975 * config/sparc/sparc.md: Add m8 to the cpu attribute.
15976 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
15977 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
15979 * configure: Regenerate.
15980 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
15983 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15985 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
15987 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
15988 ("*movdi_insn_sp32"): Do not set v3pipe.
15989 ("*movsi_insn"): Likewise.
15990 ("*movdi_insn_sp64"): Likewise.
15991 ("*movsf_insn"): Likewise.
15992 ("*movdf_insn_sp32"): Likewise.
15993 ("*movdf_insn_sp64"): Likewise.
15994 ("*zero_extendsidi2_insn_sp64"): Likewise.
15995 ("*sign_extendsidi2_insn"): Likewise.
15996 ("*mov<VM32:mode>_insn"): Likewise.
15997 ("*mov<VM64:mode>_insn_sp64"): Likewise.
15998 ("*mov<VM64:mode>_insn_sp32"): Likewise.
15999 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16000 ("<vlop:code><VL:mode>3"): Likewise.
16001 ("*not_<vlop:code><VL:mode>3"): Likewise.
16002 ("*nand<VL:mode>_vis"): Likewise.
16003 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
16004 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
16005 ("one_cmpl<VL:mode>2"): Likewise.
16006 ("faligndata<VM64:mode>_vis"): Likewise.
16007 ("alignaddrsi_vis"): Likewise.
16008 ("alignaddrdi_vis"): Likweise.
16009 ("alignaddrlsi_vis"): Likewise.
16010 ("alignaddrldi_vis"): Likewise.
16011 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16012 ("bmaskdi_vis"): Likewise.
16013 ("bmasksi_vis"): Likewise.
16014 ("bshuffle<VM64:mode>_vis"): Likewise.
16015 ("cmask8<P:mode>_vis"): Likewise.
16016 ("cmask16<P:mode>_vis"): Likewise.
16017 ("cmask32<P:mode>_vis"): Likewise.
16018 ("pdistn<P:mode>_vis"): Likewise.
16019 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16021 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16023 * config/sparc/sparc.md ("subtype"): New insn attribute.
16024 ("*wrgsr_sp64"): Set insn subtype.
16025 ("*rdgsr_sp64"): Likewise.
16026 ("alignaddrsi_vis"): Likewise.
16027 ("alignaddrdi_vis"): Likewise.
16028 ("alignaddrlsi_vis"): Likewise.
16029 ("alignaddrldi_vis"): Likewise.
16030 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16031 ("fexpand_vis"): Likewise.
16032 ("fpmerge_vis"): Likewise.
16033 ("faligndata<VM64:mode>_vis"): Likewise.
16034 ("bshuffle<VM64:mode>_vis"): Likewise.
16035 ("cmask8<P:mode>_vis"): Likewise.
16036 ("cmask16<P:mode>_vis"): Likewise.
16037 ("cmask32<P:mode>_vis"): Likewise.
16038 ("fchksm16_vis"): Likewise.
16039 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
16040 ("fmean16_vis"): Likewise.
16041 ("fp<plusminus_insn>64_vis"): Likewise.
16042 ("<plusminus_insn>v8qi3"): Likewise.
16043 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16044 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
16045 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
16046 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
16047 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
16048 ("*movqi_insn"): Likewise.
16049 ("*movhi_insn"): Likewise.
16050 ("*movsi_insn"): Likewise.
16051 ("movsi_pic_gotdata_op"): Likewise.
16052 ("*movdi_insn_sp32"): Likewise.
16053 ("*movdi_insn_sp64"): Likewise.
16054 ("movdi_pic_gotdata_op"): Likewise.
16055 ("*movsf_insn"): Likewise.
16056 ("*movdf_insn_sp32"): Likewise.
16057 ("*movdf_insn_sp64"): Likewise.
16058 ("*zero_extendhisi2_insn"): Likewise.
16059 ("*zero_extendqihi2_insn"): Likewise.
16060 ("*zero_extendqisi2_insn"): Likewise.
16061 ("*zero_extendqidi2_insn"): Likewise.
16062 ("*zero_extendhidi2_insn"): Likewise.
16063 ("*zero_extendsidi2_insn_sp64"): Likewise.
16064 ("ldfsr"): Likewise.
16065 ("prefetch_64"): Likewise.
16066 ("prefetch_32"): Likewise.
16067 ("tie_ld32"): Likewise.
16068 ("tie_ld64"): Likewise.
16069 ("*tldo_ldub_sp32"): Likewise.
16070 ("*tldo_ldub1_sp32"): Likewise.
16071 ("*tldo_ldub2_sp32"): Likewise.
16072 ("*tldo_ldub_sp64"): Likewise.
16073 ("*tldo_ldub1_sp64"): Likewise.
16074 ("*tldo_ldub2_sp64"): Likewise.
16075 ("*tldo_ldub3_sp64"): Likewise.
16076 ("*tldo_lduh_sp32"): Likewise.
16077 ("*tldo_lduh1_sp32"): Likewise.
16078 ("*tldo_lduh_sp64"): Likewise.
16079 ("*tldo_lduh1_sp64"): Likewise.
16080 ("*tldo_lduh2_sp64"): Likewise.
16081 ("*tldo_lduw_sp32"): Likewise.
16082 ("*tldo_lduw_sp64"): Likewise.
16083 ("*tldo_lduw1_sp64"): Likewise.
16084 ("*tldo_ldx_sp64"): Likewise.
16085 ("*mov<VM32:mode>_insn"): Likewise.
16086 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16087 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16089 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16091 * config/sparc/sparc.md ("type"): New insn type viscmp.
16092 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
16094 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
16095 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
16096 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16097 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
16099 ("n7_vis_logical_11cycle"): Likewise.
16100 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
16101 * config/sparc/niagara2.md ("niag3_vis": Likewise.
16102 * config/sparc/niagara.md ("niag_vis"): Likewise.
16103 * config/sparc/ultra3.md ("us3_fga"): Likewise.
16104 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
16106 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16108 * config/sparc/sparc.md: New instruction type `bmask'.
16109 (bmaskdi_vis): Use the `bmask' type.
16110 (bmasksi_vis): Likewise.
16111 * config/sparc/ultra3.md (us3_array): Likewise.
16112 * config/sparc/niagara7.md (n7_array): Likewise.
16113 * config/sparc/niagara4.md (n4_array): Likewise.
16114 * config/sparc/niagara2.md (niag2_vis): Likewise.
16115 (niag3_vis): Likewise.
16116 * config/sparc/niagara.md (niag_vis): Likewise.
16118 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16120 * ipa-comdats.c: Remove optimize check from gate.
16121 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
16122 for functions not optimized.
16123 (ipa_fn_summary_read): Skip optimize check.
16124 (ipa_fn_summary_write): Likewise.
16125 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
16127 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
16129 (can_inline_edge_p): Check flag_pcc_struct_return for match.
16130 (check_callers): Give up on caller which is not optimized.
16131 (inline_small_functions): Likewise.
16132 (ipa_inline): Do not give up when not optimizing.
16133 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
16134 away unoptimizes cdtors.
16135 (whole_program_function_and_variable_visibility): Do
16136 ipa_discover_readonly_nonaddressable_vars in LTO mode.
16137 * ipa.c (process_references): Do not check optimize.
16138 (symbol_table::remove_unreachable_nodes): Update optimize check.
16139 (set_writeonly_bit): Update optimize check.
16140 (pass_ipa_cdtor_merge::gate): Do not check optimize.
16141 (pass_ipa_single_use::gate): Remove.
16143 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16145 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
16146 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
16147 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
16148 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
16149 permute_load, permute_store, adjust_extract, adjust_splat,
16150 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
16151 replace_swap_with_copy, dump_swap_insn_table,
16152 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
16153 recombine_lvx_pattern, recombine_stvx_pattern,
16154 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
16155 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
16156 to file rs6000-p8swap.c.
16157 * config/rs6000/rs6000-p8swap.c: New file.
16158 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
16159 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
16160 and rs6000*-*-* targets.
16162 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16164 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
16166 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16168 * lto-wrapper.c (merge_and_complain): Do not merge
16169 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
16170 fsigned_zeros, ftrapping_math, fwrapv.
16171 (append_compiler_options): Do not track these options.
16172 (append_linker_options): Likewie
16174 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16176 * cgraphunit.c (cgraph_node::finalize_function): When
16177 !flag_toplevel_reorde set no_reorder flag.
16178 (varpool_node::finalize_decl): Likewise.
16179 (symbol_table::compile): Drop no toplevel reorder path.
16181 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16183 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
16184 edges; zero probability is not better than uninitialized.
16186 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
16188 * asan.h (asan_sanitize_allocas_p): Declare.
16189 * asan.c (asan_sanitize_allocas_p): New function.
16190 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
16191 (handle_builtin_alloca): Likewise.
16192 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
16193 if !asan_sanitize_allocas_p.
16194 * params.def (asan-instrument-allocas): Add new option.
16195 * params.h (ASAN_PROTECT_ALLOCAS): Define.
16196 * opts.c (common_handle_option): Disable allocas sanitization for
16199 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
16201 * asan.c: Include gimple-fold.h.
16202 (get_last_alloca_addr): New function.
16203 (handle_builtin_stackrestore): Likewise.
16204 (handle_builtin_alloca): Likewise.
16205 (asan_emit_allocas_unpoison): Likewise.
16206 (get_mem_refs_of_builtin_call): Add new parameter, remove const
16207 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
16208 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
16209 (instrument_builtin_call): Pass gimple iterator to
16210 get_mem_refs_of_builtin_call.
16211 (last_alloca_addr): New global.
16212 * asan.h (asan_emit_allocas_unpoison): Declare.
16213 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
16214 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
16215 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
16216 if function calls alloca.
16217 * gimple-fold.c (replace_call_with_value): Remove static keyword.
16218 * gimple-fold.h (replace_call_with_value): Declare.
16219 * internal-fn.c: Include asan.h.
16220 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
16221 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
16223 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16225 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
16226 (C_SELFTEST_FLAGS): New.
16227 (CPP_SELFTEST_FLAGS): New.
16228 (SELFTEST_DEPS): New, from deps of s-selftest.
16229 (C_SELFTEST_DEPS): New, from deps of s-selftest.
16230 (CPP_SELFTEST_DEPS): New.
16231 (selftest): Add dependency on s-selftest-c++.
16232 (s-selftest): Rename to...
16233 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
16234 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
16235 than SELFTEST_FLAGS.
16236 (selftest-gdb): Rename to...
16237 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
16239 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
16240 (selftest-valgrind): Rename to...
16241 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
16243 (selftest-valgrind): Reintroduce as an alias for
16244 selftest-c-valgrind.
16245 (s-selftest-c++): New.
16246 (selftest-c++-gdb): New.
16247 (selftest-c++-valgrind): New.
16249 2017-07-06 Olivier Hainque <hainque@adacore.com>
16251 * gcc.c (process_command): When deciding if undefined variables
16252 should be ignored when processing specs, accept "gcc -v" as well.
16254 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16256 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
16257 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
16259 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
16261 * config/arm/arm-cpus.in (armv8-r): Add new entry.
16262 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
16263 * config/arm/arm-tables.opt: Regenerate.
16264 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
16266 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
16268 2017-07-06 Carl Love <cel@us.ibm.com>
16270 * ChangeLog: Clean up from mid air collision
16272 2017-07-06 Carl Love <cel@us.ibm.com>
16274 * config/rs6000/rs6000-c.c: Add support for built-in functions
16275 vector signed int vec_subc (vector signed int, vector signed int);
16276 vector signed __int128 vec_subc (vector signed __int128,
16277 vector signed __int128);
16278 vector unsigned __int128 vec_subc (vector unsigned __int128,
16279 vector unsigned __int128);
16280 vector signed int vec_sube (vector signed int, vector signed int,
16281 vector signed int);
16282 vector unsigned int vec_sube (vector unsigned int,
16283 vector unsigned int,
16284 vector unsigned int);
16285 vector signed __int128 vec_sube (vector signed __int128,
16286 vector signed __int128,
16287 vector signed__int128);
16288 vector unsigned __int128 vec_sube (vector unsigned __int128,
16289 vector unsigned __int128,
16290 vector unsigned __int128);
16291 vector signed int vec_subec (vector signed int, vector signed int,
16292 vector signed int);
16293 vector unsigned int vec_subec (vector unsigned int,
16294 vector unsigned int,
16295 vector unsigned int);
16296 vector signed __int128 vec_subec (vector signed __int128,
16297 vector signed __int128,
16298 vector signed__int128);
16299 vector unsigned __int128 vec_subec (vector unsigned __int128,
16300 vector unsigned __int128,
16301 vector unsigned __int128);
16302 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
16303 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
16304 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
16305 BU_ALTIVEC_OVERLOAD_X definitions.
16306 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
16307 * doc/extend.texi: Update the built-in documentation file for the new
16308 built-in functions.
16310 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16313 * diagnostic-show-locus.c (layout::layout): Use start and finish
16314 spelling location for the start and finish of each range.
16315 * genmatch.c (linemap_client_expand_location_to_spelling_point):
16316 Add unused aspect param.
16317 * input.c (expand_location_1): Add "aspect" param, and use it
16318 to access the correct part of the location.
16319 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
16321 (expand_location_to_spelling_point): Likewise.
16322 (linemap_client_expand_location_to_spelling_point): Add "aspect"
16323 param, and pass it to expand_location_1.
16325 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
16327 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
16328 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
16329 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
16330 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
16331 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
16332 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
16333 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
16334 _mm_maskz_getmant_ss): New intrinsics.
16335 (__builtin_ia32_getexpss128_mask): Changed to ...
16336 __builtin_ia32_getexpss128_round ... this.
16337 (__builtin_ia32_getexpsd128_mask): Changed to ...
16338 __builtin_ia32_getexpsd128_round ... this.
16339 * config/i386/i386-builtin-types.def
16340 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
16341 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
16342 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
16343 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
16344 __builtin_ia32_getmantss_mask_round): New builtins.
16345 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
16346 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
16347 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
16348 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
16349 * config/i386/sse.md
16350 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
16351 avx512f_sgetexp<mode><mask_scalar_name>
16352 <round_saeonly_scalar_name> ... this.
16353 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
16354 %0, %1, %2<round_saeonly_op3>}): Changed to ...
16355 vgetexp<ssescalarmodesuffix>
16356 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16357 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
16358 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
16359 avx512f_vgetmant<mode><mask_scalar_name>
16360 <round_saeonly_scalar_name> ... this.
16361 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
16362 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
16363 vgetmant<ssescalarmodesuffix>
16364 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
16365 %0<mask_scalar_operand4>, %1, %2
16366 <round_saeonly_scalar_mask_op4>, %3} ... this.
16367 * config/i386/subst.md (mask_scalar_operand4,
16368 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
16369 round_saeonly_scalar_nimm_predicate): New subst attributes.
16371 2017-07-06 Julia Koval <julia.koval@intel.com>
16373 * config/i386/i386.c (ix86_erase_embedded_rounding):
16374 Remove code for old rounding pattern.
16376 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
16378 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
16380 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
16382 * doc/sourcebuild.texi (Test Directives, Variants of
16383 dg-require-support): Add documentation for dg-require-stack-check.
16385 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
16387 * config/i386/subst.md (mask_scalar, round_scalar,
16388 round_saeonly_scalar): New meta-templates.
16389 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
16390 round_scalar_mask_operand3, round_scalar_mask_op3,
16391 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
16392 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
16393 round_saeonly_scalar_constraint,
16394 round_saeonly_scalar_prefix): New subst attribute.
16395 * config/i386/sse.md
16396 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
16397 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
16398 <round_scalar_name> ... this.
16399 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
16400 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
16401 <round_scalar_name> ... this.
16402 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
16403 <sse>_vm<code><mode>3<mask_scalar_name>
16404 <round_saeonly_scalar_name> ... this.
16405 (v<plusminus_mnemonic><ssescalarmodesuffix>
16406 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
16407 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
16408 v<plusminus_mnemonic><ssescalarmodesuffix>
16409 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16410 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
16411 (v<multdiv_mnemonic><ssescalarmodesuffix>
16412 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
16413 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
16414 v<multdiv_mnemonic><ssescalarmodesuffix>
16415 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16416 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
16417 (v<maxmin_float><ssescalarmodesuffix>
16418 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
16419 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
16420 v<maxmin_float><ssescalarmodesuffix>
16421 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16422 %0<mask_scalar_operand3>, %1, %<iptr>2
16423 <round_saeonly_scalar_mask_op3>} ... this.
16425 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
16427 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
16428 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
16430 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
16431 Alan Hayward <alan.hayward@arm.com>
16432 David Sherwood <david.sherwood@arm.com>
16434 * combine.c (simplify_if_then_else): Remove "enum" before
16436 * compare-elim.c (can_eliminate_compare): Likewise.
16437 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
16439 (aarch64_lookup_simd_builtin_type): Likewise.
16440 (aarch64_simd_builtin_type): Likewise.
16441 (aarch64_init_simd_builtin_types): Likewise.
16442 (aarch64_simd_expand_args): Likewise.
16443 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
16445 (aarch64_reverse_mask): Likewise.
16446 (aarch64_simd_emit_reg_reg_move): Likewise.
16447 (aarch64_gen_adjusted_ldpstp): Likewise.
16448 (aarch64_ccmp_mode_to_code): Likewise.
16449 (aarch64_operands_ok_for_ldpstp): Likewise.
16450 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16451 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
16453 (aarch64_min_divisions_for_recip_mul): Likewise.
16454 (aarch64_reassociation_width): Likewise.
16455 (aarch64_get_condition_code_1): Likewise.
16456 (aarch64_simd_emit_reg_reg_move): Likewise.
16457 (aarch64_simd_attr_length_rglist): Likewise.
16458 (aarch64_reverse_mask): Likewise.
16459 (aarch64_operands_ok_for_ldpstp): Likewise.
16460 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16461 (aarch64_gen_adjusted_ldpstp): Likewise.
16462 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
16464 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
16465 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
16466 (arm_lookup_simd_builtin_type): Likewise.
16467 (arm_simd_builtin_type): Likewise.
16468 (arm_init_simd_builtin_types): Likewise.
16469 (arm_expand_builtin_args): Likewise.
16470 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
16471 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
16472 (ft32_setup_incoming_varargs): Likewise.
16473 (ft32_function_arg): Likewise.
16474 (ft32_function_arg_advance): Likewise.
16475 (ft32_pass_by_reference): Likewise.
16476 (ft32_arg_partial_bytes): Likewise.
16477 (ft32_valid_pointer_mode): Likewise.
16478 (ft32_addr_space_pointer_mode): Likewise.
16479 (ft32_addr_space_legitimate_address_p): Likewise.
16480 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
16482 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
16483 (ix86_emit_outlined_ms2sysv_restore): Likewise.
16484 (iamcu_alignment): Likewise.
16485 (canonicalize_vector_int_perm): Likewise.
16486 (ix86_noce_conversion_profitable_p): Likewise.
16487 (ix86_mpx_bound_mode): Likewise.
16488 (ix86_operands_ok_for_move_multiple): Likewise.
16489 * config/microblaze/microblaze-protos.h
16490 (microblaze_expand_conditional_branch_reg): Likewise.
16491 * config/microblaze/microblaze.c
16492 (microblaze_expand_conditional_branch_reg): Likewise.
16493 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
16495 (rs6000_reassociation_width): Likewise.
16496 (rs6000_invalid_binary_op): Likewise.
16497 (fusion_p9_p): Likewise.
16498 (emit_fusion_p9_load): Likewise.
16499 (emit_fusion_p9_store): Likewise.
16500 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
16502 (riscv_hard_regno_mode_ok_p): Likewise.
16503 (riscv_address_insns): Likewise.
16504 (riscv_split_symbol): Likewise.
16505 (riscv_legitimize_move): Likewise.
16506 (riscv_function_value): Likewise.
16507 (riscv_hard_regno_nregs): Likewise.
16508 (riscv_expand_builtin): Likewise.
16509 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
16510 (riscv_build_integer): Likewise.
16511 (riscv_split_integer): Likewise.
16512 (riscv_legitimate_constant_p): Likewise.
16513 (riscv_cannot_force_const_mem): Likewise.
16514 (riscv_regno_mode_ok_for_base_p): Likewise.
16515 (riscv_valid_base_register_p): Likewise.
16516 (riscv_valid_offset_p): Likewise.
16517 (riscv_valid_lo_sum_p): Likewise.
16518 (riscv_classify_address): Likewise.
16519 (riscv_legitimate_address_p): Likewise.
16520 (riscv_address_insns): Likewise.
16521 (riscv_load_store_insns): Likewise.
16522 (riscv_force_binary): Likewise.
16523 (riscv_split_symbol): Likewise.
16524 (riscv_force_address): Likewise.
16525 (riscv_legitimize_address): Likewise.
16526 (riscv_move_integer): Likewise.
16527 (riscv_legitimize_const_move): Likewise.
16528 (riscv_legitimize_move): Likewise.
16529 (riscv_address_cost): Likewise.
16530 (riscv_subword): Likewise.
16531 (riscv_output_move): Likewise.
16532 (riscv_canonicalize_int_order_test): Likewise.
16533 (riscv_emit_int_order_test): Likewise.
16534 (riscv_function_arg_boundary): Likewise.
16535 (riscv_pass_mode_in_fpr_p): Likewise.
16536 (riscv_pass_fpr_single): Likewise.
16537 (riscv_pass_fpr_pair): Likewise.
16538 (riscv_get_arg_info): Likewise.
16539 (riscv_function_arg): Likewise.
16540 (riscv_function_arg_advance): Likewise.
16541 (riscv_arg_partial_bytes): Likewise.
16542 (riscv_function_value): Likewise.
16543 (riscv_pass_by_reference): Likewise.
16544 (riscv_setup_incoming_varargs): Likewise.
16545 (riscv_print_operand): Likewise.
16546 (riscv_elf_select_rtx_section): Likewise.
16547 (riscv_save_restore_reg): Likewise.
16548 (riscv_for_each_saved_reg): Likewise.
16549 (riscv_register_move_cost): Likewise.
16550 (riscv_hard_regno_mode_ok_p): Likewise.
16551 (riscv_hard_regno_nregs): Likewise.
16552 (riscv_class_max_nregs): Likewise.
16553 (riscv_memory_move_cost): Likewise.
16554 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
16555 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
16556 (rl78_addr_space_address_mode): Likewise.
16557 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16559 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
16560 (rs6000_reassociation_width): Likewise.
16561 (rs6000_invalid_binary_op): Likewise.
16562 (fusion_p9_p): Likewise.
16563 (emit_fusion_p9_load): Likewise.
16564 (emit_fusion_p9_store): Likewise.
16565 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
16566 (ok_for_simple_move_operands): Likewise.
16567 (ok_for_simple_move_strict_operands): Likewise.
16568 (ok_for_simple_arith_logic_operands): Likewise.
16569 (visium_legitimize_reload_address): Likewise.
16570 (visium_select_cc_mode): Likewise.
16571 (output_cbranch): Likewise.
16572 (visium_split_double_move): Likewise.
16573 (visium_expand_copysign): Likewise.
16574 (visium_expand_int_cstore): Likewise.
16575 (visium_expand_fp_cstore): Likewise.
16576 * config/visium/visium.c (visium_pass_by_reference): Likewise.
16577 (visium_function_arg): Likewise.
16578 (visium_function_arg_advance): Likewise.
16579 (visium_libcall_value): Likewise.
16580 (visium_setup_incoming_varargs): Likewise.
16581 (visium_legitimate_constant_p): Likewise.
16582 (visium_legitimate_address_p): Likewise.
16583 (visium_legitimize_address): Likewise.
16584 (visium_secondary_reload): Likewise.
16585 (visium_register_move_cost): Likewise.
16586 (visium_memory_move_cost): Likewise.
16587 (prepare_move_operands): Likewise.
16588 (ok_for_simple_move_operands): Likewise.
16589 (ok_for_simple_move_strict_operands): Likewise.
16590 (ok_for_simple_arith_logic_operands): Likewise.
16591 (visium_function_value_1): Likewise.
16592 (rtx_ok_for_offset_p): Likewise.
16593 (visium_legitimize_reload_address): Likewise.
16594 (visium_split_double_move): Likewise.
16595 (visium_expand_copysign): Likewise.
16596 (visium_expand_int_cstore): Likewise.
16597 (visium_expand_fp_cstore): Likewise.
16598 (visium_split_cstore): Likewise.
16599 (visium_select_cc_mode): Likewise.
16600 (visium_split_cbranch): Likewise.
16601 (output_cbranch): Likewise.
16602 (visium_print_operand_address): Likewise.
16603 * expmed.c (flip_storage_order): Likewise.
16604 * expmed.h (emit_cstore): Likewise.
16605 (flip_storage_order): Likewise.
16606 * genrecog.c (validate_pattern): Likewise.
16607 * hsa-gen.c (gen_hsa_addr): Likewise.
16608 * internal-fn.c (expand_arith_overflow): Likewise.
16609 * ira-color.c (allocno_copy_cost_saving): Likewise.
16610 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16611 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
16612 (process_invariant_for_inheritance): Likewise.
16613 * lra-eliminations.c (move_plus_up): Likewise.
16614 * omp-low.c (lower_oacc_reductions): Likewise.
16615 * simplify-rtx.c (simplify_subreg): Likewise.
16616 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
16617 (TARGET_CHKP_BOUND_MODE): Likewise..
16618 * targhooks.c (default_chkp_bound_mode): Likewise.
16619 (default_setup_incoming_vararg_bounds): Likewise.
16620 * targhooks.h (default_chkp_bound_mode): Likewise.
16621 (default_setup_incoming_vararg_bounds): Likewise.
16622 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
16623 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
16624 (have_whole_vector_shift): Likewise.
16625 * tree-vect-stmts.c (vectorizable_load): Likewise.
16626 * doc/tm.texi: Regenerate.
16628 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
16630 Graceful degrade if Binutils PR21472 is not available.
16633 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
16634 .rodata in flash test fails.
16635 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
16636 * confgure: Regenerate.
16637 * config.in: Regenerate.
16638 * config/avr/avr.c (avr_asm_named_section)
16639 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
16640 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
16641 (avr_asm_init_sections): Same.
16643 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16645 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
16646 (fma<VH:mode>4_intrinsic): Likewise.
16647 (*fmsub<VCVTF:mode>4): Likewise.
16648 (*fmsub<VH:mode>4_intrinsic): Likewise.
16650 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
16653 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
16654 Don't depend on "optimize > 0".
16655 (out_movhi_r_mr, out_movqi_mr_r): Same.
16656 (out_movhi_mr_r, out_movqi_r_mr): Same.
16657 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
16658 io_address_operand on "optimize > 0".
16660 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16662 * tree-loop-distribution.c: Add general explanantion on the pass.
16663 (generate_loops_for_partition): Mark distributed loop.
16664 (pg_add_dependence_edges): New parameter. Handle alias data
16665 dependence specially and record it in the parameter if asked.
16666 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
16667 (init_partition_graph_vertices, add_partition_graph_edge): New.
16668 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
16669 (free_partition_graph_vdata, build_partition_graph): New.
16670 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
16671 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
16672 (data_ref_segment_size, latch_dominated_by_data_ref): New.
16673 (compute_alias_check_pairs, version_loop_by_alias_check): New.
16674 (version_for_distribution_p, finalize_partitions): New.
16675 (distribute_loop): Handle alias data dependence specially. Factor
16676 out loop fusion code as functions and call these functions.
16678 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16680 * tree-loop-distribution.c (classify_partition): New parameter and
16681 better handle reduction statement.
16682 (rdg_build_partitions): Revise comment.
16683 (distribute_loop): Compute statements in all partitions and pass it
16684 to classify_partition.
16686 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16688 * tree-loop-distribution.c (enum partition_type): New.
16689 (struct partition): New field type.
16690 (partition_merge_into): Add parameter. Update partition type.
16691 (data_dep_in_cycle_p, update_type_for_merge): New functions.
16692 (build_rdg_partition_for_vertex): Compute partition type.
16693 (rdg_build_partitions): Dump partition type.
16694 (distribute_loop): Update calls to partition_merge_into.
16696 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16698 * tree-loop-distribution.c (struct ddr_hasher): New.
16699 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
16701 (classify_partition): Call get_data_dependence.
16702 (pg_add_dependence_edges): Ditto.
16703 (distribute_loop): Release data dependence hash table.
16705 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16707 * tree-loop-distribution.c (ref_base_address): Delete.
16708 (similar_memory_accesses): Rename ...
16709 (share_memory_accesses): ... to this. Check if partitions access
16710 the same memory reference.
16711 (distribute_loop): Call share_memory_accesses.
16713 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16715 * tree-loop-distribution.c (struct partition): New field recording
16716 its data reference.
16717 (partition_alloc, partition_free): Init and release data refs.
16718 (partition_merge_into): Merge data refs.
16719 (build_rdg_partition_for_vertex): Collect data refs for partition.
16720 (pg_add_dependence_edges): Change parameters from vector to bitmap.
16722 (distribute_loop): Remve data refs from vertice data of partition
16725 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16727 * tree-loop-distribution.c (params.h): Include header file.
16728 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
16729 (datarefs_vec): New global var.
16730 (create_rdg_vertices): Use datarefs_vec directly.
16731 (free_rdg): Don't free data references.
16732 (build_rdg): Update use. Don't free data references.
16733 (distribute_loop): Compute global variable for data references.
16734 Bail out if there are too many data references.
16736 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16738 * tree-loop-distribution.c (loop_nest): New global var.
16739 (build_rdg): Use loop directly, rather than loop nest.
16740 (pg_add_dependence_edges): Remove loop nest parameter. Use global
16742 (distribute_loop): Compute global variable loop nest. Update use.
16744 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16746 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
16747 (partition_merge_into): New parameter. Dump reason for fusion.
16748 (distribute_loop): Update use of partition_merge_into.
16750 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16752 * tree-loop-distribution.c (bb_top_order_index): New.
16753 (bb_top_order_index_size, bb_top_order_cmp): New.
16754 (stmts_from_loop): Use topological order.
16755 (pass_loop_distribution::execute): Compute and release topological
16756 order for basic blocks.
16758 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16760 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
16763 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16765 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
16766 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
16767 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
16768 * internal-fn.def (LOOP_DIST_ALIAS): New.
16769 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
16770 (fold_loop_internal_call): ... this.
16771 (vect_loop_dist_alias_call): New function.
16772 (set_uid_loop_bbs): Call fold_loop_internal_call.
16773 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
16776 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
16779 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
16780 Require dead FLAGS_REG at the beginning of a peephole.
16782 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
16785 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
16786 arguments in the call to __builtin_ia32_sbb_u32.
16787 (_subborrow_u64): Swap _X and _Y arguments in the call to
16788 __builtin_ia32_sbb_u64.
16790 2017-07-04 Jakub Jelinek <jakub@redhat.com>
16793 * tree-vrp.c (compare_assert_loc): Turn into a function template
16794 with stable template parameter. Only test if a->e is NULL,
16795 !a->e == !b->e has been verified already. Use e == NULL or
16796 e != NULL instead of e or ! e tests. If stable is true, don't use
16797 iterative_hash_expr, on the other side allow a or b or both NULL
16798 and sort the NULLs last.
16799 (process_assert_insertions): Sort using compare_assert_loc<false>
16800 instead of compare_assert_loc, later sort using
16801 compare_assert_loc<true> before calling process_assert_insertions_for
16802 in a loop. Use break instead of continue once seen NULL pointer.
16804 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
16806 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
16807 Cortex-R7 and Cortex-R8 processors.
16809 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
16811 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
16812 uninitialized while src is not.
16814 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
16816 * common/config/arm/arm-common.c: Adjust include path for
16818 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
16819 (arm-cpu.h): Create in build directory. Adjust dependency rules.
16820 (arm-cpu-data.h): Likewise.
16821 (arm-cpu-cdata.h): Likewise.
16822 * config/arm/arm-cpu.h: Delete.
16823 * config/arm/arm-cpu-cdata.h: Delete.
16824 * config/arm/arm-cpu-data.h: Delete.
16826 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
16828 * config/arm/arm-cpus.in (cortex-a55): New.
16829 (cortex-a75): Likewise.
16830 (cortex-a75.cortex-a55): Likewise.
16831 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
16833 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
16834 * config/arm/arm-cpu-cdata.h: Regenerate.
16835 * config/arm/arm-cpu-data.h: Regenerate.
16836 * config/arm/arm-cpu.h: Regenerate.
16837 * config/arm/arm-tables.opt: Regenerate.
16838 * config/arm/arm-tune.md: Regenerate.
16840 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
16842 * haifa-sched.c (sched_create_recovery_edges): Update profile.
16844 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
16846 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
16849 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
16851 PR tree-optimization/81292
16852 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
16853 full_string_p, also call adjust_related_strinfos if the adjustment
16854 is simple, otherwise invalidate related strinfos.
16856 2017-07-04 Martin Liska <mliska@suse.cz>
16859 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
16860 newly created variable as DECL_IGNORED_P.
16862 2017-07-04 Martin Liska <mliska@suse.cz>
16865 * ipa-inline.c (inline_small_functions):
16866 Use xstrdup_for_dump.
16868 2017-07-04 Tom de Vries <tom@codesourcery.com>
16870 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
16872 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
16875 * config/darwin.c (darwin_function_switched_text_sections):
16878 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
16880 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
16882 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
16884 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
16886 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16888 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
16889 min_profitable_iters, and th as inclusive lower bounds.
16890 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
16891 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
16892 for min_profitable_iters and min_profitable_estimate.
16893 (vect_transform_loop): Treat th as an inclusive lower bound.
16894 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
16896 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
16899 * config/darwin.c (darwin_function_switched_text_sections):
16900 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
16901 in two pieces, and suppress the use of buf.
16903 2017-07-03 Nathan Sidwell <nathan@acm.org>
16905 * hash-table.h (hash_table_mod1): Fix indentation.
16907 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16909 PR middle-end/81290
16910 * predict.c (force_edge_cold): Be more careful about propagation
16912 * profile-count.h (profile_probability::guessed,
16913 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
16915 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
16917 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
16919 * doc/invoke.texi (rcpc architecture extension): Document it.
16921 2017-07-03 Richard Biener <rguenther@suse.de>
16923 PR tree-optimization/60510
16924 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
16925 the scalar reduction PHI and use it.
16926 (vectorizable_reduction): Properly guard the single_defuse_cycle
16927 path for non-SLP reduction chains where we cannot use it.
16928 Rework reduc_def/index and vector type deduction. Rework
16929 vector operand gathering during reduction op code-gen.
16930 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
16931 chains dissolve the chain and leave it to non-SLP reduction
16934 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16936 * tree-data-ref.h (dr_alignment): Declare.
16937 * tree-data-ref.c (dr_alignment): New function.
16938 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
16939 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
16941 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
16943 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16945 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
16946 and base_misalignment fields.
16947 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
16948 * tree-data-ref.c: Include builtins.h.
16949 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
16950 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
16951 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
16952 * tree-vect-data-refs.c: Include tree-cfg.h.
16953 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
16954 fields instead of calculating an alignment here.
16955 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
16956 innermost_loop_behavior fields.
16958 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16960 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
16962 (DR_STEP_ALIGNMENT): New macro.
16963 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
16964 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
16965 (create_data_ref): Print it.
16966 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
16967 to tell whether the step preserves vector (mis)alignment.
16968 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
16969 Move the check for an integer step and generalise to all INTEGER_CST.
16970 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
16971 Print the outer step alignment.
16973 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16975 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
16976 with offset_alignment.
16977 (DR_ALIGNED_TO): Delete.
16978 (DR_OFFSET_ALIGNMENT): New macro.
16979 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
16980 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
16981 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
16982 (create_data_ref): Likewise.
16983 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
16984 (vect_analyze_data_refs): Likewise.
16985 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
16986 creating dummy innermost behavior.
16988 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16990 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
16991 with a "innermost_loop_behavior *" and refeence tree.
16992 * tree-data-ref.c (dr_analyze_innermost): Likewise.
16993 (create_data_ref): Update call accordingly.
16994 * tree-predcom.c (find_looparound_phi): Likewise.
16996 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16998 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
16999 fields with dr_wrt_vec_loop.
17000 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
17001 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
17002 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
17003 (vect_dr_behavior): New function.
17004 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17005 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
17006 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
17007 track whether the step preserves the misalignment.
17008 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17009 Use vect_dr_behavior.
17010 (vect_setup_realignment): Update call accordingly.
17011 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
17012 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
17013 call to vect_create_addr_base_for_vector_ref.
17014 (vect_create_cond_for_align_checks): Likewise.
17015 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
17016 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
17017 (vect_recog_mask_conversion_pattern): Likewise.
17018 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
17019 (new_stmt_vec_info): Remove redundant zeroing.
17021 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17023 * common/config/arm/arm-common.c (arm_be8_option): New function.
17024 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
17025 (ISA_ARMv6): Add isa_bit_be8.
17026 * config/arm/arm.h (arm_be8_option): Add prototype.
17027 (BE8_SPEC_FUNCTION): New define.
17028 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
17029 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
17030 (mlittle-endian): Similarly.
17031 (mbe8, mbe32): New options.
17032 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
17033 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
17035 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17037 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
17039 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17041 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
17042 (cleanup_tree_cfg_bb): Use it.
17043 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
17045 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
17047 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17050 * loop-doloop.c (add_test): Update profile.
17052 2017-07-03 Martin Liska <mliska@suse.cz>
17055 * sanopt.c (rewrite_usage_of_param): New function.
17056 (sanitize_rewrite_addressable_params): Likewise.
17057 (pass_sanopt::execute): Call rewrite_usage_of_param.
17059 2017-07-03 Richard Biener <rguenther@suse.de>
17061 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
17062 back to using VIEW_CONVERT_EXPR.
17064 2017-07-03 Martin Liska <mliska@suse.cz>
17067 * doc/extend.texi: Document when a resolver function is
17068 generated for target_clones.
17070 2017-07-03 Martin Liska <mliska@suse.cz>
17072 * asan.c (asan_emit_stack_protection): Unpoison just red zones
17073 and shadow memory of auto variables which are subject of
17074 use-after-scope sanitization.
17075 (asan_expand_mark_ifn): Add do set only when is_poison.
17077 2016-07-03 Richard Biener <rguenther@suse.de>
17079 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
17081 (vect_force_simple_reduction): Record reduction def -> phi mapping.
17082 (vectorizable_reduction): Perform reduction PHI creation when
17083 visiting a reduction PHI and adjust and simplify code generation
17084 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
17085 (vect_transform_loop): Visit reduction PHIs.
17086 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
17087 defs into the SLP tree.
17088 (vect_build_slp_tree): Reduction defs terminate the recursion.
17089 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
17091 (vect_get_vec_defs_for_stmt_copy): Export.
17092 (vect_get_vec_defs): Likewise.
17093 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
17095 (vect_get_vec_defs_for_stmt_copy): Declare.
17096 (vect_get_vec_defs): Likewise.
17098 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17100 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
17101 parameter with a "loop" parameter and use it instead of the
17102 loop containing DR_STMT. Don't check simple_iv when doing
17103 BB analysis. Describe the two analysis modes in the comment.
17105 2017-07-03 Tom de Vries <tom@codesourcery.com>
17107 PR tree-optimization/69468
17108 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
17109 (find_same_succ_bb): Handle ignore_edge_flags.
17111 2017-07-03 Tom de Vries <tom@codesourcery.com>
17113 PR tree-optimization/81192
17114 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
17116 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
17118 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
17120 2017-07-03 Tom de Vries <tom@codesourcery.com>
17122 PR tree-optimization/81192
17123 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
17124 BB_SAME_SUCC (bb) == NULL.
17126 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17128 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
17131 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17133 * dumpfile.c: Include profile-count.h
17134 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
17136 (insert_cond_bb): Update profile.
17137 * tree-cfg.h (insert_cond_bb): Update prototype.
17138 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
17139 * tree-dump.c: Do not include tree-cfg.
17141 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17143 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
17145 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17147 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
17150 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17152 * tree-complex.c (expand_complex_div_wide): update profile.
17154 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17155 Alan Hayward <alan.hayward@arm.com>
17156 David Sherwood <david.sherwood@arm.com>
17158 * Makefile.in (MACHMODE_H): Remove insn-modes.h
17159 (CORETYPES_H): New define.
17160 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
17161 (insn-modes-inline.h, s-modes-inline-h): New rules.
17162 (generated_files): Add insn-modes-inline.h.
17163 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
17164 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
17165 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
17166 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
17167 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
17168 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
17169 (build/gencodes.o, build/genconditions.o): Likewise.
17170 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
17171 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
17172 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
17173 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
17174 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
17175 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
17176 * coretypes.h: Include everything up to real.h for generators.
17177 Include insn-modes.h first. Include wide-int-print.h after
17178 wide-int.h. Include insn-modes-inline.h and then machmode.h.
17179 * machmode.h: Don't include insn-modes.h here.
17180 * function-tests.c: Remove includes of signop.h, machmode.h,
17181 double-int.h and wide-int.h.
17183 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
17185 * optc-save-gen.awk: Likewise.
17186 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
17187 * godump.c: Remove include of wide-int-print.h.
17188 * pretty-print.h: Likewise.
17189 * wide-int-print.cc: Likewise.
17190 * wide-int.cc: Likewise.
17191 * hash-map-tests.c: Remove include of signop.h.
17192 * hash-set-tests.c: Likewise.
17193 * rtl-tests.c: Likewise.
17194 * mkconfig.sh: Remove include of machmode.h.
17195 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
17197 (emit_insn_modes_inline_h): ...this new function. Emit the code
17198 into an insn-modes-inline.h header file, adding appropriate
17199 include guards and end comments.
17200 (emit_insn_modes_c_header): Remove include of machmode.h.
17201 (emit_min_insn_modes_c_header): Include coretypes.h rather than
17203 (main): Handle -i flag and call emit_insn_modes_inline_h when
17206 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17208 * tree-ssa-strlen.c (strinfo): Rename the length field to
17209 nonzero_chars. Add a full_string_p field.
17210 (compare_nonzero_chars, zero_length_string_p): New functions.
17211 (get_addr_stridx): Add an offset_out parameter.
17212 Use compare_nonzero_chars.
17213 (get_stridx): Update accordingly. Use compare_nonzero_chars.
17214 (new_strinfo): Update after above changes to strinfo.
17215 (set_endptr_and_length): Set full_string_p.
17216 (get_string_length): Update after above changes to strinfo.
17217 (unshare_strinfo): Update call to new_strinfo.
17218 (maybe_invalidate): Likewise.
17219 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
17220 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
17221 as a uhwi instead of an shwi. Update after above changes to
17222 strinfo and new_strinfo.
17223 (zero_length_string): Assert that chainsi contains full strings.
17224 Use zero_length_string_p. Update call to new_strinfo.
17225 (adjust_related_strinfos): Update after above changes to strinfo.
17226 Copy full_string_p from origsi.
17227 (adjust_last_stmt): Use zero_length_string_p.
17228 (handle_builtin_strlen): Update after above changes to strinfo and
17229 new_strinfo. Install the lhs as the string length if the previous
17230 entry didn't describe a full string.
17231 (handle_builtin_strchr): Update after above changes to strinfo
17233 (handle_builtin_strcpy): Likewise.
17234 (handle_builtin_strcat): Likewise.
17235 (handle_builtin_malloc): Likewise.
17236 (handle_pointer_plus): Likewise.
17237 (handle_builtin_memcpy): Likewise. Track nonzero characters
17238 that aren't necessarily followed by a nul terminator.
17239 (handle_char_store): Likewise.
17241 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17243 PR tree-optimization/80769
17244 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
17245 for malloc and calloc. Document the new invariant that all related
17246 strinfos have delayed lengths or none do.
17247 (verify_related_strinfos): Move earlier in file.
17248 (set_endptr_and_length): New function, split out from...
17249 (get_string_length): ...here. Also set the lengths of related
17251 (zero_length_string): Assert that chainsi has known (rather than
17253 (adjust_related_strinfos): Likewise.
17255 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17257 PR tree-optimization/81136
17258 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
17259 assert that two references with the same misalignment have the same
17260 compile-time misalignment if those compile-time misalignments
17263 2017-07-01 Andi Kleen <ak@linux.intel.com>
17265 * print-tree.c (print_node): Print all attributes.
17267 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17269 * cfg.c (scale_bbs_frequencies): New function.
17270 * cfg.h (scale_bbs_frequencies): Declare it.
17271 * cfgloopanal.c (single_likely_exit): Cleanup.
17272 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
17274 (scale_loop_profile): Likewise.
17275 (loop_version): Likewise.
17276 (create_empty_loop_on_edge): Update.
17277 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
17278 scale_loop_frequencies, scale_loop_profile, loopify,
17279 loop_version): Update prototypes.
17280 * modulo-sched.c (sms_schedule): Update.
17281 * predict.c (unlikely_executed_edge_p): Also check probability.
17282 (probably_never_executed_edge_p): Fix typo.
17283 * tree-if-conv.c (version_loop_for_if_conversion): Update.
17284 * tree-parloops.c (gen_parallel_loop): Update.
17285 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
17286 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
17287 * tree-ssa-loop-split.c (split_loop): Update.
17288 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
17289 * tree-vect-loop-manip.c (vect_do_peeling): Update.
17290 (vect_loop_versioning): Update.
17291 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
17293 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17295 * trans-mem.c (split_bb_make_tm_edge): Update profile.
17297 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17299 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
17300 to keep profile consistent.
17302 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17304 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
17305 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
17306 * profile-count.h (max_safe_multiplier): Make unsigned.
17307 (profile_count::guessed_zero): New.
17309 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17311 * bb-reorder.c (fix_up_crossing_landing_pad,
17312 fix_crossing_conditional_branches): Use make_single_succ_edge
17313 to keep profile consistent.
17315 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17317 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
17320 2017-07-01 Jakub Jelinek <jakub@redhat.com>
17323 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
17324 the right scopes, make sure cond_jump isn't preserved between multiple
17325 iterations. Search for fallthru edge whenever there are 3+ edges and
17326 use find_fallthru_edge for it.
17328 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17330 Patch by Alexander Monakov <amonakov@ispras.ru>
17331 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
17332 probabilities consistently.
17334 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17336 * pa.c (pa_expand_compare_and_swap_loop): Update call of
17337 emit_cmp_and_jump_insns.
17339 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17342 * tree-inline.c (expand_call_inline): Combine profile statuses.
17344 2017-06-30 Andrew Pinski <apinski@cavium.com>
17346 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
17347 fold_stmt returned true.
17349 2017-06-30 Nathan Sidwell <nathan@acm.org>
17351 * ggc.h (empty_string): Delete.
17352 * cfgexpand.c (expand_asm_stmt): Use plain "".
17353 * optabs.c (expand_asm_memory_barrier): Likewise.
17354 * stringpool.c (empty_string): Delete.
17355 (digit_vector, digit_string): Delete.
17356 (ggc_alloc_string): Use plain "", don't optimize single digit
17357 strings. Use ggc_alloc_atomic.
17359 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
17361 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
17362 comparison set and one other set, use the cost of the non-comparison
17365 2017-06-30 Nathan Sidwell <nathan@acm.org>
17367 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
17370 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
17372 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
17373 loops. Remove now unneeded calls to gimple_switch_set_label() that
17374 just set removed labels to NULL_TREE.
17376 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
17378 * tree-ssanames.c (set_range_info_raw): Abstract from ...
17379 (set_range_info): ...here. Only call set_range_info_raw if domain
17381 (set_nonzero_bits): Call set_range_info_raw.
17382 * tree-ssanames.h (set_range_info_raw): New.
17384 2017-06-30 Jakub Jelinek <jakub@redhat.com>
17387 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
17388 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
17389 of nonimmediate_operand and <store_mask_constraint> instead of m
17390 for the input operand. For V8FI iterator, always split if input
17391 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
17392 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
17393 <store_mask_predicate> instead of register_operand and
17394 <store_mask_constraint> instead of v for the input operand. Make
17395 sure both operands aren't MEMs for if not <mask_applied>.
17397 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
17399 * lto-wrapper.c (copy_file) Close both file descriptors before
17402 2017-06-30 Martin Liska <mliska@suse.cz>
17405 * multiple_target.c (create_dispatcher_calls): Make ifunc
17406 also for function that don't have calls or are not referenced.
17408 2017-06-30 Richard Biener <rguenther@suse.de>
17410 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
17411 analyze the first scalar stmt. Move vector type computation
17412 for the BB case here from ...
17413 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
17414 live operation processing in the SLP case properly.
17416 2017-06-30 Richard Biener <rguenther@suse.de>
17418 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
17420 2017-06-30 Martin Liska <mliska@suse.cz>
17423 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
17424 before BUILT_IN_UNWIND_RESUME when ASAN is used.
17426 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
17428 * doc/invoke.texi (AArch64): Add missing options and remove redundant
17431 2017-06-30 Richard Biener <rguenther@suse.de>
17433 PR tree-optimization/81249
17434 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
17435 condition reduction result to original scalar type.
17437 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17439 * profile-count.h (enum profile_quality): Fix typos and whitespace
17442 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17444 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
17445 type for branch probabilities.
17447 2017-06-29 Julian Brown <julian@codesourcery.com>
17448 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17450 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
17451 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
17452 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
17453 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
17455 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17457 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
17458 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
17459 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
17460 CC usage from generic code to here.
17461 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
17462 CC usage into the target macros.
17464 2017-06-29 Maya Rashish <coypu@sdf.org>
17466 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
17469 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17471 * arm/arm-builtins.c: Include profile-count.h
17472 * except.c (sjlj_emit_function_enter): Use
17473 profile_probability::unlikely.
17475 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17477 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
17478 and tocrel_offset be pointer args rather than implicitly using
17480 (legitimate_constant_pool_address_p, rs6000_emit_move,
17481 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
17482 tocrel_offset and use in toc_relative_expr_p call.
17483 (print_operand, print_operand_address): Use static tocrel_base_oac
17484 and tocrel_offset_oac.
17485 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
17488 2017-06-29 Maya Rashish <coypu@sdf.org>
17490 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
17492 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
17494 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
17495 objects, take into account only the alignment of 'op0' and 'mode1' if
17498 2017-06-29 Steve Ellcey <sellcey@cavium.com>
17500 * ccmp.c (ccmp_tree_comparison_p): New function.
17501 (ccmp_candidate_p): Update to use above function.
17502 (get_compare_parts): New function.
17503 (expand_ccmp_next): Update to use new functions.
17504 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
17506 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
17507 take mode as argument.
17508 * ccmp.h (expand_ccmp_expr): Add mode as argument.
17509 * expr.c (expand_expr_real_1): Pass mode as argument.
17511 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
17513 * combine.c (combine_instructions): Print insns to dump_file, together
17516 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17518 * asan.c (asan_emit_stack_protection): Update.
17519 (create_cond_insert_point): Update.
17520 * auto-profile.c (afdo_propagate_circuit): Update.
17521 * basic-block.h (struct edge_def): Turn probability to
17522 profile_probability.
17523 (EDGE_FREQUENCY): Update.
17524 * bb-reorder.c (find_traces_1_round): Update.
17525 (better_edge_p): Update.
17526 (sanitize_hot_paths): Update.
17527 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
17528 (make_single_succ_edge): Update.
17529 (check_bb_profile): Update.
17530 (dump_edge_info): Update.
17531 (update_bb_profile_for_threading): Update.
17532 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
17533 probabilitycount to 0.
17534 * cfgbuild.c (compute_outgoing_frequencies): Update.
17535 * cfgcleanup.c (try_forward_edges): Update.
17536 (outgoing_edges_match): Update.
17537 (try_crossjump_to_edge): Update.
17538 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
17539 (expand_gimple_tailcall): Update.
17540 (construct_init_block): Use make_single_succ_edge.
17541 (construct_exit_block): Use make_single_succ_edge.
17542 * cfghooks.c (verify_flow_info): Update.
17543 (redirect_edge_succ_nodup): Update.
17544 (split_edge): Update.
17545 (account_profile_record): Update.
17546 * cfgloopanal.c (single_likely_exit): Update.
17547 * cfgloopmanip.c (scale_loop_profile): Update.
17548 (set_zero_probability): Remove.
17549 (duplicate_loop_to_header_edge): Update.
17550 * cfgloopmanip.h (loop_version): Update prototype.
17551 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
17552 (force_nonfallthru_and_redirect): Update.
17553 (update_br_prob_note): Update.
17554 (rtl_verify_edges): Update.
17555 (purge_dead_edges): Update.
17556 (rtl_lv_add_condition_to_bb): Update.
17557 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
17558 * cgraphunit.c (init_lowered_empty_function): Update.
17559 (cgraph_node::expand_thunk): Update.
17560 * cilk-common.c: Include profile-count.h
17561 * dojump.c (inv): Remove.
17562 (jumpifnot): Update.
17563 (jumpifnot_1): Update.
17564 (do_jump_1): Update.
17566 (do_jump_by_parts_greater_rtx): Update.
17567 (do_compare_rtx_and_jump): Update.
17568 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
17569 do_jump_1. do_compare_rtx_and_jump): Update prototype.
17570 * dwarf2cfi.c: Include profile-count.h
17571 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
17572 (sjlj_emit_dispatch_table): Likewise.
17573 * explow.c: Include profile-count.h
17574 * expmed.c (emit_store_flag_force): Update.
17575 (do_cmp_and_jump): Update.
17576 * expr.c (compare_by_pieces_d::generate): Update.
17577 (compare_by_pieces_d::finish_mode): Update.
17578 (emit_block_move_via_loop): Update.
17579 (store_expr_with_bounds): Update.
17580 (store_constructor): Update.
17581 (expand_expr_real_2): Update.
17582 (expand_expr_real_1): Update.
17583 * expr.h (try_casesi, try_tablejump): Update prototypes.
17584 * gimple-pretty-print.c (dump_probability): Update.
17585 (dump_profile): New.
17586 (dump_gimple_label): Update.
17587 (dump_gimple_bb_header): Update.
17588 * graph.c (draw_cfg_node_succ_edges): Update.
17589 * hsa-gen.c (convert_switch_statements): Update.
17590 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
17591 (find_if_case_1): Update.
17592 (find_if_case_2): Update.
17593 * internal-fn.c (expand_arith_overflow_result_store): Update.
17594 (expand_addsub_overflow): Update.
17595 (expand_neg_overflow): Update.
17596 (expand_mul_overflow): Update.
17597 (expand_vector_ubsan_overflow): Update.
17598 * ipa-cp.c (good_cloning_opportunity_p): Update.
17599 * ipa-split.c (split_function): Use make_single_succ_edge.
17600 * ipa-utils.c (ipa_merge_profiles): Update.
17601 * loop-doloop.c (add_test): Update.
17602 (doloop_modify): Update.
17603 * loop-unroll.c (compare_and_jump_seq): Update.
17604 (unroll_loop_runtime_iterations): Update.
17605 * lra-constraints.c (lra_inheritance): Update.
17606 * lto-streamer-in.c (input_cfg): Update.
17607 * lto-streamer-out.c (output_cfg): Update.
17608 * mcf.c (adjust_cfg_counts): Update.
17609 * modulo-sched.c (sms_schedule): Update.
17610 * omp-expand.c (expand_omp_for_init_counts): Update.
17611 (extract_omp_for_update_vars): Update.
17612 (expand_omp_ordered_sink): Update.
17613 (expand_omp_for_ordered_loops): Update.
17614 (expand_omp_for_generic): Update.
17615 (expand_omp_for_static_nochunk): Update.
17616 (expand_omp_for_static_chunk): Update.
17617 (expand_cilk_for): Update.
17618 (expand_omp_simd): Update.
17619 (expand_omp_taskloop_for_outer): Update.
17620 (expand_omp_taskloop_for_inner): Update.
17621 * omp-simd-clone.c (simd_clone_adjust): Update.
17622 * optabs.c (expand_doubleword_shift): Update.
17623 (expand_abs): Update.
17624 (emit_cmp_and_jump_insn_1): Update.
17625 (expand_compare_and_swap_loop): Update.
17626 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
17627 * predict.c (predictable_edge_p): Update.
17628 (edge_probability_reliable_p): Update.
17629 (set_even_probabilities): Update.
17630 (combine_predictions_for_insn): Update.
17631 (combine_predictions_for_bb): Update.
17632 (propagate_freq): Update.
17633 (estimate_bb_frequencies): Update.
17634 (force_edge_cold): Update.
17635 * profile-count.c (profile_count::dump): Add missing space into dump.
17636 (profile_count::debug): Add newline.
17637 (profile_count::differs_from_p): Explicitly convert to unsigned.
17638 (profile_count::stream_in): Update.
17639 (profile_probability::dump): New member function.
17640 (profile_probability::debug): New member function.
17641 (profile_probability::differs_from_p): New member function.
17642 (profile_probability::differs_lot_from_p): New member function.
17643 (profile_probability::stream_in): New member function.
17644 (profile_probability::stream_out): New member function.
17645 * profile-count.h (profile_count_quality): Rename to ...
17646 (profile_quality): ... this one.
17647 (profile_probability): New.
17648 (profile_count): Update.
17649 * profile.c (compute_branch_probabilities): Update.
17650 * recog.c (peep2_attempt): Update.
17651 * sched-ebb.c (schedule_ebbs): Update.
17652 * sched-rgn.c (find_single_block_region): Update.
17653 (compute_dom_prob_ps): Update.
17654 (schedule_region): Update.
17655 * sel-sched-ir.c (compute_succs_info): Update.
17656 * stmt.c (struct case_node): Update.
17657 (do_jump_if_equal): Update.
17658 (get_outgoing_edge_probs): Update.
17659 (conditional_probability): Update.
17660 (emit_case_dispatch_table): Update.
17661 (expand_case): Update.
17662 (expand_sjlj_dispatch_table): Update.
17663 (emit_case_nodes): Update.
17664 * targhooks.c: Update.
17665 * tracer.c (better_p): Update.
17666 (find_best_successor): Update.
17667 * trans-mem.c (expand_transaction): Update.
17668 * tree-call-cdce.c: Update.
17669 * tree-cfg.c (gimple_split_edge): Upate.
17670 (move_sese_region_to_fn): Upate.
17671 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
17672 * tree-eh.c (lower_resx): Upate.
17673 (cleanup_empty_eh_move_lp): Upate.
17674 * tree-if-conv.c (version_loop_for_if_conversion): Update.
17675 * tree-inline.c (copy_edges_for_bb): Update.
17676 (copy_cfg_body): Update.
17677 * tree-parloops.c (gen_parallel_loop): Update.
17678 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
17679 (gimple_gen_time_profiler): Update.
17680 * tree-ssa-dce.c (remove_dead_stmt): Update.
17681 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
17682 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
17683 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
17684 (unloop_loops): Update.
17685 (try_peel_loop): Update.
17686 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
17687 * tree-ssa-loop-split.c (connect_loops): Update.
17688 (split_loop): Update.
17689 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
17690 (hoist_guard): Update.
17691 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
17692 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
17693 (value_replacement): Update.
17694 * tree-ssa-reassoc.c (branch_fixup): Update.
17695 * tree-ssa-tail-merge.c (replace_block_by): Update.
17696 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
17697 (create_edge_and_update_destination_phis): Update.
17698 (compute_path_counts): Update.
17699 (recompute_probabilities): Update.
17700 (update_joiner_offpath_counts): Update.
17701 (freqs_to_counts_path): Update.
17702 (duplicate_thread_path): Update.
17703 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
17704 (struct switch_conv_info): Update.
17705 (gen_inbound_check): Update.
17706 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
17707 (vect_do_peeling): Update.
17708 (vect_loop_versioning): Update.
17709 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
17710 (optimize_mask_stores): Update.
17711 * ubsan.c (ubsan_expand_null_ifn): Update.
17712 * value-prof.c (gimple_divmod_fixed_value): Update.
17713 (gimple_divmod_fixed_value_transform): Update.
17714 (gimple_mod_pow2): Update.
17715 (gimple_mod_pow2_value_transform): Update.
17716 (gimple_mod_subtract): Update.
17717 (gimple_mod_subtract_transform): Update.
17718 (gimple_ic): Update.
17719 (gimple_stringop_fixed_value): Update.
17720 (gimple_stringops_transform): Update.
17721 * value-prof.h: Update.
17723 2017-06-29 Carl Love <cel@us.ibm.com>
17725 * config/rs6000/rs6000-c.c: Add support for built-in functions
17726 vector signed int vec_signed (vector float);
17727 vector signed long long vec_signed (vector double);
17728 vector signed int vec_signed2 (vector double, vector double);
17729 vector signed int vec_signede (vector double);
17730 vector signed int vec_signedo (vector double);
17731 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
17732 instruction generator.
17733 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
17734 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
17735 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
17737 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
17738 vunsignede_v2df): Add define_expands.
17739 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
17740 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
17741 VEC_UNSIGNEDO): Add definitions.
17742 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
17743 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
17744 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
17745 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
17746 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
17747 * config/rs6000/altivec.h (vec_signed, vec_signed2,
17748 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
17749 vec_unsignede, vec_unsignedo): Add builtin defines.
17750 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
17752 * doc/extend.texi: Update the built-in documentation file for the
17753 new built-in functions.
17755 2017-06-29 Richard Biener <rguenther@suse.de>
17757 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
17758 reduction chains to LOOP_VINFO_REDUCTIONS.
17759 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
17760 SLP reductions after processing reduction chains.
17762 2017-06-29 Nathan Sidwell <nathan@acm.org>
17764 * builtins.c (fold_builtin_FUNCTION): Use
17765 lang_hooks.decl_printable_name.
17767 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
17769 PR middle-end/81194
17770 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
17771 with only one label.
17772 * stmt.c (expand_case): Assert NCASES is greater than one.
17774 2017-06-29 Richard Biener <rguenther@suse.de>
17776 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
17778 (group_case_labels): Likewise.
17779 (find_taken_edge): Push sanity checking on val to workers...
17780 (find_taken_edge_cond_expr): ... here
17781 (find_taken_edge_switch_expr): ... and here, handle cases
17782 with just a default label.
17783 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
17784 (group_case_labels): Likewise.
17785 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
17786 group_case_labels does anything cleanup the CFG again.
17788 2017-06-29 Bin Cheng <bin.cheng@arm.com>
17790 PR tree-optimization/81196
17791 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
17792 exit condition comparing two IVs.
17794 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
17796 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
17797 profile to the dummy entry at the end of the list of architectures.
17798 * config/arm/arm-cpu-cdata.h: Regenerated.
17800 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17801 Michael Collison <michael.collison@arm.com>
17804 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
17806 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
17807 (*aarch64_reg_<mode>3_minus_mask): New pattern.
17808 (*aarch64_<optab>_reg_di3_mask2): New pattern.
17809 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
17810 of shift when the shift amount is masked with constant equal to
17811 the size of the mode.
17812 * config/aarch64/predicates.md (subreg_lowpart_operator): New
17815 2017-06-29 Martin Liska <mliska@suse.cz>
17817 * config/i386/i386.opt: Change range from [1,5] to [0,5].
17819 2017-06-29 Yury Gribov <tetra2005@gmail.com>
17822 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
17825 (edge_growth_cache_entry::edge_growth_cache_entry): New
17827 (reset_edge_growth_cache): Update to use constructor.
17829 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17831 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
17832 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
17833 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
17835 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
17837 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
17838 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
17840 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
17842 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
17843 (*-linux-uclibc*): Add t-uclibc tmake_file.
17844 * config/t-musl: New.
17845 * config/t-uclibc: New.
17847 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
17849 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
17851 (gen_comm_data): Emit architectural setting of arch_prof.
17852 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
17854 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
17855 (armv8-m.base, armv8-m.main): Likewise.
17856 * arm-protos.h (arm_build_target): Add profile field.
17857 (arch_option): Likewise.
17858 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
17860 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
17861 arm_active_target.profile.
17863 2017-06-28 Richard Biener <rguenther@suse.de>
17865 PR middle-end/81227
17866 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
17867 TYPE_OVERFLOW_WRAPS.
17868 * match.pd (negate_expr_p): Likewise.
17869 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
17870 fold_build2, not fold_binary.
17872 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
17874 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
17875 Convert memory address to Pmode.
17876 (aarch64_print_operand): Assert MEM operands are always Pmode.
17878 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
17881 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
17882 Remove redundant if.
17883 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
17884 * config/arm/aarch-common-protos.h
17885 (aarch_forward_to_shift_is_not_shifted_re): Remove.
17886 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
17888 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
17891 * multiple_target.c (create_dispatcher_calls): Set the default
17892 clone to be static, not public.
17894 2017-06-28 Richard Biener <rguenther@suse.de>
17896 * tree-vect-loop.c (vectorizable_reduction): Move special
17897 cond reduction IV var creation ...
17898 (vect_create_epilog_for_reduction): ... here. Remove induction_index
17899 parameter. Use STMT_VINFO_VECTYPE.
17900 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
17903 2017-06-28 Martin Liska <mliska@suse.cz>
17906 * ipa-visibility.c (non_local_p): Handle visibility.
17908 2017-06-28 Martin Liska <mliska@suse.cz>
17911 * common.opt: Add IntegerRange to various options.
17912 * opt-functions.awk (integer_range_info): New function.
17913 * optc-gen.awk: Add integer_range_info to cl_options struct.
17914 * opts-common.c (decode_cmdline_option): Handle
17915 CL_ERR_INT_RANGE_ARG.
17916 (cmdline_handle_error): Likewise.
17917 * opts.c (print_filtered_help): Show valid interval in
17918 when --help is provided.
17919 * opts.h (struct cl_option): Add range_min and range_max fields.
17920 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
17922 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
17924 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
17925 (x * C EQ/NE y * C): New transformation.
17927 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
17929 * genmultilib (combination_space): Accept '+' in option names.
17931 2017-06-28 Martin Liska <mliska@suse.cz>
17934 * asan.c (instrument_derefs): Bail out inner references
17935 that are hard register variables.
17937 2017-06-28 Jakub Jelinek <jakub@redhat.com>
17940 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
17941 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
17943 2017-06-28 Richard Biener <rguenther@suse.de>
17945 * tree-vectorizer.h (vect_get_vec_defs): Remove.
17946 (vect_get_slp_defs): Adjust.
17947 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
17949 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
17951 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
17952 get_initial_defs_for_reduction instead of vect_get_vec_defs.
17953 (vectorizable_reduction): Adjust.
17954 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
17956 (vect_get_slp_defs): Likewise.
17957 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
17958 (vectorizable_bswap): Adjust.
17959 (vectorizable_call): Likewise.
17960 (vectorizable_conversion): Likewise.
17961 (vectorizable_assignment): Likewise.
17962 (vectorizable_shift): Likewise.
17963 (vectorizable_operation): Likewise.
17964 (vectorizable_store): Likewise.
17965 (vectorizable_condition): Likewise.
17966 (vectorizable_comparison): Likewise.
17968 2017-06-28 Michael Collison <michael.collison@arm.com>
17971 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
17973 (arm_gen_movmemqi): Removed unused variable 'i'.
17974 Convert 'for' loop into 'while' loop.
17975 (arm_expand_prologue): Remove last unnecessary set of insn.
17976 (thumb_pop): Remove unused variable 'pushed_words'.
17977 (thumb_exit): Remove last unnecessary set of regs_to_pop.
17979 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17981 * config/s390/predicates.md: Use s390_rel_address_ok_p.
17982 * config/s390/s390-protos.h: Add prototype of
17983 s390_rel_address_ok_p.
17984 * config/s390/s390.c (s390_got_symbol): New function.
17985 (s390_rel_address_ok_p): New function.
17986 (legitimize_pic_address): Use s390_rel_address_ok_p.
17987 (s390_load_got): Use s390_got_symbol.
17988 (s390_option_override): Issue error if
17989 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
17990 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
17992 * config/s390/s390.opt: New option mpic-data-is-text-relative.
17994 2017-06-27 Andrew Pinski <apinski@cavium.com>
17996 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
17997 (X * copysign (1.0, X)): New pattern.
17998 (X * copysign (1.0, -X)): New pattern.
17999 (copysign (-1.0, CST)): New pattern.
18001 2017-06-27 Joseph Myers <joseph@codesourcery.com>
18003 * genmultilib (combination_space): Remove variable.
18004 Validate reuse rules against regular expression for any sequence
18005 of multilib options in any order.
18007 2017-06-27 Michael Collison <michael.collison@arm.com>
18009 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
18010 call aarch64_split_simd_combine.
18011 * (aarch64_combine_internal<mode>): Delete pattern.
18012 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
18013 Allow register and subreg operands.
18015 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18017 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
18018 specific need, just fallback on defaults.
18019 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
18021 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18022 Olivier Hainque <hainque@adacore.com>
18024 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
18026 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
18027 targets. Pick a default if no particular attempt applied.
18028 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
18031 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18033 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
18034 (x86_64-wrs-vxworks7): Likewise.
18036 2017-06-27 Marek Polacek <polacek@redhat.com>
18039 * ubsan.c (instrument_null): Check get_base_address's result for null.
18041 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18043 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
18045 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18047 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
18048 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
18049 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
18050 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
18051 New function types.
18052 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
18053 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
18054 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
18055 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
18056 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
18057 BUILT_IN_FEUPDATEENV): New builtins.
18058 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
18059 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
18060 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
18061 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
18063 (builtin_structptr_types): Adjust size.
18064 * tree.c (builtin_structptr_types): Add four entries.
18066 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18067 Olivier Hainque <hainque@adacore.com>
18069 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
18070 (TLS_SYM): New local macro, forcing reference to __tls__ on
18071 link command lines for VxWorks 7 RTPs, triggering initialization
18073 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
18074 OS features TLS support, true for RTPs on VxWorks 7.
18075 * config/vxworks.c (vxworks_override_options): Setup emutls
18078 2017-06-27 Jakub Jelinek <jakub@redhat.com>
18080 * predict.c (test_prediction_value_range): Use -1U instead of -1
18081 to avoid narrowing conversion warning.
18082 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
18083 to avoid narrowing conversion warning.
18084 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
18086 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
18088 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18090 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
18091 64bit configurations.
18092 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
18093 (SIZE_TYPE): Likewise.
18094 * config/vxworks.c (vxworks_emutls_var_fields): Use
18095 long_unsigned_type_node instead of unsigned_type_node as the offset
18096 field type, which is "pointer" mode in emutls.c.
18098 2017-06-27 Jakub Jelinek <jakub@redhat.com>
18101 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
18103 PR middle-end/81207
18104 * gimple-fold.c (replace_call_with_call_and_fold): Handle
18105 gimple_vuse copying separately from gimple_vdef copying.
18107 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18109 * value-prof.c (free_hist): Remove call to memset and the enclosing if
18112 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18113 Olivier Hainque <hainque@adacore.com>
18115 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
18116 for all vxworks7 targets.
18117 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
18118 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
18119 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
18120 variations for VX6/VX7 and 32/64bits later on in ...
18121 (VXWORKS_LIB_SPEC): Leverage new macros.
18122 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
18123 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
18125 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18127 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
18128 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
18130 2017-06-26 Carl Love <cel@us.ibm.com>
18132 * config/rs6000/rs6000-c.c: Add support for built-in functions
18133 vector bool char vec_reve (vector bool char);
18134 vector signed char vec_reve (vector signed char);
18135 vector unsigned char vec_reve (vector unsigned char);
18136 vector bool int vec_reve (vector bool int);
18137 vector signed int vec_reve (vector signed int);
18138 vector unsigned int vec_reve (vector unsigned int);
18139 vector bool long long vec_reve (vector bool long long);
18140 vector signed long long vec_reve (vector signed long long);
18141 vector unsigned long long vec_reve (vector unsigned long long);
18142 vector bool short vec_reve (vector bool short);
18143 vector signed short vec_reve (vector signed short);
18144 vector double vec_reve (vector double);
18145 vector float vec_reve (vector float);
18146 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
18147 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
18148 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
18149 (altivec_vreve): New pattern.
18150 * config/rs6000/altivec.h (vec_reve): New define.
18151 * doc/extend.texi (vec_rev): Update the built-in documentation file
18152 for the new built-in functions.
18154 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18156 PR tree-optimization/71815
18157 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
18159 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
18161 (slsr_process_phi): Likewise.
18162 (replace_uncond_cands_and_profitable_phis): Don't replace a
18163 multiply candidate with a stride of 1 (copy or cast).
18164 (phi_incr_cost): Call uses_consumed_by_stmt rather than
18166 (lowest_cost_path): Likewise.
18167 (total_savings): Likewise.
18169 2017-06-26 Richard Biener <rguenther@suse.de>
18172 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
18173 Use def_builtin_pure for all gather builtins.
18175 2017-06-26 Richard Biener <rguenther@suse.de>
18177 PR tree-optimization/81203
18178 * tree-tailcall.c (find_tail_calls): Do not move stmts into
18179 non-dominating BBs.
18181 2017-06-26 Marek Polacek <polacek@redhat.com>
18184 * doc/invoke.texi: Document -Wmultistatement-macros.
18186 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
18188 * doc/sourcebuild.texi (ARM-specific attributes): Document new
18189 arm_neon_ok_no_float_abi effective target.
18191 2017-06-26 Richard Biener <rguenther@suse.de>
18193 PR tree-optimization/80928
18194 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
18195 (copy_bbs): Set BB_DUPLICATED flag early.
18196 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
18198 (execute_on_shrinking_pred): Likewise.
18199 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
18200 BB_DUPLICATED blocks.
18201 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
18202 iterate over all PHIs considering removal of *gsi.
18204 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
18206 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
18209 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18211 * config/rs6000/rs6000-string.c: (expand_block_clear,
18212 do_load_for_compare, select_block_compare_mode,
18213 compute_current_alignment, expand_block_compare,
18214 expand_strncmp_align_check, expand_strn_compare,
18215 expand_block_move, rs6000_output_load_multiple)
18216 Move functions related to string/block move/compare
18217 to a separate file.
18218 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
18219 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
18220 for this function which is now used in two files.
18221 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
18222 * config.gcc: Add rs6000-string.o to extra_objs for
18223 targets powerpc*-*-* and rs6000*-*-*.
18225 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
18228 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
18229 32-bit, since indexed is not valid for DImode.
18230 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
18231 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
18232 (define_peephole2 for Altivec d-form load): Add 32-bit support.
18233 (define_peephole2 for Altivec d-form store): Likewise.
18236 * multiple_target.c (create_dispatcher_calls): Only create the
18237 dispatcher call if the function is the default clone of a
18238 versioned function.
18240 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
18242 PR middle-end/80902
18243 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
18244 a call, force the call to not be a tail call.
18246 2017-06-23 Jeff Law <law@redhat.com>
18248 * doc/contrib.texi: Add entry for Steven Pemberton's work on
18251 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
18253 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
18254 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
18255 handling for early expansion of vector shifts (sl,sr,sra,rl).
18256 (builtin_function_type): Add vector shift right instructions
18257 to the unsigned argument list.
18259 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
18261 rtl-optimizatoin/79286
18262 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
18263 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
18264 trap. PIC register plus a const unspec without offset can never trap.
18266 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
18268 * tree.h (builtin_structptr_type): New type.
18269 (builtin_structptr_types): Declare new array.
18270 * tree.c (builtin_structptr_types): New array.
18271 (free_lang_data, build_common_tree_nodes): Use it.
18273 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
18276 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
18279 2017-06-22 Matt Turner <mattst88@gmail.com>
18281 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
18282 Lake models to skylake case. Assume skylake for unknown
18283 models with clflushopt.
18285 2017-06-22 Jeff Law <law@redhat.com>
18287 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
18288 frame sizes that do not satisfy aarch64_uimm12_shift.
18290 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
18292 * profile-count.h (apply_probability,
18293 apply_scale, probability_in): Fix checks for zero.
18295 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18297 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
18298 * doc/cppdiropts.texi (-I @var{dir}): Document it.
18300 2016-06-22 Richard Biener <rguenther@suse.de>
18302 * tree-vect-loop.c (vect_model_reduction_cost): Handle
18303 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
18304 REDUC_MAX_EXPR support.
18305 (vectorizable_reduction): Likewise.
18306 (vect_create_epilog_for_reduction): Likewise.
18308 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
18310 * match.pd (A / (1 << B) -> A >> B): New.
18311 * generic-match-head.c: Include optabs-tree.h.
18312 * gimple-match-head.c: Likewise.
18313 * optabs-tree.h (target_supports_op_p): New.
18314 * optabs-tree.c (target_supports_op_p): New.
18316 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18318 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
18319 $gcc_cv_ld --help output.
18320 (gcc_cv_ld_demangle): Likewise.
18321 (gcc_cv_ld_eh_frame_hdr): Likewise.
18322 (gcc_cv_ld_pie): Likewise.
18323 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
18324 (gcc_cv_ld_buildid): Likewise.
18325 (gcc_cv_ld_sysroot): Likewise.
18326 (ld_bndplt_support): Likewise.
18327 (ld_pushpopstate_support): Likewise.
18328 * configure: Regenerate.
18329 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
18331 2017-06-21 Jakub Jelinek <jakub@redhat.com>
18334 * config/i386/sse.md (round<mode>2): Renumber match_dup and
18335 operands indexes to avoid gap between operands and match_dups.
18337 2017-06-21 Andrew Pinski <apinski@cavium.com>
18339 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
18340 Increment Arith_shift and Arith_shift_reg by 1.
18341 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
18343 * config/aarch64/aarch64.c (thunderx_tunings): Enable
18344 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
18345 (aarch64_strip_extend): Add new argument and test for it.
18346 (aarch64_cheap_mult_shift_p): New function.
18347 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
18348 add a cost if it is true.
18349 Update calls to aarch64_strip_extend.
18350 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
18352 2017-06-21 Andrew Pinski <apinski@cavium.com>
18354 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
18356 (thunderxt88): Likewise.
18357 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
18358 (thunderx_prefetch_tune): New variable.
18359 (thunderx2t99_prefetch_tune): Update for the correct values.
18360 (thunderxt88_tunings): New variable.
18361 (thunderx_tunings): Use thunderx_prefetch_tune instead of
18362 generic_prefetch_tune.
18363 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
18365 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18367 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
18368 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
18369 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
18370 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
18371 (aarch64_atomic_cas<mode>, GPI): Likewise.
18373 2017-06-21 Martin Liska <mliska@suse.cz>
18375 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
18376 statements on cold and hot labels.
18377 * predict.c (tree_estimate_probability_bb): Remove the
18378 prediction from this place.
18380 2017-06-21 Martin Liska <mliska@suse.cz>
18382 PR tree-optimization/79489
18383 * gimplify.c (maybe_add_early_return_predict_stmt): New
18385 (gimplify_return_expr): Call the function.
18386 * predict.c (tree_estimate_probability_bb): Remove handling
18388 * predict.def: Update comment about early return predictor.
18389 * gimple-predict.h (is_gimple_predict): New function.
18390 * predict.def: Change default value of early return to 66.
18391 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
18393 * passes.def: Put pass_strip_predict_hints to the beginning of
18396 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
18398 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
18399 FUNCTION_DECL declarations.
18400 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
18402 (dwaf2out_decl): Likewise.
18403 * godump.c (go_early_global_decl): Skip call to the real debug hook
18404 for FUNCTION_DECL declarations.
18405 * passes.c (rest_of_decl_compilation): Skip call to the
18406 early_global_decl debug hook for FUNCTION_DECL declarations, unless
18407 -fdump-go-spec is passed.
18409 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
18411 * config/i386/i386.c (struct builtin_isa): New field pure_p.
18412 Reorder for compactness.
18413 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
18414 (def_builtin_pure, def_builtin_pure2): New functions.
18415 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
18417 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
18419 * match.pd (nop_convert): New predicate.
18420 ((A +- CST1) +- CST2): Allow some NOP conversions.
18422 2017-06-21 Jakub Jelinek <jakub@redhat.com>
18425 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
18426 with ctors/dtors if GOVD_SHARED is set.
18428 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
18430 * config/aarch64/aarch64.md (movti_aarch64):
18431 Emit mov rather than orr.
18432 (movtf_aarch64): Likewise.
18433 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
18434 Emit mov rather than orr.
18436 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
18438 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
18439 Swap alternatives, make integer dup more expensive.
18441 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
18443 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
18444 Return true for non-tls symbols.
18446 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
18448 * config/aarch64/aarch64-cores.def (cortex-a55): New.
18449 (cortex-a75): Likewise.
18450 (cortex-a75.cortex-a55): Likewise.
18451 * config/aarch64/aarch64-tune.md: Regenerate.
18452 * doc/invoke.texi (-mtune): Document new values for -mtune.
18454 2017-06-21 Tom de Vries <tom@codesourcery.com>
18456 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
18457 stack_size feature.
18458 (Effective-Target Keywords, Other attributes): Suggest using
18459 dg-add-options stack_size feature to get stack limit in stack_size
18460 effective target documentation.
18462 2017-06-21 Julian Brown <julian@codesourcery.com>
18463 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18465 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
18466 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
18467 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
18469 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
18470 attribute type list for neon_multiply.
18471 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
18472 attribute type list for neon_multiply.
18473 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
18474 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
18475 attribute type list for neon_multiply.
18476 * config/arm/types.md (crypto_pmull): Add.
18477 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
18478 attribute type list.
18480 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
18482 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
18485 2017-06-20 Jakub Jelinek <jakub@redhat.com>
18487 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
18488 to make sure not to dereference a NULL cost_classes_ptr pointer.
18490 2017-06-20 Carl Love <cel@us.ibm.com>
18492 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18493 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
18494 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
18495 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
18496 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
18497 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
18498 VMULOSW): New enum "unspec" values.
18499 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
18500 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
18501 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
18502 altivec_vmulosw): New patterns.
18503 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
18504 VMULOSW): Add definitions.
18506 2017-06-20 Julia Koval <julia.koval@intel.com>
18508 * config/i386/i386.c: Fix rounding expand for new pattern.
18509 * config/i386/subst.md: Fix pattern (parallel -> unspec).
18511 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
18513 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
18514 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
18516 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
18518 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
18521 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
18523 * config/aarch64/aarch64-cores.def: Rearrange to sort by
18524 architecture, then by implementer ID.
18525 * config/aarch64/aarch64-tune.md: Regenerate.
18527 2017-06-20 Richard Biener <rguenther@suse.de>
18529 PR middle-end/81097
18530 * fold-const.c (split_tree): Fold to type before negating.
18532 2017-06-20 David Malcolm <dmalcolm@redhat.com>
18534 * diagnostic-show-locus.c
18535 (selftest::test_fixit_deletion_affecting_newline): New function.
18536 (selftest::diagnostic_show_locus_c_tests): Call it.
18538 2017-06-20 Andreas Schwab <schwab@suse.de>
18541 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
18544 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18546 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
18547 __ARM_FEATURE_COPROC according to support.
18549 2017-06-20 Jakub Jelinek <jakub@redhat.com>
18551 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
18552 Rewritten to avoid overflow for > 32-bit pointers.
18555 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
18556 by removing enum keyword.
18557 (ubsan_type_descriptor): Likewise. Formatting fix.
18560 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
18561 splitter): Require TARGET_SSE2 in the condition.
18563 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
18566 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
18567 for doing vector set of SFmode on ISA 3.0.
18568 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
18569 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
18571 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
18572 SFmode value into a V4SF variable that was extracted from another
18573 V4SF variable without converting the element to double precision
18574 and back to single precision vector format.
18575 (vsx_insert_extract_v4sf_p9_2): Likewise.
18577 2017-06-19 Jakub Jelinek <jakub@redhat.com>
18579 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
18580 in UWHI to avoid undefined overflow.
18583 * ubsan.h (enum ubsan_encode_value_phase): New.
18584 (ubsan_encode_value): Change second argument to
18585 enum ubsan_encode_value_phase with default value of
18586 UBSAN_ENCODE_VALUE_GENERIC.
18587 * ubsan.c (ubsan_encode_value): Change second argument to
18588 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
18589 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
18590 create_tmp_var_raw instead of create_tmp_var and use a
18592 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
18593 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
18594 ubsan_encode_value callers.
18597 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
18598 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
18599 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
18601 2017-06-19 Richard Biener <rguenther@suse.de>
18603 PR middle-end/81118
18604 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
18605 estimates if we changed anything.
18607 2017-06-19 Richard Biener <rguenther@suse.de>
18609 PR tree-optimization/80887
18610 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
18611 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
18612 simplified lookups, then reset mprts_hook.
18613 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
18615 (try_to_simplify): Likewise.
18617 2017-06-19 Martin Liska <mliska@suse.cz>
18620 * gimplify.c (gimplify_switch_expr):
18621 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
18623 2017-06-19 Martin Liska <mliska@suse.cz>
18625 * doc/install.texi: Document that PGO runs in 4 stages.
18627 2017-06-19 Martin Liska <mliska@suse.cz>
18630 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
18631 to dispatcher function name.
18632 * multiple_target.c (replace_function_decl): New function.
18633 (create_dispatcher_calls): Redirect both edges and references.
18635 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
18637 * profile-count.c (profile_count::dump): Dump quality.
18638 (profile_count::differs_from_p): Update for unsigned val.
18639 * profile-count.h (profile_count_quality): New enum.
18640 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
18642 2017-06-19 Richard Biener <rguenther@suse.de>
18644 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
18645 struct function as arg.
18646 (estimate_numbers_of_iterations): Export overload with loop arg.
18647 (free_numbers_of_iterations_estimates_loop): Use an overload of
18648 free_numbers_of_iterations_estimates instead.
18649 * tree-cfg.c (remove_bb): Adjust.
18650 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
18651 * tree-parloops.c (gen_parallel_loop): Likewise.
18652 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
18654 (tree_unroll_loops_completely): Likewise.
18655 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
18656 Use an overload instead and export.
18657 (estimated_loop_iterations): Adjust.
18658 (max_loop_iterations): Likewise.
18659 (likely_max_loop_iterations): Likewise.
18660 (estimate_numbers_of_iterations): Take struct function as arg
18662 (loop_exits_before_overflow): Adjust.
18663 (free_numbers_of_iterations_estimates_loop): Use an overload.
18664 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
18665 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
18667 2017-06-19 Richard Biener <rguenther@suse.de>
18670 * ipa-prop.c (find_constructor_constant_at_offset): Handle
18671 RANGE_EXPR conservatively.
18673 2017-06-16 Carl Love <cel@us.ibm.com>
18675 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18676 definitions for vec_float, vec_float2, vec_floato,
18677 vec_floate built-ins.
18678 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
18679 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
18681 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
18682 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
18683 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
18684 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
18685 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
18686 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
18687 vec_floato): Add builtin defines.
18688 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
18689 Update the built-in documentation file for the new built-in
18692 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18694 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
18695 (mthumb): Mark as the negative of -marm.
18697 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18699 * doc/invoke.texi (ARM Options, -mcpu): Document supported
18701 (ARM Options, -mtune): Document that this accepts the same
18702 extension options as -mcpu.
18703 (ARM Options, -mfpu): Document addition of -mfpu=auto.
18705 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18707 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
18708 permitted extensions.
18710 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18712 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
18713 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
18714 (armv8-m.main): Add option +nodsp.
18715 * config/arm/arm-cpu-cdata.h: Regenerated.
18717 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18719 * config/arm/t-fuchsia: New file.
18720 * config.gcc (arm*-*-fuchsia*): Use it.
18722 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18724 * config/arm/t-symbian: Rewrite for new option infrastructure.
18726 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18728 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
18729 (MULTILIB_REQUIRED): Likewise.
18731 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18733 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
18734 (MULTILIB_RESUE): Likewise.
18735 (MULTILIB_MATCHES): Likewise.
18736 (MULTLIB_REQUIRED): Likewise.
18738 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18740 * config/arm/t-rtems: Rewrite for new option framework.
18742 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18744 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
18745 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
18746 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
18747 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
18748 * config/arm/t-multilib: ... here.
18749 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
18750 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
18751 armv7-a and armv8*-a when A-profile libraries have not been built.
18752 * config/arm/t-rmprofile: Rewrite.
18754 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18756 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
18757 with a backslash. Remove the backslash after substituting unescaped
18759 * doc/fragments.texi (MULTILIB_REUSE): Document it.
18761 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18763 * config.gcc: (arm*-*-*): When building a-profile libraries, force
18764 the driver to pass through the default setting of -mfloat-abi.
18765 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
18767 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
18768 (all_feat_combs): New rule.
18769 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
18771 * config/arm/t-aprofile: Rewrite.
18773 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18775 * config/arm/arm.h (FPUTYPE_AUTO): Define.
18776 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
18777 fpu is not specified by the user/command-line.
18778 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
18779 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
18780 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
18781 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
18782 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
18783 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
18785 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18787 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
18788 * config/arm/t-arm-elf: Rewritten.
18790 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18792 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
18793 have some floating-point instructions.
18794 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
18795 (TARGET_MAYBE_HARD_FLOAT): New macro.
18796 * config/arm/arm-builtins.c (arm_init_builtins): Use
18797 TARGET_MAYBE_HARD_FLOAT.
18798 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
18800 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18802 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
18803 (configargs.h): Include it.
18804 (arm_print_hint_for_fpu_option): New function.
18805 (arm_parse_fpu_option): New function.
18806 (candidate_extension): New class.
18807 (arm_canon_for_multilib): New function.
18808 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
18809 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
18810 (ARCH_CANONICAL_SPECS): New macro.
18811 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
18813 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18815 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
18816 are set after handling multilib fragments. Set target_cpu_default2
18819 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18821 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
18823 (arm*-*-*): Set target_cpu_default2 to a quoted string.
18824 * config/arm/parsecpu.awk (check_cpu): Validate any extension
18826 (check_arch): Likewise.
18827 * config/arm/arm.c (arm_configure_build_target): Handle
18828 TARGET_CPU_DEFAULT being a string constant. Scan any feature
18829 options in the default.
18831 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18833 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
18834 when an option is an alias of another.
18835 * config/arm/parsecpu.awk (optalias): New parser token.
18836 (gen_comm_data): Mark non-alias options as such. Emit entries
18837 for extension aliases.
18838 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
18839 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
18840 (armv6kz, armv6zk, armv6t2): Likewise.
18841 (armv7): Make vfpv3-d16 an alias.
18842 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
18844 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
18845 Sort in canonical order.
18846 (armv8-a): Sort in canonical order.
18847 (armv8.1-a, armv8.2-a): Likewise.
18848 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
18850 (cortex-a9): Sort in canonical order.
18851 * config/arm/arm.c (selftests.h): Include it.
18852 (arm_test_cpu_arch_data): New function.
18853 (arm_run_self_tests): New function.
18854 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
18855 (targetm): Move declaration to the end of the file.
18856 * arm-cpu-cdata.h: Regenerated.
18858 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18860 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
18861 call to target_mode_check describing the type of option passed.
18862 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
18863 (arm_target_thumb_only): Use arm_parse_arch_option_name or
18864 arm_parse_cpu_option_name to match parameters against list of
18866 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
18867 arm_arch_core_flags data structure.
18868 * config/arm/arm-cpu_cdata.h: Regenerated.
18870 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18872 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
18874 (arm_print_hint_for_cpu_option): Likewise.
18875 (arm_print_hint_for_arch_option): Likewise.
18876 (arm_parse_cpu_option_name): Likewise.
18877 (arm_parse_arch_option_name): Likewise.
18878 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
18879 of entries in the all_fpus list.
18880 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
18881 (arm_parse_cpu_option_name): Declare.
18882 (arm_parse_arch_option_name): Declare.
18883 (arm_parse_option_features): Declare.
18884 (arm_intialize_isa): Declare.
18885 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
18887 (gen_comm_data): ... here. Make definitions non-static.
18888 * config/arm/arm-cpu-data.h: Regenerated.
18889 * config/arm/arm-cpu-cdata.h: Regenerated.
18891 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18893 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
18894 (cpu_arch_extension): New structure.
18895 (cpu_arch_option, arch_option, cpu_option): New structures.
18896 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
18897 architecture types.
18898 (gen_data): Generate new format data tables.
18899 * config/arm/arm.c (cpu_tune): New structure.
18900 (cpu_option, processors): Delete.
18901 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
18902 (arm_print_hint_for_cpu_option): ... this and ...
18903 (arm_print_hint_for_arch_option): ... this.
18904 (arm_parse_arch_cpu_name): Delete. Replace with ...
18905 (arm_parse_cpu_option_name): ... this and ...
18906 (arm_parse_arch_option_name): ... this.
18907 (arm_unrecognized_feature): Change type of target parameter to
18909 (arm_parse_arch_cpu_features): Delete. Replace with ...
18910 (arm_parse_option_features): ... this.
18911 (arm_configure_build_target): Rework to use new configuration data
18913 (arm_print_tune_info): Rework for new configuration data tables.
18914 * config/arm/arm-cpu-data.h: Regenerated.
18915 * config/arm/arm-cpu.h: Regenerated.
18917 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18919 * Makefile.in (OBJS): Move sbitmap.o from here ...
18920 (OBJS-libcommon): ... to here.
18922 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18924 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
18925 (ISA_ALL_CRYPTO): New macro.
18926 (ISA_ALL_SIMD): New macro
18927 (ISA_ALL_FP): New macro.
18928 * config/arm/arm.c (fpu_bitlist): Update initializer.
18929 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
18931 (arm9e): Add fpu. Add option for nofp
18932 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
18933 (arm926ej-s, arm1026ej-s): Likewise.
18934 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
18935 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
18936 neon-fp16, neon-vfpv4, nofp and nosimd.
18937 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
18938 (cortex-a8): Add fpu. Add option for nofp.
18939 (cortex-a9): Add fpu. Add options for nosimd and nofp.
18940 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
18941 (cortex-r4f): Add fpu.
18942 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
18943 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
18945 (cortex-r8): Likewise.
18946 (cortex-m4): Add fpu. Add option for nofp.
18947 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
18948 (cortex-a17.cortex-a7): Likewise.
18949 (cortex-a32): Add fpu. Add options for crypto and nofp.
18950 (cortex-a35, cortex-a53): Likewise.
18951 (cortex-a57): Add fpu. Add option for crypto.
18952 (cortex-a72, cortex-a73): Likewise.
18953 (exynos-m1): Likewise.
18954 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
18955 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
18956 (cortex-m33): Add fpu. Add option for nofp.
18957 * config/arm/arm-cpu-cdata.h: Regenerated
18958 * config/arm/arm-cpu-data.h: Regenerated.
18960 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18962 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
18963 (armv5te, armv5tej): Likewise.
18964 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
18965 (armv7): Add options fp and vfpv3-d16.
18966 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
18967 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
18969 (armv7ve): Likewise.
18970 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
18971 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
18972 (armv8-a): Add nocrypto option.
18973 (armv8.1-a, armv8.2-a): Likewise.
18974 (armv8-m.main): add options fp, fp.dp and nofp.
18976 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18978 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
18980 (armv8-a+crc): Delete.
18981 (armv8.1-a): Add options simd, crypto and nofp.
18982 (armv8.2-a): Add options fp16, simd, crypto and nofp.
18983 (armv8.2-a+fp16): Delete.
18984 (armv8-m.main): Add option dsp.
18985 (armv8-m.main+dsp): Delete.
18986 (cortex-a8): Add fpu. Add nofp option.
18987 (cortex-a9): Add fpu. Add nofp and nosimd options.
18988 * config/arm/parsecpu.awk (gen_data): Generate option tables and
18989 link to main cpu and architecture data structures.
18990 (gen_comm_data): Only put isa attributes from the main architecture
18992 (option): New statement for architecture and CPU entries.
18993 * arm.c (struct cpu_option): New structure.
18994 (struct processors): Add entry for options.
18995 (arm_unrecognized_feature): New function.
18996 (arm_parse_arch_cpu_name): Ignore any characters after the first
18998 (arm_parse_arch_cpu_feature): New function.
18999 (arm_configure_build_target): Separate out any CPU and architecture
19000 features and parse separately. Don't error out if -mfpu=auto is
19001 used with only an architecture string.
19002 (arm_print_asm_arch_directives): New function.
19003 (arm_file_start): Call it.
19004 * config/arm/arm-cpu-cdata.h: Regenerated.
19005 * config/arm/arm-cpu-data.h: Likewise.
19006 * config/arm/arm-tables.opt: Likewise.
19008 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19010 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
19011 assembler when it is not -mfpu=auto.
19013 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19015 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
19016 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
19017 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
19018 (ASM_CPU_SPEC): Rewrite.
19019 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
19020 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
19021 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
19022 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
19023 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
19024 copied string is NUL-terminated. Also strip any characters prefixed
19026 (arm_rewrite_selected_arch): New function.
19027 (arm_rewrite_march): New function.
19029 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19031 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
19032 (x_arm_cpu_string, x_arm_tune_string): Likewise.
19033 (march, mcpu, mtune): Convert to string-based options.
19034 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
19035 (arm_parse_arch_cpu_name): New function.
19036 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
19037 identify selected architecture or CPU.
19038 (arm_option_save): New function.
19039 (TARGET_OPTION_SAVE): Redefine.
19040 (arm_option_restore): Restore string options.
19041 (arm_option_print): Print string options.
19043 2017-06-16 Martin Sebor <msebor@redhat.com>
19045 PR tree-optimization/80933
19046 PR tree-optimization/80934
19047 * builtins.c (fold_builtin_3): Do not handle bcmp here.
19048 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
19049 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
19050 (gimple_fold_builtin): Call them.
19052 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19054 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
19055 as unlikely; update profile.
19057 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19059 * predict.c (force_edge_cold): Handle declaring edges impossible
19062 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19064 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
19066 (try_unroll_loop_completely): Fix reporting.
19068 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19070 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
19072 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
19075 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
19076 if given a non-constant argument for an intrinsic which requires a
19079 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19081 * profile.c (compare_freqs): New function.
19082 (branch_prob): Sort edge list.
19083 (find_spanning_tree): Assume that the list is priority sorted.
19085 2017-06-16 Richard Biener <rguenther@suse.de>
19087 PR tree-optimization/81090
19088 * passes.def (pass_record_bounds): Remove.
19089 * tree-pass.h (make_pass_record_bounds): Likewise.
19090 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
19091 make_pass_record_bounds): Likewise.
19092 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
19093 not free niter estimates at the beginning but at the end.
19094 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
19096 2017-06-16 Richard Biener <rguenther@suse.de>
19098 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
19099 initializer to workaround ICE in host GCC 4.8.
19101 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19103 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
19105 (clone_inlined_nodes): Update.
19107 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19109 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
19110 prefetch settings, and enable prefetching by default at -O3.
19112 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19114 * config/aarch64/aarch64.c (aarch64_override_options_internal):
19115 Set flag_prefetch_loop_arrays according to tuning data.
19117 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19119 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
19120 New tune structure.
19121 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
19122 [Unrelated to main purpose of the patch] Place the pointer field last
19123 to enable type checking errors when tune structure are wrongly merged.
19124 * config/aarch64/aarch64.c (generic_prefetch_tune,)
19125 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
19126 (thunderx2t99_prefetch_tune): New tune constants.
19127 (tune_params *_tunings): Update all tunings (no functional change).
19128 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
19129 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
19130 from tunings structures.
19132 2017-06-16 Jakub Jelinek <jakub@redhat.com>
19135 * ubsan.c (instrument_null): Add T argument, use it instead
19136 of computing it based on IS_LHS.
19137 (instrument_object_size): Likewise.
19138 (pass_ubsan::execute): Adjust instrument_null and
19139 instrument_object_size callers to pass gimple_get_lhs or
19140 gimple_assign_rhs1 result to it. Use instrument_null instead of
19141 calling get_base_address and instrument_mem_ref. Handle
19142 aggregate call arguments for object-size sanitization.
19144 2017-06-16 Yury Gribov <tetra2005@gmail.com>
19146 PR tree-optimization/81089
19147 * tree-vrp.c (is_masked_range_test): Validate operands of
19150 2017-06-15 Martin Sebor <msebor@redhat.com>
19153 * dumpfile.c (dump_register): Avoid calling memset to initialize
19154 a class with a default ctor.
19155 * gcc.c (struct compiler): Remove const qualification.
19156 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
19157 * hash-table.h: Ditto.
19158 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
19160 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
19161 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
19163 * params.h (struct param_info): Make struct members non-const.
19164 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
19165 with default initialization.
19166 * vec.h (vec_copy_construct, vec_default_construct): New helper
19168 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
19169 with vec_copy_construct.
19170 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
19171 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
19172 * doc/invoke.texi (-Wclass-memaccess): Document.
19174 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19176 * emit-rtl.h (is_leaf): Update comment about local
19177 register allocator.
19179 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19182 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
19183 for a variable to have a section before checking if the section has a
19185 Set section to.persistent if persistent attribute is set.
19186 Warn if .persistent attribute is used on an automatic variable.
19188 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
19190 PR rtl-optimization/80474
19191 * reorg.c (update_block): Do not ignore instructions in a delay slot.
19193 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
19195 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
19198 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
19200 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
19201 (casesi): Emit bounds checking as RTL.
19202 (casesi_internal_mips16_<mode>): Remove bounds checking.
19204 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
19206 * config/xtensa/xtensa.c (xtensa_option_override): Append
19207 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
19208 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
19209 xtensa_doloop_hooks): Define unconditionally.
19210 (xtensa_reorg_loops): Only call reorg_loops in the presence of
19212 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
19213 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
19214 for it in xtensa_option_override.
19215 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
19216 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
19218 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
19220 * doc/cppopts.texi: Document '-' special value to -MF.
19222 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
19224 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
19225 (cortex_a53_fconst): Likewise.
19226 (cortex_a53_fpmul): Likewise.
19227 (cortex_a53_f_load_64): Likewise.
19228 (cortex_a53_f_load_many): Likewise.
19229 (cortex_a53_advsimd_alu): Likewise.
19230 (cortex_a53_advsimd_alu_q): Likewise.
19231 (cortex_a53_advsimd_mul): Likewise.
19232 (cortex_a53_advsimd_mul_q): Likewise.
19233 (fpmac bypass): Add new bypass for fpmac-fpmac case.
19234 Add missing fmul, r2f_cvt and fconst cases.
19236 2017-06-14 Richard Biener <rguenther@suse.de>
19238 PR middle-end/81088
19239 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
19241 (fold_binary_loc): When associating do not treat pre-existing
19242 TREE_OVERFLOW on literal constants as a reason to allow
19243 TREE_OVERFLOW on associated literal constants.
19245 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19247 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
19248 (MASK_FEATURES): New macro.
19249 * config/sparc/sparc.c (sparc_option_override): Remove the special
19250 handling of -mfpu and generalize it to all MASK_FEATURES switches.
19252 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19254 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
19255 a division of 0 if non-call exceptions are enabled.
19257 2017-06-14 Andrew Pinski <apinski@cavium.com>
19258 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19261 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
19262 Improve vector initialization code gen for only variable case.
19264 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19266 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
19268 2017-06-14 Richard Biener <rguenther@suse.de>
19270 PR tree-optimization/81083
19271 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
19274 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19276 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
19277 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
19278 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
19279 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
19280 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
19281 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
19283 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19285 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
19286 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
19288 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19290 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
19292 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19294 * config/rs6000/t-rtems: Don't handle SPE.
19296 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19298 * config/rs6000/t-linux: Don't handle SPE.
19300 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19302 * config/rs6000/eabispe.h: Delete file.
19304 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19306 * config/rs6000/t-spe: Delete file.
19308 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19310 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
19311 (rs6000_legitimate_offset_address_p): Return false for anything in
19312 V2SImode or V2SFmode.
19314 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19316 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
19317 except V2SF and V2SI. Rearrange the vector modes, and add comments.
19318 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
19320 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
19321 (rs6000_legitimate_offset_address_p): Ditto.
19322 (rs6000_emit_move): Ditto.
19323 (rs6000_init_builtins): Remove V4HI_type_node.
19325 2017-06-13 Martin Liska <mliska@suse.cz>
19328 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
19329 (gate_asan): Likewise.
19330 * asan.h (asan_no_sanitize_address_p): Remove the function.
19331 (sanitize_flags_p): New function.
19332 * builtins.def: Fix coding style.
19333 * common.opt: Use renamed enum value.
19334 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
19335 * doc/extend.texi: Document no_sanitize attribute.
19336 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
19337 to SANITIZE_UNDEFINED_NONDEFAULT.
19338 * gcc.c (sanitize_spec_function): Use the renamed enum value.
19339 * gimple-fold.c (optimize_atomic_compare_exchange_p):
19340 Use sanitize_flags_p.
19341 * gimplify.c (gimplify_function_tree): Likewise.
19342 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
19343 * opts.c (parse_no_sanitize_attribute): New function.
19344 (common_handle_option): Use renamed enum value.
19345 * opts.h (parse_no_sanitize_attribute): Declare.
19346 * tree.c (sanitize_flags_p): New function.
19347 * tree.h: Declared here.
19348 * tsan.c: Use sanitize_flags_p.
19349 * ubsan.c (ubsan_expand_null_ifn): Likewise.
19350 (instrument_mem_ref): Likewise.
19351 (instrument_bool_enum_load): Likewise.
19352 (do_ubsan_in_current_function): Remove the function.
19353 (pass_ubsan::execute): Use sanitize_flags_p.
19354 * ubsan.h: Remove do_ubsan_in_current_function
19355 * tree-cfg.c (print_no_sanitize_attr_value): New function.
19356 (dump_function_to_file): Use it here.
19358 2017-06-13 Martin Jambor <mjambor@suse.cz>
19360 PR tree-optimization/80803
19361 PR tree-optimization/81063
19362 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
19363 (propagate_subaccesses_across_link): Enqueue subtree whenever
19364 necessary instead of relying on the caller.
19366 2017-06-13 Martin Jambor <mjambor@suse.cz>
19368 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
19369 that have a first_link.
19370 (sort_and_splice_var_accesses): Do not check first_link before
19372 (subtree_mark_written_and_enqueue): Likewise.
19373 (propagate_all_subaccesses): Likewise and do not stop at first
19374 parent with a first_link.
19376 2017-06-13 Martin Jambor <mjambor@suse.cz>
19378 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
19381 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19383 * match.pd: New pattern.
19385 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19387 * tree-vrp.c (is_masked_range_test): New function.
19388 (register_edge_assert_for): Determine ranges for
19391 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19393 PR tree-optimization/67328
19394 * fold-const.c (maskable_range_p): New function.
19395 (build_range_check): Generate bittests if possible.
19397 2017-06-13 Martin Liska <mliska@suse.cz>
19399 * gimple-pretty-print.c (dump_probability): Add new argument.
19400 (dump_edge_probability): Dump both probability and count.
19401 (dump_gimple_label): Likewise.
19402 (dump_gimple_bb_header): Likewise.
19404 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
19407 * config/avr/avr-devices.c: Fix indentation.
19408 * config/avr/gen-avr-mmcu-specs.c: Dito.
19410 2017-06-13 Richard Biener <rguenther@suse.de>
19412 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
19413 instead get vector type from stmt_info.
19414 (vectorizable_reduction): Adjust. Remove dead code.
19416 2017-06-13 Richard Biener <rguenther@suse.de>
19418 PR middle-end/81065
19419 * fold-const.c (extract_muldiv_1): Remove bogus distribution
19420 case of C * (x * C2 + C3).
19421 (fold_addr_of_array_ref_difference): Properly fold index difference.
19423 2017-06-12 David S. Miller <davem@davemloft.net>
19426 * config/sparc/sparc.md (return expander): Emit frame blockage if
19427 function uses alloca.
19429 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
19431 * combine.c (make_field_assignment): Check len rather than the mode
19432 precision when calling force_to_mode.
19434 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
19436 Support multilibs and devices that see flash in RAM address range.
19439 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
19440 (avr_mcu_t) <flash_pm_offset>: New field.
19441 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
19442 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
19443 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
19444 (AVR_TINY_PM_OFFSET): Remove macro.
19445 * config/avr/avr.opt (-mshort-calls): New option.
19446 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
19447 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
19448 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
19449 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
19450 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
19451 instead of avr_arch->have_jmp_call.
19452 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
19453 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
19454 avr_arch->flash_pm_offset to define.
19455 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
19456 new field flash_pm_offset. Add entry for avrxmega3.
19457 (avr_texinfo): Add entry for avrxmega3.
19458 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
19459 attiny212, attiny214,
19460 attiny412, attiny414, attiny416, attiny417,
19461 attiny814, attiny816, attiny817,
19462 attiny1614, attiny1616, attiny1617,
19463 attiny3214, attiny3216, attiny3217.
19464 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
19465 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
19466 (avr_print_operand_address) [AVR_TINY]: Same.
19467 (avr_asm_init_sections) <readonly_data_section>: Only patch
19468 callback if avr_arch->flash_pm_offset = 0.
19469 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
19470 for rodata if avr_arch->flash_pm_offset != 0.
19471 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
19472 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
19473 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
19474 (m_options): Append opt_rcall.
19475 (m_dirnames): Append dir_rcall.
19476 * config/avr/t-multilib: Regenerate.
19478 * configure.ac [target=avr]: Check whether avrxmega3 default
19479 linker description file works as needed.
19480 * configure: Regenerate.
19481 * doc/avr-mmcu.texi: Regenerate.
19482 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
19483 <__AVR_ARCH__>: Document avrxmega3 and 103.
19484 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
19485 <__AVR_SHORT_CALLS__>: Document it.
19486 <__AVR_PM_BASE_ADDRESS__>: Document it.
19487 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
19488 (AVR Variable Attributes) <progmem>: Document this is
19489 not needed for avrxmega3.
19490 (AVR Named Address Spaces) <__flash>: Dito.
19492 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
19494 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
19496 2017-06-12 Doug Rupp <rupp@adacore.com>
19498 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
19499 Append vxworks-stdint.h to the tm_file list.
19500 * config/vxworks-stdint.h: New file.
19502 2017-06-12 Martin Liska <mliska@suse.cz>
19504 PR tree-optimization/81041
19505 * tree-profile.c (gimple_gen_ic_func_profiler):
19506 Create an extra BB in profile-generate
19507 (gimple_gen_time_profiler): Likewise.
19509 2017-06-12 Jakub Jelinek <jakub@redhat.com>
19511 PR tree-optimization/81003
19512 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
19513 (update_range_test): Use it instead of force_gimple_operand_gsi.
19515 2017-06-12 Richard Biener <rguenther@suse.de>
19517 PR tree-optimization/81053
19518 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
19519 with backedge value not defined in loop. Simplify def stmt
19522 2017-06-11 Tom de Vries <tom@codesourcery.com>
19525 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
19527 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
19528 nvptx_cannot_force_const_mem.
19530 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19532 * opts.c (finish_options): Move test for flag_split_stack after
19533 it has been initialized.
19535 2017-06-11 Jason Merrill <jason@redhat.com>
19537 * tree.h (id_equal): New.
19538 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
19539 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
19540 instead of strcmp of IDENTIFIER_POINTER.
19542 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19544 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
19545 (mark_all_inlined_calls_cdtor): Fix formating.
19546 (inline_transform): Rescale profile before inlining.
19548 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19550 * cgraph.h (cgraph_edge::clone): Update prototype.
19551 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
19552 (cgraph_node::create_clone): Update.
19553 (cgraph_node::create_version_clone): Update.
19554 * tree-inline.c (copy_bb): Update.
19555 (expand_call_inline): Update.
19557 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
19559 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
19560 factored out from ...
19561 (rs6000_emit_prologue): ... here.
19563 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
19565 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
19566 factored out from ...
19567 (rs6000_emit_prologue): ... here.
19569 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19571 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
19573 (handle_missing_profiles): Fix computation of tp_first_run.
19574 (counts_to_freqs): Do not touch freqs when count is 0.
19576 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19578 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
19581 2017-06-10 Tom de Vries <tom@codesourcery.com>
19583 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
19584 attributes): Document signal effective target.
19586 2017-06-10 Tom de Vries <tom@codesourcery.com>
19588 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
19589 Document effective target stack_size.
19591 2017-06-09 David Malcolm <dmalcolm@redhat.com>
19593 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
19594 to the edit_context if they can be auto-applied.
19596 2017-06-9 Ian Lance Taylor <iant@golang.org>
19598 * opts.c (finish_options): If -fsplit-stack, disable implicit
19599 -forder-blocks-and-partition.
19600 * doc/invoke.texi (Optimize Options): Document that when using
19601 -fsplit-stack -forder-blocks-and-partition is not implicitly
19604 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
19606 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
19607 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
19608 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
19609 * builtins.def (abort, trap, unreachable): Declare cold.
19610 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
19611 * tree-core.h (ECF_COLD): New.
19612 * tree.c (set_call_expr_flags): Handle ECF_COLD.
19613 (build_common_builtin_nodes): Mark unreachable and abort as cold.
19615 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
19617 * predict.c (unlikely_executed_stmt_p): Cleanup.
19619 2017-06-09 Richard Biener <rguenther@suse.de>
19621 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
19622 model if the ref is always written to.
19624 2017-06-09 Tamar Christina <tamar.christina@arm.com>
19626 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
19628 2017-06-09 Tamar Christina <tamar.christina@arm.com>
19630 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
19633 2017-06-09 Tom de Vries <tom@codesourcery.com>
19636 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
19637 "target cannot support label values" when encountering LABEL_REF.
19639 2017-06-09 Martin Liska <mliska@suse.cz>
19641 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
19642 (gimple_gen_ic_func_profiler): Emit direct comparison
19643 of __gcov_indirect_call_callee with NULL.
19644 (gimple_gen_time_profiler): Change probability from
19645 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
19647 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
19649 * profile.c (edge_gcov_counts): Turn to pointer.
19650 (compute_branch_probabilities, compute_branch_probabilities): Update.
19651 (branch_prob): Do not clear edge_gcov_count.
19652 * profile.h (edge_gcov_counts): Turn to pointer.
19653 (edge_gcov_count): Update.
19655 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
19657 * gimple.h (gimple_check_failed): Mark cold.
19659 2017-06-09 Richard Biener <rguenther@suse.de>
19661 PR tree-optimization/66623
19662 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
19663 refactor check_reduction into two parts, properly computing
19664 whether we have to check reduction validity for outer loop
19667 2017-06-09 Richard Biener <rguenther@suse.de>
19669 PR tree-optimization/79483
19670 * graphite-scop-detection.c (order): New global.
19671 (get_order): Compute bb to order mapping that satisfies code
19672 generation constraints.
19673 (cmp_pbbs): New helper.
19674 (build_scops): Start domwalk at entry block, sort generated
19677 2017-06-09 Richard Biener <rguenther@suse.de>
19679 PR middle-end/81007
19680 * ipa-polymorphic-call.c
19681 (ipa_polymorphic_call_context::restrict_to_inner_class):
19682 Skip FIELD_DECLs with error_mark_node type.
19683 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
19686 2017-06-09 Martin Liska <mliska@suse.cz>
19688 * predict.c (struct branch_predictor): New struct.
19689 (test_prediction_value_range): New test.
19690 (predict_c_tests): New function.
19691 * selftest-run-tests.c (selftest::run_tests): Run the function.
19692 * selftest.h: Declare new tests.
19694 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
19697 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
19698 gen_add3_insn did not fail.
19699 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
19700 r0, construct that number in a temporary reg and add that reg to r0.
19701 If asked to put the result in r0 as well, fail.
19703 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
19705 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
19706 for early expansion of vec_eqv.
19708 2017-06-08 Jakub Jelinek <jakub@redhat.com>
19710 PR middle-end/81005
19711 * ubsan.c (instrument_null): Avoid pointless code temporary.
19712 (pass_ubsan::execute): Instrument aggregate arguments of calls.
19714 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
19718 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
19721 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
19724 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19726 * predict.c (unlikely_executed_edge_p): Move ahead.
19727 (probably_never_executed_edge_p): Use it.
19729 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
19731 PR middle-end/79988
19732 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
19733 gimple_call_builtin_p call.
19735 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19737 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
19738 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
19739 rtl_check_failed_type2, rtl_check_failed_code1,
19740 rtl_check_failed_code2, rtl_check_failed_code_mode,
19741 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
19742 rtvec_check_failed_bounds, rtl_check_failed_flag,
19743 _fatal_insn_not_found, _fatal_insn): Likewise.
19744 * tree.h (tree_contains_struct_check_failed,
19745 tree_check_failed, tree_not_check_failed,
19746 tree_class_check_failed, tree_range_check_failed,
19747 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
19748 tree_vec_elt_check_failed, phi_node_elt_check_failed,
19749 tree_operand_check_failed, omp_clause_check_failed,
19750 omp_clause_operand_check_failed, omp_clause_range_check_failed):
19753 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19755 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
19756 flag_branch_probabilities.
19757 * ipa-inline.c (edge_badness): Likewise.
19758 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
19759 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
19760 * predict.c (maybe_hot_frequency_p): Likewise.
19761 (probably_never_executed): Likewise.
19762 * sched-ebb.c (schedule_ebbs): Likewise.
19763 * sched-rgn.c (find_single_block_region): Likewise.
19764 * tracer.c (tail_duplicate): Likewise.
19766 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19768 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
19769 longer requires x_flag_profile_use.
19771 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19773 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
19774 instead of flag_reorder_blocks_and_partition.
19775 * dbxout.c (dbxout_function_end): Likewise.
19776 * dwarf2out.c (gen_subprogram_die): Likewise.
19777 * haifa-sched.c (sched_create_recovery_edges): Likewise.
19778 * hw-doloop.c (reorg_loops): Likewise.
19779 * varasm.c (assemble_start_function,
19780 assemble_end_function): Likewise.
19781 (decide_function_section): Do not check for
19782 flag_reorder_blocks_and_partition.
19784 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
19786 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
19788 (chkp_get_hard_register_fake_addr_expr): Ditto.
19789 (chkp_build_addr_expr): Add check for hard reg case.
19790 (chkp_parse_array_and_component_ref): Ditto.
19791 (chkp_find_bounds_1): Ditto.
19792 (chkp_process_stmt): Don't generate bounds store for
19795 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19797 * predict.c (maybe_hot_bb_p): Do not check profile status.
19798 (maybe_hot_edge_p): Likewise.
19799 (probably_never_executed): Check for zero counts even if profile
19801 (unlikely_executed_edge_p): New function.
19802 (unlikely_executed_stmt_p): New function.
19803 (unlikely_executed_bb_p): New function.
19804 (set_even_probabilities): Use unlikely predicates.
19805 (combine_predictions_for_bb): Likewise.
19806 (predict_paths_for_bb): Likewise.
19807 (predict_paths_leading_to_edge): Likewise.
19808 (determine_unlikely_bbs): New function.
19809 (estimate_bb_frequencies): Use it.
19810 (compute_function_frequency): Use zero counts even if profile is
19812 * profile-count.h: Fix typo.
19814 2017-08-08 Julia Koval <julia.koval@intel.com>
19816 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
19817 _mm512_mask_cvtsepi16_storeu_epi8,
19818 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
19819 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
19820 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
19821 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
19822 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
19823 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
19824 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
19825 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
19826 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
19827 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
19828 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
19829 __builtin_ia32_pmovuswb256mem_mask,
19830 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
19831 __builtin_ia32_pmovwb512mem_mask): New builtins.
19833 2017-08-08 Julia Koval <julia.koval@intel.com>
19835 PR target/73350,80862
19836 * config/i386/subst.md (round): Fix round pattern.
19837 * config/i386/i386.c (ix86_erase_embedded_rounding):
19838 Fix erasing rounding for the fixed pattern.
19840 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19842 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
19844 2017-06-08 Martin Liska <mliska@suse.cz>
19846 PR gcov-profile/80911
19847 * gcov.c (block_info::block_info): New constructor.
19849 2017-06-07 Carl Love <cel@us.ibm.com>
19851 * config/rs6000/rs6000-c: The return type of the following
19852 built-in functions was implemented as int not long long. Fix sign
19853 of return value for the unsigned version of vec_mulo and vec_mule.
19854 vector unsigned long long vec_bperm (vector unsigned long long,
19855 vector unsigned char)
19856 vector signed long long vec_mule (vector signed int,
19858 vector unsigned long long vec_mule (vector unsigned int,
19859 vector unsigned int)
19860 vector signed long long vec_mulo (vector signed int,
19862 vector unsigned long long vec_mulo (vector unsigned int,
19863 vector unsigned int)
19864 * doc/extend.texi: Fix the documentation for the built-in
19867 2017-06-07 Carl Love <cel@us.ibm.com>
19870 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
19873 2017-06-07 Carl Love <cel@us.ibm.com>
19875 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
19876 support, Generate doublehv for signed int/float for BE case only.
19878 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
19880 * doc/invoke.texi (mcx16): Rewrite.
19882 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19884 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
19885 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
19886 *mov<mode>_softfloat, and an anonymous splitter): Use
19887 nonimmediate_operand instead of rs6000_nonimmediate_operand.
19889 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19891 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
19893 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
19894 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
19895 (rs6000_debug_reg_global): Adjust.
19896 (rs6000_init_hard_regno_mode_ok): Adjust.
19897 (rs6000_dbx_register_number): Adjust.
19898 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
19899 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
19900 Remove SPE_ACC and SPEFSCR.
19901 (REG_ALLOC_ORDER): Ditto.
19902 (FRAME_POINTER_REGNUM): Change to 111.
19903 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
19904 (REG_CLASS_NAMES): Ditto.
19905 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
19906 (REGISTER_NAMES): Ditto.
19907 (ADDITIONAL_REG_NAMES): Ditto.
19908 (rs6000_reg_names): Ditto.
19909 * config/rs6000/rs6000.md: Renumber some register number
19912 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19914 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
19916 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
19917 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
19919 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
19920 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
19921 Delete the SPE high registers.
19922 (REG_ALLOC_ORDER): Ditto.
19923 (enum reg_class): Remove SPE_HIGH_REGS.
19924 (REG_CLASS_NAMES): Ditto.
19925 (REG_CLASS_CONTENTS): Delete the SPE high registers.
19926 (REGISTER_NAMES): Ditto.
19927 (rs6000_reg_names): Ditto.
19928 * doc/tm.texi.in: Remove SPE as example.
19929 * doc/tm.texi: Regenerate.
19931 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19933 * config/rs6000/8540.md (ppc8540_brinc): Delete.
19934 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
19935 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
19936 * config/rs6000/rs6000.md (type): Remove "brinc".
19938 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19940 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
19941 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
19942 * config/rs6000/linuxspe.h: Delete file.
19943 * config/rs6000/rs6000.md: Don't include spe.md.
19944 * config/rs6000/spe.h: Delete file.
19945 * config/rs6000/spe.md: Delete file.
19946 * config/rs6000/t-rs6000: Remove spe.md.
19948 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19950 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
19951 (reg_or_none500mem_operand): Delete.
19952 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
19953 instead of reg_or_none500mem_operand.
19955 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19957 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
19958 handling of SPE flags.
19959 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
19961 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19963 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
19965 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
19966 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
19967 paired_divv2sf3): Similar.
19968 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
19969 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
19970 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
19972 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
19973 Rename the paired_* instruction patterns.
19974 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
19976 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
19977 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
19978 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
19979 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
19980 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
19981 PAIRED_VECTOR_MODE.
19982 (struct machine_function): Delete field spe_insn_chain_scanned_p.
19983 (spe_func_has_64bit_regs_p): Delete.
19984 (spe_expand_predicate_builtin): Delete.
19985 (spe_expand_evsel_builtin): Delete.
19986 (TARGET_DWARF_REGISTER_SPAN): Do not define.
19987 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
19988 (invalid_e500_subreg): Delete.
19989 (rs6000_legitimize_address): Always force_reg op2 as well, for
19990 paired single memory accesses.
19991 (rs6000_member_type_forces_blk): Delete.
19992 (rs6000_spe_function_arg): Delete.
19993 (rs6000_expand_unop_builtin): Delete SPE handling.
19994 (rs6000_expand_binop_builtin): Ditto.
19995 (spe_expand_stv_builtin): Delete.
19996 (bdesc_2arg_spe): Delete.
19997 (spe_expand_builtin): Delete.
19998 (spe_expand_predicate_builtin): Delete.
19999 (spe_expand_evsel_builtin): Delete.
20000 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
20001 (spe_init_builtins): Delete.
20002 (spe_func_has_64bit_regs_p): Delete.
20003 (savres_routine_name): Delete "info" parameter. Adjust callers.
20004 (rs6000_emit_stack_reset): Ditto.
20005 (rs6000_dwarf_register_span): Delete.
20006 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
20007 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
20008 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
20010 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
20012 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
20013 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
20014 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
20015 mulv2sf3, divv2sf3): Delete expanders.
20017 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20019 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
20021 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20023 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
20024 * config/rs6000/rs6000.c: Ditto.
20026 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20028 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
20029 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
20030 comparison_operator.
20032 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20034 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
20035 * config/rs6000/rs6000.opt: Ditto.
20036 * config/rs6000/t-rtems: Ditto.
20038 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20040 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
20041 TARGET_E500_SINGLE by 0, simplify.
20042 * config/rs6000/rs6000.c: Ditto.
20043 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
20044 (spe_build_register_parallel): Delete.
20045 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
20046 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
20047 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
20048 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
20049 (E500_CONVERT): Delete.
20050 * config/rs6000/spe.md: Remove many patterns and all define_constants.
20052 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20054 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
20055 * config/rs6000/dfp.md: Ditto.
20056 (negdd2, *negdd2_fpr): Merge.
20057 (absdd2, *absdd2_fpr): Merge.
20058 (negtd2, *negtd2_fpr): Merge.
20059 (abstd2, *abstd2_fpr): Merge.
20060 * config/rs6000/e500.h: Delete file.
20061 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
20062 TARGET_FPRS by 1 and simplify.
20063 * config/rs6000/rs6000-c.c: Ditto.
20064 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
20065 TARGET_DF_SPE by 0.
20066 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
20068 * config/rs6000/rs6000.md: Ditto.
20069 (floatdidf2, *floatdidf2_fpr): Merge.
20070 (move_from_CR_gt_bit): Delete.
20071 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
20072 (E500_CR_IOR_COMPARE): Delete.
20073 (All patterns that require !TARGET_FPRS): Delete.
20074 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
20076 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20078 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
20080 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20082 * graphds.c (add_edge): Intitialize edge's attached data.
20083 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
20084 pointer parameter. Call pointed function on each edge during
20085 graph traversing. Skip traversing the edge when the function
20087 (graphds_dfs, graphds_scc): Ditto.
20088 (for_each_edge): New parameter. Pass the new parameter to callback
20090 * graphds.h (skip_edge_callback): New function pointer type.
20091 (graphds_dfs, graphds_scc): New function pointer parameter.
20092 (graphds_edge_callback, for_each_edge): New parameter.
20094 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20096 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
20097 out code checking if runtime alias check is possible to below ...
20098 Call the new function.
20099 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
20100 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
20102 2017-06-07 Marek Polacek <polacek@redhat.com>
20105 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
20106 TYPE_OVERFLOW_WRAPS check.
20108 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20110 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
20111 if versioning is required.
20112 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
20113 peeling with the check for versioning.
20115 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20117 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
20118 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
20119 Set true to new parameter if new ssa variable is defined.
20120 (vect_gen_vector_loop_niters): Refactor. Set range information
20121 for the new vector loop bound variable.
20122 (vect_do_peeling): Ditto.
20124 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20126 * tree-affine.c (ssa.h): Include header file.
20127 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
20128 has wrapping overflow behavior.
20130 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20132 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
20134 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20136 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
20137 (tree_to_aff_combination): ... here.
20139 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20141 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
20142 reg_pressure model function.
20143 (ivopts_global_cost_for_size): Delete.
20144 (determine_set_costs, iv_ca_recount_cost): Call new model function
20145 ivopts_estimate_reg_pressure.
20147 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20149 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
20150 expensive than udiv. Remove floating point cases from mod.
20152 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20154 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
20155 Increase idiv cost.
20157 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20159 * config/aarch64/aarch64.md
20160 (copysignsf3): Fix mask generation.
20162 2017-06-07 Jakub Jelinek <jakub@redhat.com>
20164 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
20166 (class dump_manager): Add register_dumps method.
20167 * dumpfile.c: Include langhooks.h.
20168 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
20169 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
20170 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
20171 (dump_manager::dump_register): Start with 512 entries instead of 32.
20172 (dump_manager::register_dumps): New method.
20173 * toplev.c (general_init): Instead of invoking register_dumps
20174 langhook, invoke register_dumps method on the dump manager.
20175 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
20178 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
20180 * doc/md.texi: Clarify the restrictions on a define_insn condition.
20181 Say that # requires an associated define_split to exist, and that
20182 the define_split must be suitable for use after register allocation.
20184 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20186 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
20187 (compute_outgoing_frequencies): Also initialize zero counts.
20188 (find_many_sub_basic_blocks): Do not produce uninitialized profile
20189 around loops; preserve more of profile when nothing changes.
20191 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
20193 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
20195 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
20196 * config/arm/arm-cpu-cdata.h: Regenerate.
20197 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
20198 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
20199 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
20200 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
20201 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
20203 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
20205 * config/arm/t-rmprofile: Likewise.
20206 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
20208 2017-06-06 David S. Miller <davem@davemloft.net>
20211 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
20212 blockage if function uses alloca.
20214 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20216 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
20217 New "uid" fields to hold pretty-print IDs of group and ref.
20218 Memory references are now identified as <group_id>:<ref_id>
20219 instead of using [random] addresses.
20220 (dump_mem_details): Simplify, no functional change.
20221 (dump_mem_ref): Simplify and make output more concise.
20222 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
20223 (find_or_create_group): Initialize group uid.
20224 (record_ref): Initialize ref uid. Improve debug output.
20225 (prune_group_by_reuse, should_issue_prefetch_p,)
20226 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
20227 (mark_nontemporal_store, determine_loop_nest_reuse):
20228 Improve debug output.
20230 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20232 * dbgcnt.def (prefetch): New debug counter.
20233 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
20234 (schedule_prefetches): Stop issueing prefetches if debug counter
20237 2017-06-06 Tom de Vries <tom@codesourcery.com>
20239 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
20240 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
20242 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20244 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
20245 Use aarch64_reg_or_zero predicate for operand 4.
20246 (aarch64_compare_and_swap<mode> define_insn_and_split):
20247 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
20248 (aarch64_store_exclusive<mode>): Likewise for operand 2.
20250 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
20252 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
20253 (arm_compute_save_core_reg_mask): This.
20254 (thumb1_compute_save_reg_mask): Rename into ...
20255 (thumb1_compute_save_core_reg_mask): This.
20256 (arm_compute_save_reg0_reg12_mask): Adapt comment.
20257 (arm_compute_frame_layout): Likewise.
20259 2017-06-06 Richard Biener <rguenther@suse.de>
20261 PR tree-optimization/80974
20262 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
20263 keep or clear leaders SSA info.
20265 2017-06-06 Tom de Vries <tom@codesourcery.com>
20267 * config/nvptx/nvptx.c (split_mode_p): New function.
20268 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
20270 2017-06-06 Tom de Vries <tom@codesourcery.com>
20272 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
20274 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20277 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
20280 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20282 * shrink-wrap.c (handle_simple_exit): Update profile.
20283 (try_shrink_wrapping): Upate profile.
20285 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20287 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
20288 (tree_guess_outgoing_edge_probabilities): New.
20289 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
20290 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
20292 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20294 * ipa-split.c (split_function): Initialize return bb profile.
20296 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20298 * profile.c (compute_branch_probabilities): Also initialize
20299 EXIT_BLOCK profile.
20301 2017-06-06 Richard Biener <rguenther@suse.de>
20303 PR tree-optimization/80928
20304 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
20305 (vect_analyze_loop_operations): Properly guard analysis for
20307 (vect_transform_loop): Likewise.
20308 (vect_analyze_loop_2): Also reset SLP type on PHIs.
20309 (vect_model_induction_cost): Do not cost for pure SLP.
20310 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
20311 of induction in inner loop vectorization.
20312 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
20313 (vect_get_and_check_slp_defs): Handle vect_induction_def.
20314 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
20316 (vect_analyze_slp_cost_1): Cost induction.
20317 (vect_detect_hybrid_slp_stmts): Handle PHIs.
20318 (vect_get_slp_vect_defs): Likewise.
20319 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
20320 (vect_transform_stmt): Handle SLP reductions.
20321 * tree-vectorizer.h (vectorizable_induction): Adjust.
20323 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
20325 * config/rs6000/rs6000.c (make_resolver_func): Update
20326 init_lowered_empty_function call.
20328 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
20330 * doc/invoke.texi: Document the -fprofile-abs-path option.
20331 * common.opt (fprofile-abs-path): New option.
20332 * gcov-io.h (gcov_write_filename): Declare.
20333 * gcov-io.c (gcov_write_filename): New function.
20334 * coverage.c (coverage_begin_function): Use gcov_write_filename.
20335 * profile.c (output_location): Likewise.
20337 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20339 * shring-wrap.c: Revert accidental commit.
20341 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
20343 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
20345 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20347 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
20349 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
20350 profile in callgraph edge.
20351 * profile-count.h (apply_probability): If THIS is 0, then result is 0
20352 (apply_scale): Likewise.
20353 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
20354 Also scale profile when inlining function with zero profile.
20355 (initialize_cfun): Update exit block profile even when it is zero.
20356 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
20357 when profile is read.
20359 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
20361 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
20362 (CLONE_*): New constants to define the processors we can generate
20363 code for with the target_clone attribute.
20364 (rs6000_clone_map): New array to identify which clone processors
20365 the current program is running on.
20366 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
20367 target_clone attribute.
20368 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
20369 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
20370 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
20371 (cpu_expand_builtin): Add support for target_clone attribute.
20372 (rs6000_valid_attribute_p): Allow "default" attribute.
20373 (get_decl_name): New debug function to simplify printing the
20374 current function name in debugging statements.
20375 (rs6000_clone_priority): New functions to support the target_clone
20376 attribute, and be able to generate code to switch between ISA 2.05
20377 through ISA 3.0 (power6 through power9).
20378 (rs6000_compare_version_priority): Likewise.
20379 (rs6000_get_function_versions_dispatcher): Likewise.
20380 (make_resolver_func): Likewise.
20381 (add_condition_to_bb): Likewise.
20382 (dispatch_function_versions): Likewise.
20383 (rs6000_generate_version_dispatcher_body): Likewise.
20384 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
20385 (fusion_gpr_load_p): Fix a spacing issue.
20386 * doc/extend.texi (Common Function Attributes): Document that the
20387 PowerPC supports the target_clone attribute.
20389 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
20391 * config/arm/arm.h: explain F symbol found in description of ARM
20392 register allocation in its legend.
20394 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20396 * config/mips/frame-header-opt.c: Include profile-count.h.
20397 * config/riscv/riscv.c: Include profile-count.h
20399 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20401 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
20403 (sm_set_flag_if_changed): Add bbs field.
20404 (execute_sm_if_changed_flag_set): Pass BBS.
20405 (execute_sm): Update.
20407 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20409 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
20412 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20414 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
20415 (peephole2): New peephole2 to emit the above.
20416 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
20418 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20420 * config/aarch64/aarch64.c (define_peephole2 above
20421 *sub_<shift>_<mode>): New peephole.
20423 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
20425 * config/i386/i386.c (make_resolver_func): Update.
20426 * Makefile.in: Add profile-count.h and profile-count.o
20427 * auto-profile.c (afdo_indirect_call): Update to new API.
20428 (afdo_set_bb_count): Update.
20429 (afdo_propagate_edge): Update.
20430 (afdo_propagate_circuit): Update.
20431 (afdo_calculate_branch_prob): Update.
20432 (afdo_annotate_cfg): Update.
20433 * basic-block.h: Include profile-count.h
20434 (struct edge_def): Turn count to profile_count.
20435 (struct basic_block_def): Likewie.
20436 (REG_BR_PROB_BASE): Move to profile-count.h
20437 (RDIV): Move to profile-count.h
20438 * bb-reorder.c (max_entry_count): Turn to profile_count.
20439 (find_traces): Update.
20440 (rotate_loop):Update.
20441 (connect_traces):Update.
20442 (sanitize_hot_paths):Update.
20443 * bt-load.c (migrate_btr_defs): Update.
20444 * cfg.c (RDIV): Remove.
20445 (init_flow): Use alloc_block.
20446 (alloc_block): Uninitialize count.
20447 (unchecked_make_edge): Uninitialize count.
20448 (check_bb_profile): Update.
20449 (dump_edge_info): Update.
20450 (dump_bb_info): Update.
20451 (update_bb_profile_for_threading): Update.
20452 (scale_bbs_frequencies_int): Update.
20453 (scale_bbs_frequencies_gcov_type): Update.
20454 (scale_bbs_frequencies_profile_count): New.
20455 * cfg.h (update_bb_profile_for_threading): Update.
20456 (scale_bbs_frequencies_profile_count): Declare.
20457 * cfgbuild.c (compute_outgoing_frequencies): Update.
20458 (find_many_sub_basic_blocks): Update.
20459 * cfgcleanup.c (try_forward_edges): Update.
20460 (try_crossjump_to_edge): Update.
20461 * cfgexpand.c (expand_gimple_tailcall): Update.
20462 (construct_exit_block): Update.
20463 * cfghooks.c (verify_flow_info): Update.
20464 (dump_bb_for_graph): Update.
20465 (split_edge): Update.
20466 (make_forwarder_block): Update.
20467 (duplicate_block): Update.
20468 (account_profile_record): Update.
20469 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
20470 (get_estimated_loop_iterations): Update.
20471 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
20472 (single_likely_exit): Update.
20473 * cfgloopmanip.c (scale_loop_profile): Update.
20475 (set_zero_probability): Update.
20476 (lv_adjust_loop_entry_edge): Update.
20477 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
20478 (purge_dead_edges): Update.
20479 (rtl_account_profile_record): Update.
20480 * cgraph.c (cgraph_node::create): Uninitialize count.
20481 (symbol_table::create_edge): Uninitialize count.
20482 (cgraph_update_edges_for_call_stmt_node): Update.
20483 (cgraph_edge::dump_edge_flags): Update.
20484 (cgraph_node::dump): Update.
20485 (cgraph_edge::maybe_hot_p): Update.
20486 * cgraph.h: Include profile-count.h
20487 (create_clone), create_edge, create_indirect_edge): Update.
20488 (cgraph_node): Turn count to profile_count.
20489 (cgraph_edge0: Likewise.
20490 (make_speculative, clone): Update.
20491 (create_edge): Update.
20492 (init_lowered_empty_function): Update.
20493 * cgraphclones.c (cgraph_edge::clone): Update.
20494 (duplicate_thunk_for_node): Update.
20495 (cgraph_node::create_clone): Update.
20496 * cgraphunit.c (cgraph_node::analyze): Update.
20497 (cgraph_node::expand_thunk): Update.
20498 * final.c (dump_basic_block_info): Update.
20499 * gimple-streamer-in.c (input_bb): Update.
20500 * gimple-streamer-out.c (output_bb): Update.
20501 * graphite.c (print_global_statistics): Update.
20502 (print_graphite_scop_statistics): Update.
20503 * hsa-brig.c: Include basic-block.h.
20504 * hsa-dump.c: Include basic-block.h.
20505 * hsa-gen.c (T sum_slice): Update.
20506 (convert_switch_statements):Update.
20507 * hsa-regalloc.c: Include basic-block.h.
20508 * ipa-chkp.c (chkp_produce_thunks): Update.
20509 * ipa-cp.c (struct caller_statistics): Update.
20510 (init_caller_stats): Update.
20511 (gather_caller_stats): Update.
20512 (ipcp_cloning_candidate_p): Update.
20513 (good_cloning_opportunity_p): Update.
20514 (get_info_about_necessary_edges): Update.
20515 (dump_profile_updates): Update.
20516 (update_profiling_info): Update.
20517 (update_specialized_profile): Update.
20518 (perhaps_add_new_callers): Update.
20519 (decide_about_value): Update.
20520 (ipa_cp_c_finalize): Update.
20521 * ipa-devirt.c (struct odr_type_warn_count): Update.
20522 (struct decl_warn_count): Update.
20523 (struct final_warning_record): Update.
20524 (possible_polymorphic_call_targets): Update.
20525 (ipa_devirt): Update.
20526 * ipa-fnsummary.c (redirect_to_unreachable): Update.
20527 * ipa-icf.c (sem_function::merge): Update.
20528 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
20529 * ipa-inline.c (compute_uninlined_call_time): Update.
20530 (compute_inlined_call_time): Update.
20531 (want_inline_small_function_p): Update.
20532 (want_inline_self_recursive_call_p): Update.
20533 (edge_badness): Update.
20534 (lookup_recursive_calls): Update.
20535 (recursive_inlining): Update.
20536 (inline_small_functions): Update.
20537 (dump_overall_stats): Update.
20538 (dump_inline_stats): Update.
20539 * ipa-profile.c (ipa_profile_generate_summary): Update.
20540 (ipa_propagate_frequency): Update.
20541 (ipa_profile): Update.
20542 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
20543 * ipa-utils.c (ipa_merge_profiles): Update.
20544 * loop-doloop.c (doloop_modify): Update.
20545 * loop-unroll.c (report_unroll): Update.
20546 (unroll_loop_runtime_iterations): Update.
20547 * lto-cgraph.c (lto_output_edge): Update.
20548 (lto_output_node): Update.
20549 (input_node): Update.
20550 (input_edge): Update.
20551 (merge_profile_summaries): Update.
20552 * lto-streamer-in.c (input_cfg): Update.
20553 * lto-streamer-out.c (output_cfg): Update.
20554 * mcf.c (create_fixup_graph): Update.
20555 (adjust_cfg_counts): Update.
20556 (sum_edge_counts): Update.
20557 * modulo-sched.c (sms_schedule): Update.
20558 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
20559 * predict.c (maybe_hot_count_p): Update.
20560 (probably_never_executed): Update.
20561 (dump_prediction): Update.
20562 (combine_predictions_for_bb): Update.
20563 (propagate_freq): Update.
20564 (handle_missing_profiles): Update.
20565 (counts_to_freqs): Update.
20566 (rebuild_frequencies): Update.
20567 (force_edge_cold): Update.
20568 * predict.h: Include profile-count.h
20569 (maybe_hot_count_p, counts_to_freqs): UPdate.
20570 * print-rtl-function.c: Do not include cfg.h
20571 * print-rtl.c: Include basic-block.h
20572 * profile-count.c: New file.
20573 * profile-count.h: New file.
20574 * profile.c (is_edge_inconsistent): Update.
20575 (correct_negative_edge_counts): Update.
20576 (is_inconsistent): Update.
20577 (set_bb_counts): Update.
20578 (read_profile_edge_counts): Update.
20579 (compute_frequency_overlap): Update.
20580 (compute_branch_probabilities): Update; Initialize and deinitialize
20582 (branch_prob): Update.
20583 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
20584 (edge_gcov_count): New.
20585 (bb_gcov_count): New.
20586 * shrink-wrap.c (try_shrink_wrapping): Update.
20587 * tracer.c (better_p): Update.
20588 * trans-mem.c (expand_transaction): Update.
20589 (ipa_tm_insert_irr_call): Update.
20590 (ipa_tm_insert_gettmclone_call): Update.
20591 * tree-call-cdce.c: Update.
20592 * tree-cfg.c (gimple_duplicate_sese_region): Update.
20593 (gimple_duplicate_sese_tail): Update.
20594 (gimple_account_profile_record): Update.
20595 (execute_fixup_cfg): Update.
20596 * tree-inline.c (copy_bb): Update.
20597 (copy_edges_for_bb): Update.
20598 (initialize_cfun): Update.
20599 (freqs_to_counts): Update.
20600 (copy_cfg_body): Update.
20601 (expand_call_inline): Update.
20602 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
20603 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
20604 (try_unroll_loop_completely): Update.
20605 (try_peel_loop): Update.
20606 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
20607 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
20608 * tree-ssa-loop-split.c (connect_loops): Update.
20609 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
20610 * tree-ssa-reassoc.c (branch_fixup): Update.
20611 * tree-ssa-tail-merge.c (replace_block_by): Update.
20612 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
20613 (compute_path_counts): Update.
20614 (update_profile): Update.
20615 (recompute_probabilities): Update.
20616 (update_joiner_offpath_counts): Update.
20617 (estimated_freqs_path): Update.
20618 (freqs_to_counts_path): Update.
20619 (clear_counts_path): Update.
20620 (ssa_fix_duplicate_block_edges): Update.
20621 (duplicate_thread_path): Update.
20622 * tree-switch-conversion.c (case_bit_test_cmp): Update.
20623 (struct switch_conv_info): Update.
20624 * tree-tailcall.c (decrease_profile): Update.
20625 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
20626 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
20627 * value-prof.c (check_counter): Update.
20628 (gimple_divmod_fixed_value): Update.
20629 (gimple_mod_pow2): Update.
20630 (gimple_mod_subtract): Update.
20631 (gimple_ic_transform): Update.
20632 (gimple_stringop_fixed_value): Update.
20633 * value-prof.h (gimple_ic): Update.
20635 2017-06-02 Carl Love <cel@us.ibm.com>
20637 * config/rs6000/rs6000-c: Add support for built-in functions
20638 vector double vec_doublee (vector signed int);
20639 vector double vec_doublee (vector unsigned int);
20640 vector double vec_doublee (vector float);
20641 vector double vec_doubleh (vector signed int);
20642 vector double vec_doubleh (vector unsigned int);
20643 vector double vec_doubleh (vector float);
20644 vector double vec_doublel (vector signed int);
20645 vector double vec_doublel (vector unsigned int);
20646 vector double vec_doublel (vector float);
20647 vector double vec_doubleo (vector signed int);
20648 vector double vec_doubleo (vector unsigned int);
20649 vector double vec_doubleo (vector float);.
20650 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
20651 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
20653 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
20654 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
20655 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
20657 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
20658 vec_doublel, vec_doubleh.
20659 * doc/extend.texi: Update the built-in documentation file for the
20660 new built-in functions.
20662 2017-06-02 David Malcolm <dmalcolm@redhat.com>
20665 * ipa-inline-analysis.c (free_growth_caches): Set
20666 edge_removal_hook_holder to NULL after removing it.
20668 2017-06-02 Sudakshina Das <sudi.das@arm.com>
20670 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
20671 comparision with zero.
20673 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
20674 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
20675 for early expansion of vec_min and vec_max builtins.
20676 (builtin_function_type): Add min/max unsigned variants to those
20677 identified as having unsigned arguments.
20679 2017-06-02 Olivier Hainque <hainque@adacore.com>
20681 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
20683 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20685 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
20686 Use VALL_F16 iterator rather than VALL.
20688 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20690 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
20691 Emit CBNZ inside loop when doing a strong exchange and comparing
20692 against zero. Generate the CC flags after the loop.
20694 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
20696 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
20697 (dl_section_ref): New.
20698 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
20699 On AIX, append an expression to subtract the size of the
20700 section length to dl_section_ref.
20702 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
20704 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
20705 for early expansion of vector absolute builtins.
20707 2017-06-02 Richard Biener <rguenther@suse.de>
20709 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
20710 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
20712 2017-06-02 Richard Biener <rguenther@suse.de>
20714 PR tree-optimization/80948
20715 * tree-tailcall.c (find_tail_calls): Track stmts to move in
20716 stmt order as well.
20718 2017-06-02 Richard Biener <rguenther@suse.de>
20720 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
20722 * tree-vect-stmts.c (process_use): Do not mark backedge defs
20723 for inductions as relevant.
20725 2017-06-02 Richard Biener <rguenther@suse.de>
20727 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
20728 (vectorizable_induction): ... this. Remove dead code.
20730 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
20732 * builtins. (expand_builtin_alloca): Remove second parameter and
20733 infer its value from the first parameter instead.
20734 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
20736 2017-06-02 Jakub Jelinek <jakub@redhat.com>
20738 PR rtl-optimization/80903
20739 * loop-doloop.c (add_test): Unshare sequence.
20741 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
20743 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
20745 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
20747 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
20749 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
20750 xlogue_layout::get_instance, logue_layout::xlogue_layout,
20751 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
20752 (xlogue_layout::get_stub_rtx): Make static.
20753 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
20754 (xlogue_layout::compute_stub_managed_regs): Rename to...
20755 (xlogue_layout::count_stub_managed_regs): ...this.
20756 (xlogue_layout::is_stub_managed_reg): New function.
20757 (xlogue_layout::m_stub_names): Rename to...
20758 (xlogue_layout::s_stub_names): ...this, make static.
20759 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
20760 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
20761 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
20762 xlogue_layout::s_stub_names): Instantiate statics.
20763 (stub_managed_regs): Remove.
20764 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
20765 (disable_call_ms2sysv_xlogues): Rename to...
20766 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
20767 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
20769 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
20770 change after reload_completed.
20771 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
20773 (ix86_expand_prologue): Likewise.
20774 (ix86_expand_epilogue): Likewise.
20775 (ix86_expand_split_stack_prologue): Likewise.
20776 (ix86_compute_frame_layout): Remove frame parameter ...
20777 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
20778 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
20780 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
20781 (ix86_frame): Move from here ...
20782 * config/i386/i386.h (ix86_frame): ... to here.
20783 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
20784 complete ix86_frame data structure instead. Remove some_ld_name.
20786 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
20788 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
20789 symbols that hold a DECL_VALUE_EXPR.
20791 2017-06-01 Martin Jambor <mjambor@suse.cz>
20793 PR tree-optimization/80898
20794 * tree-sra.c (process_subtree_disqualification): Removed.
20795 (disqualify_candidate): Do not acll
20796 process_subtree_disqualification.
20797 (subtree_mark_written_and_enqueue): New function.
20798 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
20799 RHS has been disqualified and re-queue LHS if necessary. Apart
20800 from that, ignore disqualified RHS.
20802 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20804 * config/s390/s390.c (s390_emit_epilogue): Disable early return
20805 address fetch for z10 or later.
20807 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20809 * config/arc/arc.md (tst_movb): Add guard when splitting.
20811 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20813 * config/arc/arc.c (arc_can_eliminate): Test against
20814 arc_frame_pointer_needed.
20816 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20818 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
20819 to prevent store reordering.
20820 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
20821 (type): Add block type.
20822 (stack_tie): Define special instruction to be used in
20825 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20827 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
20828 constraint. It is not valid for the pattern.
20829 (noncommutative_binary_comparison): Likewise.
20831 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20833 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
20836 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20838 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
20839 be used by the reg-alloc.
20841 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20843 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
20844 reg-alloc when having mul64 or mul32x16 instructions.
20845 (mulsidi3): Likewise.
20846 (umulsidi3): Likewise.
20847 (mulsi32x16): New pattern.
20848 (mulsi64): Likewise.
20849 (mulsidi64): Likewise.
20850 (umulsidi64): Likewise.
20851 (MUL32x16_REG): Define.
20852 (mul64_600): Use MUL32x16_REG.
20853 (mac64_600): Likewise.
20854 (umul64_600): Likewise.
20855 (umac64_600): Likewise.
20857 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20859 * config/arc/arc.md (mulsi3_700): Make it commutative.
20861 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
20863 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
20865 (*sign_extendsidi2_insn): Likewise for movstosw.
20867 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
20869 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
20870 the type of the input discriminant value. Convert the
20871 discriminant value of signedness vary.
20873 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
20875 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
20876 Add to -Wall section.
20878 2017-06-01 Richard Biener <rguenther@suse.de>
20880 PR middle-end/66313
20881 * fold-const.c (fold_plusminus_mult_expr): If the factored
20882 factor may be zero use a wrapping type for the inner operation.
20883 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
20884 and handle moved defs.
20885 (process_assignment): Properly guard the unary op case. Return a
20886 tri-state indicating that moving the stmt before the call may allow
20887 to continue. Pass through to_move.
20888 (find_tail_calls): Handle moving unrelated defs before
20891 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
20894 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
20895 splitter result in the canonical way.
20897 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
20899 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
20900 also for 32bit target. Update insn attributes.
20901 (zero-extendsidi2 splitter): Allow all registers for operand 1.
20903 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
20905 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
20906 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
20907 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
20908 (_mm_maskz_min_ss): New intrinsics.
20910 2017-05-31 Martin Liska <mliska@suse.cz>
20912 * tree-vect-loop.c (vect_create_epilog_for_reduction):
20913 Change comment style to one we normally use.
20914 (vectorizable_reduction): Likewise.
20915 (vectorizable_induction): Likewise.
20916 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
20917 (vectorizable_call): Likewise.
20918 (vectorizable_simd_clone_call): Likewise.
20919 (vectorizable_conversion): Likewise.
20920 (vectorizable_assignment): Likewise.
20921 (vectorizable_shift): Likewise.
20922 (vectorizable_operation): Likewise.
20923 (vectorizable_store): Likewise.
20924 (vectorizable_load): Likewise.
20925 * tree-vectorizer.h: Likewise.
20927 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
20929 * passes.c (emergency_dump_function): New.
20930 * tree-pass.h (emergency_dump_function): Declare.
20931 * plugin.c (plugins_internal_error_function): Remove.
20932 * plugin.h (plugins_internal_error_function): Remove declaration.
20933 * toplev.c (internal_error_function): New static function. Use it...
20934 (general_init): ...here.
20936 2017-05-31 Graham Markall <graham.markall@embecosm.com>
20938 * config/arc/arc.c (arc_print_operand): Handle constant operands.
20939 (arc_rtx_costs): Add costs for new patterns.
20940 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
20941 * config/arc/predicates.md: Add _1_2_3_operand predicate.
20943 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
20945 * tree-ssa-strlen.c (get_next_strinfo): New function.
20946 (get_stridx_plus_constant): Use it.
20947 (zero_length_string): Likewise.
20948 (adjust_related_strinfos): Likewise.
20949 (adjust_last_stmt): Likewise.
20951 2017-05-31 Richard Biener <rguenther@suse.de>
20954 * config/i386/i386.c (ix86_expand_builtin): Remove assert
20955 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
20957 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
20959 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
20960 loop_vinfo argument and use of dependence distance vectors.
20961 Check instead whether the two references differ only in their
20962 initial value and assume that they have the same alignment if the
20963 difference is a multiple of the vector alignment.
20964 (vect_analyze_data_refs_alignment): Update call accordingly.
20966 2017-05-31 Martin Liska <mliska@suse.cz>
20969 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
20971 2017-05-31 Bin Cheng <bin.cheng@arm.com>
20973 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
20974 (create_intersect_range_checks): Move from ...
20975 * tree-data-ref.c (create_intersect_range_checks_index)
20976 (create_intersect_range_checks): ... to here.
20977 (create_runtime_alias_checks): New function factored from ...
20978 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
20979 here. Call above function.
20980 * tree-data-ref.h (create_runtime_alias_checks): New function.
20982 2017-05-31 Bin Cheng <bin.cheng@arm.com>
20984 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
20985 segment length for dr_b and compute it in wide_int.
20987 2017-05-31 Richard Biener <rguenther@suse.de>
20989 PR tree-optimization/80906
20990 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
20991 and pass through iv_map.
20992 (copy_bb_and_scalar_dependences): Adjust.
20993 (translate_pending_phi_nodes): Likewise.
20994 (copy_loop_close_phi_args): Handle code-generating IVs instead
20997 2017-05-30 David Malcolm <dmalcolm@redhat.com>
20999 * diagnostic-color.c (color_dict): Add "type-diff".
21000 (parse_gcc_colors): Update comment.
21001 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
21002 -fdiagnostics-show-template-tree and -fno-elide-type.
21003 (GCC_COLORS): Add type-diff to example.
21005 (-fdiagnostics-show-template-tree): New.
21006 (-fno-elide-type): New.
21007 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
21008 the pp_format_decoder callback. Call any m_format_postprocessor's
21010 (pretty_printer::pretty_printer): Initialize
21011 m_format_postprocessor.
21012 (pretty_printer::~pretty_printer): Delete any
21013 m_format_postprocessor.
21014 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
21015 (class format_postprocessor): New class.
21016 (struct pretty_printer::format_decoder): Document the new parameters.
21017 (struct pretty_printer::m_format_postprocessor): New field.
21018 * tree-diagnostic.c (default_tree_printer): Update for new
21019 bool and const char ** params.
21020 * tree-diagnostic.h (default_tree_printer): Likewise.
21022 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
21024 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
21025 (lwa_operand): Delete rs6000_gen_cell_microcode test.
21026 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21027 rs6000_gen_cell_microcode code.
21028 (rs6000_final_prescan_insn): Delete.
21029 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
21030 "warn-cell-microcode" entries.
21031 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
21032 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
21033 throughout. Change cc_reg_not_micro_cr0_operand to
21034 cc_reg_not_cr0_operand throughout.
21035 (*extendhi<mode>2_noload): Delete.
21036 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
21037 (mwarn-cell-microcode): Delete.
21038 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
21039 -mgen-cell-microcode and -mwarn-cell-microcode.
21041 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
21044 * config/i386/constraints.md (Yd): New constraint.
21046 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
21047 and (?Yd, r) alternatives. Update insn attributes.
21048 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
21049 and (?*Yd, r) alternatives. Update insn attributes.
21050 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
21052 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
21054 * gimplify.c (gimplify_modify_expr): Don't create a
21055 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
21058 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
21060 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
21062 2017-05-30 Richard Biener <rguenther@suse.de>
21064 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
21065 and reduc_def fields.
21066 (STMT_VINFO_REDUC_TYPE): New define.
21067 (STMT_VINFO_REDUC_DEF): Likewise.
21068 (vect_force_simple_reduction): Adjust prototype.
21069 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
21070 (vect_is_simple_reduction): Remove check_reduction argument.
21071 (vect_force_simple_reduction): Adjust and set
21072 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21073 (vectorizable_reduction): Do not re-do reduction analysis
21074 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21075 * tree-parloops.c (gather_scalar_reductions): Adjust.
21077 2017-05-30 Richard Biener <rguenther@suse.de>
21079 PR middle-end/80901
21080 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
21083 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
21085 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
21086 Introduce unknown_misalignment parameter and remove vf.
21087 (vect_peeling_hash_get_lowest_cost):
21088 Pass unknown_misalignment parameter.
21089 (vect_enhance_data_refs_alignment):
21090 Fix unsupportable data ref treatment.
21092 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21094 * tree-vect-data-refs.c (vect_get_data_access_cost):
21095 Workaround for SLP handling.
21096 (vect_enhance_data_refs_alignment):
21097 Compute costs for doing no peeling at all, compare to the best
21098 peeling costs so far and avoid peeling if cheaper.
21100 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21102 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
21103 Return peeling info and set costs to zero for unlimited cost
21105 (vect_enhance_data_refs_alignment): Also inspect all datarefs
21106 with unknown misalignment. Compute and costs for unknown
21107 misalignment, compare them to the costs for known misalignment
21108 and choose the cheapest for peeling.
21110 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21112 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
21113 (vect_get_peeling_costs_all_drs): Create function.
21114 (vect_peeling_hash_get_lowest_cost):
21115 Use vect_get_peeling_costs_all_drs.
21116 (vect_peeling_supportable): Create function.
21117 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
21119 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21121 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
21122 DR_HAS_NEGATIVE_STEP.
21123 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
21124 (vect_enhance_data_refs_alignment): Use.
21125 (vect_duplicate_ssa_name_ptr_info): Use.
21126 * tree-vectorizer.h (dr_misalignment): Use.
21127 (known_alignment_for_access_p): Use.
21129 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21132 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
21134 (has_section_name): New function.
21136 2017-05-30 Martin Liska <mliska@suse.cz>
21139 * auto-profile.c (get_function_decl_from_block): Fix
21142 2017-05-30 Richard Biener <rguenther@suse.de>
21144 PR middle-end/80876
21145 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
21147 2017-05-30 Martin Liska <mliska@suse.cz>
21149 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
21150 * dumpfile.h (struct dump_file_info): Remove ctors.
21152 2017-05-30 Martin Liska <mliska@suse.cz>
21154 * predict.def: Fix GNU coding style.
21156 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21158 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
21159 Mark 'to' argument with ATTRIBUTE_UNUSED.
21161 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21163 * config/xtensa/xtensa.c (xtensa_emit_call): Use
21164 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
21165 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
21168 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
21170 * doc/install.texi (Options specification): Restore entry of
21171 --enable-sjlj-exceptions.
21173 2017-05-27 Michael Eager <eager@eagercon.com>
21176 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
21178 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
21180 * config/microblaze/microblaze.h
21181 (FIXED_REGISTERS): Update in macro.
21182 (CALL_USED_REGISTERS): Update in macro.
21184 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
21186 * doc/install.texi: Add links to macOS binary distributions.
21188 2017-05-27 Jakub Jelinek <jakub@redhat.com>
21192 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21194 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
21196 2017-05-26 Martin Liska <mliska@suse.cz>
21198 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
21200 2017-05-26 Martin Liska <mliska@suse.cz>
21202 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
21203 always leading ';; '.
21204 (dump_bb_info): Likewise.
21205 (brief_dump_cfg): Likewise.
21206 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
21207 * dumpfile.c: Remove usage of TDF_VERBOSE.
21208 * dumpfile.h (enum dump_kind): Likewise.
21209 (dump_gimple_bb_header): Do not use TDF_COMMENT.
21210 * print-tree.c (debug_verbose): Remove.
21211 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
21212 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
21213 * tree-diagnostic.c (default_tree_printer): Replace
21214 TDF_DIAGNOSTIC with TDF_SLIM.
21216 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21218 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
21219 in parameter loop, rather than loop_vinfo.
21220 (create_intersect_range_checks): Ditto.
21221 (vect_create_cond_for_alias_checks): Update call to above functions.
21223 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21225 PR tree-optimization/80815
21226 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
21227 for merging runtime alias checks. Handle negative DR_STEPs.
21229 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21231 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
21233 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
21234 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
21235 out code pruning runtime alias checks.
21236 * tree-data-ref.c (prune_runtime_alias_test_list): New function
21237 factored out from above.
21238 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
21240 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
21242 (prune_runtime_alias_test_list): New decalaration.
21244 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21246 * tree-vect-data-refs.c (compare_tree): Rename and move ...
21247 * tree-data-ref.c (data_ref_compare_tree): ... to here.
21248 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
21249 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
21250 (operator==, comp_dr_with_seg_len_pair): Ditto.
21251 (vect_prune_runtime_alias_test_list): Ditto.
21253 2017-05-26 Martin Liska <mliska@suse.cz>
21256 * params.def: Bound partial-inlining-entry-probability param.
21258 2017-05-26 Marek Polacek <polacek@redhat.com>
21261 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
21264 2017-05-26 Richard Biener <rguenther@suse.de>
21266 PR tree-optimization/80842
21267 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
21270 2017-05-26 Richard Biener <rguenther@suse.de>
21272 PR tree-optimization/80844
21273 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
21275 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
21277 * doc/md.texi (Machine Constraints): Update x86 family
21278 machine constraints section to match 'config/i386/constraints.md'.
21280 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
21282 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
21284 2017-05-25 Nathan Sidwell <nathan@acm.org>
21286 * doc/invoke.texi (--enable-languages): Update documentation.
21288 2017-05-25 Martin Liska <mliska@suse.cz>
21290 * dumpfile.c: Add TDF_FOLDING.
21291 * dumpfile.h (enum dump_kind): Likewise.
21292 * genmatch.c (dt_simplify::gen_1): Use it.
21294 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21296 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
21298 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21300 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
21301 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
21303 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21305 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
21306 * match.pd (X == C): Rewrite it here.
21307 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
21308 with_certain_nonzero_bits2): New predicates.
21309 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
21311 2017-05-24 Nathan Sidwell <nathan@acm.org>
21313 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
21316 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
21319 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
21321 * config/powerpcspe: New port. Files are copied from the rs6000
21322 port, with "rs6000" in filenames replaced by "powerpcspe".
21324 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
21326 PR rtl-optimization/80754
21327 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
21329 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
21331 * config/sparc/sparc.md (length): Return the correct value for -mflat
21332 sibcalls to match output_sibcall.
21334 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
21338 * config/rs6000/rs6000.c (struct machine_function): Add new field
21340 (rs6000_get_separate_components): Init that field, use it.
21341 (rs6000_components_for_bb): Use the field.
21343 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
21345 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
21347 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
21349 PR middle-end/80823
21350 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
21352 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21355 * config/s390/s390.c (s390_check_qrst_address): Check incoming
21356 address against address_operand predicate.
21357 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
21359 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
21361 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
21362 parameters passed indirectly.
21364 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
21366 * config/i386/i386.md (*movdi_internal): Remove SSE4
21367 alternative 18 (?r, *v). Update insn attributes.
21368 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
21369 Update insn attributes.
21370 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
21371 Update insn attributes.
21372 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
21373 alternative 1 (r, v). Remove isa attribute.
21374 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
21375 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
21376 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
21378 2017-05-23 Tom de Vries <tom@codesourcery.com>
21380 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
21383 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21385 * cgraphunit.c (symbol_table::process_new_functions): Update.
21386 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
21387 (inline_generate_summary): Rename to ...
21388 (ipa_fn_summary_generate): ... this one.
21389 (inline_read_summary): Rename to ...
21390 (ipa_fn_summary_read): ... this one.
21391 (inline_write_summary): Rename to ...
21392 (ipa_fn_summary_write): ... this one.
21393 (inline_free_summary): Rename to ...
21394 (ipa_free_fn_summary): ... this one.
21395 (pass_data_local_fn_summary, pass_local_fn_summary,
21396 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
21397 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
21398 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
21399 make_pass_ipa_fn_summary): New.
21400 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
21401 inline_write_summary, inline_free_summary): Remove.
21402 (ipa_free_fn_summary) : New.
21403 * ipa-inline.c (ipa_inline): Update.
21404 (pass_ipa_inline): Do not generate summaries.
21405 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
21407 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
21408 and add pass_ipa_fn_summary.
21409 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
21411 (make_pass_inline_parameters): Remove.
21413 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
21415 * omp-low.c (struct omp_context): Remove "default_kind" member.
21418 * omp-offload.c (execute_oacc_device_lower): Remove the
21419 parallelism dimensions function attributes for unparallelized
21420 OpenACC kernels constructs.
21422 2017-05-23 Martin Liska <mliska@suse.cz>
21424 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
21426 (cgraph_edge::make_speculative): Likewise.
21427 (cgraph_edge::resolve_speculation): Likewise.
21428 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
21429 (cgraph_node::dump): Likewise.
21430 * cgraph.h: Likewise.
21431 * cgraphunit.c (analyze_functions): Likewise.
21432 (symbol_table::compile): Likewise.
21433 * ipa-cp.c (print_all_lattices): Likewise.
21434 (determine_versionability): Likewise.
21435 (initialize_node_lattices): Likewise.
21436 (ipcp_verify_propagated_values): Likewise.
21437 (estimate_local_effects): Likewise.
21438 (update_profiling_info): Likewise.
21439 (create_specialized_node): Likewise.
21440 (perhaps_add_new_callers): Likewise.
21441 (decide_about_value): Likewise.
21442 (decide_whether_version_node): Likewise.
21443 (identify_dead_nodes): Likewise.
21444 (ipcp_store_bits_results): Likewise.
21445 * ipa-devirt.c (dump_targets): Likewise.
21446 (ipa_devirt): Likewise.
21447 * ipa-icf.c (sem_item::dump): Likewise.
21448 (sem_function::equals): Likewise.
21449 (sem_variable::equals): Likewise.
21450 (sem_item_optimizer::read_section): Likewise.
21451 (sem_item_optimizer::execute): Likewise.
21452 (congruence_class::dump): Likewise.
21453 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
21454 (dump_inline_summary): Likewise.
21455 (estimate_node_size_and_time): Likewise.
21456 (inline_analyze_function): Likewise.
21457 * ipa-inline-transform.c (inline_call): Likewise.
21458 * ipa-inline.c (report_inline_failed_reason): Likewise.
21459 (want_early_inline_function_p): Likewise.
21460 (edge_badness): Likewise.
21461 (update_edge_key): Likewise.
21462 (inline_small_functions): Likewise.
21463 * ipa-profile.c (ipa_profile): Likewise.
21464 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
21465 (ipa_make_edge_direct_to_target): Likewise.
21466 (remove_described_reference): Likewise.
21467 (ipa_impossible_devirt_target): Likewise.
21468 (propagate_controlled_uses): Likewise.
21469 (ipa_print_node_params): Likewise.
21470 (ipcp_transform_function): Likewise.
21471 * ipa-pure-const.c (pure_const_read_summary): Likewise.
21472 (propagate_pure_const): Likewise.
21473 * ipa-reference.c (generate_summary): Likewise.
21474 (read_write_all_from_decl): Likewise.
21475 (propagate): Likewise.
21476 (ipa_reference_read_optimization_summary): Likewise.
21477 * ipa-utils.c (ipa_merge_profiles): Likewise.
21478 * ipa.c (walk_polymorphic_call_targets): Likewise.
21479 (symbol_table::remove_unreachable_nodes): Likewise.
21480 (ipa_single_use): Likewise.
21481 * passes.c (execute_todo): Likewise.
21482 * predict.c (drop_profile): Likewise.
21483 * symtab.c (symtab_node::get_dump_name): New function.
21484 (symtab_node::dump_name): Likewise.
21485 (symtab_node::dump_asm_name): Likewise.
21486 (symtab_node::dump_references): Likewise.
21487 (symtab_node::dump_referring): Likewise.
21488 (symtab_node::dump_base): Likewise.
21489 (symtab_node::debug_symtab): Likewise.
21490 * tree-sra.c (convert_callers_for_node): Likewise.
21491 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
21492 * value-prof.c (init_node_map): Likewise.
21494 2017-05-23 Martin Liska <mliska@suse.cz>
21496 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
21497 and symtab_node::debug_symtab to symbol_table::debug.
21498 * cgraphunit.c (analyze_functions): Use the renamed function.
21499 (symbol_table::compile): Likewise.
21500 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
21501 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
21502 * passes.c (execute_todo): Likewise.
21503 * symtab.c (symbol_table::dump): New function.
21504 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
21506 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21508 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
21509 that nonconst implies exec.
21511 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21513 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
21514 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
21515 (inline_edge_summary_vec): Turn into ...
21516 (ipa_call_summaries): ... this one.
21517 (redirect_to_unreachable, edge_set_predicate,
21518 evaluate_properties_for_edge, inline_summary_alloc,
21519 reset_ipa_call_summary, reset_inline_summary,
21520 inline_summary_t::duplicate): Update.
21521 (inline_edge_duplication_hook): Turn to ...
21522 (ipa_call_summary_t::duplicate): ... this one.
21523 (inline_edge_removal_hook): Turn to ...
21524 (ipa_call_summary_t::remove): ... this one.
21525 (dump_inline_edge_summary): Turn to ...
21526 (dump_ipa_call_summary): ... this one.
21527 (estimate_function_body_sizes): Update.
21528 (inline_update_callee_summaries): Update.
21529 (remap_edge_change_prob): Update.
21530 (remap_edge_summaries): Update.
21531 (inline_merge_summary): Update.
21532 (do_estimate_edge_time): Update.
21533 (inline_generate_summary): Update.
21534 (inline_read_section): Update.
21535 (inline_read_summary): Update.
21536 (inline_free_summary): Update.
21537 * ipa-inline.c (can_inline_edge_p): Update.
21538 (compute_inlined_call_time): Update.
21539 (want_inline_small_function_p): Update.
21540 (edge_badness): Update.
21541 (early_inliner): Update.
21542 * ipa-inline.h (inline_edge_summary): Turn to ...
21543 (ipa_call_summary): ... this one.
21544 (ipa_call_summary_t): New class.
21545 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
21546 (ipa_call_summaries): New.
21547 (inline_edge_summary): Remove.
21548 (estimate_edge_growth): Update.
21549 * ipa-profile.c (ipa_propagate_frequency_1): Update.
21550 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21551 * ipa-split.c (execute_split_functions): Update.
21552 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
21554 2017-05-23 Tom de Vries <tom@codesourcery.com>
21556 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
21557 attributes): Document rdrand effective target.
21559 2017-05-23 Tom de Vries <tom@codesourcery.com>
21561 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
21562 attributes): Sort alphabetically.
21564 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
21566 * config/avr/genmultilib.awk: Use gsub instead of gensub.
21568 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
21571 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
21572 V2DF/V2DI splat into two separate patterns, one that handles
21573 registers, and the other that only handles memory. Drop support
21574 for splatting from a GPR on ISA 2.07 and then splitting the
21575 splat into direct move and splat.
21576 (vsx_splat_<mode>_reg): Likewise.
21577 (vsx_splat_<mode>_mem): Likewise.
21579 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
21581 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
21583 2017-05-22 Jakub Jelinek <jakub@redhat.com>
21585 PR middle-end/80809
21586 * omp-low.c (finish_taskreg_remap): New function.
21587 (finish_taskreg_scan): If unit size of ctx->record_type
21588 is non-constant, unshare the size expression and replace
21589 decls in it with possible outer var refs.
21591 PR middle-end/80809
21592 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
21593 GOVD_SHARED rather than GOVD_PRIVATE with it.
21594 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
21595 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
21597 PR middle-end/80853
21598 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
21599 as last argument to build_outer_var_ref for pointer bases of array
21600 section reductions.
21602 2017-05-19 Martin Sebor <msebor@redhat.com>
21604 * print-tree.c (print_node): Print DECL_READ_P flag.
21606 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21608 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
21609 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
21610 * cgraph.c: Likewise.
21611 * cgraphunit.c: Likewise.
21612 * gengtype.c: Likewise.
21613 * ipa-cp.c: Likewise.
21614 * ipa-devirt.c: Likewise.
21615 * ipa-icf.c: Likewise.
21616 * ipa-predicate.c: Likewise.
21617 * ipa-profile.c: Likewise.
21618 * ipa-prop.c: Likewise.
21619 * ipa-split.c: Likewise.
21621 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
21622 edge_predicate_pool, dump_inline_hints,
21623 inline_summary::account_size_time, redirect_to_unreachable,
21624 edge_set_predicate, set_hint_predicate,
21625 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
21626 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
21627 inline_summary_t::remove, remap_hint_predicate_after_duplication,
21628 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
21629 ipa_call_summary_t::remove, initialize_growth_caches,
21630 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
21631 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
21632 mark_modified, unmodified_parm_1, unmodified_parm,
21633 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
21634 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
21635 compute_bb_predicates, will_be_nonconstant_expr_predicate,
21636 will_be_nonconstant_predicate, record_modified_bb_info,
21637 get_minimal_bb, record_modified, param_change_prob,
21638 phi_result_unknown_predicate, predicate_for_phi_result,
21639 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
21640 estimate_function_body_sizes, compute_inline_parameters,
21641 compute_inline_parameters_for_curren, pass_data_inline_parameters,
21642 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
21643 inline_update_callee_summaries, remap_edge_change_prob,
21644 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
21645 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
21646 inline_analyze_function, inline_summary_t::insert,
21647 inline_generate_summary, read_ipa_call_summary, inline_read_section,
21648 inline_read_summary, write_ipa_call_summary, inline_write_summary,
21649 inline_free_summary): Move to ipa-fnsummary.h
21650 (predicate_t): Remove.
21651 * ipa-fnsummary.c: New file.
21652 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
21653 (enum inline_hints_vals, inline_hints, agg_position_info,
21654 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
21655 inline_summaries, ipa_call_summary, ipa_call_summary_t,
21656 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
21657 dump_inline_summary, dump_inline_hints, inline_generate_summary,
21658 inline_read_summary, inline_write_summary, inline_free_summary,
21659 inline_analyze_function, initialize_inline_failed,
21660 inline_merge_summary, inline_update_overall_summary,
21661 compute_inline_parameters): Move to ipa-fnsummary.h
21662 * ipa-fnsummary.h: New file.
21663 * ipa-inline-transform.h: Include ipa-inline.h.
21664 * ipa-inline.c: LIkewise.
21666 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21668 * ipa-inline.c (edge_badness): Use inlined_time instead of
21669 inline_summaries->get.
21671 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21673 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
21675 2017-05-22 Nathan Sidwell <nathan@acm.org>
21677 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
21678 (fdump-lang): Document 'raw' option.
21679 * dumpfile.h (TDI_tu): Delete.
21680 * dumpfile.c (dump_files): Remove translation-unit.
21681 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
21683 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
21685 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
21686 command option from $(AWK) call.
21687 * config/avr/genmultilib.awk: Simplify and rewrite so that it
21688 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
21689 [FORMAT]: Remove handling of variable.
21690 * config/avr/t-multilib: Regenerate.
21692 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21694 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
21696 (dump_inline_summary): Do not print self_time.
21697 (estimate_function_body_sizes): Do not set self_time.
21698 (compute_inline_parameters): Likewise.
21699 (inline_read_section, inline_write_summary): Do not stream self_time.
21700 * ipa-inline.h (inline_summary): Drop self_time.
21702 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21704 * ipa-inline-analysis.c (account_size_time): Rename to ...
21705 (inline_summary::account_size_time): ... this one.
21706 (reset_ipa_call_summary): Turn to ...
21707 (ipa_call_summary::reset): ... this one.
21708 (reset_inline_summary): Turn to ...
21709 (inline_summary::reset): ... this one.
21710 (inline_summary_t::remove): Update.
21711 (inline_summary_t::duplicate): Update.
21712 (ipa_call_summary_t::remove): Update.
21713 (dump_inline_summary): Update.
21714 (estimate_function_body_sizes): Update.
21715 (compute_inline_parameters): Update.
21716 (estimate_node_size_and_time): Update.
21717 (inline_merge_summary): Update.
21718 (inline_update_overall_summary): Update.
21719 (inline_read_section): Update.
21720 (inline_write_summary): Update.
21721 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
21722 add account_size_time and reset member functions.
21723 (ipa_call_summary): Add reset function.
21724 * ipa-predicate.h (predicate::operator &): Constify.
21726 2017-05-22 Richard Biener <rguenther@suse.de>
21728 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
21730 2017-05-19 Jason Merrill <jason@redhat.com>
21732 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
21734 2017-05-19 Marek Polacek <polacek@redhat.com>
21737 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
21738 TYPE_OVERFLOW_WRAPS checks.
21740 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
21742 * tree-core.h (enum omp_clause_default_kind): Add
21743 "OMP_CLAUSE_DEFAULT_PRESENT".
21744 * tree-pretty-print.c (dump_omp_clause): Handle it.
21745 * gimplify.c (enum gimplify_omp_var_data): Add
21746 "GOVD_MAP_FORCE_PRESENT".
21747 (gimplify_adjust_omp_clauses_1): Map it to
21748 "GOMP_MAP_FORCE_PRESENT".
21749 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
21751 * gimplify.c (oacc_default_clause): Clarify.
21753 2017-05-19 Nathan Sidwell <nathan@acm.org>
21755 LANG_HOOK_REGISTER_DUMPS
21756 * toplev.c (general_init): Call register dump lang hook.
21757 * doc/invoke.texi: Document -fdump-lang option family.
21758 * dumpfile.c (dump_files): Remove class dump here.
21759 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
21760 * dumpfile.h (tree_dump_index): Remove TDI_class.
21761 * langhooks-def.h (lhd_register_dumps): Declare.
21762 (LANG_HOOKS_REGISTER_DUMPS): Define.
21763 (LANG_HOOKS_INITIALIZER): Add it.
21764 * langhooks.c (lhd_register_dumps): Define.
21765 * langhooks.h (struct lang_hooks): Add register_dumps.
21767 2017-05-19 Nathan Sidwell <nathan@acm.org>
21769 * context.h (context::set_passes): New.
21770 * context.c (context::context): Do not create pass manager.
21771 * toplev.c (general_init): Create pass manager here.
21773 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
21775 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
21776 use this splitter if two add or or instructions would also work for
21777 the constant we want to generate.
21779 2017-05-19 Richard Biener <rguenther@suse.de>
21782 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
21783 predicate evaluation.
21785 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
21787 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
21789 * ipa-inline.c (want_inline_small_function_p): Do not cast to
21792 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
21794 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
21795 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
21796 (inline_edge_summary_vec): Turn into ...
21797 (ipa_call_summaries): ... this one.
21798 (redirect_to_unreachable, edge_set_predicate,
21799 evaluate_properties_for_edge, inline_summary_alloc,
21800 reset_ipa_call_summary, reset_inline_summary,
21801 inline_summary_t::duplicate): Update.
21802 (inline_edge_duplication_hook): Turn to ...
21803 (ipa_call_summary_t::duplicate): ... this one.
21804 (inline_edge_removal_hook): Turn to ...
21805 (ipa_call_summary_t::remove): ... this one.
21806 (dump_inline_edge_summary): Turn to ...
21807 (dump_ipa_call_summary): ... this one.
21808 (estimate_function_body_sizes): Update.
21809 (inline_update_callee_summaries): Update.
21810 (remap_edge_change_prob): Update.
21811 (remap_edge_summaries): Update.
21812 (inline_merge_summary): Update.
21813 (do_estimate_edge_time): Update.
21814 (inline_generate_summary): Update.
21815 (inline_read_section): Update.
21816 (inline_read_summary): Update.
21817 (inline_free_summary): Update.
21818 * ipa-inline.c (can_inline_edge_p): Update.
21819 (compute_inlined_call_time): Update.
21820 (want_inline_small_function_p): Update.
21821 (edge_badness): Update.
21822 (early_inliner): Update.
21823 * ipa-inline.h (inline_edge_summary): Turn to ...
21824 (ipa_call_summary): ... this one.
21825 (ipa_call_summary_t): New class.
21826 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
21827 (ipa_call_summaries): New.
21828 (inline_edge_summary): Remove.
21829 (estimate_edge_growth): Update.
21830 * ipa-profile.c (ipa_propagate_frequency_1): Update.
21831 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21832 * ipa-split.c (execute_split_functions): Update.
21833 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
21835 2017-05-19 Richard Biener <rguenther@suse.de>
21837 PR middle-end/80764
21838 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
21840 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
21842 * config/rs6000/rs6000.c (struct machine_function): Add field
21843 fpr_is_wrapped_separately.
21844 (rs6000_get_separate_components): Use 64 components. Handle the
21845 new FPR components.
21846 (rs6000_components_for_bb): Handle the FPR components.
21847 (rs6000_emit_prologue_components): Handle the FPR components.
21848 (rs6000_emit_epilogue_components): Handle the FPR components.
21849 (rs6000_set_handled_components): Handle the FPR components.
21850 (rs6000_emit_prologue): Don't output prologue code for those FPRs
21851 that are already separately shrink-wrapped.
21852 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
21853 that are already separately shrink-wrapped.
21855 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
21858 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
21861 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
21862 (define_peephole2 for Altivec d-form load): Add peepholes to catch
21863 cases where the register allocator uses a move and an offsettable
21864 memory operation to/from a FPR register on ISA 2.06/2.07.
21865 (define_peephole2 for Altivec d-form store): Likewise.
21867 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
21870 * config/i386/mmx.md (*mov<mode>_internal): Enable
21871 alternatives 11, 12, 13 and 14 also for 32bit targets.
21872 Remove alternatives 15, 16, 17 and 18.
21873 * config/i386/sse.md (vec_concatv2di): Change
21874 alternative (!x, *y) to (x, ?!*Yn).
21876 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
21878 * dumpfile.h (enum dump_kind): Remove stray comma.
21880 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
21882 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
21883 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
21884 predicate::num_conditions
21885 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
21886 (CHANGED): turn into predicate::changed.
21887 (agg_position_info): Move to ipa-predicate.h
21888 (add_condition, predicate::add_clause, predicate::operator &=,
21889 predicate::or_with, predicate::evaluate, predicate::probability,
21890 dump_condition, dump_clause, predicate::dump,
21891 predicate::remap_after_duplication, predicate::remap_after_inlining,
21892 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
21893 (evaluate_conditions_for_known_args): Update.
21894 (set_cond_stmt_execution_predicate): Update.
21895 * ipa-inline.h: Include ipa-predicate.h
21896 (condition, inline_param_summary, conditions, agg_position_info,
21897 predicate): Move to ipa-predicate.h
21898 * ipa-predicate.c: New file.
21899 * ipa-predicate.h: New file.
21901 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
21903 * final.c (leaf_function_p): Check we are not in a sequence.
21905 2017-05-18 Martin Liska <mliska@suse.cz>
21907 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
21908 * dumpfile.c (dump_register): Use new enum dump_kind.
21909 (get_dump_file_name): Likewise.
21910 (dump_enable_all): Likewise.
21911 (dump_switch_p_1): Likewise.
21912 (enable_rtl_dump_file): Remove usage of TDF_RTL.
21913 * dumpfile.h (enum dump_kind): New enum type.
21914 (struct dump_file_info): Create constructor and
21915 format fields and comments.
21916 * passes.c (pass_manager::register_one_dump_file):
21918 * statistics.c (statistics_early_init): Likewise.
21919 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
21920 TDF_TREE with TDF_SLIM.
21921 (gather_memory_references_ref): Likewise.
21923 2017-05-18 Martin Liska <mliska@suse.cz>
21925 * vec.h (struct vnull): Use it.
21927 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
21929 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
21930 (true_predicate, false_predicate, true_predicate_p,
21931 false_predicate_p): Remove.
21932 (single_cond_predicate, not_inlined_predicate): Turn to member function
21934 (add_condition): Update.
21935 (add_clause): Turn to...
21936 (predicate::add_clause): ... this one; update; allow passing NULL
21938 (and_predicates): Turn to ...
21939 (predicate::operator &=): ... this one.
21940 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
21941 (or_predicates): Turn to ...
21942 (predicate::or_with): ... this one.
21943 (evaluate_predicate): Turn to ...
21944 (predicate::evaluate): ... this one.
21945 (predicate_probability): Turn to ...
21946 (predicate::probability): ... this one.
21947 (dump_condition): Update.
21948 (dump_predicate): Turn to ...
21949 (predicate::dump): ... this one.
21950 (account_size_time): Update.
21951 (edge_set_predicate): Update.
21952 (set_hint_predicate): UPdate.
21953 (evaluate_conditions_for_known_args): Update.
21954 (evaluate_properties_for_edge): Update.
21955 (remap_predicate_after_duplication): Turn to...
21956 (predicate::remap_after_duplication): ... this one.
21957 (remap_hint_predicate_after_duplication): Update.
21958 (inline_summary_t::duplicate): UPdate.
21959 (dump_inline_edge_summary): Update.
21960 (dump_inline_summary): Update.
21961 (set_cond_stmt_execution_predicate): Update.
21962 (set_switch_stmt_execution_predicate): Update.
21963 (compute_bb_predicates): Update.
21964 (will_be_nonconstant_expr_predicate): Update.
21965 (will_be_nonconstant_predicate): Update.
21966 (phi_result_unknown_predicate): Update.
21967 (predicate_for_phi_result): Update.
21968 (array_index_predicate): Update.
21969 (estimate_function_body_sizes): Update.
21970 (estimate_node_size_and_time): Update.
21971 (estimate_ipcp_clone_size_and_time): Update.
21972 (remap_predicate): Rename to ...
21973 (predicate::remap_after_inlining): ... this one.
21974 (remap_hint_predicate): Update.
21975 (inline_merge_summary): Update.
21976 (inline_update_overall_summary): Update.
21977 (estimate_size_after_inlining): Update.
21978 (read_predicate): Rename to ...
21979 (predicate::stream_in): ... this one.
21980 (read_inline_edge_summary): Update.
21981 (write_predicate): Rename to ...
21982 (predicate::stream_out): ... this one.
21983 (write_inline_edge_summary): Update.
21984 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
21985 (clause_t): Turn to uint32_t
21986 (predicate): Turn to class; implement constructor and operators
21988 (size_time_entry): Update.
21989 (inline_summary): Update.
21990 (inline_edge_summary): Update.
21992 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
21994 * fold-const.c (fold_binary_loc): Move transformation...
21995 * match.pd (C - X CMP X): ... here.
21997 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
21999 * config/sparc/sparc.c (sparc_option_override): Set function
22000 alignment for -mcpu=niagara7 to 64 to match the I$ line.
22001 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
22003 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
22005 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
22007 2017-05-18 Marek Polacek <polacek@redhat.com>
22010 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
22011 (pass_ubsan::execute): Call gimple_assign_single_p instead of
22012 gimple_assign_load_p.
22014 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
22016 PR middle-end/80692
22017 * real.c (do_compare): Give decimal_do_compare preference over
22018 comparing just the signs.
22020 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
22022 * doc/md.texi (Canonicalization of Instructions): Describe the
22023 canonical form of instructions that inherently set a condition
22026 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
22028 PR middle-end/80775
22029 * tree-cfg.c: Move deletion of unreachable case statements to after
22030 the merging of consecutive case labels.
22032 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
22034 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
22035 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
22036 restoring of callee-saved registers.
22038 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
22040 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
22041 * config/visium/visium.c (single_set_and_flags): Likewise.
22042 * config/visium/visium.md (Substitutions): Likewise.
22044 2017-05-17 Martin Liska <mliska@suse.cz>
22046 * cfg.c: Introduce dump_flags_t type and
22047 use it instead of int type.
22049 * cfghooks.c: Likewise.
22050 * cfghooks.h (struct cfg_hooks): Likewise.
22051 * cfgrtl.c: Likewise.
22052 * cfgrtl.h: Likewise.
22053 * cgraph.c (cgraph_node::get_body): Likewise.
22054 * coretypes.h: Likewise.
22055 * domwalk.c: Likewise.
22056 * domwalk.h: Likewise.
22057 * dumpfile.c (struct dump_option_value_info): Likewise.
22058 (dump_enable_all): Likewise.
22059 (dump_switch_p_1): Likewise.
22060 (opt_info_switch_p): Likewise.
22061 * dumpfile.h (enum tree_dump_index): Likewise.
22062 (struct dump_file_info): Likewise.
22063 * genemit.c: Likewise.
22064 * generic-match-head.c: Likewise.
22065 * gengtype.c (open_base_files): Likewise.
22066 * gimple-pretty-print.c: Likewise.
22067 * gimple-pretty-print.h: Likewise.
22068 * graph.c (print_graph_cfg): Likewise.
22069 * graphite-scop-detection.c (dot_all_sese): Likewise.
22070 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
22071 * loop-unroll.c (report_unroll): Likewise.
22072 * passes.c (pass_manager::register_one_dump_file): Likewise.
22073 * print-tree.c: Likewise.
22074 * statistics.c: Likewise.
22075 * tree-cfg.c: Likewise.
22076 * tree-cfg.h: Likewise.
22077 * tree-dfa.c: Likewise.
22078 * tree-dfa.h: Likewise.
22079 * tree-dump.c (dump_function): Likewise.
22080 * tree-dump.h (struct dump_info): Likewise.
22081 * tree-pretty-print.c: Likewise.
22082 * tree-pretty-print.h: Likewise.
22083 * tree-ssa-live.c: Likewise.
22084 * tree-ssa-live.h: Likewise.
22085 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
22086 * tree-vect-loop.c: Likewise.
22087 * tree-vect-slp.c: Likewise.
22089 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
22090 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22092 PR tree-optimization/80457
22093 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
22094 of all arguments to a statement as scalar_to_vec operations.
22095 (vectorizable_call): Adjust call to vect_model_simple_cost for
22097 (vectorizable_conversion): Likewise.
22098 (vectorizable_assignment): Likewise.
22099 (vectorizable_shift): Likewise.
22100 (vectorizable_operation): Likewise.
22101 (vectorizable_comparison): Likewise.
22102 (vect_is_simple_cond): Record the def types for operands.
22103 (vectorizable_condition): Likewise, call vect_model_simple_cost.
22104 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
22105 for statement argument count.
22107 2017-05-16 Carl Love <cel@us.ibm.com>
22109 * config/rs6000/rs6000-c: Add support for built-in functions
22110 vector unsigned long long vec_bperm (vector unsigned long long,
22111 vector unsigned char)
22112 vector signed long long vec_mule (vector signed int,
22114 vector unsigned long long vec_mule (vector unsigned int,
22115 vector unsigned int)
22116 vector signed long long vec_mulo (vector signed int,
22118 vector unsigned long long vec_mulo (vector unsigned int,
22119 vector unsigned int)
22120 vector signed char vec_sldw (vector signed char,
22121 vector signed char,
22123 vector unsigned char vec_sldw (vector unsigned char,
22124 vector unsigned char,
22126 vector signed short vec_sldw (vector signed short,
22127 vector signed short,
22129 vector unsigned short vec_sldw (vector unsigned short,
22130 vector unsigned short,
22132 vector signed int vec_sldw (vector signed int,
22135 vector unsigned int vec_sldw (vector unsigned int,
22136 vector unsigned int,
22138 vector signed long long vec_sldw (vector signed long long,
22139 vector signed long long,
22141 vector unsigned long long vec_sldw (vector unsigned long long,
22142 vector unsigned long long,
22144 * config/rs6000/rs6000-c: Add support for built-in functions
22145 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
22146 * config/rs6000/altivec.h: Add defintion for vec_sldw.
22147 * doc/extend.texi: Update the built-in documentation for the
22148 new built-in functions.
22150 2017-05-16 Marek Polacek <polacek@redhat.com>
22154 * tree.c (save_expr): Don't fold the expression.
22156 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
22158 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
22159 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
22160 and (?*y,m). Update insn attributes.
22162 2017-05-16 Martin Liska <mliska@suse.cz>
22164 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
22165 flags argument of print_gimple_stmt, print_gimple_expr,
22166 print_generic_stmt and print_generic_expr.
22167 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
22168 * coretypes.h: Likewise.
22169 * except.c (dump_eh_tree): Likewise.
22170 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
22171 * gimple-pretty-print.h: Likewise.
22172 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
22173 (backprop::push_to_worklist): Likewise.
22174 (backprop::pop_from_worklist): Likewise.
22175 (backprop::process_use): Likewise.
22176 (backprop::intersect_uses): Likewise.
22177 (note_replacement): Likewise.
22178 * gimple-ssa-store-merging.c
22179 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
22180 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
22181 (pass_store_merging::execute): Likewise.
22182 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
22183 (ssa_base_cand_dump_callback): Likewise.
22184 (dump_incr_vec): Likewise.
22185 (replace_refs): Likewise.
22186 (replace_mult_candidate): Likewise.
22187 (create_add_on_incoming_edge): Likewise.
22188 (create_phi_basis): Likewise.
22189 (insert_initializers): Likewise.
22190 (all_phi_incrs_profitable): Likewise.
22191 (introduce_cast_before_cand): Likewise.
22192 (replace_one_candidate): Likewise.
22193 * gimplify.c (gimplify_expr): Likewise.
22194 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
22195 (set_rename): Likewise.
22196 (rename_uses): Likewise.
22197 (copy_loop_phi_nodes): Likewise.
22198 (add_close_phis_to_merge_points): Likewise.
22199 (copy_loop_close_phi_args): Likewise.
22200 (copy_cond_phi_args): Likewise.
22201 (graphite_copy_stmts_from_block): Likewise.
22202 (translate_pending_phi_nodes): Likewise.
22203 * graphite-poly.c (print_pdr): Likewise.
22204 (dump_gbb_cases): Likewise.
22205 (dump_gbb_conditions): Likewise.
22206 (print_scop_params): Likewise.
22207 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
22208 (build_cross_bb_scalars_use): Likewise.
22209 (gather_bbs::before_dom_children): Likewise.
22210 * hsa-dump.c (dump_hsa_immed): Likewise.
22211 * ipa-cp.c (print_ipcp_constant_value): Likewise.
22212 (get_replacement_map): Likewise.
22213 * ipa-inline-analysis.c (dump_condition): Likewise.
22214 (estimate_function_body_sizes): Likewise.
22215 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
22216 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
22217 * ipa-prop.c (ipa_dump_param): Likewise.
22218 (ipa_print_node_jump_functions_for_edge): Likewise.
22219 (ipa_modify_call_arguments): Likewise.
22220 (ipa_modify_expr): Likewise.
22221 (ipa_dump_param_adjustments): Likewise.
22222 (ipa_dump_agg_replacement_values): Likewise.
22223 (ipcp_modif_dom_walker::before_dom_children): Likewise.
22224 * ipa-pure-const.c (check_stmt): Likewise.
22225 (pass_nothrow::execute): Likewise.
22226 * ipa-split.c (execute_split_functions): Likewise.
22227 * omp-offload.c (dump_oacc_loop_part): Likewise.
22228 (dump_oacc_loop): Likewise.
22229 * trans-mem.c (tm_log_emit): Likewise.
22230 (tm_memopt_accumulate_memops): Likewise.
22231 (dump_tm_memopt_set): Likewise.
22232 (dump_tm_memopt_transform): Likewise.
22233 * tree-cfg.c (gimple_verify_flow_info): Likewise.
22234 (print_loop): Likewise.
22235 * tree-chkp-opt.c (chkp_print_addr): Likewise.
22236 (chkp_gather_checks_info): Likewise.
22237 (chkp_get_check_result): Likewise.
22238 (chkp_remove_check_if_pass): Likewise.
22239 (chkp_use_outer_bounds_if_possible): Likewise.
22240 (chkp_reduce_bounds_lifetime): Likewise.
22241 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
22242 (chkp_mark_completed_bounds): Likewise.
22243 (chkp_register_incomplete_bounds): Likewise.
22244 (chkp_mark_invalid_bounds): Likewise.
22245 (chkp_maybe_copy_and_register_bounds): Likewise.
22246 (chkp_build_returned_bound): Likewise.
22247 (chkp_get_bound_for_parm): Likewise.
22248 (chkp_build_bndldx): Likewise.
22249 (chkp_get_bounds_by_definition): Likewise.
22250 (chkp_generate_extern_var_bounds): Likewise.
22251 (chkp_get_bounds_for_decl_addr): Likewise.
22252 * tree-chrec.c (chrec_apply): Likewise.
22253 * tree-data-ref.c (dump_data_reference): Likewise.
22254 (dump_subscript): Likewise.
22255 (dump_data_dependence_relation): Likewise.
22256 (analyze_overlapping_iterations): Likewise.
22257 * tree-inline.c (expand_call_inline): Likewise.
22258 (tree_function_versioning): Likewise.
22259 * tree-into-ssa.c (dump_defs_stack): Likewise.
22260 (dump_currdefs): Likewise.
22261 (dump_names_replaced_by): Likewise.
22262 (dump_update_ssa): Likewise.
22263 (update_ssa): Likewise.
22264 * tree-object-size.c (pass_object_sizes::execute): Likewise.
22265 * tree-parloops.c (build_new_reduction): Likewise.
22266 (try_create_reduction_list): Likewise.
22267 (ref_conflicts_with_region): Likewise.
22268 (oacc_entry_exit_ok_1): Likewise.
22269 (oacc_entry_exit_single_gang): Likewise.
22270 * tree-pretty-print.h: Likewise.
22271 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
22272 (get_scalar_evolution): Likewise.
22273 (add_to_evolution): Likewise.
22274 (get_loop_exit_condition): Likewise.
22275 (analyze_evolution_in_loop): Likewise.
22276 (analyze_initial_condition): Likewise.
22277 (analyze_scalar_evolution): Likewise.
22278 (instantiate_scev): Likewise.
22279 (number_of_latch_executions): Likewise.
22280 (gather_chrec_stats): Likewise.
22281 (final_value_replacement_loop): Likewise.
22282 (scev_const_prop): Likewise.
22283 * tree-sra.c (dump_access): Likewise.
22284 (disqualify_candidate): Likewise.
22285 (create_access): Likewise.
22286 (reject): Likewise.
22287 (maybe_add_sra_candidate): Likewise.
22288 (create_access_replacement): Likewise.
22289 (analyze_access_subtree): Likewise.
22290 (analyze_all_variable_accesses): Likewise.
22291 (sra_modify_assign): Likewise.
22292 (initialize_constant_pool_replacements): Likewise.
22293 (find_param_candidates): Likewise.
22294 (decide_one_param_reduction): Likewise.
22295 (replace_removed_params_ssa_names): Likewise.
22296 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
22297 * tree-ssa-copy.c (dump_copy_of): Likewise.
22298 (copy_prop_visit_cond_stmt): Likewise.
22299 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
22300 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
22301 (record_equivalences_from_stmt): Likewise.
22302 * tree-ssa-dse.c (compute_trims): Likewise.
22303 (delete_dead_call): Likewise.
22304 (delete_dead_assignment): Likewise.
22305 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
22306 (forward_propagate_into_cond): Likewise.
22307 (pass_forwprop::execute): Likewise.
22308 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
22309 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
22311 (move_computations_worker): Likewise.
22312 (execute_sm): Likewise.
22313 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
22314 (remove_exits_and_undefined_stmts): Likewise.
22315 (remove_redundant_iv_tests): Likewise.
22316 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
22317 (adjust_iv_update_pos): Likewise.
22318 * tree-ssa-math-opts.c (bswap_replace): Likewise.
22319 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
22320 (value_replacement): Likewise.
22321 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
22322 * tree-ssa-pre.c (print_pre_expr): Likewise.
22323 (get_representative_for): Likewise.
22324 (create_expression_by_pieces): Likewise.
22325 (insert_into_preds_of_block): Likewise.
22326 (eliminate_insert): Likewise.
22327 (eliminate_dom_walker::before_dom_children): Likewise.
22328 (eliminate): Likewise.
22329 (remove_dead_inserted_code): Likewise.
22330 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
22331 * tree-ssa-reassoc.c (get_rank): Likewise.
22332 (eliminate_duplicate_pair): Likewise.
22333 (eliminate_plus_minus_pair): Likewise.
22334 (eliminate_not_pairs): Likewise.
22335 (undistribute_ops_list): Likewise.
22336 (eliminate_redundant_comparison): Likewise.
22337 (update_range_test): Likewise.
22338 (optimize_range_tests_var_bound): Likewise.
22339 (optimize_vec_cond_expr): Likewise.
22340 (rewrite_expr_tree): Likewise.
22341 (rewrite_expr_tree_parallel): Likewise.
22342 (linearize_expr): Likewise.
22343 (break_up_subtract): Likewise.
22344 (linearize_expr_tree): Likewise.
22345 (attempt_builtin_powi): Likewise.
22346 (attempt_builtin_copysign): Likewise.
22347 (transform_stmt_to_copy): Likewise.
22348 (transform_stmt_to_multiply): Likewise.
22349 (dump_ops_vector): Likewise.
22350 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
22351 (print_scc): Likewise.
22352 (set_ssa_val_to): Likewise.
22353 (visit_reference_op_store): Likewise.
22354 (visit_use): Likewise.
22355 (sccvn_dom_walker::before_dom_children): Likewise.
22356 (run_scc_vn): Likewise.
22357 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
22359 (expr_hash_elt::print): Likewise.
22360 (const_and_copies::pop_to_marker): Likewise.
22361 (const_and_copies::record_const_or_copy_raw): Likewise.
22362 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
22363 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
22364 (dump_predicates): Likewise.
22365 (find_uninit_use): Likewise.
22366 (warn_uninitialized_phi): Likewise.
22367 (pass_late_warn_uninitialized::execute): Likewise.
22368 * tree-ssa.c (verify_vssa): Likewise.
22369 (verify_ssa): Likewise.
22370 (maybe_optimize_var): Likewise.
22371 * tree-vrp.c (dump_value_range): Likewise.
22372 (dump_all_value_ranges): Likewise.
22373 (dump_asserts_for): Likewise.
22374 (register_edge_assert_for_2): Likewise.
22375 (vrp_visit_cond_stmt): Likewise.
22376 (vrp_visit_switch_stmt): Likewise.
22377 (vrp_visit_stmt): Likewise.
22378 (vrp_visit_phi_node): Likewise.
22379 (simplify_cond_using_ranges_1): Likewise.
22380 (fold_predicate_in): Likewise.
22381 (evrp_dom_walker::before_dom_children): Likewise.
22382 (evrp_dom_walker::push_value_range): Likewise.
22383 (evrp_dom_walker::pop_value_range): Likewise.
22384 (execute_early_vrp): Likewise.
22386 2017-05-16 Richard Biener <rguenther@suse.de>
22388 * dwarf2out.c (loc_list_from_tree_1): Do not create
22389 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
22391 2017-05-16 Richard Biener <rguenther@suse.de>
22393 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
22395 (note_variable_value_in_expr): If we resolved the decl ref
22396 do not push to the stack.
22398 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
22400 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
22401 operations in fast-math mode.
22402 (vaddq_f16): Likewise.
22403 (vmul_f16): Likewise.
22404 (vmulq_f16): Likewise.
22405 (vsub_f16): Likewise.
22406 (vsubq_f16): Likewise.
22407 * config/arm/neon.md (add<mode>3): New.
22409 (fma:<VH:mode>3): New. Also remove outdated comment.
22412 2017-05-16 Martin Liska <mliska@suse.cz>
22416 * ipa-devirt.c (warn_types_mismatch): Fix typo.
22417 (odr_types_equivalent_p): Likewise.
22419 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
22421 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
22423 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
22426 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
22427 non-interunit SSE move alternatives with '?'.
22428 (zero-extendsidi peephole2): New peephole to skip intermediate
22429 general register in SSE zero-extend sequence.
22431 2017-05-15 Jeff Law <law@redhat.com>
22433 * reorg.c (relax_delay_slots): Create a new variable to hold
22434 the temporary target rather than clobbering TARGET_LABEL.
22436 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
22437 missing argument to extract_bit_field call.
22438 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
22440 2017-05-15 Martin Liska <mliska@suse.cz>
22443 * gcc.c (process_command): Do not allow empty argument of -o option.
22445 2017-05-15 Renlin Li <renlin.li@arm.com>
22447 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
22448 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
22449 * config/aarch64/constraints.md (Usf): Add long call check.
22450 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
22451 (call_value): Likewise.
22452 (sibcall): Likewise.
22453 (sibcall_value): Likewise.
22455 (call_value_insn): New.
22456 (sibcall_insn): Update rtx pattern.
22457 (sibcall_value_insn): Likewise.
22458 (call_internal): Remove.
22459 (call_value_internal): Likewise.
22460 (sibcall_internal): Likewise.
22461 (sibcall_value_internal): Likewise.
22462 (call_reg): Likewise.
22463 (call_symbol): Likewise.
22464 (call_value_reg): Likewise.
22465 (call_value_symbol): Likewise.
22467 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
22470 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
22472 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
22474 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
22475 compatible with CCGOCmode and with CCZmode.
22477 2017-05-14 Martin Sebor <msebor@redhat.com>
22479 PR middle-end/77671
22480 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
22481 (gimple_fold_builtin_snprintf): Same.
22482 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
22483 (gimple_fold_builtin_snprintf): Same.
22484 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
22485 of character types.
22486 (is_call_safe): New function.
22487 (try_substitute_return_value): Call it.
22488 (try_simplify_call): New function.
22489 (pass_sprintf_length::handle_gimple_call): Call it.
22491 2017-05-14 Martin Sebor <msebor@redhat.com>
22493 PR middle-end/80669
22494 * builtins.c (expand_builtin_stpncpy): Simplify.
22496 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
22498 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
22499 * config/i386/i386.h
22500 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
22501 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
22502 (struct machine_function): Add new members call_ms2sysv,
22503 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
22504 (struct machine_frame_state): New fields sp_realigned and
22505 sp_realigned_offset.
22506 * config/i386/i386.c
22507 (enum xlogue_stub): New enum.
22508 (enum xlogue_stub_sets): New enum.
22509 (class xlogue_layout): New class.
22510 (struct ix86_frame): New fields stack_realign_allocate_offset,
22511 stack_realign_offset and outlined_save_offset. Modify comments to
22512 detail stack layout when using out-of-line stubs.
22513 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
22514 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
22515 -mcall-ms2sysv-xlogues.
22516 (stub_managed_regs): New static variable.
22517 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
22518 registers managed by out-of-line stub.
22519 (disable_call_ms2sysv_xlogues): New function.
22520 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
22521 m->call_ms2sysv when appropriate and compute frame layout for
22523 (sp_valid_at, fp_valid_at): New inline functions.
22524 (choose_basereg): New function.
22525 (choose_baseaddr): Add align parameter, use choose_basereg and modify
22527 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
22528 Use align parameter of choose_baseaddr to generated aligned SSE movs
22530 (pro_epilogue_adjust_stack): Modify to track
22531 machine_frame_state::sp_realigned.
22532 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
22533 (ix86_nsaved_sseregs): Likewise.
22534 (ix86_emit_save_regs): Likewise.
22535 (ix86_emit_save_regs_using_mov): Likewise.
22536 (ix86_emit_save_sse_regs_using_mov): Likewise.
22537 (get_scratch_register_on_entry): Likewise.
22538 (gen_frame_set): New function.
22539 (gen_frame_load): Likewise.
22540 (gen_frame_store): Likewise.
22541 (emit_outlined_ms2sysv_save): Likewise.
22542 (emit_outlined_ms2sysv_restore): Likewise.
22543 (ix86_expand_prologue): Modify stack re-alignment code and call
22544 emit_outlined_ms2sysv_save when appropriate.
22545 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
22546 parameter rtx_insn *insn, which allows the function to be used to only
22547 generate the notes.
22548 (ix86_expand_epilogue): Modify validity checks of frame and stack
22549 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
22550 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
22551 * config/i386/predicates.md
22552 (save_multiple): New predicate.
22553 (restore_multiple): Likewise.
22554 * config/i386/sse.md
22555 (save_multiple<mode>): New pattern.
22556 (save_multiple_realign<mode>): Likewise.
22557 (restore_multiple<mode>): Likewise.
22558 (restore_multiple_and_return<mode>): Likewise.
22559 (restore_multiple_leave_return<mode>): Likewise.
22560 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
22562 2017-05-14 Julia Koval <julia.koval@intel.com>
22564 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
22565 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
22566 (__builtin_ia32_xsetbv): New builtins.
22567 * config/i386/i386.c (ix86_expand_special_args_builtin):
22569 (ix86_expand_builtin): Special expand for new intrinsics.
22570 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
22571 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
22572 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
22574 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22576 * cfganal.c (inverted_post_order_compute): Change argument type
22578 * cfganal.h (inverted_post_order_compute): Adjust prototype.
22579 * df-core.c (rest_of_handle_df_initialize): Adjust.
22580 (rest_of_handle_df_finish): Likewise.
22581 (df_analyze_1): Likewise.
22582 (df_analyze): Likewise.
22583 (loop_inverted_post_order_compute): Change argument to be a vec *.
22584 (df_analyze_loop): Adjust.
22585 (df_get_n_blocks): Likewise.
22586 (df_get_postorder): Likewise.
22587 * df.h (struct df_d): Change field to be a vec.
22588 * lcm.c (compute_laterin): Adjust.
22589 (compute_available): Likewise.
22590 * lra-lives.c (lra_create_live_ranges_1): Likewise.
22591 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
22592 * tree-ssa-pre.c (compute_antic): Likewise.
22594 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22596 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
22597 (depth_first_search::depth_first_search): Change structure init
22598 function to this constructor.
22599 (depth_first_search::add_bb): Rename function to this member.
22600 (depth_first_search::execute): Likewise.
22601 (flow_dfs_compute_reverse_finish): Adjust.
22603 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22605 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
22606 (longest_simple_path): Likewise.
22607 * shrink-wrap.c (spread_components): Likewise.
22608 (disqualify_problematic_components): Likewise.
22609 (emit_common_heads_for_components): Likewise.
22610 (emit_common_tails_for_components): Likewise.
22611 (insert_prologue_epilogue_for_components): Likewise.
22613 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22615 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
22618 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22620 * df-core.c (df_set_blocks): Start using auto_bitmap.
22621 (df_compact_blocks): Likewise.
22622 * df-problems.c (df_rd_confluence_n): Likewise.
22623 * df-scan.c (df_insn_rescan_all): Likewise.
22624 (df_process_deferred_rescans): Likewise.
22625 (df_update_entry_block_defs): Likewise.
22626 (df_update_exit_block_uses): Likewise.
22627 (df_entry_block_bitmap_verify): Likewise.
22628 (df_exit_block_bitmap_verify): Likewise.
22629 (df_scan_verify): Likewise.
22630 * lra-constraints.c (lra_constraints): Likewise.
22631 (undo_optional_reloads): Likewise.
22632 (lra_undo_inheritance): Likewise.
22633 * lra-remat.c (calculate_gen_cands): Likewise.
22634 (do_remat): Likewise.
22635 * lra-spills.c (assign_spill_hard_regs): Likewise.
22636 (spill_pseudos): Likewise.
22637 * tree-ssa-pre.c (bitmap_set_and): Likewise.
22638 (bitmap_set_subtract_values): Likewise.
22640 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22642 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
22643 management with auto_bitmap.
22644 (fix_inter_tick): Likewise.
22645 (fix_recovery_deps): Likewise.
22646 * ira.c (add_store_equivs): Likewise.
22647 (find_moveable_pseudos): Likewise.
22648 (split_live_ranges_for_shrink_wrap): Likewise.
22649 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
22650 (rtx_reuse_manager::seen_def_p): Likewise.
22651 (rtx_reuse_manager::set_seen_def): Likewise.
22652 * print-rtl.h (class rtx_reuse_manager): Likewise.
22654 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22656 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
22658 (migrate_btr_def): Likewise.
22659 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
22660 * df-core.c (loop_post_order_compute): Likewise.
22661 (loop_inverted_post_order_compute): Likewise.
22662 * hsa-common.h: Likewise.
22663 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
22664 * init-regs.c (initialize_uninitialized_regs): Likewise.
22665 * ipa-inline.c (resolve_noninline_speculation): Likewise.
22666 (inline_small_functions): Likewise.
22667 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
22668 * ira.c (combine_and_move_insns): Likewise.
22669 (build_insn_chain): Likewise.
22670 * loop-invariant.c (find_invariants): Likewise.
22671 * lower-subreg.c (propagate_pseudo_copies): Likewise.
22672 * predict.c (tree_predict_by_opcode): Likewise.
22673 (predict_paths_leading_to): Likewise.
22674 (predict_paths_leading_to_edge): Likewise.
22675 (estimate_loops_at_level): Likewise.
22676 (estimate_loops): Likewise.
22677 * shrink-wrap.c (try_shrink_wrapping): Likewise.
22678 (spread_components): Likewise.
22679 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
22680 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
22681 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
22682 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
22683 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
22684 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
22685 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
22686 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
22687 (mark_threaded_blocks): Likewise.
22688 (thread_through_all_blocks): Likewise.
22689 * tree-ssa.c (verify_ssa): Likewise.
22690 (execute_update_addresses_taken): Likewise.
22691 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
22693 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22695 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
22697 (post_order_compute): Likewise.
22698 (inverted_post_order_compute): Likewise.
22699 (pre_and_rev_post_order_compute_fn): Likewise.
22701 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22703 * genrecog.c (int_set::int_set): Explicitly construct our
22704 auto_vec base class.
22705 * vec.h (auto_vec::auto_vec): New constructor.
22707 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22709 * bitmap.h (class auto_bitmap): New constructor taking
22710 bitmap_obstack * argument.
22712 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22714 * bitmap.h (class auto_bitmap): Change type of m_bits to
22715 bitmap_head, and adjust ctor / dtor and member operators.
22717 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
22719 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
22720 when returned register mode doesn't match original mode.
22722 2017-05-12 Jeff Law <law@redhat.com>
22723 Jakub Jelinek <jakub@redhat.com>
22725 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
22726 we look for cc setter after the compare-elim changes.
22727 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
22728 within the vector to match what compare-elim now expects.
22729 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
22730 (xorsi3_flags, one_cmplsi2_flags): Likewise.
22732 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
22733 after the compare-elim changes.
22734 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
22735 the vector to match what compare-elim now expects.
22736 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
22737 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
22738 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
22739 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
22740 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
22742 * config/visium/visium.c (single_set_and_flags): Fix where
22743 we look for cc setter after the compare-elim changes.
22744 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
22745 with the vector to match what compare-elim now expects.
22746 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
22747 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
22748 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
22749 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
22750 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
22751 (neg<mode>2_insn_set_overflow): Likewise.
22753 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
22755 PR middle-end/79794
22756 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
22757 maybe_expand_insn call, set ops[0].target. If still set after call,
22758 set alt_rtl. Add extra arg to recursive calls.
22759 (extract_bit_field): Add alt_rtl argument. Pass to
22761 * expmed.h (extract_bit_field): Fix prototype.
22762 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
22763 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
22764 to extract_bit_field_calls.
22765 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
22766 Pass alt_rtl to extract_bit_field calls.
22767 * calls.c (store_unaligned_arguments_into_psuedos)
22768 load_register_parameters): Pass extra NULL to extract_bit_field calls.
22769 * optabs.c (maybe_legitimize_operand): Clear op->target when call
22771 * optabs.h (struct expand_operand): Add target bitfield.
22773 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
22775 * compare-elim.c (try_eliminate_compare): Canonicalize
22776 operation with embedded compare to
22777 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
22778 (set (reg) (operation)].
22780 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
22782 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
22785 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
22786 cost of adding a carry flag for ADC instruction.
22787 [case MINUS]: Ignore the cost of subtracting a carry flag
22788 for SBB instruction.
22790 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
22792 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
22794 * config/rs6000/bmiintrin.h: New file.
22795 * config/rs6000/bmi2intrin.h: New file.
22796 * config/rs6000/x86intrin.h: New file.
22798 2017-05-12 Jeff Law <law@redhat.com>
22800 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
22803 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
22805 PR middle-end/80707
22806 * tree-cfg.c: Remove cfg edges of unreachable case statements.
22808 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
22810 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
22811 early expansion of vector divide builtins.
22812 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
22813 builtins identified as having unsigned arguments.
22815 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
22817 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
22818 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
22819 expansion of vector logical operations (and, andc, or, xor,
22822 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
22824 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
22825 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
22827 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
22829 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
22830 early GIMPLE expansion of vector multiplies.
22832 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22834 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
22835 TARGET_HAVE_MOVT conditional.
22836 (movt splitter): Likewise.
22838 2017-05-12 Richard Biener <rguenther@suse.de>
22840 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
22841 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
22842 Fold all stmts not inplace.
22844 2017-05-12 Richard Biener <rguenther@suse.de>
22846 PR tree-optimization/80713
22847 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
22848 inserted_exprs bit for not removed stmts.
22850 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
22852 PR middle-end/69921
22853 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
22854 parallelized" attribute for parallelized OpenACC kernels.
22855 * omp-offload.c (execute_oacc_device_lower): Use it.
22857 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
22858 Set "oacc kernels" attribute.
22859 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
22860 parameter. Adjust all users.
22861 (oacc_fn_attrib_kernels_p): Remove function.
22862 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
22863 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
22864 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
22865 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
22866 assert "oacc kernels" attribute is set.
22868 2017-05-11 Carl Love <cel@us.ibm.com>
22870 * config/rs6000/rs6000-c: Add support for built-in functions
22871 vector unsigned char vec_popcnt (vector signed char)
22872 vector unsigned char vec_popcnt (vector unsigned char)
22873 vector unsigned short vec_popcnt (vector signed short)
22874 vector unsigned short vec_popcnt (vector unsigned short)
22875 vector unsigned int vec_popcnt (vector signed int)
22876 vector unsigned int vec_popcnt (vector unsigned int)
22877 vector unsigned long long vec_popcnt (vector signed long long)
22878 vector unsigned long long vec_popcnt (vector unsigned long long)
22879 vector signed long long vec_slo (vector signed long long,
22880 vector signed char)
22881 vector signed long long vec_slo (vector signed long long,
22882 vector unsigned char)
22883 vector unsigned long long vec_slo (vector unsigned long long,
22884 vector signed char)
22885 vector unsigned long long vec_slo (vector unsigned long long,
22886 vector unsigned char)
22887 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
22888 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
22889 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
22890 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
22891 * doc/extend.texi: Update the built-in documentation file for the
22892 new built-in functions.
22894 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
22896 * attribs.h (sorted_attr_string): Move machine independent
22897 functions for target clone support from the i386 port to common
22898 code. Rename ix86_function_versions to common_function_versions.
22899 Rename make_name to make_unique_name.
22900 (common_function_versions): Likewise.
22901 (make_unique_name): Likewise.
22902 (make_dispatcher_decl): Likewise.
22903 (is_function_default_version): Likewise.
22904 * attribs.c (attr_strcmp): Likewise.
22905 (sorted_attr_string): Likewise.
22906 (common_function_versions): Likewise.
22907 (make_unique_name): Likewise.
22908 (make_dispatcher_decl): Likewise.
22909 (is_function_default_version): Likewise.
22910 * config/i386/i386.c (attr_strcmp): Likewise.
22911 (sorted_attr_string): Likewise.
22912 (ix86_function_versions): Likewise.
22913 (make_name): Likewise.
22914 (make_dispatcher_decl): Likewise.
22915 (is_function_default_version): Likewise.
22916 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
22918 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22921 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
22922 Account for direct move costs for vec_construct of integer
22925 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
22928 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
22929 (UNSPEC_STX_ATOMIC): Ditto.
22930 (loaddi_via_sse): New insn.
22931 (storedi_via_sse): Ditto.
22932 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
22933 Update corresponding peephole2 patterns.
22934 (atomic_storedi_fpu): Ditto.
22936 2017-05-11 Julia Koval <julia.koval@intel.com>
22938 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
22939 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
22941 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
22942 (__builtin_ia32_rsqrt14ss_mask): New builtins.
22943 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
22945 2017-05-11 Nathan Sidwell <nathan@acm.org>
22947 * graphite-poly.c: Include dumpfile.h.
22949 * dumpfle.h (dump_function): Declare here ...
22950 * tree-dump.h (dump_function): ... not here.
22951 * dumpfile.c: #include tree-cfg.h.
22952 (dump_function): Move here from ...
22953 * tree-dump.c (dump_function): ... here.
22954 * gimplify.c: #include splay-tree.h, not tree-dump.h.
22955 * graphite-poly.c: Don't include tree-dump.h.
22956 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
22957 * print-tree.c: Likewise.
22958 * stor-layout.c: Likewise.
22959 * tree-nested.c: Likewise.
22961 * dumpfile.c (dump_start): Use TDF_FLAGS.
22962 (dump_enable_all): Fix TDF_KIND check thinko.
22964 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
22966 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22967 array entries to represent two legal parameterizations of the
22968 overloaded __builtin_cmpb function, as represented by the
22969 P6_OV_BUILTIN_CMPB constant.
22970 (altivec_resolve_overloaded_builtin): Add special case handling
22971 for the __builtin_cmpb function, as represented by the
22972 P6_OV_BUILTIN_CMPB constant.
22973 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
22974 (BU_P6_64BIT_2): New macro.
22975 (BU_P6_OVERLOAD_2): New macro
22976 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
22977 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
22978 (CMPB): Add overload support to represent both 32-bit and 64-bit
22979 compare-bytes function.
22980 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
22981 support for TARGET_CMPB.
22982 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
22983 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
22984 documentation of the __builtin_cmpb overloaded built-in function.
22986 2017-05-11 Richard Biener <rguenther@suse.de>
22988 PR tree-optimization/80705
22989 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
22990 bases are not vectorizable.
22992 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22994 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
22995 when counting register pressure.
22997 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22999 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
23000 (iv_ca_more_deps): Renamed to ...
23001 (iv_ca_compare_deps): ... this.
23002 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
23004 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23006 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
23008 (determine_group_iv_costs): ... here.
23009 (find_inv_vars_cb): Record inv var if it's not recorded before.
23011 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23013 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
23014 (get_shiftadd_cost): Ditto.
23016 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23018 * tree-ssa-address.c: Include header file.
23019 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
23021 (add_to_parts): Refactor.
23022 (addr_to_parts): New parameter. Update use of move_hint_to_base.
23023 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
23026 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23028 PR tree-optimization/53090
23029 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
23031 (extract_cond_operands): Detect condition with IV on both sides
23032 and return COMP_IV_EXPR_2.
23033 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
23034 (rewrite_use_compare): Simplify by removing call to function
23035 extract_cond_operands.
23037 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23039 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
23040 (extract_cond_operands): Detect condition comparing against non-
23041 invariant bound and return appropriate enum value.
23042 (find_interesting_uses_cond): Update use of extract_cond_operands.
23043 Handle its return value accordingly.
23044 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
23046 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23048 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
23049 nonlinear iv_use computation in loop invariant sensitive way.
23051 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23053 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
23054 (find_iv_candidates): Call relate_compare_use_with_all_cands.
23056 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23058 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
23059 (dump_cand): Support iv_cand.inv_exprs.
23060 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
23062 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
23065 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23067 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
23069 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
23070 as local function. Include necessary header files.
23071 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
23073 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23075 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
23077 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23079 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
23080 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
23081 RSHIFT_EXPR and BIT_NOT_EXPR.
23083 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23085 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
23086 (adjust_setup_cost): New parameter supporting round up adjustment.
23087 (struct address_cost_data): Delete.
23088 (force_expr_to_var_cost): Don't bound cost with spill_cost.
23089 (split_address_cost, ptr_difference_cost): Delete.
23090 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
23091 (struct ainc_cost_data): New struct.
23092 (get_address_cost_ainc): New function.
23093 (get_address_cost, get_computation_cost): Reimplement.
23094 (determine_group_iv_cost_address): Record inv_expr for all uses of
23096 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
23097 (iv_ca_has_deps): Reimplemented to ...
23098 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
23100 (iv_ca_extend): Call iv_ca_more_deps.
23102 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23104 * tree-ssa-address.c (struct mem_address): Move to header file.
23105 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
23106 * tree-ssa-address.h (struct mem_address): Move from C file.
23107 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
23109 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23111 * tree-affine.h (aff_combination_type): New interface.
23112 (aff_combination_zero_p): Remove static.
23113 (aff_combination_const_p): New interface.
23114 (aff_combination_singleton_var_p): New interfaces.
23116 2017-05-11 Richard Biener <rguenther@suse.de>
23118 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23119 Skip unreachable blocks and destinations.
23120 (eliminate): Move stmt removal and fixup ...
23121 (fini_eliminate): ... here. Skip inserted exprs.
23122 (pass_pre::execute): Move fini_pre after fini_eliminate.
23123 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
23124 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
23125 PRE to get rid of dead code that has invalid SSA form and
23126 split critical edges again.
23128 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23130 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
23132 2017-05-11 Richard Biener <rguenther@suse.de>
23134 * passes.c (execute_function_todo): Verify loops if they are
23135 said to be up-to-date.
23136 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
23137 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
23139 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
23142 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
23143 handle calling assemble_external ourself.
23146 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
23147 modes with zero size. Enhance comment.
23149 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23151 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
23152 built-ins for vec_xl and vec_xst with short and char pointer
23155 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
23157 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
23158 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
23159 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
23160 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
23161 (_mm_maskz_min_round_ss): New intrinsics.
23162 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
23163 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
23164 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
23165 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
23166 (__builtin_ia32_minss_mask_round): New builtins.
23167 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23168 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23169 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
23171 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
23172 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
23174 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
23177 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
23179 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
23180 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
23181 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
23182 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
23183 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
23184 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
23185 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
23186 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23187 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
23188 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
23189 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
23190 (__builtin_ia32_mulss_mask_round): New builtins.
23191 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23192 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23193 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
23195 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
23196 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
23198 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
23201 2017-05-10 Julia Koval <julia.koval@intel.com>
23203 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
23204 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
23205 (_mm256_setr_m128i): New intrinsics.
23207 2017-05-10 Julia Koval <julia.koval@intel.com>
23209 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
23210 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
23211 (_mm_maskz_rcp14_ss): New intrinsics.
23212 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
23213 (__builtin_ia32_rcp14ss_mask): New builtins.
23214 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
23216 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
23218 PR tree-optimization/51513
23219 * tree-cfg.c (gimple_seq_unreachable_p): New function.
23220 (assert_unreachable_fallthru_edge_p): Use it.
23221 (group_case_labels_stmt): Likewise.
23222 * tree-cfg.h: Prototype it.
23223 * stmt.c: Include cfghooks.h and tree-cfg.h.
23224 (emit_case_dispatch_table) <gap_label>: New local variable.
23225 Use it to fill dispatch table gaps.
23226 Test for default_label before updating probabilities.
23227 (expand_case) <default_label>: Remove unneeded initialization.
23228 Test for unreachable default case statement and remove its edge.
23229 Set default_label accordingly.
23230 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
23232 2017-05-10 Carl Love <cel@us.ibm.com>
23234 * config/rs6000/rs6000-c: Add support for built-in functions
23235 vector signed char vec_neg (vector signed char)
23236 vector signed short int vec_neg (vector short int)
23237 vector signed int vec_neg (vector signed int)
23238 vector signed long long vec_neg (vector signed long long)
23239 vector float vec_neg (vector float)
23240 vector double vec_neg (vector double)
23241 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
23243 * config/rs6000/altivec.h: Add define for vec_neg
23244 * doc/extend.texi: Update the built-in documentation for the
23245 new built-in functions.
23247 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23249 PR tree-optimization/77644
23250 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
23252 2017-05-10 Nathan Sidwell <nathan@acm.org>
23254 * dumpfile.h (TDI_lang_all): New.
23255 (TDF_KIND): New. Renumber others
23256 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
23258 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
23260 (get_dump_file_name): Adjust suffix generation.
23261 (dump_enable_all): Use TDF_KIND.
23262 * doc/invoke.texi (-fdump-lang-all): Document.
23264 * dumpfile.h: Tabify.
23266 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
23269 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
23270 Move member access before delete.
23272 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
23274 * tree-inline.c (expand_call_inline): Split block at stmt
23277 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
23280 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
23281 are now unused after splitting mov{sf,sd}_hardfloat.
23282 (f32_lr2): Likewise.
23283 (f32_lm): Likewise.
23284 (f32_lm2): Likewise.
23285 (f32_li): Likewise.
23286 (f32_li2): Likewise.
23287 (f32_lv): Likewise.
23288 (f32_sr): Likewise.
23289 (f32_sr2): Likewise.
23290 (f32_sm): Likewise.
23291 (f32_sm2): Likewise.
23292 (f32_si): Likewise.
23293 (f32_si2): Likewise.
23294 (f32_sv): Likewise.
23295 (f32_dm): Likewise.
23296 (f32_vsx): Likewise.
23297 (f32_av): Likewise.
23298 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
23299 For movsf, order stores so the VSX stores occur before the GPR
23300 store which encourages the register allocator to use a traditional
23301 FPR instead of a GPR. For movsd, order the stores so that the GPR
23302 store comes before the VSX stores to allow the power6 to work.
23303 This is due to the power6 not having a 32-bit integer store
23304 instruction from a FPR.
23305 (movsf_hardfloat): Likewise.
23306 (movsd_hardfloat): Likewise.
23308 2017-05-09 Martin Sebor <msebor@redhat.com>
23310 PR translation/80280
23311 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
23314 PR translation/80280
23315 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
23316 data member added in r247778.
23317 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
23319 2017-05-09 Nathan Sidwell <nathan@acm.org>
23321 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
23323 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
23326 2017-05-09 Marek Polacek <polacek@redhat.com>
23328 * doc/invoke.texi: Fix typo.
23330 2017-05-09 Richard Biener <rguenther@suse.de>
23332 * tree-vrp.c (vrp_val_is_max): Adjust comment.
23333 (vrp_val_is_min): Likewise.
23334 (set_value_range_to_value): Likewise.
23335 (set_value_range_to_nonnegative): Likewise.
23336 (gimple_assign_nonzero_p): Likewise.
23337 (gimple_stmt_nonzero_p): Likewise.
23338 (vrp_int_const_binop): Likewise. Remove unreachable case.
23339 (adjust_range_with_scev): Adjust comments.
23340 (compare_range_with_value): Likewise.
23341 (extract_range_from_phi_node): Likewise.
23342 (test_for_singularity): Likewise.
23344 2017-05-09 Richard Biener <rguenther@suse.de>
23346 * tree-vrp.c (get_single_symbol): Add assert that we don't
23347 get overflowed constants as invariant part.
23348 (compare_values_warnv): Add comment before the TREE_NO_WARNING
23349 checks. Use wi::cmp instead of recursing for integer constants.
23350 (compare_values): Just ignore whether we assumed undefined
23351 overflow instead of failing the compare.
23352 (extract_range_for_var_from_comparison_expr): Add comment before the
23353 TREE_NO_WARNING sets.
23354 (test_for_singularity): Likewise.
23355 (extract_range_from_comparison): Do not disable optimization
23356 when we assumed undefined overflow.
23357 (extract_range_basic): Remove init of unused var.
23359 2017-05-09 Richard Biener <rguenther@suse.de>
23361 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
23362 (extract_range_from_multiplicative_op_1): Adjust.
23363 (extract_range_from_binary_expr_1): Use int_const_binop.
23365 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
23368 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
23369 rs6000_store_data_bypass_p in seven define_bypass directives and
23370 in several comments.
23371 * config/rs6000/rs6000-protos.h: Add prototype for
23372 rs6000_store_data_bypass_p function.
23373 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
23374 function implements slightly different (rs6000-specific) semantics
23375 than store_data_bypass_p, returning false rather than aborting
23376 with assertion error when arguments do not satisfy the
23377 requirements of store data bypass.
23378 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
23379 rs6000_store_data_bypass_p.
23381 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
23383 * config/xtensa/xtensa-protos.h
23384 (xtensa_initial_elimination_offset): New declaration.
23385 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
23386 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
23387 macro definition, add case for FRAME_POINTER_REGNUM when
23388 FRAME_GROWS_DOWNWARD.
23389 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
23390 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
23391 xtensa_initial_elimination_offset.
23393 2017-05-08 Nathan Sidwell <nathan@acm.org>
23395 * doc/invoke.texi: Alphabetize -fdump options.
23397 2017-05-08 Martin Sebor <msebor@redhat.com>
23399 PR translation/80280
23400 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
23402 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
23404 * target.def (compute_frame_layout): New optional target hook.
23405 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
23406 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
23407 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
23409 * reload1.c (verify_initial_elim_offsets): Likewise.
23410 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
23411 (use_simple_return_p): Call arm_compute_frame_layout if needed.
23412 (arm_get_frame_offsets): Split up into this ...
23413 (arm_compute_frame_layout): ... and this function.
23415 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
23417 * config/aarch64/constraints.md (Usa): New constraint.
23418 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
23420 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
23422 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
23423 with_multilib_list after it has been checked.
23425 2017-05-08 Richard Biener <rguenther@suse.de>
23427 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
23428 (bitmap_set_subtract_values): Likewise.
23430 2017-05-08 Richard Biener <rguenther@suse.de>
23432 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
23433 (gimple_assign_nonzero): ... this and remove strict_overflow_p
23435 (gimple_stmt_nonzero_warnv_p): Rename to ...
23436 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
23438 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
23439 (extract_range_basic): Adjust, do not disable propagation on
23440 strict overflow sensitive simplification.
23441 (vrp_visit_cond_stmt): Likewise.
23443 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
23445 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
23446 body size unconditionally.
23448 2017-05-07 Jeff Law <law@redhat.com>
23451 2017-05-06 Jeff Law <law@redhat.com>
23452 PR tree-optimization/78496
23453 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
23456 PR tree-optimization/78496
23457 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
23458 (simplify_stmt_using_ranges): Call it.
23459 (vrp_dom_walker::before_dom_children): Extract equivalences
23460 from an ASSERT_EXPR with an equality comparison against a
23463 2017-05-06 Jeff Law <law@redhat.com>
23465 PR tree-optimization/78496
23466 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
23469 PR tree-optimization/78496
23470 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
23471 (simplify_stmt_using_ranges): Call it.
23472 (vrp_dom_walker::before_dom_children): Extract equivalences
23473 from an ASSERT_EXPR with an equality comparison against a
23476 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
23478 * lra-constraints.c (lra_copy_reg_equiv): New function.
23479 (split_reg): Use it to copy equivalence information from the
23480 original register to the spill register.
23482 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
23484 PR rtl-optimization/75964
23485 * simplify-rtx.c (simplify_const_relational_operation): Remove
23486 invalid handling of comparisons of integer ABS.
23488 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
23490 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
23491 initialize to zero.
23492 (init_regs): Remove declaration.
23493 (function_arg_advance_32): Initialize error_p as boolean variable.
23495 2017-05-05 Nathan Sidwell <nathan@acm.org>
23497 * store-motion.c (remove_reachable_equiv_notes): Reformat long
23498 lines. Use for (;;).
23500 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23502 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
23503 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
23504 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
23505 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
23506 VF=2 that require versioning.
23508 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23510 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
23513 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23515 * diagnostic.h (diagnostic_override_option_index): Convert from
23516 macro to inline function.
23518 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23520 * diagnostic.c (last_module_changed_p): New function.
23521 (set_last_module): New function.
23522 (diagnostic_report_current_module): Convert macro usage to
23523 the above functions.
23524 * diagnostic.h (diagnostic_context::last_module): Strengthen
23525 from const line_map * to const line_map_ordinary *.
23526 (diagnostic_last_module_changed): Delete macro.
23527 (diagnostic_set_last_module): Delete macro.
23529 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23531 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
23532 with diagnostic_report_diagnostic.
23533 (diagnostic_n_impl_richloc): Likewise.
23534 * diagnostic.h (report_diagnostic): Delete macro.
23535 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
23536 with diagnostic_report_diagnostic.
23537 * substring-locations.c (format_warning_va): Likewise.
23539 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23541 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
23542 save/restor of format_spec. Move option-printing code to...
23543 (print_option_information): ...this new function, and
23544 reimplement by simply printing to the pretty_printer,
23545 rather than appending to the format string.
23547 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23549 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
23550 handling logic into...
23551 (update_effective_level_from_pragmas): ...this new function.
23553 2017-05-04 Andrew Waterman <andrew@sifive.com>
23555 * config/riscv/riscv.opt (mstrict-align): New option.
23556 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
23557 (SLOW_UNALIGNED_ACCESS): Define.
23558 (riscv_slow_unaligned_access): Declare.
23559 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
23561 (riscv_slow_unaligned_access): New variable.
23562 (rocket_tune_info): Set slow_unaligned_access to true.
23563 (optimize_size_tune_info): Set slow_unaligned_access to false.
23564 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
23565 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
23566 (riscv_option_override): Set riscv_slow_unaligned_access.
23567 * doc/invoke.texi: Add -mstrict-align to RISC-V.
23569 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
23571 * config/riscv/riscv.md: Unify indentation.
23573 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
23578 * config/rs6000/rs6000.md (u code attribute): Add FIX and
23580 (extendsi<mode>2): Add support for doing sign extension via
23581 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
23582 don't have ISA 3.0 instructions.
23583 (extendsi<mode>2 splitter): Likewise.
23584 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
23585 generate the normal insns since SImode can now go in vector
23586 registers. Disallow the special UNSPECs needed for previous
23587 machines to hide SImode being used. Add new insns
23588 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
23589 (fix_trunc<mode>si2_stfiwx): Likewise.
23590 (fix_trunc<mode>si2_internal): Likewise.
23591 (fixuns_trunc<mode>si2): Likewise.
23592 (fixuns_trunc<mode>si2_stfiwx): Likewise.
23593 (fctiw<u>z_<mode>_smallint): Likewise.
23594 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
23595 of floating point to 32-bit integer from doing a direct move to
23596 the GPR registers to do a store.
23597 (fctiwz_<mode>): Break long line.
23599 2017-05-05 Bin Cheng <bin.cheng@arm.com>
23601 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
23602 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
23603 (addr_list, addr_offset_valid_p): New.
23604 (split_address_groups): Check offset validity with above function.
23605 (gt-tree-ssa-loop-ivopts.h): Include header file.
23607 2017-05-05 Nathan Sidwell <nathan@acm.org>
23609 * config.gcc (arm*-*-*): Add missing 'fi'.
23611 2017-05-05 Steve Ellcey <sellcey@cavium.com>
23613 * doc/invoke.texi (-fopt-info): Explicitly say order of options
23614 included in -fopt-info does not matter.
23615 * doc/optinfo.texi (-fopt-info): Fix description of default
23616 behavour. Explicitly say order of options included in -fopt-info
23619 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
23621 * config.gcc: Allow combinations of aprofile and rmprofile values for
23622 --with-multilib-list.
23623 * config/arm/t-multilib: New file.
23624 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
23625 variables. Remove setting of ISA and floating-point ABI in
23626 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
23627 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
23628 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
23629 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
23631 * config/arm/t-rmprofile: Likewise except for the matches changes.
23632 * doc/install.texi (--with-multilib-list): Document the combination of
23633 aprofile and rmprofile values and warn about pitfalls in doing that.
23635 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
23637 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
23638 (movdi_aarch64): Likewise.
23640 2017-05-05 Jakub Jelinek <jakub@redhat.com>
23642 PR tree-optimization/80632
23643 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
23645 (build_arrays): Initialize it for virtual phis.
23646 (fix_phi_nodes): Use it for virtual phis.
23648 PR tree-optimization/80558
23649 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
23650 [x, y] op z into [x op, y op z] for op & or | if conditions
23653 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
23654 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23657 * config/arm/arm.md (use_literal_pool): Remove.
23658 (64-bit immediate split): No longer takes cost into consideration
23659 if arm_disable_literal_pool is enabled.
23660 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
23661 used when arm_disable_literal_pool is enabled.
23662 (arm_max_const_double_inline_cost): Remove use of
23663 arm_disable_literal_pool.
23664 (push_minipool_fix): Add assert.
23665 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
23666 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
23667 (no_literal_pool_sf_immediate): New.
23669 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23671 PR tree-optimization/80613
23672 * tree-ssa-dce.c (propagate_necessity): Remove cases for
23673 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
23675 2017-05-05 Richard Biener <rguenther@suse.de>
23677 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
23679 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
23681 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
23682 of this flag from insn conditions due to removal from r247495.
23684 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
23686 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
23688 (arm_early_store_addr_dep_ptr): Likewise.
23689 * config/arm/aarch-common-protos.h
23690 (arm_early_load_addr_dep_ptr): Add prototype.
23691 (arm_early_store_addr_dep_ptr): Likewise.
23692 * config/arm/cortex-a53.md: Add new bypasses.
23694 2017-05-05 Jakub Jelinek <jakub@redhat.com>
23696 * tree.c (next_type_uid): Change type to unsigned.
23697 (type_hash_canon): Decrement back next_type_uid if
23698 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
23699 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
23702 2017-05-04 Martin Sebor <msebor@redhat.com>
23704 * builtins.c: Fix a trivial typo in a comment.
23706 PR middle-end/79234
23707 * builtins.c (check_sizes): Adjust to handle reading past the end.
23708 Avoid printing excessive upper bound of ranges. Use %E to print
23709 tree nodes instead of converting them to %wu.
23710 (expand_builtin_memchr): New function.
23711 (compute_dest_size): Rename...
23712 (compute_objsize): ...to this.
23713 (expand_builtin_memcpy): Adjust.
23714 (expand_builtin_mempcpy): Adjust.
23715 (expand_builtin_strcat): Adjust.
23716 (expand_builtin_strcpy): Adjust.
23717 (check_strncat_sizes): Adjust.
23718 (expand_builtin_strncat): Adjust.
23719 (expand_builtin_strncpy): Adjust and simplify.
23720 (expand_builtin_memset): Adjust.
23721 (expand_builtin_bzero): Adjust.
23722 (expand_builtin_memcmp): Adjust.
23723 (expand_builtin): Handle memcmp.
23724 (maybe_emit_chk_warning): Check strncat just once.
23726 2017-05-04 Martin Sebor <msebor@redhat.com>
23728 PR preprocessor/79214
23729 PR middle-end/79222
23730 PR middle-end/79223
23731 * builtins.c (check_sizes): Add inlining context and issue
23732 warnings even when -Wno-system-headers is set.
23733 (check_strncat_sizes): Same.
23734 (expand_builtin_strncat): Same.
23735 (expand_builtin_memmove): New function.
23736 (expand_builtin_stpncpy): Same.
23737 (expand_builtin): Handle memmove and stpncpy.
23739 2017-05-04 Bin Cheng <bin.cheng@arm.com>
23741 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
23742 which is not used any more.
23744 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
23746 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
23748 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
23750 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
23751 (cortexa53_tunings): Likewise.
23752 (cortexa57_tunings): Likewise.
23753 (cortexa72_tunings): Likewise.
23754 (cortexa73_tunings): Likewise.
23756 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
23758 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
23759 Set loop alignment to 8.
23761 2017-05-04 Martin Sebor <msebor@redhat.com>
23763 PR translation/80280
23764 * builtins.c (expand_builtin_object_size): Add missing quoting to
23765 %D and like directives.
23766 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
23767 (hsa_type_for_tree_type): Same.
23768 (verify_function_arguments): Same.
23769 * symtab.c (symbol_table::change_decl_assembler_name): Same.
23770 * varasm.c (get_section): Same.
23773 2017-05-04 Martin Sebor <msebor@redhat.com>
23775 PR translation/80280
23776 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
23778 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
23780 * config/aarch64/aarch64.c (generic_addrcost_table):
23781 Change HI/TI mode setting.
23783 2017-05-04 Martin Jambor <mjambor@suse.cz>
23785 PR tree-optimization/80622
23786 * tree-sra.c (comes_initialized_p): New function.
23787 (build_accesses_from_assign): Only set write lazily when
23788 comes_initialized_p is false.
23789 (analyze_access_subtree): Use comes_initialized_p.
23790 (propagate_subaccesses_across_link): Assert !comes_initialized_p
23791 instead of testing for PARM_DECL.
23793 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23795 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
23796 constraint on operand 0 to allow more general addressing modes.
23797 Adjust output template.
23798 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
23800 * config/aarch64/aarch64-protos.h
23801 (aarch64_address_valid_for_prefetch_p): Declare prototype.
23802 * config/aarch64/constraints.md (Dp): New address constraint.
23803 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
23806 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
23808 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
23809 update use of estimate_ipcp_clone_size_and_time.
23810 (estimate_local_effects): Update use of
23811 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
23812 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
23813 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
23814 Return nonspecialized time.
23816 2017-05-04 Richard Biener <rguenther@suse.de>
23818 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
23819 for the last VUSE which def dominates the PHI. Directly call
23821 (get_continuation_for_phi_1): Remove.
23823 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
23825 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
23826 to explain the use of truncating division. Cap the number of
23827 iterations to the maximum given by nb_iterations_upper_bound,
23830 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
23832 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
23833 * configure: Regenerate.
23834 * config.in: Regenerate.
23835 * config/i386/driver-mingw32.c: new file.
23836 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
23837 * config.host: Link driver-mingw32.o on MinGW host.
23838 * doc/install.texi: Document new --enable-mingw-wildcard configure
23841 2017-05-04 Marek Polacek <polacek@redhat.com>
23843 PR tree-optimization/80612
23844 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
23846 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23847 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
23849 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
23850 (movt splitter): Likewise.
23851 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
23852 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
23853 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
23854 block for Thumb-1 with MOVT.
23855 (thumb2_legitimate_address_p): Move code block ...
23856 (can_avoid_literal_pool_for_label_p): ... into this new function.
23857 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
23859 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
23860 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
23861 "M-profile targets with the MOVT instruction".
23863 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23865 * config/arm/arm-builtins.c (arm_init_builtins): Rename
23866 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
23867 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
23869 2017-05-04 Martin Liska <mliska@suse.cz>
23871 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
23872 variable cond_code.
23874 2017-05-04 Richard Biener <rguenther@suse.de>
23876 * tree.c (array_at_struct_end_p): Handle arrays at struct
23877 end with flexarrays more conservatively. Refactor and treat
23878 arrays of arrays or aggregates more strict. Fix
23879 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
23880 * tree.c (array_at_struct_end_p): Adjust prototype.
23881 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
23882 * gimple-fold.c (get_range_strlen): Likewise.
23883 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
23885 2017-05-04 Richard Biener <rguenther@suse.de>
23887 PR tree-optimization/31130
23888 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
23890 (supports_overflow_infinity): Likewise.
23891 (is_negative_overflow_infinity): Likewise.
23892 (is_positive_overflow_infinity): Likewise.
23893 (is_overflow_infinity): Likewise.
23894 (stmt_overflow_infinity): Likewise.
23895 (overflow_infinity_range_p): Likewise.
23896 (usable_range_p): Remove as always returning true.
23897 (make_overflow_infinity): Remove.
23898 (negative_overflow_infinity): Likewise.
23899 (positive_overflow_infinity): Likewise.
23900 (avoid_overflow_infinity): Likewise.
23901 (set_value_range): Adjust accordingly.
23902 (set_value_range_to_nonnegative): Likewise, remove now unused
23903 overflow_infinity arg.
23904 (vrp_operand_equal_p): Adjust.
23905 (update_value_range): Likewise.
23906 (range_int_cst_singleton_p): Likewise.
23907 (operand_less_p): Likewise.
23908 (compare_values_warnv): Likewise.
23909 (extract_range_for_var_from_comparison_expr): Likewise.
23910 (vrp_int_const_binop): Likewise.
23911 (zero_nonzero_bits_from_vr): Likewise.
23912 (extract_range_from_multiplicative_op_1): Likewise.
23913 (extract_range_from_binary_expr_1): Likewise.
23914 (extract_range_from_unary_expr): Likewise.
23915 (extract_range_from_comparison): Likewise.
23916 (extract_range_basic): Likewise.
23917 (adjust_range_with_scev): Likewise.
23918 (compare_ranges): Likewise.
23919 (compare_range_with_value): Likewise.
23920 (dump_value_range): Likewise.
23921 (test_for_singularity): Likewise, remove strict_overflow_p parameter
23923 (simplify_cond_using_ranges): Adjust.
23925 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
23927 * brig-builtins.def: Added a builtin for class_f64.
23928 * builtin-types.def: Added a builtin type needed by class_f64.
23930 2017-05-03 Jason Merrill <jason@redhat.com>
23932 * timevar.def: Add TV_CONSTEXPR.
23934 2017-05-03 David Malcolm <dmalcolm@redhat.com>
23936 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
23938 2017-05-03 Martin Jambor <mjambor@suse.cz>
23940 * ipa-prop.c (ipa_update_after_lto_read): Removed.
23941 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
23942 * ipa-cp.c (ipcp_propagate_stage): Do not call
23943 ipa_update_after_lto_read.
23944 * ipa-inline.c (ipa_inline): Likewise.
23946 2017-05-03 Martin Jambor <mjambor@suse.cz>
23948 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
23949 tag. Added a default constructor and a destructor.
23950 (ipa_edge_args_sum_t): New class;
23951 (ipa_edge_args_sum): Declare.
23952 (ipa_edge_args_vector): Remove declaration.
23953 (IPA_EDGE_REF): Use ipa_edge_args_sum.
23954 (ipa_free_edge_args_substructures): Remove declaration.
23955 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
23956 (ipa_edge_args_info_available_for_edge_p): Likewise.
23957 * ipa-prop.c (ipa_edge_args_vector): Removed.
23958 (edge_removal_hook_holder): Likewise.
23959 (edge_duplication_hook_holder): Likewise.
23960 (ipa_edge_args_sum): New variable.
23961 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
23962 ipa_edge_args_vector.
23963 (ipa_free_edge_args_substructures): Likewise.
23964 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
23965 ipa_edge_args_vector.
23966 (ipa_edge_removal_hook): Turned into method
23967 ipa_edge_args_sum_t::remove.
23968 (ipa_edge_duplication_hook): Turned into method
23969 ipa_edge_args_sum_t::duplicate.
23970 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
23971 registering edge hooks.
23972 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
23973 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
23974 ipa_edge_args_sum instead of ipa_edge_args_vector.
23975 * ipa-profile.c (ipa_profile): Likewise.
23977 2017-05-03 Martin Jambor <mjambor@suse.cz>
23979 * symbol-summary.h (function_summary): New method exists.
23980 (function_summary::symtab_removal): Deallocate through release.
23981 (call_summary): New class.
23982 (gt_ggc_mx): New overload.
23983 (gt_pch_nx): Likewise.
23984 (gt_pch_nx): Likewise.
23986 2017-05-03 Jeff Law <law@redhat.com>
23988 PR tree-optimization/78496
23989 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
23990 from simplify_cond_using_ranges. Split off code to walk
23991 backwards through casts into ...
23992 (simplify_cond_using_ranges_2): New function.
23993 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
23994 (execute_vrp): After identifying jump threads, call
23995 simplify_cond_using_ranges_2.
23997 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
24000 * ipa-inline.h (inline_summary): Add ctor.
24001 (create_ggc): Do not use ggc_cleared_alloc.
24003 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
24004 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24006 * gcc.c (handle_braces): Support escaping in switch matching
24008 * doc/invoke.texi (Spec Files): Document it.
24009 Remove superfluous @code markup in items.
24011 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24013 * diagnostic-show-locus.c (struct column_range): New struct.
24014 (get_affected_columns): New function.
24015 (get_printed_columns): New function.
24016 (struct correction): New struct.
24017 (correction::ensure_capacity): New function.
24018 (correction::ensure_terminated): New function.
24019 (struct line_corrections): New struct.
24020 (line_corrections::~line_corrections): New dtor.
24021 (line_corrections::add_hint): New function.
24022 (layout::print_trailing_fixits): Reimplement in terms of the new
24024 (selftest::test_overlapped_fixit_printing): New function.
24025 (selftest::diagnostic_show_locus_c_tests): Call it.
24027 2017-05-03 Nathan Sidwell <nathan@acm.org>
24029 Canonicalize canonical type hashing
24030 * tree.h (type_hash_canon_hash): Declare.
24031 * tree.c (type_hash_list, attribute_hash_list): Move into
24032 type_hash_canon_hash.
24033 (build_type_attribute_qual_variant): Break out hash code calc into
24034 type_hash_canon_hash.
24035 (type_hash_canon_hash): New. Generic type hash computation.
24036 (build_range_type_1, build_array_type_1, build_function_type,
24037 build_method_type_directly, build_offset_type, build_complex_type,
24038 make_vector_type): Call it.
24040 2017-05-03 Richard Biener <rguenther@suse.de>
24042 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
24043 When all DRs have unknown misaligned do not always peel
24044 when there is a store but apply the same costing model as if
24045 there were only loads.
24047 2017-05-03 Richard Biener <rguenther@suse.de>
24050 PR tree-optimization/80492
24051 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
24052 compare_base_decls returning dont-know properly.
24054 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
24056 * config/arm/iterators.md (CCSI): New mode iterator.
24057 (arch): New mode attribute.
24058 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
24059 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
24060 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
24061 code iterator for success result mode.
24062 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
24063 the corresponding new insn generators.
24065 2017-05-03 Bin Cheng <bin.cheng@arm.com>
24068 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24069 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24071 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
24073 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
24074 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
24075 (DDR_A): Wrap DDR argument in brackets.
24076 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
24077 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
24078 (DDR_REVERSED_P): Likewise.
24080 2017-05-03 Jakub Jelinek <jakub@redhat.com>
24082 PR tree-optimization/79472
24083 * tree-switch-conversion.c (struct switch_conv_info): Add
24084 contiguous_range and default_case_nonstandard fields.
24085 (collect_switch_conv_info): Compute contiguous_range and
24086 default_case_nonstandard fields, don't clear final_bb if
24087 contiguous_range and only the default case doesn't have the required
24089 (check_all_empty_except_final): Set default_case_nonstandard instead
24090 of failing if contiguous_range and the default case doesn't have empty
24092 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
24093 and only the default case doesn't have the required constants. Skip
24095 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
24096 if default_case_nonstandard.
24097 (build_constructors): Build constant 1 just once. Assert that default
24098 values aren't inserted in between cases if contiguous_range. Skip
24100 (build_arrays): Skip virtual phis.
24101 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
24102 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
24103 Handle virtual phis.
24104 (gen_inbound_check): Handle default_case_nonstandard case.
24105 (process_switch): Adjust check_final_bb caller. Call
24106 gather_default_values with the first non-default case instead of
24107 default case if default_case_nonstandard.
24109 2017-05-02 Nathan Sidwell <nathan@acm.org>
24111 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
24112 check. Fix formatting.
24114 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
24116 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
24117 errors when comparing specialized and unspecialized times.
24119 2017-05-02 David Malcolm <dmalcolm@redhat.com>
24121 * diagnostic-show-locus.c
24122 (layout::should_print_annotation_line_p): Make private.
24123 (layout::print_annotation_line): Make private.
24124 (layout::annotation_line_showed_range_p): Make private.
24125 (layout::show_ruler): Make private.
24126 (layout::print_source_line): Make private. Pass in line and
24127 line_width, rather than calling location_get_source_line. Drop
24129 (layout::print_leading_fixits): New method.
24130 (layout::print_any_fixits): Rename to...
24131 (layout::print_trailing_fixits): ...this, and make private.
24132 Don't print newline fixits.
24133 (diagnostic_show_locus): Move logic for printing one row into...
24134 (layout::print_line): ...this new function. Move the
24135 location_get_source_line call and error-handling from
24136 print_source_line to here. Call print_leading_fixits, and rename
24137 print_any_fixits to print_trailing_fixits.
24138 (selftest::test_fixit_insert_containing_newline): Update now that
24139 newlines are partially supported.
24140 (selftest::test_fixit_insert_containing_newline_2): New test.
24141 (selftest::test_fixit_replace_containing_newline): Update comments.
24142 (selftest::diagnostic_show_locus_c_tests): Call the new test.
24143 * edit-context.c (class added_line): New class.
24144 (class edited_line): Describe newline handling in comment.
24145 (edited_line::actually_edited_p): New method.
24146 (edited_line::print_content): Delete redundant decl.
24147 (edited_line::m_predecessors): New field.
24148 (edited_file::print_content): Call edited_line::print_content.
24149 (edited_file::print_diff): Update to support newlines.
24150 (edited_file::print_diff_hunk): Likewise.
24151 (edited_file::print_run_of_changed_lines): New function.
24152 (edited_file::print_diff_line): Convert to...
24153 (print_diff_line): ...this.
24154 (edited_file::get_effective_line_count): New function.
24155 (edited_line::edited_line): Initialize new field m_predecessors.
24156 (edited_line::~edited_line): Clean up m_predecessors.
24157 (edited_line::apply_fixit): Handle newlines.
24158 (edited_line::get_effective_line_count): New function.
24159 (edited_line::print_content): New function.
24160 (edited_line::print_diff_lines): New function.
24161 (selftest::test_applying_fixits_insert_containing_newline): New
24163 (selftest::test_applying_fixits_replace_containing_newline): New
24165 (selftest::insert_line): New function.
24166 (selftest::test_applying_fixits_multiple_lines): Add example of
24168 (selftest::edit_context_c_tests): Call the new tests.
24170 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24172 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
24173 parameter cand. Update dump information.
24174 (get_computation_cost): Update uses.
24176 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24178 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
24179 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
24180 (get_computation_at, rewrite_use_address): Update use of
24181 get_computation_aff.
24183 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24185 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
24186 (get_computation): Delete.
24187 (get_computation_cost): Implement like get_computation_cost_at.
24188 Use get_computation_at.
24189 (get_computation_cost_at): Delete.
24190 (rewrite_use_nonlinear_expr): Use get_computation_at.
24191 (rewrite_use_compare, remove_unused_ivs): Ditto.
24193 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24195 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
24197 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24199 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
24200 (ivopts_global_cost_for_size): Rename parameter and update uses.
24201 (iv_ca_recount_cost): Update uses.
24202 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
24203 candidates seperately in n_invs and n_cands.
24204 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
24206 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24208 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
24209 (find_inv_vars_cb): New.
24210 (find_depends): Renamed to ...
24211 (find_inv_vars): ... this.
24212 (add_candidate_1, force_var_cost): Call find_inv_vars.
24213 (split_address_cost, determine_group_iv_cost_cond): Ditto.
24215 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24217 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
24218 inv_vars. Add inv_exprs.
24219 (struct iv_cand): Rename depends_on to inv_vars.
24220 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
24221 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
24222 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
24223 (dump_cand): Dump inv_vars.
24224 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
24225 (record_invariant, find_depends, add_candidate_1): Ditto.
24226 (set_group_iv_cost, force_var_cost): Ditto.
24227 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
24228 (get_computation_cost_at, get_computation_cost): Ditto.
24229 (determine_group_iv_cost_generic): Ditto.
24230 (determine_group_iv_cost_address): Ditto.
24231 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
24232 (determine_group_iv_costs): Ditto.
24233 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
24234 (iv_ca_set_remove_invariants): Renamed to ...
24235 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
24236 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
24237 (iv_ca_set_add_invariants): Renamed to ...
24238 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
24239 (iv_ca_set_cp): Use iv_ca_set_add_invs.
24240 (iv_ca_has_deps): Support inv_vars and inv_exprs.
24241 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
24242 (create_new_ivs): Remove useless dump.
24244 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24246 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
24248 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
24249 (iv_ca_set_no_cp, create_new_iv): Ditto.
24251 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24253 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24255 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
24257 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
24258 function tree_check2.
24260 2017-05-02 Martin Liska <mliska@suse.cz>
24262 * doc/gcov.texi: Add missing preposition.
24263 * gcov.c (function_info::function_info): Properly fill up
24264 all member variables.
24266 2017-05-02 Tamar Christina <tamar.christina@arm.com>
24268 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
24270 2017-05-02 Tamar Christina <tamar.christina@arm.com>
24272 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
24274 2017-05-02 Martin Liska <mliska@suse.cz>
24277 * lto-streamer-in.c (lto_read_tree_1): Remove
24278 LTO_STREAMER_DEBUG.
24279 * lto-streamer.c (struct tree_hash_entry): Likewise.
24280 (struct tree_entry_hasher): Likewise.
24281 (tree_entry_hasher::hash): Likewise.
24282 (tree_entry_hasher::equal): Likewise.
24283 (lto_streamer_init): Likewise.
24284 (lto_orig_address_map): Likewise.
24285 (lto_orig_address_get): Likewise.
24286 (lto_orig_address_remove): Likewise.
24287 * lto-streamer.h: Likewise.
24288 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
24289 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
24291 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
24293 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
24294 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
24295 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
24296 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
24297 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
24298 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
24299 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
24300 (mm_maskz_sub_ss): New intrinsics.
24301 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24302 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
24303 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
24304 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
24305 (__builtin_ia32_subss_mask_round): New builtins.
24306 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24307 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24308 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
24310 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
24311 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
24313 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
24316 2017-05-02 Martin Jambor <mjambor@suse.cz>
24318 PR tree-optimization/78687
24319 * tree-sra.c (access): New field parent.
24320 (process_subtree_disqualification): New function.
24321 (disqualify_candidate): Call it.
24322 (build_accesses_from_assign): Reset write flag if creating an
24324 (build_access_subtree): Fill in parent field and also prpagate
24325 down grp_write flag.
24326 (create_artificial_child_access): New parameter set_grp_write, set
24327 grp_write to its value.
24328 (propagate_subaccesses_across_link): Also propagate grp_write flag
24330 (propagate_all_subaccesses): Push the closest parent back to work
24331 queue if add_access_to_work_queue returned true.
24333 2017-05-02 Richard Biener <rguenther@suse.de>
24335 * common.opt (fstrict-overflow): Alias negative to fwrapv.
24336 * doc/invoke.texi (fstrict-overflow): Remove all traces of
24337 -fstrict-overflow documentation.
24338 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
24339 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
24340 flag_strict_overflow.
24341 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
24342 * lto-opts.c (lto_write_options): Do not stream it.
24343 * lto-wrapper.c (merge_and_complain): Do not handle it.
24344 * opts.c (default_options_table): Do not set -fstrict-overflow.
24345 (finish_options): Likewise do not clear it when sanitizing.
24346 * simplify-rtx.c (simplify_const_relational_operation): Do not
24347 test flag_strict_overflow.
24349 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
24351 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
24352 using enabled attribute.
24353 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
24354 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
24355 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
24356 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
24357 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
24358 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
24359 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
24360 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
24361 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
24362 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
24364 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
24366 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
24368 2017-05-02 Richard Biener <rguenther@suse.de>
24370 PR tree-optimization/80591
24372 2017-04-10 Richard Biener <rguenther@suse.de>
24374 * tree-ssa-structalias.c (find_func_aliases): Properly handle
24377 2017-05-02 Richard Biener <rguenther@suse.de>
24379 PR tree-optimization/80549
24380 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
24381 (cleanup_tree_cfg_noloop): Create forwarders to known loop
24382 headers if they do not have a preheader.
24384 2017-05-02 Martin Liska <mliska@suse.cz>
24387 * common.opt: Fix typo.
24388 * doc/invoke.texi: Likewise.
24390 2017-05-01 Jan Beulich <jbeulich@suse.com>
24392 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
24393 swapping, add (x,x,m,x,n) alternative.
24395 2017-05-01 Nathan Sidwell <nathan@acm.org>
24397 * calls.c (combine_pending_stack_adjustment_and_call): Remove
24398 unnecessary unadjusted_alignment check.
24400 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
24403 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
24405 * gimplify.c (gimplify_cilk_detach): New function.
24406 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
24407 * tree-core.h: Document EXPR_CILK_SPAWN.
24408 * tree.h (EXPR_CILK_SPAWN): Define.
24410 2017-05-01 David Malcolm <dmalcolm@redhat.com>
24412 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
24413 to use new fixit_hint representation, using the "replace" logic.
24414 (get_line_span_for_fixit_hint): Likewise.
24415 (layout::print_any_fixits): Likewise.
24416 (selftest::test_one_liner_many_fixits): Rename to...
24417 (selftest::test_one_liner_many_fixits_1): ...this, and update
24418 comment and expected output to reflect that the multiple fix-it
24419 hints are now consolidated into one insertion.
24420 (selftest::test_one_liner_many_fixits_2): New test.
24421 (selftest::test_diagnostic_show_locus_one_liner): Update for
24423 (selftest::test_fixit_consolidation): Update for fix-it API
24425 * diagnostic.c (print_parseable_fixits): Likewise.
24426 * edit-context.c (edited_line::m_line_events): Convert from
24427 auto_vec <line_event *> to auto_vec <line_event>.
24428 (class line_event): Convert from abstract base class to a concrete
24429 class, taking over the role of replace_event.
24430 (class insert_event): Delete.
24431 (class replace_event): Rename to class line_event. Convert to
24433 (edit_context::add_fixits): Reimplement.
24434 (edit_context::apply_insert): Delete.
24435 (edit_context::apply_replace): Rename to...
24436 (edit_context::apply_fixit): ...this. Convert to half-open range.
24437 (edited_file::apply_insert): Delete.
24438 (edited_file::apply_replace): Rename to...
24439 (edited_file::apply_fixit): ...this.
24440 (edited_line::~edited_line): Drop deletion of events.
24441 (edited_line::apply_insert): Delete.
24442 (edited_line::apply_replace): Rename to...
24443 (edited_line::apply_fixit): ...this. Convert to half-open range.
24444 Update for change to type of m_line_events.
24445 * edit-context.h (edit_context::apply_insert): Delete.
24446 (edit_context::apply_replace): Rename to...
24447 (edit_context::apply_fixit): ...this.
24449 2017-05-01 Martin Sebor <msebor@redhat.com>
24451 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
24454 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
24457 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
24458 __get_cpuid_max returns 0.
24459 (__get_cpuid_count): Ditto.
24461 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
24463 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
24464 replacement expression is another instance of one of its arguments.
24466 2017-05-01 Jakub Jelinek <jakub@redhat.com>
24469 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
24470 check for stack push/pop autoinc.
24471 * config/i386/i386.c (ix86_agi_dependent): Return false
24472 if the only reason why modified_in_p returned true is that
24473 addr is SP based and set_insn is a push or pop.
24475 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
24477 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
24480 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
24483 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
24484 (account_size_time): Use two predicates - exec_pred and
24486 (evaluate_conditions_for_known_args): Compute both clause and
24488 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
24489 (inline_summary_t::duplicate): Update.
24490 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
24492 (compute_inline_parameters): Likewise.
24493 (estimate_edge_size_and_time): Update caluclation of time.
24494 (estimate_node_size_and_time): Compute both time and nonspecialized
24496 (estimate_ipcp_clone_size_and_time): Update.
24497 (inline_merge_summary): Update.
24498 (do_estimate_edge_time): Update.
24499 (do_estimate_edge_size): Update.
24500 (do_estimate_edge_hints): Update.
24501 (inline_read_section, inline_write_summary): Stream both new predicates.
24502 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
24504 (compute_inlined_call_time): Cleanup.
24505 (big_speedup_p): Update.
24506 (edge_badness): Update.
24507 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
24508 (size_time_entry): Replace predicate by exec_predicate and
24509 nonconst_predicate.
24510 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
24511 (estimate_edge_time): Return also nonspec_time.
24512 (reset_edge_growth_cache): Update.
24514 2017-04-29 Jakub Jelinek <jakub@redhat.com>
24516 PR rtl-optimization/80491
24517 * ifcvt.c (noce_process_if_block): When looking for x setter
24518 with missing else_bb, don't check only the insn right before
24519 cond_earliest, but look for the last insn that x is modified in
24520 within the same bb.
24522 PR rtl-optimization/80491
24523 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
24525 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
24527 PR tree-optimization/80487
24528 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
24530 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24532 PR tree-optimization/79697
24533 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
24534 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
24535 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
24537 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
24538 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
24540 2017-04-28 Martin Sebor <msebor@redhat.com>
24542 PR tree-optimization/80523
24543 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
24544 (init_target_to_host_charmap, target_to_host, target_strtol10): New
24546 (maybe_warn, format_directive, parse_directive): Use new functions.
24547 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
24549 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
24551 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
24553 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
24555 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
24556 target_header_dir): Set correctly.
24557 * configure: Regenerated.
24558 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
24559 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
24560 instead of SYSTEM_HEADER_DIR.
24562 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
24564 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
24565 (estimate_local_effects): Likewise.
24566 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
24567 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
24568 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
24569 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
24570 do_estimate_edge_time, estimate_edge_time): Likewise.
24571 * ipa-inline-analysis.c (estimate_node_size_and_time,
24572 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
24573 (estimate_time_after_inlining): Remove.
24575 2017-04-28 Martin Liska <mliska@suse.cz>
24577 * doc/gcov.texi: Enhance documentation of gcov.
24579 2017-04-28 Martin Liska <mliska@suse.cz>
24581 * doc/gcov.texi: Sort options in alphabetic order.
24582 * doc/gcov-dump.texi: Likewise.
24583 * doc/gcov-tool.texi: Likewise.
24584 * gcov.c (print_usage): Likewise.
24585 * gcov-dump.c (print_usage): Likewise.
24586 * gcov-tool.c (print_merge_usage_message): Likewise.
24587 (print_rewrite_usage_message): Likewise.
24588 (print_overlap_usage_message): Likewise.
24590 2017-04-28 Martin Liska <mliska@suse.cz>
24592 PR gcov-profile/53915
24593 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
24595 2017-04-28 Martin Liska <mliska@suse.cz>
24597 PR gcov-profile/79891
24598 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
24599 is marked by compiler as living on a line.
24600 (get_cycles_count): Remove usage of the union.
24601 (output_intermediate_file): Likewise.
24602 (find_source): Fix GNU coding style.
24603 (accumulate_line_counts): Remove old non-all block mode.
24604 (output_lines): Remove usage of the union.
24605 * profile.c (output_location): Include all BBs, even if
24606 belonging to a same line (and file) as a previous BB.
24608 2017-04-28 Martin Liska <mliska@suse.cz>
24610 * gcov.c (process_args): Handle new argument 'w'.
24611 (read_graph_file): Assign ID to BBs.
24612 (output_branch_count): Display BB # if verbose flag is set.
24613 (output_lines): Likewise for arcs.
24614 (print_usage): Add '--verbose' option help.
24615 * doc/gcov.texi: Document --verbose (-w) option.
24617 2017-04-28 Martin Liska <mliska@suse.cz>
24619 * gcov.c (struct block_location_info): New struct.
24620 (process_file): Fill up the new structure.
24621 (read_graph_file): Replace usage of encoding by the newly added
24623 (add_line_counts): Likewise.
24624 (accumulate_line_counts): Remove usage of the union.
24625 (function_info::function_info): New function.
24626 (function_info::~function_info): Likewise.
24627 (process_file): Call delete instead of release_function.
24628 (release_function): Release the function.
24629 (release_structures): Call delete instead of release_function.
24630 (solve_flow_graph): Replace usage of num_blocks.
24631 (find_exception_blocks): Likewise.
24632 (output_lines): Fix GNU coding style.
24634 2017-04-28 Martin Liska <mliska@suse.cz>
24637 * coverage.c (coverage_remove_note_file): New function.
24638 * coverage.h: Declare the function.
24639 * toplev.c (finalize): Clean if an error has been seen.
24641 2017-04-28 Martin Liska <mliska@suse.cz>
24643 PR gcov-profile/80031
24644 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
24645 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
24646 * gcov.c (read_graph_file): Read just number of blocks.
24647 * profile.c (branch_prob): Do not stream 0 flags per a basic
24650 2017-04-28 Martin Liska <mliska@suse.cz>
24652 * gcov-dump.c (tag_*): Add new argument to declarations.
24653 (dump_gcov_file): Likewise.
24654 (tag_blocks): Add and use new argument depth.
24655 (tag_arcs): Likewise.
24656 (tag_lines): Likewise.
24657 (tag_counters): Likewise.
24658 (tag_summary): Likewise.
24659 (dump_working_sets): Use depth to do a proper indentation.
24661 2017-04-28 Jakub Jelinek <jakub@redhat.com>
24664 * cgraph.h (symtab_node::debug_symtab): No longer inline.
24665 * symtab.c (symtab_node::debug_symtab): Move definition here.
24667 2017-04-28 Richard Biener <rguenther@suse.de>
24669 * lto-streamer.h (LTO_major_version): Bump to 7.
24671 2017-04-28 Richard Biener <rguenther@suse.de>
24673 * tree-vrp.c (assert_info): New struct.
24674 (add_assert_info): New helper.
24675 (register_edge_assert_for_2): Refactor to add asserts to a vector
24677 (register_edge_assert_for_1): Likewise.
24678 (register_edge_assert_for): Likewise.
24679 (finish_register_edge_assert_for): New helper actually registering
24680 asserts where live on edge.
24681 (find_conditional_asserts): Adjust.
24682 (find_switch_asserts): Likewise.
24683 (evrp_dom_walker::try_find_new_range): Generalize.
24684 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
24686 2017-04-27 Marek Polacek <polacek@redhat.com>
24689 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
24690 arg10 and arg11 to itype.
24692 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
24694 * doc/extend.texi (Object Size Checking): Improve grammar.
24696 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
24699 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
24700 that the logic for permitting reciprocal estimates matches that
24703 2017-04-27 Jakub Jelinek <jakub@redhat.com>
24706 * tree.c (type_cache_hasher::equal): Only compare
24707 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
24708 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
24709 non-aggregate element types.
24710 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
24711 about the flag on ARRAY_TYPEs in the comment, formatting fix.
24713 2017-04-27 Richard Biener <rguenther@suse.de>
24715 PR middle-end/80533
24716 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
24717 stripping ARRAY_REFs from MEM_EXPR make sure we're not
24718 keeping a reference to a trailing array.
24720 2017-04-27 Richard Biener <rguenther@suse.de>
24722 PR middle-end/80539
24723 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
24724 being in loop-closed SSA form conservatively.
24725 (chrec_fold_multiply_poly_poly): Likewise.
24727 2017-04-27 Tamar Christina <tamar.christina@arm.com>
24729 PR middle-end/79665
24730 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
24731 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
24733 2017-04-27 Jakub Jelinek <jakub@redhat.com>
24736 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
24737 (aarch64_function_arg_alignment): Return unsigned int again, but still
24738 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
24739 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
24740 Don't emit -Wpsabi note.
24741 (aarch64_function_arg_boundary): Likewise.
24742 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
24745 2017-04-26 Nathan Sidwell <nathan@acm.org>
24747 * tree.h (crc32_unsigned_n): Declare.
24748 (crc32_unsigned, crc32_unsigned): Make inline.
24749 * tree.c (crc32_unsigned_bits): Replace with ...
24750 (crc32_unsigned_n): ... this.
24751 (crc32_unsigned, crc32_byte): Remove.
24752 (crc32_string): Remove unnecessary braces.
24754 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
24756 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
24757 * ipa-inline-analysis.c (MAX_TIME): Remove.
24758 (account_size_time): Use sreal for time.
24759 (dump_inline_summary): Update.
24760 (estimate_function_body_sizes): Update.
24761 (estimate_edge_size_and_time): Update.
24762 (estimate_calls_size_and_time): Update.
24763 (estimate_node_size_and_time): Update.
24764 (inline_merge_summary): Update.
24765 (inline_update_overall_summary): Update.
24766 (estimate_time_after_inlining): Update.
24767 (inline_read_section): Update.
24768 (inline_write_summary): Update.
24769 * ipa-inline.c (compute_uninlined_call_time): Update.
24770 (compute_inlined_call_time): Update.
24771 (recursive_inlining): Update.
24772 (inline_small_functions): Update.
24773 (dump_overall_stats): Update.
24774 * ipa-inline.h: Include sreal.h.
24775 (size_time_entry): Turn time to sreal.
24776 (inline_summary): Turn self_time nad time to sreal.
24778 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
24780 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
24782 (sreal::stream_out, sreal::stream_in): New.
24783 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
24785 2017-04-25 Jakub Jelinek <jakub@redhat.com>
24787 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
24790 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
24793 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
24795 (dimode_scalar_chain::compute_convert_gain): Ditto.
24796 (dimode_scalar_chain::make_vector_copies): Ditto.
24797 (dimode_scalar_chain::convert_reg): Ditto.
24798 (dimode_scalar_chain::convert_insn): Ditto.
24799 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
24800 (VI248_AVX512BW_1): New mode iterator.
24801 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
24802 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
24805 2017-04-25 Martin Sebor <msebor@redhat.com>
24807 PR tree-optimization/80497
24808 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
24809 constants are representable in HOST_WIDE_INT.
24810 (parse_directive): Ditto.
24812 2017-04-25 Martin Sebor <msebor@redhat.com>
24815 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
24816 (new_zero_array): Adjust signature.
24817 (dom_info::dom_init): Used unsigned rather that size_t.
24818 (dom_info::dom_info): Same.
24820 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24821 Jakub Jelinek <jakub@redhat.com>
24824 * config/arm/arm.c: Include gimple.h.
24825 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
24826 returns negative, increment ncrn only if it returned positive.
24827 (arm_needs_doubleword_align): Return int instead of bool,
24828 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
24829 members, but if there is any such non-FIELD_DECL
24830 > PARM_BOUNDARY aligned decl, return -1 instead of false.
24831 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
24832 returns negative, increment nregs only if it returned positive.
24833 (arm_setup_incoming_varargs): Likewise.
24834 (arm_function_arg_boundary): Emit -Wpsabi note if
24835 arm_needs_doubleword_align returns negative, return
24836 DOUBLEWORD_ALIGNMENT only if it returned positive.
24838 2017-04-25 Marek Polacek <polacek@redhat.com>
24841 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
24842 first argument to type.
24844 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
24847 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
24848 type checks to test for compatibility instead of equality.
24850 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24851 Jakub Jelinek <jakub@redhat.com>
24854 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
24856 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
24857 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
24858 the alignment computation, but return their maximum in warn_alignment.
24859 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
24860 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
24862 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
24863 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
24866 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24868 * config/arc/simdext.md (dmpyh): Fix typo.
24870 2017-04-25 Richard Biener <rguenther@suse.de>
24872 PR tree-optimization/80492
24873 * alias.c (compare_base_decls): Handle registers with asm
24874 specification conservatively.
24875 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
24876 compare_base_decls returning dont-know properly.
24878 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24880 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
24881 (legitimate_offset_address_p): New function.
24882 (arc_legitimate_address_p): Use above function.
24884 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24886 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
24888 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24890 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
24891 ACCH registers whenever they are available.
24893 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24895 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
24896 double regs fix when not used.
24898 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24900 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
24902 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
24903 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
24905 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24907 * config/arc/arc.c (arc_output_addsi): Check for h-register class
24908 when emitting short ADD instructions.
24910 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24912 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
24914 (cmpsi_cc_c_insn): Likewise.
24915 (cbranchsi4_scratch): Compute proper instruction length using
24916 compact_hreg_operand.
24917 * config/arc/predicates.md (compact_hreg_operand): New predicate.
24919 2017-04-25 Richard Biener <rguenther@suse.de>
24921 PR middle-end/80509
24922 * passes.c (pass_manager::pass_manager): Initialize
24923 m_name_to_pass_map.
24925 2017-04-25 Richard Biener <rguenther@suse.de>
24927 PR tree-optimization/79201
24928 * tree-ssa-sink.c (statement_sink_location): Handle calls.
24930 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24933 * config/s390/vector.md: Split MEM->GPR vector moves for
24934 non-s_operand addresses.
24936 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24939 * config/s390/predicates.md (reload_const_wide_int_operand): New
24941 * config/s390/s390.md ("movti"): Remove d/P alternative.
24942 ("movti_bigconst"): New pattern definition.
24944 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
24947 * s390-protos.h (s390_expand_cs_hqi): Removed.
24948 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
24949 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
24950 modes as well as CCZ1mode and CCZmode.
24951 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
24952 signature of s390_emit_compare_and_swap.
24953 (s390_expand_cs_hqi): Likewise, make static.
24954 (s390_expand_cs_tdsi): Generate an explicit compare before trying
24955 compare-and-swap, in some cases.
24956 (s390_expand_cs): Wrapper function.
24957 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
24959 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
24960 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
24961 patterns for small and large integers. Forbid symref memory operands.
24962 Move expander to s390.c. Require cc register.
24963 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
24964 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
24965 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
24966 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
24967 symref memory operands. Remove CC mode and call s390_match_ccmode
24969 ("atomic_exchange<mode>"): Allow and implement all integer modes.
24971 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
24973 * config/s390/s390.md (define_peephole2): New peephole to help
24974 combining the load-and-test pattern with volatile memory.
24976 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
24978 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
24979 with CCZmode for TARGET_Z196.
24981 2017-04-25 Jakub Jelinek <jakub@redhat.com>
24983 PR rtl-optimization/80501
24984 * combine.c (make_compound_operation_int): Set subreg_code to SET
24985 even for AND with mask of the sign bit of mode.
24987 PR rtl-optimization/80500
24988 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
24989 sum's initial value.
24991 2017-04-25 Julian Brown <julian@codesourcery.com>
24992 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24994 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
24996 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
24998 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
25000 2017-04-25 Julian Brown <julian@codesourcery.com>
25001 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25003 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
25004 (thunderx2t99_sha): New Reservation.
25006 2017-04-25 Julian Brown <julian@codesourcery.com>
25007 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25009 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
25010 type for 1-element load.
25012 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
25014 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
25016 2017-04-24 Martin Jambor <mjambor@suse.cz>
25018 PR tree-optimization/80293
25019 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
25020 char arrays not totally scalarizable if it is false.
25021 (analyze_all_variable_accesses): Pass correct value in the new
25022 parameter. Add a statistics counter.
25024 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
25026 PR middle-end/79931
25027 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
25029 2017-04-24 Richard Biener <rguenther@suse.de>
25031 PR tree-optimization/80494
25032 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
25033 out for complex types.
25035 2017-04-24 Richard Biener <rguenther@suse.de>
25037 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
25038 * tree-ssa-sccvn.c (print_scc): Print SCC size.
25039 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
25040 (DFS): Adjust and never fail.
25041 (sccvn_dom_walker::fail): Remove.
25042 (sccvn_dom_walker::before_dom_children): Adjust.
25043 (run_scc_vn): Likewise and never fail.
25044 * tree-ssa-pre.c (pass_pre::execute): Adjust.
25045 (pass_fre::execute): Likewise.
25047 2017-04-24 Richard Biener <rguenther@suse.de>
25049 PR tree-optimization/79725
25050 * tree-ssa-sink.c (statement_sink_location): Return whether
25051 failure reason was zero uses. Move that check later.
25052 (sink_code_in_bb): Deal with zero uses by removing the stmt
25055 2017-04-24 Richard Biener <rguenther@suse.de>
25058 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
25059 pointer-based references.
25061 2017-04-24 Richard Biener <rguenther@suse.de>
25064 * pass_manager.h (pass_manager::operator new): Remove.
25065 (pass_manager::operator delete): Likewise.
25066 * passes.c (pass_manager::operator new): Remove.
25067 (pass_manager::operator delete): Likewise.
25068 (pass_manager::pass_manager): Zero individual pass members.
25070 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
25073 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
25074 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
25075 Check "XEXP (src, 1)" operand here.
25076 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
25077 Check "XEXP (src, 1)" operand here.
25078 (dimode_scalar_chain::make_vector_copies): Detect count register
25079 of a shift instruction. Zero extend count register from QImode
25080 to DImode to satisfy vector shift pattern count operand predicate.
25081 Substitute vector shift count operand with a DImode copy.
25082 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
25085 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
25087 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
25088 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
25089 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
25090 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
25091 (UNSPEC_NOREX_MEM): Remove definition.
25093 2017-04-21 Richard Biener <rguenther@suse.de>
25095 PR tree-optimization/79547
25096 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
25097 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
25098 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
25099 without any constraints.
25101 2017-04-21 Richard Biener <rguenther@suse.de>
25103 PR tree-optimization/78847
25104 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
25106 2017-04-21 Richard Biener <rguenther@suse.de>
25108 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
25109 (build_distinct_type_copy): Likewise.
25110 (build_variant_type_copy): Likewise.
25111 * tree.c (build_qualified_type): Pass down mem-stat info.
25112 (build_distinct_type_copy): Likewise.
25113 (build_variant_type_copy): Likewise.
25115 2017-04-21 Richard Biener <rguenther@suse.de>
25117 PR tree-optimization/80237
25118 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
25120 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
25121 for a simplified result.
25123 2016-04-21 Richard Biener <rguenther@suse.de>
25125 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
25126 sth as strict as a simple_iv but a chrec without symbols and an
25127 operand defined in the loop we are peeling (and not some subloop).
25128 (propagate_constants_for_unrolling): Propagate all constants.
25130 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25133 * config/i386/i386.c (print_reg): Remove assert for disalowed
25134 regno values, call output_operand_lossage instead.
25136 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25139 * config/i386/constraints.md (Yc): New register constraint.
25140 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
25141 Use Yc constraint for alternative 2 of operand 0. Remove
25142 preferred_for_speed attribute.
25144 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
25146 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
25147 lastprivate clauses in SIMT case.
25149 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
25151 * doc/invoke.texi (-Wextra-semi): Document new warning option.
25153 2017-04-20 Richard Biener <rguenther@suse.de>
25155 PR tree-optimization/57796
25156 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
25157 as N scalar stores.
25158 (vect_model_load_cost): Cost gathers as N scalar loads.
25160 2017-04-20 Richard Biener <rguenther@suse.de>
25162 * ggc-page.c (ggc_allocated_p): Rename to ...
25163 (safe_lookup_page_table_entry): ... this and return the lookup
25165 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
25167 2017-04-20 Richard Biener <rguenther@suse.de>
25169 PR tree-optimization/80453
25170 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
25171 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
25172 from the conditions.
25173 (vn_phi_eq): Pass them down.
25174 (vn_phi_lookup): Record them.
25175 (vn_phi_insert): Likewise.
25177 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
25179 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
25180 uninitialized variable warning to avoid buffer overrun.
25182 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
25185 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
25186 is suppressed for '{ 0 }' in C.
25188 2017-04-20 Jakub Jelinek <jakub@redhat.com>
25190 * BASE-VER: Set to 8.0.0.
25192 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
25194 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
25195 priority .init_array and .fini_array section with SECTION_NOTYPE
25198 2017-04-20 Jakub Jelinek <jakub@redhat.com>
25200 PR middle-end/80423
25201 * tree.h (build_array_type): Add typeless_storage default argument.
25202 * tree.c (type_cache_hasher::equal): Also compare
25203 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
25204 (build_array_type): Add typeless_storage argument, set
25205 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
25207 (build_nonshared_array_type): Adjust build_array_type_1 caller.
25208 (build_array_type): Likewise. Add typeless_storage argument.
25210 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
25211 Jakub Jelinek <jakub@redhat.com>
25213 PR tree-optimization/80426
25214 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
25215 operation on symbolic operands, also compute the overflow for the
25216 invariant part when the operation degenerates into a negation.
25218 2017-04-19 Jakub Jelinek <jakub@redhat.com>
25221 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
25222 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
25225 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
25227 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
25230 * config/avr/avr.c (tree.h): Include it.
25231 (cgraph.h): Include it.
25232 (avr_encode_section_info): Don't warn for uninitialized progmem
25233 variable if it's just an alias.
25235 2017-04-19 Richard Biener <rguenther@suse.de>
25238 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
25239 when needed by AutoPGO.
25241 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
25244 * doc/lto.texi: Remove an extra 'that'.
25246 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
25248 PR rtl-optimization/80429
25249 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
25250 are only used in debug insns.
25252 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
25253 Vladimir Makarov <vmakarov@redhat.com>
25255 * config/sparc/predicates.md (input_operand): Add comment. Return
25256 true for any memory operand when LRA is in progress.
25257 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
25259 2017-04-18 Jeff Law <law@redhat.com>
25262 * mips.md ({return,simple_return}_internal): Do not overwrite
25265 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25267 PR tree-optimization/80443
25268 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
25269 instead of adding 1, subtract -1 and similarly instead of subtracting
25272 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
25274 PR rtl-optimization/80357
25275 * haifa-sched.c (tmp_bitmap): New variable.
25276 (model_recompute): Handle duplicate use records.
25277 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
25278 (free_global_sched_pressure_data): Free it.
25280 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
25283 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
25284 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25285 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25286 instead of SYSTEM_HEADER_DIR.
25288 2017-04-18 Jeff Law <law@redhat.com>
25290 PR middle-end/80422
25291 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
25292 predecessors after walking up the insn chain.
25294 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25297 * dwarf2out.c (modified_type_die): Try harder not to emit internal
25298 sizetype type into debug info.
25300 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
25303 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
25304 unneeded test for TARGET_UPPER_REGS_SF.
25305 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
25307 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25310 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
25311 instead of gsi_after_labels.
25313 2017-04-18 Jeff Law <law@redhat.com>
25315 * regcprop.c (maybe_mode_change): Avoid creating copies of the
25319 2017-04-13 Jeff Law <law@redhat.com>
25320 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
25321 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
25323 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
25326 * config/avr/avr.c (intl.h): Include it.
25327 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
25329 2017-04-18 Martin Liska <mliska@suse.cz>
25331 PR gcov-profile/78783
25332 * gcov-tool.c (gcov_output_files): Validate that destination
25333 file is either removed by the tool or by a user.
25335 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
25336 Guy Benyei <guybe@mellanox.com>
25338 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
25339 block, and do not negate it, the stored id is already negative.
25341 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
25343 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
25345 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
25348 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
25349 masks of options that should be turned off if the VSX vector
25350 options are turned off.
25351 (OTHER_P8_VECTOR_MASKS): Likewise.
25352 (OTHER_VSX_VECTOR_MASKS): Likewise.
25353 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
25354 rs6000_disable_incompatible_switches to validate no type switches
25356 (rs6000_incompatible_switch): New function to disallow turning on
25357 other vector options if -mno-vsx, -mno-power8-vector, or
25358 -mno-power9-vector are specified.
25360 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25362 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
25364 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25366 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
25367 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
25368 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
25369 (ARG_POINTER_CFA_OFFSET): Likewise.
25371 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25373 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
25374 conditions to take advantage of various optimizations.
25376 2017-04-13 Jeff Law <law@redhat.com>
25378 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
25379 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
25380 (zero_extendsidi2_dext): Likewise.
25382 2017-04-13 Jakub Jelinek <jakub@redhat.com>
25385 * fold-const.c (fold_ternary_loc): Revert
25386 use op0 instead of fold_convert_loc (loc, type, arg0) part of
25389 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
25391 PR rtl-optimization/80343
25392 * lra-remat.c (update_scratch_ops): Assign original hard reg to
25393 new scratch pseudo.
25395 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
25398 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
25399 to ubsan_encode_value.
25401 2017-04-13 Jeff Law <law@redhat.com>
25403 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
25404 appearing in DEBUG_INSNs.
25406 2017-04-13 Martin Liska <mliska@suse.cz>
25408 PR gcov-profile/80413
25409 * gcov-io.c (gcov_write_string): Copy to buffer just when
25410 allocated size is greater than zero.
25412 2017-04-13 Jakub Jelinek <jakub@redhat.com>
25415 * dwarf2out.c (decls_for_scope): Ignore declarations of
25416 current_function_decl in BLOCK_NONLOCALIZED_VARS.
25418 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
25421 * ipa-visibility.c (non_local_p): Fix typos.
25422 (localize_node): When localizing symbol in same comdat group,
25423 dissolve the group only when we know external symbols are going
25425 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
25427 2017-04-12 Jakub Jelinek <jakub@redhat.com>
25429 PR tree-optimization/79390
25430 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
25431 order does not result in usable sequence, retry with reversed operand
25437 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
25438 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
25439 op0 instead of fold_convert_loc (loc, type, arg0).
25441 2017-04-12 Jeff Law <law@redhat.com>
25443 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
25444 has a delay slot in the generated code.
25446 * config/cris/cris.md (cris_preferred_reload_class): Return
25447 GENNONACR_REGS rather than GENERAL_REGS.
25449 2017-04-12 Jakub Jelinek <jakub@redhat.com>
25452 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
25453 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
25454 signedness of the result type.
25456 2017-04-12 Richard Biener <rguenther@suse.de>
25457 Jeff Law <law@redhat.com>
25459 PR tree-optimization/80359
25460 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
25461 trim stores to TARGET_MEM_REFs.
25463 2017-04-12 Richard Biener <rguenther@suse.de>
25465 PR tree-optimization/79390
25466 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
25467 threading case even more.
25469 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
25472 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
25473 for quad_address_p for TImode, instead of just not indexed_address.
25475 2017-04-12 Richard Biener <rguenther@suse.de>
25476 Bernd Edlinger <bernd.edlinger@hotmail.de>
25478 PR middle-end/79671
25479 * alias.c (component_uses_parent_alias_set_from): Handle
25480 TYPE_TYPELESS_STORAGE.
25481 (get_alias_set): Likewise.
25482 * tree-core.h (tree_type_common): Add typeless_storage flag.
25483 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
25484 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
25485 for types containing members with TYPE_TYPELESS_STORAGE.
25486 (place_field): Likewise.
25487 (layout_type): Likewise for ARRAY_TYPE.
25488 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
25489 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
25490 TYPE_TYPELESS_STORAGE.
25491 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
25493 2017-04-12 Jakub Jelinek <jakub@redhat.com>
25496 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
25497 first argument to type.
25499 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25503 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
25504 CONST0_RTX (mode) rather than const0_rtx where appropriate.
25505 (rs6000_expand_binop_builtin): Likewise.
25506 (rs6000_expand_ternop_builtin): Likewise; also add missing
25507 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
25508 vshasigma built-ins.
25509 * doc/extend.texi: Document that vec_xxpermdi's third argument
25510 must be a constant.
25512 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
25514 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
25515 Use shift_const cost parameter when calculating gain of STV shifts.
25517 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
25519 PR rtl-optimization/70478
25520 * lra-constraints.c (process_alt_operands): Check memory for
25521 disfavoring memory insn operand.
25523 2017-04-11 Jakub Jelinek <jakub@redhat.com>
25525 PR middle-end/80100
25526 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
25527 left shift in unsigned HOST_WIDE_INT type.
25529 PR rtl-optimization/80385
25530 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
25531 (not (neg X)) into (plus X -1) for complex or non-integral modes.
25534 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
25535 if they have any depend clauses.
25537 2017-04-11 Martin Liska <mliska@suse.cz>
25540 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
25541 * ipa-split.c (split_function): Create a local comdat symbol
25542 if caller is in a comdat group.
25544 2017-04-11 Martin Liska <mliska@suse.cz>
25547 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
25550 2017-04-11 Martin Sebor <msebor@redhat.com>
25552 PR middle-end/80364
25553 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
25554 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
25556 (directive::set_width, directive::set_precision, format_character):
25558 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
25561 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
25564 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
25565 conflict, set target->arch_name instead of target->cpu_name.
25567 2017-04-11 Richard Biener <rguenther@suse.de>
25569 PR tree-optimization/80374
25570 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
25571 build_zero_cst, remove fold_convertible_p check again.
25573 2017-04-11 Martin Liska <mliska@suse.cz>
25576 * ubsan.c (instrument_object_size): Do not instrument register
25579 2017-04-11 Jakub Jelinek <jakub@redhat.com>
25582 * config/i386/i386-builtin-types.def
25583 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
25584 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
25585 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
25586 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
25587 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
25588 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
25589 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
25590 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
25591 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
25592 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
25593 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
25594 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
25595 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
25596 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
25597 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
25598 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
25599 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
25600 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
25601 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
25602 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
25603 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
25604 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
25605 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
25606 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
25607 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
25608 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
25609 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
25610 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
25611 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
25612 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
25613 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
25614 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
25615 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
25616 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
25617 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
25618 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
25619 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
25620 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
25621 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
25622 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
25623 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
25624 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
25625 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
25626 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
25627 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
25628 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
25630 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
25631 flag to second_arg_count, handle 4 argument function type _COUNT
25632 aliases, handle second_arg_count on second argument rather than last.
25634 2017-04-10 Jeff Law <law@redhat.com>
25636 PR tree-optimization/80374
25637 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
25638 record anything if we can not convert integer_zero_node to the
25641 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
25644 * config/rs6000/rs6000.c (rs6000_option_override_internal):
25645 Enhance special handling given to the TARGET_P9_MINMAX option in
25646 relation to certain other options.
25648 2017-04-10 Bin Cheng <bin.cheng@arm.com>
25650 PR tree-optimization/80153
25651 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
25652 remove POINTER_PLUS_EXPR's base part directly, rather than through
25655 2017-04-10 Richard Biener <rguenther@suse.de>
25656 Bin Cheng <bin.cheng@arm.com>
25658 PR tree-optimization/80153
25659 * tree-affine.c (aff_combination_to_tree): Get base pointer from
25660 the first element of pointer type aff_tree. Build result expr in
25662 (add_elt_to_tree): Convert to type unconditionally. Remove other
25663 fold_convert calls.
25664 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
25665 (rewrite_use_nonlinear_expr): Check invariant using iv information.
25667 2017-04-10 Richard Biener <rguenther@suse.de>
25669 * tree-ssa-structalias.c (find_func_aliases): Properly handle
25672 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
25674 PR rtl-optimization/70478
25675 * lra-constraints.c (curr_small_class_check): New.
25676 (update_and_check_small_class_inputs): New.
25677 (process_alt_operands): Update curr_small_class_check. Disfavor
25678 alternative insn memory operands. Check available regs for small
25681 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
25684 * config/mips/mips.opt (-mvirt): Update description.
25685 * doc/invoke.texi (-mvirt): Likewise.
25687 2017-04-10 Richard Biener <rguenther@suse.de>
25689 PR middle-end/80362
25690 * fold-const.c (fold_binary_loc): Look at unstripped ops when
25691 looking for NEGATE_EXPR in -A / -B to A / B folding.
25693 2017-04-10 Martin Liska <mliska@suse.cz>
25695 PR gcov-profile/80224
25696 * gcov.c (print_usage): Fix usage string.
25697 (get_gcov_intermediate_filename): Remove.
25698 (output_gcov_file): Use both for normal and intermediate format.
25699 (generate_results): Do not initialize special file for
25700 intermediate format.
25702 2017-04-10 Richard Biener <rguenther@suse.de>
25704 PR tree-optimization/80304
25705 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
25708 2017-04-10 Nathan Sidwell <nathan@acm.org>
25711 * config/rs6000/rs6000.c (rs6000_vector_type): New.
25712 (rs6000_init_builtins): Use it.
25714 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25716 * config/arm/arm.md (<mrc>): Add mode to SET source.
25717 (<mrrc>): Likewise.
25719 2017-04-10 Richard Biener <rguenther@suse.de>
25721 PR middle-end/80344
25722 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
25724 2017-04-10 Jakub Jelinek <jakub@redhat.com>
25727 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
25728 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
25729 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
25730 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
25731 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
25732 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
25733 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
25734 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
25735 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
25736 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
25737 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
25738 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
25739 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
25740 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
25741 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
25742 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
25743 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
25744 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
25745 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
25746 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
25747 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
25748 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
25749 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
25751 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
25753 PR rtl-optimization/70478
25754 * lra-constraints.c: Reverse the last patch.
25756 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
25758 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
25759 Add comment for WCHAR_T.
25761 2017-04-08 Martin Liska <mliska@suse.cz>
25764 2017-04-07 Martin Liska <mliska@suse.cz>
25767 * ipa-split.c (split_function): Add function part to a same comdat
25770 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25773 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
25775 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
25777 * rs6000/rs6000.c (vec_load_pendulum): Rename...
25778 (vec_pairing): ...to this.
25779 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
25780 (rs6000_sched_init): Adjust for name change.
25781 (struct rs6000_sched_context): Likewise.
25782 (rs6000_init_sched_context): Likewise.
25783 (rs6000_set_sched_context): Likewise.
25785 2017-04-07 Jakub Jelinek <jakub@redhat.com>
25791 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
25793 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
25794 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
25795 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
25797 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
25799 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
25801 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
25803 PR rtl-optimization/70703
25804 * ira-color.c (update_conflict_hard_regno_costs): Use
25805 int64_t instead of HOST_WIDE_INT.
25807 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
25809 PR rtl-optimization/70478
25810 * lra-constraints.c (process_alt_operands): Disfavor alternative
25811 insn memory operands.
25813 2017-04-07 Jeff Law <law@redhat.com>
25815 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
25816 CALL and NOTE_INSN_CALL_ARG_LOCATION.
25818 2017-04-07 Martin Liska <mliska@suse.cz>
25821 * config/aarch64/aarch64.c (aarch64_process_target_attr):
25822 Show error message instead of an ICE.
25824 2017-04-07 Martin Liska <mliska@suse.cz>
25827 * ipa-split.c (split_function): Add function part to a same comdat
25830 2017-04-07 Richard Biener <rguenther@suse.de>
25832 PR middle-end/80341
25833 * tree.c (get_unwidened): Also handle ! for_type case for
25835 * convert.c (do_narrow): Split out from ...
25836 (convert_to_integer_1): ... here. Do not pass final truncation
25837 type to get_unwidened for TRUNC_DIV_EXPR.
25839 2017-04-07 Richard Biener <rguenther@suse.de>
25841 * tree-affine.c (wide_int_ext_for_comb): Take type rather
25843 (aff_combination_const): Adjust.
25844 (aff_combination_scale): Likewise.
25845 (aff_combination_add_elt): Likewise.
25846 (aff_combination_add_cst): Likewise.
25847 (aff_combination_convert): Likewise.
25848 (add_elt_to_tree): Likewise. Remove unused argument.
25849 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
25851 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
25853 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
25855 * config/arm/arm.c (arm_default_short_enums): Use
25856 ARM_DEFAULT_SHORT_ENUMS.
25857 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
25859 2017-04-06 Jakub Jelinek <jakub@redhat.com>
25862 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
25863 members with redundant out-of-class redeclaration.
25865 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
25868 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
25869 * config/i386/i386.md (*zero_extendsidi2):
25870 Add (?*x,*x) and (?*v,*v) alternatives.
25872 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
25875 * config/i386/i386.c (ix86_expand_builtin)
25876 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
25877 mode from insn data. Convert operands to insn operand mode.
25878 Copy operands that don't satisfy insn predicate to a register.
25880 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
25882 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
25885 2017-04-06 Richard Biener <rguenther@suse.de>
25887 PR tree-optimization/80334
25888 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
25889 preserve alignment of accesses.
25891 2017-04-06 Richard Biener <rguenther@suse.de>
25893 PR tree-optimization/80262
25894 * tree-sra.c (build_ref_for_offset): Preserve address-space
25896 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
25897 Drop useless address-space information on MEM_REF offsets.
25899 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
25901 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
25903 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
25905 PR rtl-optimization/70703
25906 * ira-color.c (update_conflict_hard_regno_costs): Use
25907 HOST_WIDE_INT instead of long.
25909 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
25912 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
25913 not defined for x86_64 target. Add -mmmx target option when __SSE2__
25915 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
25916 for x86_64 target. Handle -m3dnowa option.
25918 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
25920 PR rtl-optimization/70703
25921 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
25922 (update_conflict_hard_regno_costs): Use long instead of unsigned
25923 arithmetic for cost calculation.
25925 2017-04-05 Jakub Jelinek <jakub@redhat.com>
25926 Bernd Edlinger <bernd.edlinger@hotmail.de>
25929 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
25932 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
25935 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
25936 ptr_mode with Pmode throughout.
25937 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
25938 into probe_stack_range and use DImode.
25940 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
25943 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
25944 call_eh_return is true.
25946 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25948 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
25949 Initialize last_match_fntype_index.
25951 2017-04-05 Jakub Jelinek <jakub@redhat.com>
25954 * tree-nvr.c: Include internal-fn.h.
25955 (pass_return_slot::execute): Ignore internal calls without
25958 2017-04-04 Jakub Jelinek <jakub@redhat.com>
25959 Richard Biener <rguenther@suse.de>
25962 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
25963 captures used multiple times, except for the last use.
25964 * generic-match-head.c: Include gimplify.h.
25966 2017-04-04 Jakub Jelinek <jakub@redhat.com>
25968 PR tree-optimization/79390
25969 * target.h (struct noce_if_info): Declare.
25970 * targhooks.h (default_noce_conversion_profitable_p): Declare.
25971 * target.def (noce_conversion_profitable_p): New target hook.
25972 * ifcvt.h (struct noce_if_info): New type, moved from ...
25973 * ifcvt.c (struct noce_if_info): ... here.
25974 (noce_conversion_profitable_p): Renamed to ...
25975 (default_noce_conversion_profitable_p): ... this. No longer
25977 (noce_try_store_flag_constants, noce_try_addcc,
25978 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
25979 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
25980 instead of noce_conversion_profitable_p.
25981 * config/i386/i386.c: Include ifcvt.h.
25982 (ix86_option_override_internal): Don't override
25983 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
25984 (ix86_noce_conversion_profitable_p): New function.
25985 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
25986 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
25987 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
25988 * doc/tm.texi: Regenerated.
25990 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25992 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
25995 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25998 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
25999 instructions for small multiply cores.
26001 2017-04-04 Jeff Law <law@redhat.com>
26003 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
26005 (mips_expand_vec_perm_const): Initialize elements in orig_perm
26006 that are not set by the loop over the elements.
26008 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26011 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
26012 int mode, convert_modes it to mode as unsigned, otherwise use
26013 lowpart_subreg to mode rather than SImode.
26014 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
26015 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
26016 Use DImode instead of SImode for the shift count operand.
26017 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
26020 2017-04-04 Richard Biener <rguenther@suse.de>
26022 PR middle-end/80281
26023 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
26024 arithmetic done for the negate or the plus. Simplify.
26025 (A - (-B) -> A + B): Likewise.
26026 * fold-const.c (split_tree): Make sure to not negate pointers.
26028 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
26030 PR rtl-optimization/60818
26031 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
26032 a compare of comparisons with the thing compared if this results
26033 in a different machine mode.
26035 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
26037 * alias.c (base_alias_check): Fix typo in comment.
26038 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
26039 * cgraphunit.c (symbol_table::compile): Likewise.
26040 * collect2.c (maybe_run_lto_and_relink): Likewise.
26041 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
26042 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
26043 * config/avr/avr.c (avr_map_op_t): Likewise.
26044 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
26045 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
26046 * config/epiphany/epiphany.md (movcc): Likewise.
26047 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
26048 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
26050 * config/mips/mips.c (mips_save_restore_reg): Likewise.
26051 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
26052 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
26053 * config/sh/sh.c (sh_rtx_costs): Likewise.
26054 * fold-const.c (fold_truth_andor): Likewise.
26055 * genautomata.c (collapse_flag): Likewise.
26056 * gengtype.h (struct type::u::s): Likewise.
26057 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
26058 * input.c (FORMAT_AMOUNT): Likewise.
26059 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
26060 (known_aggs_to_agg_replacement_list): Likewise.
26061 * ipa-inline-analysis.c: Likewise.
26062 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
26063 * ipa-polymorphic-call.c
26064 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
26065 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
26066 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
26068 * modulo-sched.c (apply_reg_moves): Likewise.
26069 * omp-expand.c (build_omp_regions_1): Likewise.
26070 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
26071 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
26072 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
26073 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
26074 * value-prof.c: Likewise.
26075 * var-tracking.c (val_reset): Likewise.
26077 2017-04-03 Richard Biener <rguenther@suse.de>
26079 PR tree-optimization/80275
26080 * fold-const.c (split_address_to_core_and_offset): Handle
26083 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
26085 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
26086 descriptors is at least equal to that of functions.
26088 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
26090 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
26092 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
26095 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
26096 (mov<IMOD4:mode>): New expander.
26097 (*mov<IMOD4:mode>_internal): New insn and split pattern.
26099 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
26101 PR rtl-optimization/79405
26102 * fwprop.c (propagations_left): New variable.
26103 (forward_propagate_into): Decrement it.
26104 (fwprop_init): Initialize it.
26105 (fw_prop): If the variable has reached zero, stop propagating.
26106 (fwprop_addr): Ditto.
26108 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26111 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
26112 a FUNCTION_DECL, pass it as decl instead of origin to
26115 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
26117 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
26120 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
26123 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
26124 TARGET_VSX_SMALL_INTEGER.
26126 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26128 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
26129 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
26131 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26133 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
26134 extraction from odd-numbered MSA register.
26136 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26138 PR middle-end/80173
26139 * expmed.c (store_bit_field_1): Don't attempt to create
26140 a word subreg out of hard registers wider than word if they
26141 have HARD_REGNO_NREGS of 1 for their mode.
26143 PR middle-end/80163
26144 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
26145 conversions to integer types wider than word and pointer.
26148 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
26149 (rtx_equal_for_cselib_p): Pass 0 to it.
26150 * cselib.c (cselib_hasher::equal): Likewise.
26151 (rtx_equal_for_cselib_1): Add depth argument. If depth
26152 is 128, don't look up VALUE locs and punt. Increment
26153 depth in recursive calls when walking VALUE locs.
26155 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
26157 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
26158 (make_gcov_file_name): Use the canonical path name for generating
26160 (read_line): Fix handling of files with ascii null bytes.
26162 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
26164 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
26165 to initialise a vector register instead
26166 of using a const_int.
26168 2017-03-30 Jakub Jelinek <jakub@redhat.com>
26170 PR translation/80189
26171 * gimplify.c (omp_default_clause): Use %qs instead of %s in
26172 diagnostic messages.
26174 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
26177 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
26178 (dfp_diex_<mode>): Update mode of operand 1.
26179 * doc/extend.texi (dxex, dxexq): Document change to return type.
26180 (diex, diexq): Document change to argument type.
26182 2017-03-30 Martin Jambor <mjambor@suse.cz>
26185 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
26186 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
26187 it reflects the signature changes performed at the callee side.
26188 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
26189 to cgraph_build_function_type_skip_args.
26190 (build_function_decl_skip_args): Adjust call to the above function.
26192 2017-03-30 Jakub Jelinek <jakub@redhat.com>
26195 * config/i386/sse.md
26196 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
26197 register as dest whenever it is a MEM not rtx_equal_p to the
26198 corresponding dup operand, and when forcing into reg move the
26199 reg into the memory afterwards.
26200 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
26201 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
26202 for the force_reg mode.
26203 (avx512vl_vextractf128<mode>): Use register as dest either
26204 always when a MEM, or when it is a MEM not rtx_equal_p to the
26205 corresponding dup operand, or even not when it is a CONST_VECTOR
26206 depending on the mode and lo vs. hi.
26207 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
26209 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
26210 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
26211 Likewise. Require that operands[2] is even.
26212 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
26213 Remove extraneous parens. Require that operands[2] is a multiple
26215 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
26216 operands[0] is a MEM if <mask_applied>, the predicates/constraints
26217 disallow memory then.
26219 2017-03-30 Richard Biener <rguenther@suse.de>
26221 PR tree-optimization/77498
26222 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
26223 to non-constants over backedges.
26225 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
26227 PR rtl-optimization/80233
26228 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
26229 as last_combined_insn. Do not test for BARRIER_P separately.
26231 2017-03-29 Andreas Schwab <schwab@suse.de>
26234 * calls.c (prepare_call_address): Convert funexp to Pmode before
26235 copying to temp reg.
26237 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26239 PR tree-optimization/80158
26240 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
26241 Handle possible future case of more than one alternate
26243 (replace_rhs_if_not_dup): Likewise.
26244 (replace_one_candidate): Likewise.
26246 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
26248 PR rtl-optimization/80193
26249 * ira.c (ira): Do not check allocation for LRA.
26251 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
26253 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
26254 (nvptx_output_simt_exit): Declare.
26255 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
26256 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
26257 (init_softstack_frame): Move initialization of crtl->is_leaf to...
26258 (nvptx_declare_function_name): ...here. Emit declaration of local
26259 memory space buffer for omp_simt_enter insn.
26260 (nvptx_output_unisimt_switch): New.
26261 (nvptx_output_softstack_switch): New.
26262 (nvptx_output_simt_enter): New.
26263 (nvptx_output_simt_exit): New.
26264 * config/nvptx/nvptx.h (struct machine_function): New fields
26265 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
26266 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
26267 (UNSPECV_SIMT_EXIT): Ditto.
26268 (omp_simt_enter_insn): New insn.
26269 (omp_simt_enter): New expansion.
26270 (omp_simt_exit): New insn.
26271 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
26273 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
26274 (expand_GOMP_SIMT_ENTER_ALLOC): New.
26275 (expand_GOMP_SIMT_EXIT): New.
26276 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
26277 (GOMP_SIMT_ENTER_ALLOC): Ditto.
26278 (GOMP_SIMT_EXIT): Ditto.
26279 * target-insns.def (omp_simt_enter): New insn.
26280 (omp_simt_exit): Ditto.
26281 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
26283 (lower_rec_simd_input_clauses): Implement SIMT privatization.
26284 (lower_rec_input_clauses): Likewise.
26285 (lower_lastprivate_clauses): Handle SIMT privatization.
26287 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
26288 (ompdevlow_adjust_simt_enter): New.
26289 (find_simtpriv_var_op): New.
26290 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
26291 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
26293 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
26294 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
26295 (copy_decl_for_dup_finish): Ditto.
26297 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
26299 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
26302 * config/i386/i386.c (ix86_option_override_internal): Always
26303 allow -mpreferred-stack-boundary=3 for 64-bit targets.
26305 2017-03-28 Bin Cheng <bin.cheng@arm.com>
26307 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
26309 2017-03-28 Bin Cheng <bin.cheng@arm.com>
26311 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
26312 mark new edge's irreducible flag accordign to it.
26313 (vect_do_peeling): Check loop preheader edge's irreducible flag
26314 and pass it to function slpeel_add_loop_guard.
26316 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
26318 PR tree-optimization/80218
26319 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
26320 Update block frequencies and counts.
26322 2017-03-28 Richard Biener <rguenther@suse.de>
26324 PR tree-optimization/78644
26325 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
26326 of a simplification result we may not use it at all.
26328 2017-03-28 Richard Biener <rguenther@suse.de>
26331 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
26332 without arguments, generate default definition of a SSA name.
26334 2017-03-28 Richard Biener <rguenther@suse.de>
26336 PR middle-end/80222
26337 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
26338 TYPE_REF_CAN_ALIAS_ALL references.
26339 * fold-const.c (fold_indirect_ref_1): Likewise.
26341 2017-03-28 Martin Liska <mliska@suse.cz>
26344 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
26345 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
26347 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
26348 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
26350 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
26351 (EXTRA_SPECS): Define.
26352 (SUBTARGET_EXTRA_SPECS): Likewise.
26353 (SUBTARGET_CPP_SPEC): Likewise.
26354 * config/arc/elf.h (EXTRA_SPECS): Renamed to
26355 SUBTARGET_EXTRA_SPECS.
26356 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
26358 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
26360 * config/arc/simdext.md (vst64_insn): Update pattern.
26361 (vld32wh_insn): Likewise.
26362 (vld32wl_insn): Likewise.
26363 (vld64_insn): Likewise.
26364 (vld32_insn): Likewise.
26366 2017-03-28 Marek Polacek <polacek@redhat.com>
26369 * fold-const.c (fold_comparison): Use protected_set_expr_location
26370 instead of SET_EXPR_LOCATION.
26372 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
26374 * tree.c (add_expr): Avoid name lookup warning.
26376 2017-03-27 Jeff Law <law@redhat.com>
26378 PR tree-optimization/80216
26379 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
26380 function name. Limit recursion depth.
26381 (record_temporary_equivalences): Corresponding changes.
26383 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
26385 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
26388 2017-03-27 Jakub Jelinek <jakub@redhat.com>
26391 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
26393 * cfgcleanup.c (reg_note_cfa_p): New array.
26394 (insns_have_identical_cfa_notes): New function.
26395 (old_insns_match_p): Don't cross-jump in between /f
26396 and non-/f instructions. If both i1 and i2 are frame related,
26397 verify all CFA notes, their order and content.
26399 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
26402 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
26403 HImode and SImode with zero extend to DImode to one insn.
26404 (bswap<mode>2_extenddi): Likewise.
26405 (bswapsi2_extenddi): Likewise.
26406 (bswaphi2_extendsi): Likewise.
26407 (bswaphi2): Combine bswap HImode and SImode into one insn.
26408 Separate memory insns from swapping register.
26409 (bswapsi2): Likewise.
26410 (bswap<mode>2): Likewise.
26411 (bswaphi2_internal): Delete, no longer used.
26412 (bswapsi2_internal): Likewise.
26413 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
26414 store, and gpr<-gpr swap insns.
26415 (bswap<mode>2_store): Likewise.
26416 (bswaphi2_reg): Register only splitter, combine with the splitter.
26417 (bswaphi2 splitter): Likewise.
26418 (bswapsi2_reg): Likewise.
26419 (bswapsi2 splitter): Likewise.
26420 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
26421 the insns into load, store, and register/register insns.
26422 (bswapdi2_ldbrx): Likewise.
26423 (bswapdi2_load): Likewise.
26424 (bswapdi2_store): Likewise.
26425 (bswapdi2_reg): Likewise.
26427 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
26429 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
26430 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
26432 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
26435 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
26437 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
26438 special handling for target option conflicts between dform
26439 options (-mpower9-dform, -mpower9-dform-vector,
26440 -mpower9-dform-scalar) and -mno-direct-move.
26442 2017-03-27 Richard Biener <rguenther@suse.de>
26444 PR tree-optimization/80181
26445 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
26447 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26449 * config/arc/predicates.md (move_double_src_operand): Replace the
26450 call to move_double_src_operand with a call to address_operand.
26452 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26454 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
26455 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
26456 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
26458 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26460 * config/arc/predicates.md (long_immediate_loadstore_operand):
26461 Consider scaled addresses cases.
26463 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26465 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
26466 restored when in interrupt.
26467 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
26468 doesn't have delay slot.
26470 2017-03-27 Richard Biener <rguenther@suse.de>
26473 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
26474 inlined thunk clones.
26476 2017-03-27 Jakub Jelinek <jakub@redhat.com>
26479 * asan.c (instrument_derefs): Copy over last operand from
26480 original COMPONENT_REF to the new COMPONENT_REF with
26481 DECL_BIT_FIELD_REPRESENTATIVE.
26482 * ubsan.c (instrument_object_size): Likewise.
26484 2017-03-27 Richard Biener <rguenther@suse.de>
26486 PR tree-optimization/80170
26487 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
26488 sure DR/SCEV didnt fold in constants we do not see when looking
26489 at the reference base alignment.
26491 2017-03-27 Richard Biener <rguenther@suse.de>
26493 PR middle-end/80171
26494 * gimple-fold.c (fold_ctor_reference): Properly guard against
26495 NULL return value from canonicalize_constructor_val.
26497 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
26500 * config/i386/i386.c (ix86_expand_builtin)
26501 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
26502 flags reg setting and flags reg using instructions.
26503 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
26504 clobbering instructions to zero extend op2.
26506 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
26508 * doc/install.texi (Configuration) <--with-aix-soname>:
26509 Update link to AIX ld.
26511 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
26513 PR rtl-optimization/80160
26514 PR rtl-optimization/80159
26515 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
26516 reg_alternate_class into account.
26518 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
26521 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
26522 to consider in curr_insn_transform.
26524 2017-03-24 Jakub Jelinek <jakub@redhat.com>
26526 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
26527 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
26528 and emit_mode_inner.
26530 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26532 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
26533 argument to the overloaded builtin variants. Use the new flag to
26534 deprecate certain builtin variants.
26535 * config/s390/s390-builtin-types.def: Add new builtin types.
26536 * config/s390/s390-builtins.h: Support new flags field for
26537 overloaded builtins.
26538 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
26539 (s390_macro_to_expand): Enable vector float data type.
26540 (s390_cpu_cpp_builtins_internal): Indicate support of the new
26541 builtins by incrementing the __VEC__ version number.
26542 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
26544 (s390_resolve_overloaded_builtin): Emit error messages depending
26545 on the builtin flags.
26546 * config/s390/s390.c (s390_expand_builtin): Support additional
26547 flags argument. Change error message to match the messages
26548 emitted in s390-c.c.
26549 * config/s390/s390.md: New UNSPEC_* constants.
26550 (op_type): Add new instruction types.
26551 * config/s390/vecintrin.h: Add new builtins and test data class
26553 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
26554 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
26555 (VEC_INEXACT, VEC_NOINEXACT): New constants.
26556 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
26557 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
26558 ("vec_mergel<mode>"): V_HW -> VEC_HW.
26560 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
26561 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
26562 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
26563 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
26565 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
26566 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
26567 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
26568 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
26570 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
26571 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
26572 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
26573 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
26574 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
26575 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
26576 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
26578 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
26579 ("vec_scatter_element<V_HW_4:mode>_DI")
26580 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
26581 ("vec_fpint<mode>", "vflls")
26582 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
26583 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
26584 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
26585 ("*vec_cmphe<mode>_cc"): ... these.
26587 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
26588 mode constant instead of magic value.
26590 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26592 * config/s390/s390.c (s390_expand_vec_compare): Support other
26593 vector floating point modes than just V2DF.
26594 (s390_expand_vcond): Likewise.
26595 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
26596 (s390_cannot_change_mode_class): Prevent mode changes between TF
26597 and V1TF in vector registers.
26598 * config/s390/s390.md (DF, SF): New mode attributes.
26599 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
26600 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
26601 SFmode support for VRs.
26602 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
26604 (VFT, VF_HW): New mode iterators.
26605 (vw, sdx): New mode attributes.
26606 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
26607 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
26608 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
26609 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
26610 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
26611 also the new vector floating point modes. Renaming to ...
26613 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
26614 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
26615 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
26616 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
26617 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
26618 ("vec_unordered<mode>"): ... these.
26620 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
26621 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
26622 ("*vec_extendv2df"): New insn definitions.
26624 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26626 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
26627 ("mulditi3_2", "*muldi3_sign"): New patterns.
26628 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
26629 rename the pattern definition.
26631 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26633 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
26635 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
26637 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26639 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
26640 instruction if possible.
26641 * config/s390/vector.md (vec_halfnumelts): New mode
26643 ("*vec_vllezlf<mode>"): New pattern.
26645 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26647 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
26648 ("popcountv4si2", "popcountv2di2"): Rename to ...
26649 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
26650 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
26652 ("popcount<mode>2_vxe"): New pattern.
26654 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26656 * common/config/s390/s390-common.c (processor_flags_table): Add
26658 * config.gcc: Add arch12.
26659 * config/s390/driver-native.c (s390_host_detect_local_cpu):
26660 Default to arch12 for unknown CPU model numbers.
26661 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
26662 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
26663 PROCESSOR_max sanity check.
26664 * config/s390/s390-opts.h (enum processor_type): Add
26666 * config/s390/s390.c (processor_table): Add arch12.
26667 (s390_expand_builtin): Add check for B_VXE flag.
26668 (s390_issue_rate): Add PROCESSOR_ARCH12.
26669 (s390_get_sched_attrmask): Likewise.
26670 (s390_get_unit_mask): Likewise.
26671 (s390_sched_score): Enable z13 scheduling for arch12.
26672 (s390_sched_reorder): Likewise.
26673 (s390_sched_variable_issue): Likewise.
26674 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
26676 (s390_tune_attr): Use z13 scheduling also for arch12.
26677 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
26678 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
26679 (TARGET_VXE_P): New macros.
26680 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
26681 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
26682 * config/s390/s390.opt: Add arch12 as processor_type.
26684 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26686 * config/s390/s390.md
26687 ("fixuns_truncdddi2", "fixuns_trunctddi2")
26688 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
26689 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
26691 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
26692 Rename expanders to ...
26694 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
26695 ("fixuns_truncdddi2_emu"): ... these.
26697 ("fixuns_trunc<mode>si2_emu"): New expander.
26699 ("*fixuns_truncdfdi2_z13"): Rename to ...
26700 ("*fixuns_truncdfdi2_vx"): ... this.
26702 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26704 * config/s390/2964.md: Remove the single element vector compare
26705 instructions which are no longer used.
26706 * config/s390/s390.c (s390_select_ccmode): Remove handling of
26708 (s390_canonicalize_comparison): Remove handling of DFmode
26710 (s390_expand_vec_compare_scalar): Remove function.
26711 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
26712 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
26714 ("*cmp<mode>_ccs"): Add wfcdb instruction.
26716 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26718 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
26720 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
26721 will anyway by matched by mov<mode>_64dfp.
26723 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26725 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
26726 vlef/vstef. Add missing operand to vleif.
26728 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26730 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
26731 pair for all vector types with 64 bit elements.
26732 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
26733 * config/s390/vector.md (V_HW_64): ... here.
26734 (V_128_NOSINGLE): New mode iterator.
26735 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
26736 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
26737 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
26738 ("*vec_load_pairv2di"): Change to ...
26739 ("*vec_load_pair<mode>"): ... this one.
26741 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26743 * config/s390/constraints.md: Add comments.
26744 (jKK): Reject element sizes > 8 bytes.
26745 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
26747 * config/s390/s390.md: Add the s_operand checks formerly in
26748 s390_split_ok_p to various splitters where they are still
26750 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
26751 for 128 bit vectors. Plus two splitters.
26753 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26755 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
26758 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26761 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
26762 error if the boundary argument is not constant.
26764 2017-03-24 Jakub Jelinek <jakub@redhat.com>
26766 PR rtl-optimization/80112
26767 * loop-doloop.c (doloop_condition_get): Don't check condition
26768 if cmp isn't SET with IF_THEN_ELSE src.
26770 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26772 PR tree-optimization/80158
26773 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
26774 replacing a candidate statement, also replace it for the
26775 candidate's alternate interpretation.
26776 (replace_rhs_if_not_dup): Likewise.
26777 (replace_one_candidate): Likewise.
26779 2017-03-24 Richard Biener <rguenther@suse.de>
26781 PR tree-optimization/80167
26782 * graphite-isl-ast-to-gimple.c
26783 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
26785 (translate_isl_ast_to_gimple::get_rename): Likewise.
26787 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
26789 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
26790 handling of certain combinations of target options, including the
26791 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
26792 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
26794 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26797 * config/arm/arm.md (*load_multiple): Add reload_completed to
26798 matching condition.
26800 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26801 Richard Biener <rguenth@suse.de>
26803 PR tree-optimization/79908
26804 PR tree-optimization/80136
26805 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
26806 been cast away, gimplify_and_add suffices.
26808 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
26810 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
26812 2017-03-23 Richard Biener <rguenther@suse.de>
26814 PR tree-optimization/80032
26815 * gimplify.c (gimple_push_cleanup): Forced unconditional
26816 cleanups still have to go to the conditional_cleanups
26819 2017-03-22 Jakub Jelinek <jakub@redhat.com>
26821 PR tree-optimization/80072
26822 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
26824 (next_operand_entry_id): Change type to unsigned int.
26825 (sort_by_operand_rank): Make sure to return the right return value
26826 even if unsigned fields are bigger than INT_MAX.
26827 (struct oecount): Change cnt and id type to unsigned int.
26828 (oecount_hasher::equal): Formatting fix.
26829 (oecount_cmp): Make sure to return the right return value
26830 even if unsigned fields are bigger than INT_MAX.
26831 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
26834 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
26835 TREE_READONLY on result if writing it more than once.
26838 * doc/invoke.texi (-fsanitize=thread): Document that with
26839 -fnon-call-exceptions atomics are not able to throw
26843 * tsan.c: Include tree-eh.h.
26844 (instrument_builtin_call): Call maybe_clean_eh_stmt or
26845 maybe_clean_or_replace_eh_stmt where needed.
26846 (instrument_memory_accesses): Add cfg_changed argument.
26847 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
26848 if it returned true.
26849 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
26851 PR rtl-optimization/63191
26852 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
26853 wrapper function, moved the whole old content into ...
26854 (ix86_delegitimize_address_1): ... this. New inline function.
26855 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
26856 true as last argument instead of ix86_delegitimize_address.
26858 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
26860 * config/aarch64/aarch64.c (generic_branch_cost): Copy
26861 cortexa57_branch_cost.
26863 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
26865 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
26867 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26870 * doc/md.texi (Constraints): Document wA constraint.
26871 * config/rs6000/constraints.md (wA): New.
26872 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
26873 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
26874 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
26875 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
26877 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
26880 * gimplify.c (is_oacc_declared): New function.
26881 (oacc_default_clause): Use it to set default flags for acc declared
26882 variables inside parallel regions.
26883 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
26884 declared variables.
26885 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
26886 declare attribute to any decl as necessary.
26888 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
26891 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
26892 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
26893 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
26894 (arm_arch_lpae): This.
26895 * config/arm/arm.c (arm_arch7ve): Rename into ...
26896 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
26897 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
26900 2017-03-22 Martin Liska <mliska@suse.cz>
26903 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
26904 error message instead of an ICE.
26906 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26908 * doc/extend.texi (6.11 Additional Floating Types): Revise.
26910 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
26912 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
26914 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
26917 2017-03-21 Martin Sebor <msebor@redhat.com>
26919 * doc/extend.texi: Use "cannot" instead of "can't."
26920 * doc/hostconfig.texi: Same.
26921 * doc/install.texi: Same.
26922 * doc/invoke.texi: Same.
26923 * doc/loop.texi: Same.
26924 * doc/md.texi: Same.
26925 * doc/objc.texi: Same.
26926 * doc/rtl.texi: Same.
26927 * doc/tm.texi: Same.
26928 * doc/tm.texi.in: Same.
26929 * doc/trouble.texi: Same.
26931 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
26934 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
26935 (collect_checksum_attributes): Set it.
26936 (die_checksum_ordered): Use it.
26938 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26940 PR tree-optimization/79908
26941 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
26942 change: For a VA_ARG whose LHS has been cast away, use
26943 force_gimple_operand to construct the side effects.
26945 2017-03-21 David Malcolm <dmalcolm@redhat.com>
26947 PR translation/80001
26948 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
26949 more amenable to translation.
26950 (oacc_loop_auto_partitions): Likewise.
26952 2017-03-21 Marek Polacek <polacek@redhat.com>
26953 Martin Sebor <msebor@redhat.com>
26955 PR tree-optimization/80109
26956 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
26957 on INTEGRAL_TYPE_P.
26959 2017-03-21 Jakub Jelinek <jakub@redhat.com>
26960 Segher Boessenkool <segher@kernel.crashing.org>
26963 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
26964 check reg_used_between_p between insn and one of succ or succ2
26965 depending on if succ is artificial insn not inserted into insn
26968 2017-03-21 Martin Liska <mliska@suse.cz>
26970 PR gcov-profile/80081
26971 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
26972 * doc/gcc.texi: Include gcov-dump stuff.
26973 * doc/gcov-dump.texi: New file.
26975 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
26977 PR rtl-optimization/79150
26978 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
26979 conditional jump, if the jump is the last insn of the loop.
26981 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26982 Richard Biener <rguenth@suse.de>
26984 PR tree-optimization/79908
26985 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
26986 been cast away, use force_gimple_operand to construct the side
26989 2017-03-21 Martin Liska <mliska@suse.cz>
26991 PR libfortran/79956
26992 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
26995 2017-03-21 Brad Spengler <spender@grsecurity.net>
26998 * plugin.c (htab_hash_plugin): New function.
26999 (add_new_plugin): Use it and adjust.
27000 (parse_plugin_arg_opt): Adjust.
27001 (init_one_plugin): Likewise.
27003 2017-03-21 Richard Biener <rguenther@suse.de>
27005 PR tree-optimization/80032
27006 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
27007 if set force the cleanup to happen unconditionally.
27008 (gimplify_target_expr): Push inserted clobbers with force_uncond
27009 to avoid them being removed by control-dependent DCE.
27011 2017-03-21 Richard Biener <rguenther@suse.de>
27013 PR tree-optimization/80122
27014 * tree-inline.c (copy_bb): Do not expans va-arg packs or
27015 va_arg_pack_len when the inlined call stmt requires pack
27017 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
27019 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27022 * tsan.c (instrument_builtin_call): If the memory model argument
27023 is not a constant, assume it is valid.
27026 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
27029 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27031 PR rtl-optimization/79910
27032 * combine.c (can_combine_p): Do not allow combining an I0 or I1
27033 if its dest is used by an insn before I2 (other than the combined
27034 insns themselves, which are properly handled already).
27036 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27039 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27041 * combine.c (record_used_regs): New static function.
27042 (try_combine): Handle situations where there is an additional
27043 instruction between I2 and I3 which needs to have a LOG_LINK
27047 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27049 * combine.c (try_combine): Delete redundant i1 test. Call
27050 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27052 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27055 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
27056 alternatives 13/14.
27058 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27060 PR tree-optimization/80054
27061 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
27062 the optimization if a PHI or any of its arguments is not dominated
27063 by the candidate's basis. Use gphi* rather than gimple* as
27065 (replace_profitable_candidates): Clean up a gimple* variable that
27066 should be a gphi* variable.
27068 2017-03-20 Martin Sebor <msebor@redhat.com>
27071 * doc/extend.texi (attribute constructor): Document present limitation.
27073 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
27076 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
27077 __POWER9_VECTOR__ #ifdef control, change template definition to
27078 use Power9-specific built-in function.
27079 (vec_any_eq): Likewise.
27080 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
27081 to control outcomes from this test.
27082 (vector_ae_<mode>p): For VEC_F modes, likewise.
27084 2017-03-20 Ian Lance Taylor <iant@google.com>
27086 * config/i386/i386.c (ix86_function_regparm): Save an extra
27087 register for -fsplit-stack with DECL_STATIC_CHAIN.
27089 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
27092 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
27093 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
27095 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
27097 * config/riscv/riscv.c (riscv_print_operand): Use "fence
27099 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
27102 2017-03-20 Marek Polacek <polacek@redhat.com>
27105 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
27107 2017-03-20 Richard Biener <rguenther@suse.de>
27109 PR tree-optimization/80113
27110 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
27111 allocate extra SSA name for PHI def.
27112 (add_close_phis_to_outer_loops): Likewise.
27113 (add_close_phis_to_merge_points): Likewise.
27114 (copy_loop_close_phi_args): Likewise.
27115 (copy_cond_phi_nodes): Likewise.
27117 2017-03-20 Martin Liska <mliska@suse.cz>
27119 PR middle-end/79753
27120 * tree-chkp.c (chkp_build_returned_bound): Do not build
27121 returned bounds for a LHS that's not a BOUNDED_P type.
27123 2017-03-20 Martin Liska <mliska@suse.cz>
27127 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
27128 COMPLEX_CST and VECTOR_CST.
27130 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27133 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
27134 target operand. A new splitter adds the clobber statement in case
27135 the target operand is dead anyway.
27137 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
27139 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
27140 to age-old versions of binutils and glibc.
27142 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
27144 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
27146 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27148 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
27150 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27152 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
27153 requirement for binutils 2.13.
27155 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27157 * combine.c (try_combine): Delete redundant i1 test. Call
27158 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27160 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
27162 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
27163 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
27165 <riscv64-*-elf>: Re-arrange section
27166 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
27167 <riscv32-*-linux>: Likewise.
27168 <riscv64-*-elf>: Likewise
27169 <riscv64-*-linux>: Likewise.
27171 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
27174 * aarch64.opt(verbose-cost-dump): Fix typo.
27176 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
27179 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
27180 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
27182 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27184 * reload.c (find_reloads): When reloading a nonoffsettable address,
27185 use RELOAD_OTHER for it and its address reloads.
27187 PR rtl-optimization/79910
27188 * combine.c (record_used_regs): New static function.
27189 (try_combine): Handle situations where there is an additional
27190 instruction between I2 and I3 which needs to have a LOG_LINK
27193 2017-03-17 Jeff Law <law@redhat.com>
27195 PR tree-optimization/71437
27196 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
27197 conditional in the hash table first.
27198 (vrp_dom_walker::before_dom_children): Extract condition from
27199 ASSERT_EXPR. Record condition, its inverion and any implied
27200 conditions as well.
27202 2017-03-17 Marek Polacek <polacek@redhat.com>
27203 Markus Trippelsdorf <markus@trippelsdorf.de>
27205 PR tree-optimization/80079
27206 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
27209 2017-03-17 Richard Biener <rguenther@suse.de>
27211 PR middle-end/80075
27212 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
27213 Properly verify the LHS before the RHS possibly claims to be
27215 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
27218 2017-03-17 Martin Jambor <mjambor@suse.cz>
27220 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
27221 (List of -O2 options): Likewise.
27222 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
27225 2017-03-17 Tom de Vries <tom@codesourcery.com>
27227 * gcov-dump.c (print_usage): Print bug_report_url.
27229 2017-03-17 Richard Biener <rguenther@suse.de>
27231 PR middle-end/80050
27232 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
27233 (parser::peek): Likewise.
27235 2017-03-17 Richard Biener <rguenther@suse.de>
27237 PR tree-optimization/80048
27238 * sese.c (free_sese_info): Properly release rename_map and
27239 copied_bb_map elements.
27241 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
27243 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
27244 Add linked-list forward and backlinks. Insert on
27245 construction, remove on destruction.
27246 (class pass_store_merging): Add m_stores_head field.
27247 (pass_store_merging::terminate_and_process_all_chains):
27248 Iterate over m_stores_head list.
27249 (pass_store_merging::terminate_all_aliasing_chains):
27251 (pass_store_merging::execute): Check for debug stmts first.
27252 Push new chains onto the m_stores_head stack.
27254 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
27257 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
27258 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
27259 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
27261 2017-03-16 Jeff Law <law@redhat.com>
27263 PR tree-optimization/71437
27264 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
27265 member function. Implementation moved into after_dom_children
27266 member function and into the threader's thread_outgoing_edges
27268 (dom_opt_dom_walker::after_dom_children): Simplify by moving
27269 some code into new thread_outgoing_edges.
27270 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
27271 definition. Simplify marker handling (do it here). Assume we always
27272 have the available expression and the const/copies tables.
27273 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
27275 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
27276 * tree-vrp.c (equiv_stack): No longer file scoped.
27277 (vrp_dom_walker): New class.
27278 (vrp_dom_walker::before_dom_children): New member function.
27279 (vrp_dom_walker::after_dom_children): Likewise.
27280 (identify_jump_threads): Setup domwalker. Use it rather than
27281 walking edges in a random order by hand. Simplify setup/finalization.
27282 (finalize_jump_threads): Remove.
27283 (vrp_finalize): Do not call identify_jump_threads here.
27284 (execute_vrp): Do it here instead and call thread_through_all_blocks
27287 PR tree-optimization/71437
27288 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
27290 (simplify_stmt_for_jump_threading): Add basic_block argument. All
27292 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
27293 (dom_opt_dom_walker::thread_across_edge): Remove
27294 handle_dominating_asserts argument. All callers changed.
27295 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
27296 changes. Remove calls to lhs_of_dominating_assert. Other
27297 uses of handle_dominating_asserts turn into unconditional code
27298 (simplify_control_stmt_condition_1): Likewise.
27299 (simplify_control_stmt_condition): Likewise.
27300 (thread_through_normal_block, thread_across_edge): Likewise.
27301 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
27302 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
27303 object if it is not an SSA_NAME.
27304 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
27305 before calling into the VRP specific simplifiers.
27306 (identify_jump_threads): Remove handle_dominating_asserts
27309 2017-03-16 Jakub Jelinek <jakub@redhat.com>
27312 * tree-diagnostic.c (default_tree_printer): No longer static.
27313 * tree-diagnostic.h (default_tree_printer): New prototype.
27315 2017-03-16 Tamar Christina <tamar.christina@arm.com>
27317 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
27318 Change ins into fmov.
27320 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27322 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
27323 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
27324 Use h_con constraint for operand 1.
27325 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
27326 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
27328 2017-03-15 Jeff Law <law@redhat.com>
27330 PR tree-optimization/71437
27331 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
27332 (record_temporary_equivalences): Use it.
27334 PR tree-optimization/71437
27335 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
27336 tree-ssa-scopedtables.
27337 (lookup_avail_expr, build_and_record_new_cond): Likewise.
27338 (record_conditions, record_cond, vuse_eq): Likewise.
27339 (record_edge_info): Adjust to API tweak of record_conditions.
27340 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
27341 (record_temporary_equivalences, optimize_stmt): Likewise.
27342 (eliminate_redundant_computations): Likewise.
27343 (record_equivalences_from_stmt): Likewise.
27344 * tree-ssa-scopedtables.c: Include options.h and params.h.
27345 (vuse_eq): New function, moved from tree-ssa-dom.c
27346 (build_and_record_new_cond): Likewise.
27347 (record_conditions): Likewise. Accept vector of conditions rather
27348 than edge_equivalence structure for first argument.
27349 for the first argument.
27350 (avail_exprs_stack::lookup_avail_expr): New member function, moved
27351 from tree-ssa-dom.c.
27352 (avail_exprs_stack::record_cond): Likewise.
27353 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
27354 from tree-ssa-dom.c.
27355 (avail_exprs_stack): Add new member functions lookup_avail_expr
27357 (record_conditions): Declare.
27359 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
27362 * lra-constraints.c (process_alt_operands): Increase reject for
27363 reloading an input/output operand.
27365 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
27368 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
27369 insns to convert from signed/unsigned char/short to IEEE 128-bit
27371 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
27373 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
27376 * config/i386/i386.c (ix86_vector_duplicate_value): Create
27377 subreg of inner mode for values already in registers.
27379 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
27381 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
27382 iteration reg is used after the loop.
27384 2017-03-14 Martin Sebor <msebor@redhat.com>
27386 PR tree-optimization/79800
27387 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
27388 precision in negative-positive range.
27389 (format_floating): Call non-const overload with adjusted precision.
27391 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
27394 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
27397 2017-03-14 Martin Sebor <msebor@redhat.com>
27399 PR middle-end/80020
27400 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
27401 * builtins.def (aligned_alloc): Use it.
27404 * Makefile.in (GTFILES): Add calls.c.
27405 * calls.c: Include "gt-calls.h".
27407 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
27409 PR rtl-optimization/79728
27410 * regs.h (struct target_regs): New field
27411 x_contains_allocatable_regs_of_mode.
27412 (contains_allocatable_regs_of_mode): New macro.
27413 * reginfo.c (init_reg_sets_1): Initialize it, and change
27414 contains_reg_of_mode so it includes global regs as well.
27415 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
27416 rather than contains_regs_of_mode.
27418 2017-03-14 Martin Liska <mliska@suse.cz>
27420 * doc/invoke.texi: Document options that can't be combined with
27421 -fcheck-pointer-bounds.
27423 2017-03-14 Martin Liska <mliska@suse.cz>
27425 PR middle-end/79831
27426 * doc/invoke.texi (-Wchkp): Document the option.
27428 2017-03-14 Martin Liska <mliska@suse.cz>
27430 * Makefile.in: Install gcov-dump.
27432 2017-03-14 Martin Liska <mliska@suse.cz>
27434 * multiple_target.c (expand_target_clones): Bail out for
27435 an invalid attribute.
27437 2017-03-14 Richard Biener <rguenther@suse.de>
27439 * alias.c (struct alias_set_entry): Pack properly.
27440 * cfgloop.h (struct loop): Likewise.
27441 * cse.c (struct set): Likewise.
27442 * ipa-utils.c (struct searchc_env): Likewise.
27443 * loop-invariant.c (struct invariant): Likewise.
27444 * lra-remat.c (struct cand): Likewise.
27445 * recog.c (struct change_t): Likewise.
27446 * rtl.h (struct address_info): Likewise.
27447 * symbol-summary.h (function_summary): Likewise.
27448 * tree-loop-distribution.c (struct partition): Likewise.
27449 * tree-object-size.c (struct object_size_info): Likewise.
27450 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
27451 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
27452 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
27453 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
27454 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
27455 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
27456 (struct _stmt_vec_info): Likewise.
27458 2017-03-14 Martin Liska <mliska@suse.cz>
27461 * multiple_target.c (create_dispatcher_calls): Check that
27462 a target can create a function dispatcher.
27464 2017-03-14 Martin Liska <mliska@suse.cz>
27467 * multiple_target.c (expand_target_clones): Drop local.local
27468 flag for default implementation.
27470 2017-03-14 Richard Biener <rguenther@suse.de>
27472 PR tree-optimization/80030
27473 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
27475 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
27477 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
27478 gcc_fallthrough() instead of __attribute__((fallthrough));
27480 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
27482 * doc/gcc.texi: Remove "up" link to (DIR).
27483 * doc/gccint.texi: Ditto.
27485 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
27487 * doc/install.texi (Specific) <avr>: Remove reference to
27490 2017-03-13 Jeff Law <law@redhat.com>
27492 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
27493 attribute rather than comments.
27495 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
27496 match_scratch operand is highest.
27498 2017-03-13 Martin Liska <mliska@suse.cz>
27500 PR middle-end/78339
27501 * ipa-pure-const.c (warn_function_noreturn): If the declarations
27502 is a CHKP clone, use original declaration.
27504 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27506 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
27507 (arc_conditional_register_usage): Use a different allocation order
27508 when optimizing for size.
27509 * common/config/arc/arc-common.c (arc_option_optimization_table):
27510 Section anchors default on when optimizing for size.
27512 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27514 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
27516 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27518 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
27519 * config/arc/arc.md (cpu_facility): Add cd variant.
27520 (*movqi_insn): Add code density variant.
27521 (*movhi_insn): Likewise.
27522 (*movqi_insn): Likewise.
27523 (*addsi3_mixed): Likewise.
27524 (subsi3_insn): Likewise.
27526 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27528 * config/arc/arc.md (movsi_cond_exec): Update constraint.
27530 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27532 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
27533 expressions with MINUS and UNARY ops.
27535 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27538 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
27540 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
27541 between vec_select and vector argument.
27542 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
27543 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
27544 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
27545 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
27546 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
27547 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
27549 2017-03-13 Richard Biener <rguenther@suse.de>
27552 * params.def (vect-max-peeling-for-alignment): Fix typo.
27554 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
27556 * doc/install.texi (Specific) <mips-*-*>: Remove description of
27557 issue that only occurred with binutils below 2.18.
27559 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
27561 * doc/install.texi (Specific) <cris-axis-elf>: No longer
27562 refer to binutils 2.11/2.12 minimum.
27564 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
27566 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
27567 ftp.kernel.org and simplify binutils requirement.
27569 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
27571 * doc/invoke.texi (Warning Options): Fix spelling of link-time
27573 (Optimize Options): Ditto. Also remove redundancy.
27575 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27577 PR translation/79848
27578 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
27580 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
27581 to G_ to avoid double translation.
27583 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27585 PR translation/79923
27586 * auto-profile.c (get_combined_location): Convert leading
27587 character of diagnostics to lower case and remove trailing period.
27588 (read_profile): Likewise for various diagnostics.
27589 * config/arm/arm.c (arm_option_override): Remove trailing period
27590 from various diagnostics.
27591 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
27592 (msp430_expand_delay_cycles): Likewise.
27594 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27597 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
27598 full command-line argument, rather than just "str".
27599 (aarch64_validate_march): Likewise.
27600 (aarch64_validate_mtune): Likewise.
27602 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
27604 PR rtl-optimization/78911
27605 * lra-assigns.c (must_not_spill_p): New function.
27606 (spill_for): Use it.
27608 2017-03-10 Jakub Jelinek <jakub@redhat.com>
27610 PR tree-optimization/79981
27611 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
27612 ATOMIC_COMPARE_EXCHANGE ifn result.
27613 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
27614 IFN_ATOMIC_COMPARE_EXCHANGE.
27616 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27619 * opts.c (parse_sanitizer_options): Add missing question mark to
27620 "did you mean" message.
27622 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27624 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
27626 (VMULEUH_UNS): Likewise.
27627 (VMULOUB_UNS): Likewise.
27628 (VMULOUH_UNS): Likewise.
27629 * config/rs6000/rs6000.c (builtin_function_type): Remove
27630 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
27632 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27635 * read-rtl-function.c (function_reader::read_rtx_operand): Update
27636 x with result of extra_parsing_for_operand_code_0.
27637 (function_reader::extra_parsing_for_operand_code_0): Convert
27638 return type from void to rtx, returning x. When reading
27639 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
27640 larger size containing struct block_symbol.
27642 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
27644 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
27645 -mfloat128-hardware without -m64.
27647 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
27650 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
27651 entries to the case statement that marks unsigned arguments to
27652 overloaded functions.
27654 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
27656 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
27657 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
27659 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
27662 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
27663 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
27665 2017-03-10 Martin Liska <mliska@suse.cz>
27669 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
27670 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
27673 2017-03-10 Olivier Hainque <hainque@adacore.com>
27675 * tree-switch-conversion (array_value_type): Start by resetting
27676 candidate type to it's main variant.
27678 2017-03-10 Jakub Jelinek <jakub@redhat.com>
27680 PR rtl-optimization/79909
27681 * combine.c (try_combine): Use simplify_replace_rtx on individual
27682 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
27683 of the whole CALL_INSN_FUNCTION_USAGE.
27685 PR tree-optimization/79972
27686 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
27687 get_range_info on SSA_NAMEs. Formatting fixes.
27689 2017-03-10 Richard Biener <rguenther@suse.de>
27690 Jakub Jelinek <jakub@redhat.com>
27692 PR tree-optimization/77975
27693 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
27694 edge to be constant.
27695 (get_val_for): For constant x return it. Formatting fix.
27696 (loop_niter_by_eval): Avoid pointless looping if the next iteration
27697 would use the same bases as the current one.
27699 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27701 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
27702 instead of vec_select for V1TImode.
27703 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
27705 (VSX_LE_128): Add V1TI to this mode iterator.
27706 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
27707 (*vsx_le_perm_store_<mode>): Likewise.
27708 (pre-reload splitter for VSX stores): Likewise.
27709 (post-reload splitter for VSX stores): Likewise.
27710 (*vsx_xxpermdi2_le_<mode>): Likewise.
27711 (*vsx_lxvd2x2_le_<mode>): Likewise.
27712 (*vsx_stxvd2x2_le_<mode>): Likewise.
27714 2017-03-09 Michael Eager <eager@eagercon.com>
27716 Correct failures with --enable-checking=yes,rtl.
27718 * config/microblaze/microblaze.c (microblaze_expand_shift):
27719 Replace GET_CODE test with CONST_INT_P and INTVAL test with
27720 test for const0_rtx.
27721 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
27722 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
27724 2017-03-09 Richard Biener <rguenther@suse.de>
27726 PR tree-optimization/79977
27727 * graphite-scop-detection.c (scop_detection::merge_sese):
27728 Handle the case of extra exits to blocks dominating the entry.
27730 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
27732 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
27735 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
27737 PR rtl-optimization/79949
27738 * lra-constraints.c (process_alt_operands): Check memory when
27739 trying to predict a cycle. Print about the overall increase.
27741 2017-03-09 Richard Biener <rguenther@suse.de>
27743 PR middle-end/79971
27744 * gimple-expr.c (useless_type_conversion_p): Preserve
27745 TYPE_SATURATING for fixed-point types.
27747 2017-03-09 Richard Biener <rguenther@suse.de>
27750 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
27751 alignment of BLKmode params.
27753 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27756 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
27757 (VALL_NO_V2Q): Likewise.
27759 * config/aarch64/aarch64-simd.md
27760 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
27762 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
27763 VALL_NO_V2Q mode iterator.
27764 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
27766 2017-03-09 Martin Liska <mliska@suse.cz>
27768 PR tree-optimization/79631
27769 * tree-chkp-opt.c (chkp_is_constant_addr): Call
27770 tree_int_cst_sign_bit just for INTEGER constants.
27772 2017-03-09 Martin Liska <mliska@suse.cz>
27776 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
27779 2017-03-09 Marek Polacek <polacek@redhat.com>
27782 * tree.c (inchash::add_expr): Handle TREE_VEC.
27784 2017-03-09 Martin Liska <mliska@suse.cz>
27787 (chkp_narrow_size_and_offset): New function.
27788 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
27789 (void chkp_parse_bit_field_ref): New function.
27790 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
27791 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
27793 2017-03-09 Martin Liska <mliska@suse.cz>
27796 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
27797 (chkp_find_bounds_1): Remove gcc_unreachable.
27799 2017-03-09 Jakub Jelinek <jakub@redhat.com>
27802 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
27803 BUILT_IN_SYNC*, determine the access type from the size suffix and
27804 always build a MEM_REF with that type. Handle forgotten
27805 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
27808 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
27809 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
27810 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
27811 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
27812 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
27813 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
27814 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
27815 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
27816 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
27817 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
27818 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
27819 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
27820 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
27821 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
27822 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
27823 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
27824 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
27825 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
27826 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
27827 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
27828 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
27829 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
27830 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
27831 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
27832 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
27833 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
27834 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
27835 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
27836 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
27837 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
27838 definitions outside of __OPTIMIZE__ guarded section.
27841 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
27842 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
27843 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
27844 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
27847 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27849 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
27850 ("vfenez<mode>"): Add missing constraints.
27852 2017-03-08 Martin Sebor <msebor@redhat.com>
27855 * config/nds32/nds32.c (nds32_option_override):
27856 Fix misspelled diagnostic.
27858 2017-03-08 Jakub Jelinek <jakub@redhat.com>
27861 * gimplify.c (gimplify_omp_for): Replace index var in outer
27862 taskloop statement with an artificial variable and add
27863 OMP_CLAUSE_PRIVATE clause for it.
27865 2017-03-08 Richard Biener <rguenther@suse.de>
27867 PR tree-optimization/79955
27868 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
27869 for accesses that are completely outside of the variable.
27871 2017-03-08 Andrew Haley <aph@redhat.com>
27873 PR tree-optimization/79943
27874 * tree-ssa-loop-split.c (compute_new_first_bound): When
27875 calculating the new upper bound, (END-BEG) should be added, not
27878 2017-03-08 Jakub Jelinek <jakub@redhat.com>
27880 * config/avr/avr.md (setmemhi): Make sure match_dup
27881 operand number comes before match_scratch.
27883 2017-03-08 Richard Biener <rguenther@suse.de>
27885 PR tree-optimization/79920
27886 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
27887 with ncopies == 1 to ...
27888 (vect_transform_slp_perm_load): ... here. Properly compute
27889 all element loads by iterating VF times over the group. Do
27890 not handle ncopies (computed in a broken way) in
27891 vect_create_mask_and_perm.
27893 2017-03-08 Jakub Jelinek <jakub@redhat.com>
27896 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
27897 is a uniform vector, use uniform_vector_p return value instead of
27898 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
27900 2017-03-07 Marek Polacek <polacek@redhat.com>
27902 PR middle-end/79809
27903 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
27904 (alloca_call_type): Likewise.
27906 2017-03-07 Martin Liska <mliska@suse.cz>
27908 * gcov.c (process_args): Put comment to correct location.
27910 2017-03-07 Martin Liska <mliska@suse.cz>
27912 PR middle-end/68270
27913 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
27914 Use array_at_struct_end_p instead of DECL_CHAIN (field).
27915 (chkp_narrow_bounds_for_field): Likewise.
27916 (chkp_parse_array_and_component_ref): Pass one more argument to
27919 2017-03-07 Richard Biener <rguenther@suse.de>
27921 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
27924 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
27926 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
27927 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
27929 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27932 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
27933 to end of description.
27934 (PARAM_MAX_STORES_TO_MERGE): Likewise.
27936 2017-03-07 Jakub Jelinek <jakub@redhat.com>
27938 PR rtl-optimization/79901
27939 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
27941 (*avx512f_<code><mode>3<mask_name>): ... this.
27942 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
27943 iterator instead of VI8_AVX2_AVX512BW.
27945 PR rtl-optimization/79901
27946 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
27947 min/max expander, expand it using expand_vec_cond_expr.
27950 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
27953 2017-03-06 Jakub Jelinek <jakub@redhat.com>
27956 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
27957 to void * for PCH reasons.
27958 * dwarf2out.c (output_loc_operands, output_die): Cast
27959 v.val_vec.array to unsigned char *.
27961 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
27964 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
27967 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
27969 PR rtl-optimization/79571
27970 * lra-constraints.c (process_alt_operands): Calculate static
27971 reject and subtract it from overall when only addresses will be
27974 2017-03-06 Julia Koval <julia.koval@intel.com>
27977 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
27978 incoming stack boundary to 128 for 64-bit targets.
27980 2017-03-06 Richard Biener <rguenther@suse.de>
27982 PR tree-optimization/79894
27983 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
27984 to NULL after folding it.
27986 2017-03-06 Richard Biener <rguenther@suse.de>
27988 PR tree-optimization/79824
27989 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
27990 check disabling peeling for gaps.
27992 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
27994 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
27995 attributes): Document gettimeofday.
27997 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
27999 * config/s390/s390.c (s390_option_override_internal): Set
28000 PARAM_MIN_VECT_LOOP_BOUND
28002 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28004 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
28005 * config/s390/s390.md: Likewise.
28007 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28010 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
28011 (<avx2_avx512>_perm<mode>): Rename to ...
28012 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
28014 (<avx512>_perm<mode>_mask): Rename to ...
28015 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
28017 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
28018 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
28019 instead of VI8F_256_512.
28020 (avx512f_perm<mode>): New define_expand.
28021 (avx512f_perm<mode>_mask): Likewise.
28022 (avx512f_perm<mode>_1<mask_name>): New define_insn.
28023 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
28025 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28027 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
28028 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
28030 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
28032 2017-03-06 Martin Liska <mliska@suse.cz>
28035 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
28036 when having a SSA NAME w/o VAR_DECL assigned to it.
28038 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28040 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
28041 msa_dpsub_<su>_d): Fix MODE for vec_select.
28043 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28045 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
28047 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
28049 2017-03-06 Richard Biener <rguenther@suse.de>
28051 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
28052 * plugin.c (register_plugin_info): Likewise.
28053 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
28055 2017-03-05 Jakub Jelinek <jakub@redhat.com>
28057 * config/i386/sse.md (sse_storehps, sse_storelps,
28058 avx_<castmode><avxsizesuffix>_<castmode>,
28059 avx512f_<castmode><avxsizesuffix>_<castmode>,
28060 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
28061 in condition that at least one operand is not a MEM.
28063 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28065 PR middle-end/79805
28066 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
28067 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
28069 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
28070 gimple_call_nothrow_p flag based on whether original builtin can throw.
28071 If it can, emit following stmts on the fallthrough edge.
28072 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
28073 don't create new bb if inserting just debug stmts on the edge, try to
28074 insert them on the fallthru bb or just reset debug stmts.
28076 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
28079 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
28080 restore recog_data (including the operand rtxes inside it) around
28081 the call to get_insn_template.
28083 2017-03-03 Martin Sebor <msebor@redhat.com>
28085 PR tree-optimization/79699
28086 * context.c (context::~context): Free MPFR caches to avoid
28087 a memory leak on program exit.
28089 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28091 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
28092 Use wide_int::ulow () instead of .elt (0).
28094 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28096 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
28097 (*pushxf): Limit oF constraint to 32bit targets and add oC
28098 constraint for 64bit targets.
28099 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
28100 (*pushdf): Change rmF constraint to rmC.
28102 2017-03-03 Martin Liska <mliska@suse.cz>
28104 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
28105 Remove unused variable.
28107 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28110 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
28111 is a memory operand, increase num_memory.
28112 (ix86_expand_args_builtin): Likewise.
28114 2017-03-03 Jan Hubicka <jh@suse.cz>
28117 * ipa-devirt.c (maybe_record_node): Properly handle
28118 __cxa_pure_virtual visibility.
28120 2017-03-03 Martin Liska <mliska@suse.cz>
28122 PR tree-optimization/79803
28123 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
28125 (pass_loop_prefetch::execute): Disabled optimization if an
28126 assumption about L1 cache size is not met.
28128 2017-03-03 Martin Liska <mliska@suse.cz>
28130 PR rtl-optimization/79574
28131 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
28132 (hash_scan_set): Likewise.
28133 (dump_hash_table): Likewise.
28134 (hoist_code): Likewise.
28136 2017-03-03 Richard Biener <rguenther@suse.de>
28138 * fixed-value.c (fixed_from_string): Restore use of elt (1)
28139 in place of uhigh ().
28140 (fixed_convert_from_real): Likewise.
28142 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28145 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
28147 2017-03-03 Richard Biener <rguenther@suse.de>
28149 PR middle-end/79818
28150 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
28151 TYPE_OVERFLOW_UNDEFINED check.
28153 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28155 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
28157 (vector_ae_<mode>_p): Likewise.
28158 (vector_nez_<mode>_p): Likewise.
28159 (vector_ne_v2di_p): Likewise.
28160 (vector_ae_v2di_p): Likewise.
28161 (vector_ne_<mode>_p): Likewise.
28162 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
28164 (vsx_tsqrt<mode>2_fe): Likewise.
28166 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
28169 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
28171 2017-03-02 Jakub Jelinek <jakub@redhat.com>
28173 PR rtl-optimization/79780
28174 * cprop.c (one_cprop_pass): When second and further conditional trap
28175 in a single basic block is turned into an unconditional trap, turn it
28176 into a deleted note to avoid RTL verification failures.
28178 2017-03-02 Richard Biener <rguenther@suse.de>
28180 * fold-const.c (const_binop): Use ulow () instead of elt (0).
28182 2017-03-02 Richard Biener <rguenther@suse.de>
28184 PR tree-optimization/79345
28186 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
28187 param and abort the walk, returning -1 if it is hit.
28188 (walk_aliased_vdefs): Take a limit param and pass it on.
28189 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
28190 defaulting to 0 and return a signed int.
28191 * tree-ssa-uninit.c (struct check_defs_data): New struct.
28192 (check_defs): New helper.
28193 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
28194 about uninitialized memory.
28195 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
28196 bogus uninitialized warning.
28197 (fixed_convert_from_real): Likewise.
28199 2017-03-02 Bin Cheng <bin.cheng@arm.com>
28201 PR tree-optimization/66768
28202 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
28203 iv_use if base object can't be determined.
28205 2017-03-02 Jakub Jelinek <jakub@redhat.com>
28207 PR tree-optimization/79345
28208 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
28209 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
28210 (get_pattern_stats): Initialize it.
28211 * genemit.c (gen_expand): Verify match_scratch numbers come after
28212 match_operand/match_dup numbers.
28213 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
28214 match_scratch numbers.
28215 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
28217 * config/s390/s390.md (trunctdsd2): Likewise.
28219 2017-03-02 Richard Biener <rguenther@suse.de>
28221 * wide-int.h (wide_int_storage::operator=): Implement in terms
28224 2017-03-02 Richard Biener <rguenther@suse.de>
28226 PR tree-optimization/79777
28227 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
28228 the to insert expression to sth existing.
28230 2017-03-01 Martin Sebor <msebor@redhat.com>
28232 PR middle-end/79692
28233 * gimple-ssa-sprintf.c
28234 (directive::known_width_and_precision): New function.
28235 (format_integer): Use it.
28236 (get_mpfr_format_length): Consider the full range of precision
28237 when computing %g output with the # flag. Set the likely byte
28238 count to 3 rather than 1 when precision is indeterminate.
28239 (format_floating): Correct the lower bound of precision.
28241 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28243 * doc/invoke.texi: Document default code model for 64-bit Linux.
28245 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28248 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
28249 udiv rather than div since input pattern is unsigned.
28251 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
28253 * config/i386/i386.c (print_reg): Warn for values of
28254 unsupported size in integer register.
28256 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
28259 * config/rs6000/predicates.md (current_file_function_operand): Do
28260 not allow self calls to be local if the function is replaceable.
28262 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
28265 * config/rs6000/altivec.h (vec_ctz and others): Change the
28266 preprocessor macro that controls conditional compilation from
28267 _ARCH_PWR9 to __POWER9_VECTOR__.
28268 (vec_all_ne): Change parameterization of __altivec_scalar_pred
28269 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
28270 control (instead of _ARCH_PWR9 control) so that template
28271 definition uses power9-specific function.
28272 (vec_any_eq): Likewise.
28273 (vec_all_ne): Change macro definition to use a power9-specific
28274 expansion under #ifdef __POWER9_VECTOR__ control (instead of
28275 _ARCH_PWR9 control).
28276 (vec_any_eq) Likewise.
28277 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
28278 expansion for CMPNEF to remove support for xvcmpnesp instruction.
28279 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
28280 support for xvcmpnedp instruction.
28281 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
28282 macro expansion so that Power9 implementation of vec_all_ne does
28283 not use the AltiVec predicate framework.
28284 (VCMPNEH_P): Likewise.
28285 (VCMPNEW_P): Likewise.
28286 (VCMPNED_P): Likewise.
28287 (VCMPNEFP_P): Likewise.
28288 (VCMPNEDP_P): Likewise.
28289 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
28290 implementation of vec_any_eq to not use AltiVec predicate
28292 (VCMPAEH_P): Likewise.
28293 (VCMPAEW_P): Likewise.
28294 (VCMPAED_P): Likewise.
28295 (VCMPAEFP_P): Likewise.
28296 (VCMPAEDP_P): Likewise.
28297 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
28298 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
28299 not use the AltiVec predicate framework.
28300 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
28301 of vec_any_eq to not use AltiVec predicate framework.
28302 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
28303 support for predefined __POWER9_VECTOR__ macro to indicate that
28304 Power9 instruction selection is enabled.
28305 (altivec_overloaded_builtins): Remove extraneous
28306 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
28307 function argument types RS6000_BTI_bool_V16QI and
28308 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
28309 entry for overloaded function argument types RS6000_BTI_bool_V4SI
28310 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
28311 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
28312 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
28313 Power9 for implementations of vec_cmpne. Change the signature for
28314 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
28315 (representing vec_all_ne) to remove the previously described first
28316 argument of type RS6000_BTI_INTSI, as this was an artifact of
28317 reliance on the AltiVec predicate framework, which is no longer
28318 used in the implementation of these functions. Add
28319 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
28320 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
28321 since, unlike the AltiVec predicate framework implementation, we
28322 do not share function descriptors between vec_alle and vec_anyeq.
28323 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
28324 set of modes that receive special treatment even when
28325 TARGET_P9_VECTOR is true. The special treatment emits code that
28326 does not depend on Power9 instructions.
28327 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
28328 define_expand to not rely on AltiVec predicate framework.
28329 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
28331 (vector_ne_v2di_p): Change this define_expand to not rely on
28332 AltiVec predicate framework.
28333 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
28335 (vector_ne_<mode>_p): Change this define_expand to not rely on
28336 AltiVec predicate framework.
28337 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
28339 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
28340 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
28341 define_insn pattern.
28342 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
28343 define_insn pattern because the xvcmpne<VSs>. instruction is not
28345 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
28346 instruction is not supported.
28348 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28350 * config/nvptx/nvptx.c: Include intl.h.
28352 2017-03-01 Martin Jambor <mjambor@suse.cz>
28355 * ipa-prop.h (ipa_bits): Removed field known.
28356 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
28357 to pointers. Adjusted their comments to warn about their sharing.
28358 (ipcp_transformation_summary): Change bits to a vector of pointers.
28359 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
28360 (ipa_get_ipa_bits_for_value): Declare.
28361 * tree-vrp.h (value_range): Mark as GTY((for_user)).
28362 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
28363 (ipa_bits_hash_table): Likewise.
28364 (ipa_vr_ggc_hash_traits): Likewise.
28365 (ipa_vr_hash_table): Likewise.
28366 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
28367 being pointers and vr_known being removed.
28368 (ipa_set_jf_unknown): Likewise.
28369 (ipa_get_ipa_bits_for_value): New function.
28370 (ipa_set_jfunc_bits): Likewise.
28371 (ipa_get_value_range): New overloaded functions.
28372 (ipa_set_jfunc_vr): Likewise.
28373 (ipa_compute_jump_functions_for_edge): Use the above functions to
28374 construct bits and vr parts of jump functions.
28375 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
28376 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
28378 (ipcp_grow_transformations_if_necessary): Also allocate
28379 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
28381 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
28382 them. Fix too long lines.
28383 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
28384 vr_known being removed.
28385 (ipa_read_jump_function): Use new setter functions to construct bits
28386 and vr parts of jump functions or set them to NULL.
28387 (write_ipcp_transformation_info): Adjust for bits being pointers.
28388 (read_ipcp_transformation_info): Likewise.
28389 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
28391 Include gt-ipa-prop.h.
28392 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
28394 (ipcp_store_bits_results): Likewise.
28395 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
28396 Do not write to existing jump functions but use a temporary instead.
28398 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28401 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
28402 attempt to use its first operand as BIT_FIELD_REF base.
28404 2017-03-01 Richard Biener <rguenther@suse.de>
28406 PR middle-end/79721
28407 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
28408 interpolating formula in wrapping arithmetic.
28409 (chrec_apply): Convert chrec_evaluate return value to wanted type.
28411 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28413 PR tree-optimization/79734
28414 * tree-vect-generic.c (expand_vector_condition): Optimize
28415 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
28416 Handle VEC_COND_EXPR where comparison has different inner width from
28417 type's inner width.
28419 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
28421 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
28422 markup, and similar issues. Remove @opindex entries for things
28423 that aren't options. Add missing -mmpy-option entries.
28425 2017-02-28 Jakub Jelinek <jakub@redhat.com>
28427 PR tree-optimization/79737
28428 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
28429 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
28430 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
28431 instead of byte_size. Formatting fix.
28432 (shift_bytes_in_array_right): Formatting fix.
28434 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
28437 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
28438 condition on optimize for the leaf function test.
28440 2017-02-28 Martin Liska <mliska@suse.cz>
28443 * read-rtl-function.c (function_reader::handle_unknown_directive):
28444 Bail out when one uses -flto.
28446 2017-02-28 Martin Liska <mliska@suse.cz>
28448 * common.opt: Replace space with tabular for options of <number>
28450 * config/i386/i386.opt: Show <number> value for
28451 -mlarge-data-threshold.
28452 * opts.c (print_filtered_help): Do not display number in hexadecimal
28455 2017-02-28 Martin Liska <mliska@suse.cz>
28457 * common.opt: Fix --help=option -Q for options which are of
28460 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
28462 * config/i386/i386.c (print_reg): Error out for values
28463 of 8-bit size in invalid integer register.
28465 2017-02-28 Martin Sebor <msebor@redhat.com>
28467 PR tree-optimization/79691
28468 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
28470 2017-02-28 Jakub Jelinek <jakub@redhat.com>
28473 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
28474 gcc_unreachable with output_operand_lossage.
28476 2017-02-28 Richard Biener <rguenther@suse.de>
28478 PR tree-optimization/79740
28479 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
28481 (visit_nary_op): Insert the nary into the hashtable if we
28482 pattern-matched sth.
28483 * tree-ssa-pre.c (eliminate_insert): Robustify.
28485 2017-02-28 Richard Biener <rguenther@suse.de>
28487 PR middle-end/79731
28488 * fold-const.c (decode_field_reference): Reject out-of-bound
28491 2017-02-28 Jakub Jelinek <jakub@redhat.com>
28493 * config/i386/i386.c: Include intl.h.
28494 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
28495 instead of just cond ? "..." : "...".
28496 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
28497 * coverage.c (read_counts_file): Likewise.
28498 * omp-offload.c: Include intl.h.
28499 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
28500 of just cond ? "..." : "...".
28501 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
28502 of just cond ? "..." : "...".
28504 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
28507 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
28509 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
28510 'tune for' CPU name.
28511 * config/arm/arm-cpu-data.h: Regenerated.
28513 2017-02-28 Richard Biener <rguenther@suse.de>
28515 PR tree-optimization/79732
28516 * tree-inline.c (expand_call_inline): Do not shadow var.
28518 2017-02-28 Richard Biener <rguenther@suse.de>
28520 PR tree-optimization/79723
28521 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
28522 address-space properly.
28524 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
28526 * doc/optinfo.texi (Optimization groups): Fix option used for
28528 * doc/invoke.texi (-fopt-info): Document "omp".
28529 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
28530 (OPTGROUP_ALL): Add OPTGROUP_OMP.
28531 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
28532 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
28533 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
28535 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
28537 * dumpfile.c (optgroup_options): Instead of "openmp", associate
28538 OPTGROUP_OMP with "omp".
28540 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
28543 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
28544 for arithmetic shift of unsigned V2DI.
28546 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
28548 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
28549 arc/linux.h headers.
28550 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
28551 (LINK_SPEC): Likewise.
28552 (ARC_TLS_EXTRA_START_SPEC): Likewise.
28553 (EXTRA_SPECS): Likewise.
28554 (STARTFILE_SPEC): Likewise.
28555 (ENDFILE_SPEC): Likewise.
28556 (LIB_SPEC): Likewise.
28557 (TARGET_SDATA_DEFAULT): Likewise.
28558 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
28559 (MULTILIB_DEFAULTS): Likewise.
28560 (DWARF2_UNWIND_INFO): Likewise.
28561 * config/arc/big.h: New file.
28562 * config/arc/elf.h: Likewise.
28563 * config/arc/linux.h: Likewise.
28564 * config/arc/t-uClibc: Remove.
28566 2017-02-27 Bin Cheng <bin.cheng@arm.com>
28568 PR tree-optimization/77536
28569 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
28570 (tree_transform_and_unroll_loop): Use above function to compute the
28571 estimated niter of unrolled loop and use it when scaling profile.
28572 Also use count info rather than frequency if it's non-zero.
28573 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
28574 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
28575 (vect_transform_loop): Call above function.
28577 2017-02-27 Richard Biener <rguenther@suse.de>
28579 PR tree-optimization/45397
28580 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
28581 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
28582 (visit_nary_op): Add pattern matching for CSEing sign-changed
28583 or truncated operations with wider ones.
28585 2017-02-27 Richard Biener <rguenther@suse.de>
28587 PR tree-optimization/79690
28588 * tree-vect-stmts.c (vectorizable_store): Use vector type
28589 built from the DR with address-space.
28591 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
28593 * doc/invoke.texi (Optimize Options): Refine the description
28594 of asan-use-after-return.
28596 2017-02-25 Alan Modra <amodra@gmail.com>
28598 PR rtl-optimization/79584
28599 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
28600 base, not ad->base_term, the reg within base. Remove assertion
28601 that ad->base == ad->base_term. Replace gen_int_mode using
28602 bogus mode with const0_rtx.
28604 2017-02-25 Jakub Jelinek <jakub@redhat.com>
28606 PR middle-end/79396
28607 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
28608 FMA_EXPR like tcc_binary or tcc_unary.
28610 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
28613 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
28615 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
28616 (output_loc_operands): Handle DW_OP_call_ref and
28617 DW_OP_GNU_variable_value.
28618 (struct variable_value_struct): New type.
28619 (struct variable_value_hasher): Likewise.
28620 (variable_value_hash): New variable.
28621 (string_types): Remove.
28622 (copy_loc_descr): New function.
28623 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
28624 (prepend_loc_descr_to_each): New function.
28625 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
28626 instead of add_loc_descr_to_each if the first argument is single
28627 location list and the second has multiple.
28628 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
28629 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
28630 when looking for variable value which doesn't have other location info.
28631 (loc_list_from_tree): Formatting fix.
28632 (gen_array_type_die): Simplify DW_AT_string_length handling.
28633 (adjust_string_types): Remove.
28634 (gen_subprogram_die): Don't call adjust_string_types nor test/set
28635 string_types. Call resolve_variable_values.
28636 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
28637 (resolve_addr_in_expr): Likewise. Add A argument.
28638 (copy_deref_exprloc): Remove deref argument. Adjust for the
28639 original expression being DW_OP_GNU_variable_value with optionally
28640 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
28641 optionally after it.
28642 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
28643 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
28644 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
28645 (variable_value_hasher::hash, variable_value_hasher::equal): New
28647 (resolve_variable_value_in_expr, resolve_variable_value,
28648 resolve_variable_values, note_variable_value_in_expr,
28649 note_variable_value): New functions.
28650 (dwarf2out_early_finish): Call note_variable_value on all toplevel
28653 2017-02-24 Jakub Jelinek <jakub@redhat.com>
28656 * opts.h (handle_generated_option): Add GENERATED_P argument.
28657 * opts-common.c (handle_option): Adjust function comment.
28658 (handle_generated_option): Add GENERATED_P argument, pass it to
28660 (control_warning_option): Pass false to handle_generated_option
28662 * opts.c (maybe_default_option): Pass true to handle_generated_option
28664 * optc-gen.awk: Likewise.
28666 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28668 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
28669 a REG, look at the REG it is a SUBREG of.
28670 (splitter for cmpeqsi_t): Ditto.
28672 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28674 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
28675 the special USEs with the pattern of the insn, not the insn itself.
28677 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
28680 * doc/invoke.texi: Document -mload-store-pairs.
28682 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28683 Sandra Loosemore <sandra@codesourcery.com>
28685 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
28686 argument isn't a CONST_INT.
28687 (nios2_alternate_compare_const): Assert op is a CONST_INT.
28688 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
28689 (nios2_validate_compare): Bypass alternate compare logic if *op2
28690 is not a CONST_INT.
28691 (ldstwm_operation_p): Return false if first_base is not a REG or
28692 if first_offset is not a CONST_INT.
28694 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28696 * config/cris/cris.md: Use correct operand in a define_peephole2.
28698 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28700 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
28702 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28704 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
28705 this_insn if it is an INSN or JUMP_INSN.
28706 (force_offsettable): Look at base, not at addr.
28707 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
28708 on things that aren't necessarily CONST_INTs.
28710 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
28712 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
28713 -mfpmath=sse is the default also for x86-32 targets with SSE2
28714 instruction set when @option{-ffast-math} is enabled
28716 2017-02-24 Jeff Law <law@redhat.com>
28718 PR rtl-optimizatoin/79286
28719 * ira.c (update_equiv_regs): Drop may_trap_p exception to
28722 2017-02-24 Richard Biener <rguenther@suse.de>
28724 PR tree-optimization/79389
28725 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
28728 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
28730 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
28731 function comment to reflect reality.
28732 (loop_exits_before_overflow): Fix typo in function description.
28734 2017-02-24 Richard Biener <rguenther@suse.de>
28736 PR tree-optimization/79389
28737 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
28738 properly that a threading opportunity exists. Detect conditional
28739 copy/constant propagation opportunities.
28741 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
28743 * config/visium/visium.md (type): Add trap.
28744 (b): New mode attribute.
28745 (*btst): Rename into...
28746 (*btst<mode>): ...this and adjust.
28747 (*cbranchsi4_btst_insn): Rename into...
28748 (*cbranch<mode>4_btst_insn): ...this and adjust.
28749 (trap): New define_insn.
28751 2017-02-23 Jakub Jelinek <jakub@redhat.com>
28753 PR tree-optimization/79389
28754 * ifcvt.c (struct noce_if_info): Add rev_cond field.
28755 (noce_reversed_cond_code): New function.
28756 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
28757 reversed_comparison_code. Formatting fix.
28758 (noce_try_store_flag): Test rev_cond != NULL in addition to
28759 reversed_comparison_code.
28760 (noce_try_store_flag_constants): Likewise.
28761 (noce_try_store_flag_mask): Likewise.
28762 (noce_try_addcc): Use rev_cond if non-NULL instead of
28763 reversed_comparison_code.
28764 (noce_try_cmove_arith): Likewise. Formatting fixes.
28765 (noce_try_minmax, noce_try_abs): Clear rev_cond.
28766 (noce_find_if_block): Initialize rev_cond.
28767 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
28768 instead of false as last argument never attempt to reverse it
28771 2017-02-23 Bin Cheng <bin.cheng@arm.com>
28773 PR tree-optimization/79663
28774 * tree-predcom.c (combine_chains): Process refs in reverse order
28775 only for ZERO length chains, and add explaining comment.
28777 2017-02-23 Jeff Law <law@redhat.com>
28779 PR tree-optimization/79578
28780 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
28781 in call to operand_equal_p.
28783 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
28786 * config/i386/cpuid.h: Fix another undefined behavior.
28788 2017-02-23 Richard Biener <rguenther@suse.de>
28790 PR tree-optimization/79683
28791 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
28792 vector types for data-refs.
28794 2017-02-23 Martin Liska <mliska@suse.cz>
28796 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
28798 2017-02-23 Jakub Jelinek <jakub@redhat.com>
28800 PR middle-end/79665
28801 * internal-fn.c (get_range_pos_neg): Moved to ...
28802 * tree.c (get_range_pos_neg): ... here. No longer static.
28803 * tree.h (get_range_pos_neg): New prototype.
28804 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
28805 are known to be in between 0 and signed maximum inclusive, try to
28806 expand both unsigned and signed divmod and use the cheaper one from
28809 2017-02-22 Jeff Law <law@redhat.com>
28811 PR tree-optimization/79578
28812 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
28813 to compare base operands.
28815 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
28818 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
28819 gpc_reg_operand instead of fpr_reg_operand.
28821 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
28823 * config/mips/mips.c (mips_return_in_memory): Force FP
28824 vector types to be returned in memory for o32 ABI.
28826 2017-02-22 Jakub Jelinek <jakub@redhat.com>
28828 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
28829 instead of DW_TAG_member for static data member declarations and don't
28830 set no_linkage_name for static inline data members.
28831 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
28834 2017-02-22 Martin Liska <mliska@suse.cz>
28836 * doc/invoke.texi: Replace inequality signs with square brackets
28839 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
28842 * lra-constraints.c (simplify_operand_subreg): Handle
28843 WORD_REGISTER_OPERATIONS targets.
28845 2017-02-22 Jakub Jelinek <jakub@redhat.com>
28848 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
28849 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
28850 elimination by swapping fld*.
28852 2017-02-22 Richard Biener <rguenther@suse.de>
28854 PR tree-optimization/79673
28855 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
28856 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
28857 irrelevant address-space qualifiers and avoiding a
28858 ADDR_SPACE_CONVERT_EXPR from fold_convert.
28860 2017-02-22 Richard Biener <rguenther@suse.de>
28862 PR tree-optimization/79666
28863 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
28864 to not symbolically negate if that may introduce undefined
28867 2017-02-22 Martin Liska <mliska@suse.cz>
28870 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
28871 * data-streamer-out.c (streamer_write_gcov_count_stream):
28873 * value-prof.c (stream_out_histogram_value): Make assert more
28874 precise based on type of counter.
28876 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
28879 * config/i386/i386.md (standard_x87sse_constant_load splitter):
28880 Use nonimmediate_operand instead of memory_operand for operand 1.
28881 (float-extend standard_x87sse_constant_load splitter): Ditto.
28883 2017-02-21 Jeff Law <law@redhat.com>
28885 PR tree-optimization/79621
28886 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
28887 blocks with edges to themselves.
28889 2017-02-21 Jakub Jelinek <jakub@redhat.com>
28892 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
28893 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
28894 Use gimple_call_builtin_p.
28897 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
28898 on temporarily removed DEBUG_INSNs.
28900 PR tree-optimization/79649
28901 * tree-loop-distribution.c (classify_partition): Give up on
28902 non-generic address space loads/stores.
28904 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
28906 * doc/loop.texi (Loop manipulation): Remove nonexistent
28907 tree_ssa_loop_version from the documentation.
28908 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
28910 2017-02-21 Jakub Jelinek <jakub@redhat.com>
28913 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
28914 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
28915 * config/rs6000/rs6000.c: Include except.h.
28916 (rs6000_expand_split_stack_prologue): Call
28917 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
28919 2017-02-21 Martin Jambor <mjambor@suse.cz>
28922 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
28923 have been analyzed.
28925 2017-02-21 Martin Jambor <mjambor@suse.cz>
28927 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
28928 for backward compatibility only.
28929 * doc/invoke.texi (Option Summary): Remove all references to
28930 -fipa-cp-alignment.
28932 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
28936 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
28938 * lra-constraints.c (curr_insn_transform): Handle
28939 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
28941 2017-02-21 Martin Liska <mliska@suse.cz>
28943 * config/i386/i386.opt: Replace -masm-dialect with -masm.
28945 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
28947 PR translation/79638
28948 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
28950 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
28953 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
28954 (arm_function_ok_for_sibcall): Return false for an indirect call by
28955 descriptor if all the argument registers are used.
28956 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
28957 alignment of the function.
28959 2017-02-21 Jakub Jelinek <jakub@redhat.com>
28961 PR tree-optimization/61441
28962 * simplify-rtx.c (simplify_const_unary_operation): For
28963 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
28964 the sNaN unmodified.
28966 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
28968 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
28969 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
28970 instead of SYSTEM_HEADER_DIR.
28972 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
28973 Martin Liška <mliska@suse.cz>
28975 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
28976 Fix typos and grammar, use active voice, and clarify.
28978 2017-02-20 Marek Polacek <polacek@redhat.com>
28980 PR middle-end/79537
28981 * gimplify.c (gimplify_expr): Handle unused *&&L;.
28984 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
28986 2017-02-20 Jakub Jelinek <jakub@redhat.com>
28989 * config/i386/i386.c (ix86_expand_builtin): Handle
28990 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
28991 ix86_builtins_isa[fcode].isa as a requirement of those
28992 flags and any other flag in the bitmask.
28993 (ix86_init_mmx_sse_builtins): Use 0 instead of
28994 ~OPTION_MASK_ISA_64BIT as mask.
28995 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
28996 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
28997 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
28998 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
29000 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29003 * lra-constraints.c (split_reg): Check requested split mode
29004 is supported by the register.
29006 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29008 * lra-constraints.c (simplify_operand_subreg): Remove early
29011 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29014 * lra-constraints.c (curr_insn_transform): Tighten condition
29015 for converting SUBREG reloads from OP_OUT to OP_INOUT.
29017 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29020 * lra-constraints.c (curr_insn_transform): Handle
29021 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29023 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
29026 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
29028 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
29030 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
29033 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
29036 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29038 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
29041 2017-02-18 Jakub Jelinek <jakub@redhat.com>
29043 * final.c (last_columnnum, override_columnnum): New variables.
29044 (final_start_function): Set last_columnnum, pass it to begin_prologue
29045 hook and pass 0 to dwarf2out_begin_prologue.
29046 (final_scan_insn): Update override_columnnum. Pass last_columnnum
29047 to source_line debug hook.
29048 (notice_source_line): Compute last_columnnum and for debug_column_info
29049 return true on column changes.
29050 * debug.h (struct gcc_debug_hooks): Add column argument to
29051 source_line and begin_prologue hooks.
29052 (debug_nothing_int_charstar_int_bool): Remove prototype.
29053 (debug_nothing_int_int_charstar,
29054 debug_nothing_int_int_charstar_int_bool): New prototypes.
29055 (dwarf2out_begin_prologue): Add column argument.
29056 * debug.c (do_nothing_debug_hooks): Adjust source_line and
29057 begin_prologue hooks.
29058 (debug_nothing_int_charstar_int_bool): Remove.
29059 (debug_nothing_int_int_charstar,
29060 debug_nothing_int_int_charstar_int_bool): New functions.
29061 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
29062 through to dwarf2out_source_line.
29063 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
29064 (dwarf2out_source_line): Add column argument, emit it if requested.
29065 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
29067 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29068 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29069 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
29070 through to dwarf2out_begin_prologue.
29071 (vmsdbgout_source_line): Add column argument, pass it through to
29072 dwarf2out_source_line.
29073 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
29074 dbxout_source_line caller.
29075 (dbxout_source_line): Add column argument.
29077 * common.opt (gno-column-info, gcolumn-info): New options.
29078 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
29079 (check_die): Also test for multiple DW_AT_decl_column attributes.
29080 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
29081 DW_AT_decl_column if requested.
29082 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
29084 (gen_variable_die): Likewise.
29085 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
29086 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
29089 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
29090 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
29091 (ix86_handle_option): Handle OPT_m3dnowa.
29092 * doc/invoke.texi (-m3dnowa): Document.
29093 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
29094 -m3dnowa instead of -m3dnow -march=athlon.
29097 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
29098 instead of gcc_assert for K, r and R code checks. Formatting fixes.
29100 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29103 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
29104 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
29105 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
29106 generator for vsx_xxpermdi_<mode>_be.
29107 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
29108 force big-endian semantics.
29109 (vsx_xxpermdi_<mode>_be): New define_expand with same
29110 implementation as previous version of vsx_xxpermdi_<mode>.
29112 2017-02-17 Jakub Jelinek <jakub@redhat.com>
29114 PR tree-optimization/79327
29115 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
29116 variable, its initialization and use.
29118 2017-02-17 Julia Koval <julia.koval@intel.com>
29120 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
29121 (OPTION_MASK_ISA_PKU_UNSET): New.
29122 (ix86_handle_option): Handle -mrdpid.
29123 * config/i386/cpuid.h (bit_RDPID): New.
29124 * config/i386/driver-i386.c (host_detect_local_cpu):
29125 Detect RDPID feature.
29126 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
29127 * config/i386/i386-c.c (ix86_target_macros_internal):
29129 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
29130 (ix86_valid_target_attribute_inner_p): Add "rdpid".
29131 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
29132 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
29133 * config/i386/i386.md (define_insn "rdpid"): New.
29134 * config/i386/i386.opt Add -mrdpid.
29135 * config/i386/immintrin.h (_rdpid_u32): New.
29137 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
29139 PR rtl-optimization/79541
29140 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
29141 instead of transforming it into USE.
29143 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
29145 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
29146 If HONOR_SNANS (SFmode) force the input to a register.
29147 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
29148 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
29149 an frsp or similar insn.
29151 2017-02-17 Martin Liska <mliska@suse.cz>
29153 PR rtl-optimization/79577
29154 * params.def (selsched-max-sched-times): Increase minimum to 1.
29156 2017-02-17 Martin Liska <mliska@suse.cz>
29158 PR rtl-optimization/79574
29159 * gcse.c (want_to_gcse_p): Prevent integer overflow.
29161 2017-02-17 Martin Liska <mliska@suse.cz>
29163 PR tree-optimization/79529
29164 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
29165 ssa_defined_default_def_p to handle cases which are implicitly
29167 * tree-ssa.c (ssa_defined_default_def_p): New function.
29168 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
29169 which are implicitly defined.
29170 * tree-ssa.h (ssa_defined_default_def_p): Declare.
29172 2017-02-17 Richard Biener <rguenther@suse.de>
29174 PR middle-end/79576
29175 * params.def (max-ssa-name-query-depth): Limit to 10.
29177 2017-02-17 Richard Biener <rguenther@suse.de>
29179 PR tree-optimization/79552
29180 * tree-ssa-structalias.c (visit_loadstore): Properly verify
29183 2017-02-17 Richard Biener <rguenther@suse.de>
29186 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
29188 2017-02-17 Marek Polacek <polacek@redhat.com>
29190 PR middle-end/79536
29191 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
29192 (fold_negate_expr): New wrapper.
29194 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
29196 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
29197 Correct terminology and de-emphasize pre-standard behavior.
29199 2017-02-16 Alan Modra <amodra@gmail.com>
29201 PR rtl-optimization/79286
29202 * ira.c (def_dominates_uses): New function.
29203 (update_equiv_regs): Don't create an equivalence for insns that
29204 may trap where the register def does not dominate the use.
29206 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
29208 PR rtl-optimization/78127
29209 * lra.c (lra): Call lra_eliminate before finish the loop after
29212 2017-02-16 Richard Biener <rguenther@suse.de>
29214 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
29216 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
29217 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
29218 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
29219 (isl_val_int_from_wi): New function.
29220 (extract_affine_gmp): Rename to ...
29221 (extract_affine_wi): ... this, take a widest_int.
29222 (extract_affine_int): Just wrap extract_affine_wi.
29223 (add_param_constraints): Use isl_val_int_from_wi.
29224 (add_loop_constraints): Likewise, and extract_affine_wi.
29226 2017-02-15 Jeff Law <law@redhat.com>
29228 PR middle-end/79521
29229 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
29230 ira_init_register_move_cost_if_necessary.
29232 2017-02-15 Martin Sebor <msebor@redhat.com>
29234 PR middle-end/32003
29235 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
29236 removed in a prior commit.
29238 2017-02-15 Bin Cheng <bin.cheng@arm.com>
29240 PR tree-optimization/79347
29241 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
29242 counters during peeling.
29244 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
29246 * Makefile.in (site.exp): Remove "set ISLVER".
29248 2017-02-15 Jakub Jelinek <jakub@redhat.com>
29251 * real.c (real_from_integer): Call real_convert even for decimal.
29253 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
29256 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
29258 2017-02-14 Andrew Pinski <apinski@cavium.com>
29260 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
29261 cores and change the partno/implementer to be correct.
29262 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
29263 the 'B" as the implementer.
29264 * config/aarch64/aarch64-tune.md: Regenerate.
29266 2017-02-14 Carl Love <cel@us.ibm.com>
29268 * config/rs6000/rs6000.c: Add case statement entry to make the
29269 xvcvuxdsp built-in argument unsigned.
29270 * config/rs6000/vsx.md: Fix the source and return operand types so they
29271 match the instruction definitions from the ISA document. Fix typo
29272 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
29275 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
29278 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
29279 member early_clobber_alts.
29280 * lra-lives.c (reg_early_clobber_p): New.
29281 (process_bb_lives): Use it.
29282 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
29283 (debug_operand_data): Initialize early_clobber_alts.
29284 (setup_operand_alternative): Set up early_clobber_alts.
29285 (collect_non_operand_hard_regs): Ditto. Pass early clobber
29286 alternatives to new_insn_reg.
29287 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
29289 (lra_update_insn_regno_info): Pass the new arg.
29291 2017-02-14 Jakub Jelinek <jakub@redhat.com>
29293 PR middle-end/79505
29294 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
29295 (new_oacc_loop_raw): Don't clear already cleared fields.
29298 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
29299 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
29300 _mm512_prefetch_i64gather_ps): New inline functions and macros.
29302 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
29305 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
29307 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
29310 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
29311 the extra instruction to the right place to store 128-bit constant
29314 2017-02-14 Martin Sebor <msebor@redhat.com>
29316 PR middle-end/79448
29317 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
29318 warning for strings of unknown length.
29320 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
29322 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
29324 2017-02-14 Jeff Law <law@redhat.com>
29327 * ira-costs.c (scan_one_insn): Initialize register move costs
29328 for pseudos seen in USE/CLOBBER insns.
29330 PR tree-optimization/79095
29331 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
29332 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
29333 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
29334 if the operands are known to be not equal, then the resulting range
29336 (intersect_ranges): If the new range is ~[0,0] and the old range is
29337 wide, then prefer ~[0,0].
29338 * tree-vrp.c (overflow_comparison_p_1): New function.
29339 (overflow_comparison_p): New function.
29340 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
29341 if NAME is used in an overflow test.
29342 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
29343 overflow check that can be expressed as an equality test, then adjust
29344 ops to be that equality test.
29346 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29348 * config/s390/s390-builtin-types.def: Remove flags argument.
29349 * config/s390/s390.c (s390_init_builtins): Likewise.
29351 2017-02-14 Martin Liska <mliska@suse.cz>
29353 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
29354 vector. Fix trailing white spaces.
29356 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
29358 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
29361 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29363 PR rtl-optimization/68664
29364 * config/arm/arm.c (arm_sched_can_speculate_insn):
29365 New function. Declare prototype.
29366 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
29368 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29370 PR rtl-optimization/68664
29371 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
29373 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
29375 2017-02-14 Amit Pawar <amit.pawar@amd.com>
29377 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
29378 max skip bytes for function, loop and jump.
29380 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29382 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
29383 ABS_EXPR for gimple dump.
29385 2017-02-14 Jakub Jelinek <jakub@redhat.com>
29388 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
29390 PR tree-optimization/79408
29391 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
29392 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
29393 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
29394 also if rhs1 is INTEGER_CST.
29396 2017-02-14 Richard Biener <rguenther@suse.de>
29398 PR middle-end/79432
29399 * tree-into-ssa.c (insert_phi_nodes): When the function can
29400 have abnormal edges rewrite SSA names with broken use-def
29401 dominance out of SSA and register them for PHI insertion.
29403 2017-02-13 Martin Sebor <msebor@redhat.com>
29405 PR middle-end/79496
29406 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
29407 clearing info.nowrite flag when snprintf size argument is a range.
29409 2017-02-13 Jakub Jelinek <jakub@redhat.com>
29411 * cprop.c (cprop_jump): Add missing space in string literal.
29412 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
29413 (get_constraint_for_component_ref): Likewise.
29414 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
29415 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
29416 * lra-constraints.c (process_alt_operands): Likewise.
29417 * ipa-inline.c (inline_small_functions): Likewise.
29418 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
29419 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
29420 * trans-mem.c (diagnose_tm_1_op): Likewise.
29421 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
29422 (grid_parallel_clauses_gridifiable): Likewise.
29424 * config/nvptx/mkoffload.c (process): Add space in between
29427 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
29428 "MOD4_SSE_REGS" and "ALL_REGS".
29430 * spellcheck.c (test_data): Add , in between "foo" and "food".
29432 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29435 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
29436 boundary crossing check and subsequent code generation agree.
29438 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29440 * config/aarch64/aarch64.c (has_memory_op): Delete.
29441 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
29444 2017-02-13 Jakub Jelinek <jakub@redhat.com>
29446 PR rtl-optimization/79388
29447 PR rtl-optimization/79450
29448 * combine.c (distribute_notes): When removing TEM_INSN for which
29449 corresponding dest has last value recorded, invalidate that last
29452 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29454 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
29455 of explicit '@'. Add missing assembly comment marker on branch costs
29458 2017-02-13 Nathan Sidwell <nathan@acm.org>
29460 * gengtype-lex.l (<in_struct>): Add '/'.
29462 2017-02-13 Martin Liska <mliska@suse.cz>
29465 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
29467 2017-02-13 Richard Biener <rguenther@suse.de>
29469 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
29471 * configure: Re-generate.
29472 * config.in: Likewise.
29473 * graphite-dependences.c: Simplify as if
29474 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
29475 * graphite-isl-ast-to-gimple.c: Likewise.
29476 * graphite-optimize-isl.c: Likewise.
29477 * graphite-poly.c: Likewise.
29478 * graphite-sese-to-poly.c: Likewise.
29479 * graphite.h: Likewise.
29480 * toplev.c: Include isl/version.h and use isl_version () for
29481 printing the ISL version.
29482 * doc/install.texi: Update ISL requirement.
29484 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
29486 * doc/standards.texi (Standards): Update reference to
29489 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
29491 * doc/extend.texi (Named Address Spaces): sourceware.org now
29493 * doc/install.texi (Binaries): Ditto.
29496 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
29498 * doc/cpp.texi: Replace "stringify"/"stringification" with C
29499 standard terminology "stringize"/"stringizing" throughout.
29500 * doc/cppinternals.texi: Likewise.
29502 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
29504 * doc/extend.texi: Fix some spelling mistakes and typos.
29505 * doc/invoke.texi: Likewise.
29507 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
29510 * params.def (inline-min-speedup) Change from 10 to 8.
29512 2017-02-11 Jakub Jelinek <jakub@redhat.com>
29514 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
29517 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
29520 * ipa-inline-analysis.c (get_minimal_bb): New function.
29521 (record_modified): Use it.
29522 (remap_edge_change_prob): Handle also ancestor functions.
29524 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
29526 * doc/contrib.texi (Contributors): Remove broken link into
29527 the Mauve CVS repository.
29529 2017-02-11 Jakub Jelinek <jakub@redhat.com>
29531 PR middle-end/79454
29532 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
29533 result computation whenever lhs doesn't have vector mode, not
29534 just when it has BLKmode.
29536 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
29538 * doc/makefile.texi (profiledbootstrap): Refer to the
29539 installation instructions only in textual form.
29541 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29544 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
29546 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
29548 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
29549 (Specific): Update mingw-w64 reference.
29551 (Specific): Remove broken link to Renesas RX processor.
29553 2017-02-10 Richard Biener <rguenther@suse.de>
29555 * toplev.c (process_options): Do not mention obsolete graphite
29556 options when printing sorry message about missing graphite support.
29557 Mention -floop-nest-optimize.
29559 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
29561 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
29562 (vtst_p16): Likewise.
29563 (vtstq_p8): Likewise.
29564 (vtstq_p16): Likewise.
29566 (vtstq_p64): Likewise.
29567 * config/arm/arm_neon.h (vgetq_lane_p64): New.
29568 (vset_lane_p64): New.
29569 (vsetq_lane_p64): New.
29571 2017-02-10 Jakub Jelinek <jakub@redhat.com>
29573 PR tree-optimization/79411
29574 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
29575 stmt operands are SSA_NAMEs used in abnormal phis.
29576 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
29579 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
29582 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
29585 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
29587 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
29589 2017-02-09 Jakub Jelinek <jakub@redhat.com>
29591 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
29595 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
29596 not arbitrary TREE_CONSTANT.
29599 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
29600 "omp declare target link" attribute unless is_global_var.
29601 * omp-offload.c (find_link_var_op): Likewise.
29603 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
29604 Chung-Lin Tang <cltang@codesourcery.com>
29606 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
29608 (gimplify_adjust_omp_clauses): Don't delete TILE.
29609 (gimplify_omp_for): Deal with TILE.
29610 * internal-fn.c (expand_GOACC_TILE): New function.
29611 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
29613 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
29614 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
29616 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
29617 avoid DIV for outermost collapse var.
29618 (expand_oacc_for): Insert tile element loop as needed. Adjust.
29619 Remove out of date comments, fix whitespace.
29620 * omp-general.c (omp_extract_for_data): Deal with tiling.
29621 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
29622 adjust OLF_DIM_BASE value.
29623 (struct omp_for_data): Add tiling field.
29624 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
29625 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
29626 for auto loops. Remove default auto determining, moved to
29627 oacc_loop_fixed_partitions.
29628 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
29629 stmts, add e_mask field.
29630 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
29631 (oacc_thread_numbers): Use oacc_dim_call.
29632 (oacc_xform_tile): New.
29633 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
29634 (finish_oacc_loop): Adjust for ifns vector.
29635 (oacc_loop_discover_walk): Append loop abstraction sites to list,
29636 add case for GOACC_TILE fns.
29637 (oacc_loop_xform_loop): Delete.
29638 (oacc_loop_process): Iterate over call list directly, and add
29639 handling for GOACC_TILE fns.
29640 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
29642 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
29643 vector partitioning to outer loops. Assign 2 partitions to loops
29644 when available. Add TILE handling.
29645 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
29646 (execite_oacc_device_lower): Process GOACC_TILE fns,
29647 ignore unknown specs.
29648 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
29649 * tree.c (omp_clause_num_ops): Adjust TILE ops.
29650 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
29652 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
29654 * configure.ac (ACX_BUGURL): Update.
29655 * configure: Regenerate.
29657 2017-02-09 Richard Biener <rguenther@suse.de>
29659 PR tree-optimization/69823
29660 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
29661 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
29663 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
29665 * config/arc/arc-c.def: Add __NPS400__ definition.
29666 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
29667 (TARGET_NPS400): Define.
29669 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
29671 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
29673 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
29674 pointer, arch_info.
29675 (arc_cpu_types): Fill the arch_info field with a pointer into the
29676 arc_arch_types table.
29677 (arc_selected_cpu): Declare.
29678 * config/arc/arc.c (arc_selected_cpu): Make global.
29679 (arc_selected_arch): Delete.
29680 (arc_base_cpu): Delete.
29681 (arc_override_options): Remove references to deleted variables,
29682 update access to arch information.
29683 (ARC_OPT): Update access to arch information.
29684 (ARC_OPTX): Likewise.
29685 * config/arc/arc.h (arc_base_cpu): Remove declaration.
29686 (TARGET_ARC600): Update access to arch information.
29687 (TARGET_ARC601): Likewise.
29688 (TARGET_ARC700): Likewise.
29689 (TARGET_EM): Likewise.
29690 (TARGET_HS): Likewise.
29691 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
29694 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
29697 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
29698 condition/operands for integer GE/LE/GEU/LEU operations.
29700 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
29702 PR translation/79397
29703 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
29706 2017-02-08 Martin Jambor <mjambor@suse.cz>
29709 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
29710 whether allocation happened.
29711 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
29712 nothing was allocated.
29714 2017-02-08 Jakub Jelinek <jakub@redhat.com>
29716 PR tree-optimization/79408
29717 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
29718 constant, but SSA_NAME with a known integer range, use the minimum
29719 of that range instead of op1 to determine if modulo can be replaced
29720 with its first operand.
29722 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29724 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
29726 2017-02-08 Richard Biener <rguenther@suse.de>
29728 PR tree-optimization/71824
29729 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
29730 Check all loops contained in the merged region.
29732 2017-02-07 Andrew Pinski <apinski@cavium.com>
29734 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
29736 2017-02-07 Andrew Pinski <apinski@cavium.com>
29738 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
29739 (thunderxt88): Likewise.
29740 (thunderxt81): Disable LSE and change v8.1 to v8.
29741 (thunderxt83): Likewise.
29743 2017-02-07 Jakub Jelinek <jakub@redhat.com>
29744 Richard Biener <rguenther@suse.de>
29746 PR middle-end/79399
29747 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
29748 type from int to size_t.
29749 * ira-costs.c (struct_costs_size): Change type from int to size_t.
29751 2017-02-07 Jakub Jelinek <jakub@redhat.com>
29753 PR rtl-optimization/79386
29754 * cprop.c (bypass_conditional_jumps): Initialize
29755 bypass_last_basic_block already before splitting bbs after
29756 unconditional traps...
29757 (bypass_conditional_jumps): ... rather than here.
29760 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
29761 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
29762 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
29763 fix -masm=intel patterns.
29765 2017-02-07 Richard Biener <rguenther@suse.de>
29767 PR tree-optimization/79256
29768 PR middle-end/79278
29769 * builtins.c (get_object_alignment_2): Use min_align_of_type
29770 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
29771 and ADJUST_FIELD_ALIGN.
29773 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
29775 * doc/tm.texi: Regenerate.
29776 * stor-layout.c (layout_decl): Adjust.
29777 (update_alignment_for_field): Likewise.
29778 (place_field): Likewise.
29779 (min_align_of_type): Likewise.
29780 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
29781 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
29782 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
29783 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
29784 * config/frv/frv.c (frv_adjust_field_align): Likewise.
29785 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
29786 * config/i386/i386.c (x86_field_alignment): Likewise.
29787 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
29788 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
29789 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
29790 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
29791 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
29792 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
29796 2017-01-30 Richard Biener <rguenther@suse.de>
29798 PR tree-optimization/79256
29799 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
29800 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
29803 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
29805 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
29806 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
29807 builtins to SImode and emit a zero-extend, if necessary.
29809 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
29811 * docs/invoke.texi (RISC-V Options): Alphabetize.
29813 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
29815 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
29818 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
29820 * config/riscv/riscv.c: New file.
29821 * common/config/riscv/riscv-common.c: Likewise.
29822 * config.gcc: Likewise.
29823 * config/riscv/constraints.md: Likewise.
29824 * config/riscv/elf.h: Likewise.
29825 * config/riscv/generic.md: Likewise.
29826 * config/riscv/linux.h: Likewise.
29827 * config/riscv/multilib-generator: Likewise.
29828 * config/riscv/peephole.md: Likewise.
29829 * config/riscv/pic.md: Likewise.
29830 * config/riscv/predicates.md: Likewise.
29831 * config/riscv/riscv-builtins.c: Likewise.
29832 * config/riscv/riscv-c.c: Likewise.
29833 * config/riscv/riscv-ftypes.def: Likewise.
29834 * config/riscv/riscv-modes.def: Likewise.
29835 * config/riscv/riscv-opts.h: Likewise.
29836 * config/riscv/riscv-protos.h: Likewise.
29837 * config/riscv/riscv.h: Likewise.
29838 * config/riscv/riscv.md: Likewise.
29839 * config/riscv/riscv.opt: Likewise.
29840 * config/riscv/sync.md: Likewise.
29841 * config/riscv/t-elf-multilib: Likewise.
29842 * config/riscv/t-linux: Likewise.
29843 * config/riscv/t-linux-multilib: Likewise.
29844 * config/riscv/t-riscv: Likewise.
29845 * configure.ac: Likewise.
29846 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
29847 Waterman as RISC-V maintainers.
29848 * doc/install.texi: Add RISC-V entries.
29849 * doc/invoke.texi: Add RISC-V options section.
29850 * doc/md.texi: Add RISC-V constraints section.
29851 * configure: Regenerated.
29853 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
29856 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
29857 false values to be constant vectors with all 0 or all 1 bits set.
29858 (vcondu<mode><mode>): Likewise.
29859 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
29861 (fpmask_comparison_operator): Update comment.
29862 (vecint_comparison_operator): New predicate.
29863 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
29864 vector conditionals when the true and false values are constant
29865 vectors with all 0 bits or all 1 bits set.
29867 2017-02-06 Martin Sebor <msebor@redhat.com>
29869 PR tree-optimization/79376
29870 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
29872 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
29874 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
29875 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
29876 to simplify split condition.
29878 2017-02-06 Jakub Jelinek <jakub@redhat.com>
29880 * omp-expand.c (oxpand_omp_atomic_fetch_op,
29881 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
29884 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
29886 PR rtl-optimization/68664
29887 * target.def (can_speculate_insn): New hook.
29888 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
29889 * doc/tm.texi: Regenerate.
29890 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
29891 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
29892 (rs6000_sched_can_speculate_insn): New function.
29894 2017-02-06 Jakub Jelinek <jakub@redhat.com>
29896 PR tree-optimization/79284
29897 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
29898 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
29899 vectorizable_mask_load_store, vectorizable_operation,
29900 vect_is_simple_cond, get_same_sized_vectype): Use it instead
29901 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
29902 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
29903 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
29904 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
29905 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
29906 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
29907 is_gimple_assign (stmt). Replace another such test with
29908 is_gimple_assign (stmt).
29910 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
29913 * config/avr/avr.c (rtl-iter.h): Include it.
29914 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
29915 (avr_legitimate_combined_insn): ...and implementation.
29917 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
29919 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
29920 * config/s390/s390.c (s390_const_operand_ok)
29921 (s390_canonicalize_comparison, s390_extract_part)
29922 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
29923 (s390_contiguous_bitmask_p, s390_rtx_costs)
29924 (legitimize_pic_address): Likewise.
29925 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
29926 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
29927 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
29928 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
29929 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
29931 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
29933 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
29934 REGNO($0) == REGNO($1).
29936 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29938 * config/s390/linux.h(SIZE_TYPE): Add comment.
29940 2017-02-06 Julian Brown <julian@codesourcery.com>
29941 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29942 Virendra Pathak <virendra.pathak@broadcom.com>
29944 * config/aarch64/aarch64-cores.def: Change the scheduler
29946 * config/aarch64/aarch64.md: Include thunderx2t99.md.
29947 * config/aarch64/thunderx2t99.md: New file.
29949 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
29951 * doc/standards.texi (Go Language): Update link to language
29954 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
29956 * tree-eh.c (lower_resx): Sanitize profile.
29957 (cleanup_empty_eh_move_lp): Likewise.
29959 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
29962 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
29964 * cfgloopmanip.h (loop_version): Update prototype.
29965 * modulo-sched.c (sms_schedule): Update call of loop_version.
29966 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
29967 * tree-parloops.c (gen_parallel_loop): Likewise.
29968 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
29969 * tree-ssa-loop-split.c (split_loop): Likewise.
29970 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
29971 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
29973 2017-02-05 Martin Liska <mliska@suse.cz>
29976 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
29978 (print_operand_address): Initialize a struct to zero.
29980 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
29982 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
29983 garbage collector only in textual form.
29985 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
29987 * doc/extend.texi (x86 specific memory model extensions for
29988 transactional memory): Simplify a phrase.
29990 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
29993 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
29994 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
29995 (atomic_storedi_1): Likewise.
29997 2017-02-04 Jakub Jelinek <jakub@redhat.com>
29999 PR tree-optimization/79338
30000 * tree-parloops.c (gather_scalar_reductions): Don't call
30001 vect_analyze_loop_form for loop->inner before destroying loop's
30004 2017-02-03 Martin Sebor <msebor@redhat.com>
30006 PR tree-optimization/79327
30007 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
30008 when precision has resulted in leading zeros.
30009 (format_integer): Adjust the likely counter to assume an unknown
30010 argument that may be zero is non-zero.
30012 2017-02-03 Jason Merrill <jason@redhat.com>
30015 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
30016 avoid copying non-taken branch.
30018 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30020 PR tree-optimization/79340
30021 * tree-vect-loop.c (vectorizable_reduction): Release
30022 vec_defs elements after safe_splicing them into other vectors.
30025 PR tree-optimization/79327
30026 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
30027 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
30029 (format_integer): Use wide_int_to_tree instead of build_int_cst
30030 + to_?hwi. If argmin is NULL, just set argmin and argmax to
30031 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
30032 of shortest and longest sequence.
30034 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
30036 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
30037 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
30039 2017-02-03 Walter Lee <walt@tilera.com>
30042 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
30043 after initial stackframe link reg save.
30044 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
30046 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30049 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
30050 wu for stxssp alternative.
30052 2017-02-03 Martin Sebor <msebor@redhat.com>
30054 PR tree-optimization/79352
30055 * gimple-fold.c (get_range_strlen): Add argument.
30056 (get_range_strlen): Change return type to bool.
30057 (get_maxval_strlen): Pass in a dummy argument.
30058 * gimple-fold.h (get_range_strlen): Change return type to bool.
30059 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
30060 * tree.h (array_at_struct_end_p): Add argument.
30061 * tree.c (array_at_struct_end_p): Handle it.
30063 2017-02-03 Martin Liska <mliska@suse.cz>
30066 * multiple_target.c (create_dispatcher_calls): Redirect edge
30067 from a caller of a dispatcher.
30068 (expand_target_clones): Make the clones local.
30069 (ipa_target_clone): Do both target clones and resolvers.
30070 (ipa_dispatcher_calls): Remove the pass.
30071 (pass_dispatcher_calls::gate): Likewise.
30072 (make_pass_dispatcher_calls): Likewise.
30073 * passes.def (pass_target_clone): Put as very first IPA early
30076 2017-02-03 Martin Liska <mliska@suse.cz>
30078 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
30079 in case of a function with ifunc attribute.
30081 2017-02-03 Martin Liska <mliska@suse.cz>
30083 * cgraph.c (cgraph_node::dump): Dump function version info.
30084 * symtab.c (symtab_node::dump_base): Add missing new line.
30086 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30088 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
30089 (ifcombine_ifandif): Use it.
30091 2017-02-03 Martin Liska <mliska@suse.cz>
30093 * doc/invoke.texi: Document default value for
30094 use-after-scope-direct-emission-threshold.
30096 2017-02-03 Martin Liska <mliska@suse.cz>
30098 PR tree-optimization/79339
30099 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
30100 (format_floating): Likewise.
30102 2017-02-03 Martin Liska <mliska@suse.cz>
30105 * ipa-prop.c (ipa_node_params_t::insert): Remove current
30107 (ipa_node_params_t::remove): Likewise.
30108 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
30109 initialization from removed ipa_node_params_t::insert.
30110 (ipa_node_params::~ipa_node_params): Move from removed
30111 ipa_node_params_t::release.
30112 * symbol-summary.h (symbol_summary::m_released): New member.
30113 Do not release a summary twice. Do not allow to call finalizer
30114 for types of a summary that live in GGC memory.
30116 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30118 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
30121 2017-02-02 Martin Sebor <msebor@redhat.com>
30123 PR middle-end/79275
30124 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
30125 (format_string): Tighten up the range of output for non-constant
30126 strings and correct the expected range for wide non-constant strings.
30128 2017-02-02 Martin Sebor <msebor@redhat.com>
30130 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
30132 PR middle-end/32003
30133 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
30135 (-fdump-tree-@var): Add to index and document how to come up
30136 with pass-specific option and dump file names.
30137 (-fdump-passes): Clarify where to look for output.
30139 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30141 PR middle-end/77445
30142 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
30143 statistics of the analyzed path; allow threading for speed when
30144 any of BBs along the path are optimized for speed.
30146 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
30148 PR middle-end/78468
30149 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
30150 settings of the virtual registers.
30153 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
30155 * explow.c (get_dynamic_stack_size): Take known alignment of stack
30156 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
30159 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30161 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
30162 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
30164 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30166 * config/s390/s390.md: Add missing comments with the expanded
30168 * config/s390/vector.md: Likewise.
30169 * config/s390/vx-builtins.md: Likewise.
30171 2017-02-02 Jakub Jelinek <jakub@redhat.com>
30174 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
30175 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
30176 conditions on a single line.
30178 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30180 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
30181 __S390_VX__ to __VX__.
30183 2017-02-01 Andrew Pinski <apinski@cavium.com>
30185 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
30186 stmt_info to record_stmt_cost.
30187 (vect_get_known_peeling_cost): Pass stmt_info if known to
30189 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
30190 cpu_vector_cost field into
30191 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
30192 field into vec_int_stmt_cost and vec_fp_stmt_cost.
30193 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
30194 splitting of scalar_stmt_cost and vec_stmt_cost.
30195 (thunderx_vector_cost): Likewise.
30196 (cortexa57_vector_cost): LIkewise.
30197 (exynosm1_vector_cost): Likewise.
30198 (xgene1_vector_cost): Likewise.
30199 (thunderx2t99_vector_cost): Improve after the splitting of the two
30201 (aarch64_builtin_vectorization_cost): Update for the splitting of
30202 scalar_stmt_cost and vec_stmt_cost.
30204 2017-02-01 Torvald Riegel <triegel@redhat.com>
30205 Richard Henderson <rth@redhat.com>
30207 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
30208 conditional on existance of a fast atomic load.
30209 * optabs-query.c (can_atomic_load_p): New function.
30210 * optabs-query.h (can_atomic_load_p): Declare it.
30211 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
30212 no fast atomic load is available for the particular size of access.
30213 (expand_atomic_compare_and_swap): Likewise.
30214 (expand_atomic_load): Likewise.
30215 (expand_atomic_store): Likewise.
30216 (expand_atomic_fetch_op): Likewise.
30217 * testsuite/lib/target-supports.exp
30218 (check_effective_target_sync_int_128): Remove x86 because it provides
30219 no fast atomic load.
30220 (check_effective_target_sync_int_128_runtime): Likewise.
30222 2017-02-01 Richard Biener <rguenther@suse.de>
30224 * graphite.c: Include tree-vectorizer.h for find_loop_location.
30225 (graphite_transform_loops): Provide opt-info for optimized nests.
30226 * tree-parloop.c (parallelize_loops): Provide opt-info for
30227 parallelized loops.
30229 2017-02-01 Richard Biener <rguenther@suse.de>
30231 PR middle-end/79315
30232 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
30233 was not set before.
30235 2017-02-01 Richard Biener <rguenther@suse.de>
30237 PR tree-optimization/71824
30238 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
30239 Verify the loops are valid in the merged SESE region.
30240 (scop_detection::can_represent_loop_1): Check analyzing the
30241 evolution of the number of iterations in the region succeeds.
30243 2017-01-31 Ian Lance Taylor <iant@golang.org>
30245 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
30246 REG_ARGS_SIZE note to 32-bit push insns and call insn.
30248 2017-01-31 David Malcolm <dmalcolm@redhat.com>
30250 PR preprocessor/79210
30251 * input.c (get_substring_ranges_for_loc): Replace line_width
30252 assertion with error-handling.
30254 2017-01-31 Richard Biener <rguenther@suse.de>
30256 PR tree-optimization/77318
30257 * graphite-sese-to-poly.c (extract_affine): Fix assert.
30258 (create_pw_aff_from_tree): Take loop parameter.
30259 (add_condition_to_pbb): Pass loop of the condition to
30260 create_pw_aff_from_tree.
30262 2017-01-31 Jakub Jelinek <jakub@redhat.com>
30264 * config/s390/s390.c (s390_asan_shadow_offset): New function.
30265 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
30267 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
30271 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
30273 (convert_int_to_float128): Likewise.
30274 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
30275 (convert_int_to_float128): Likewise.
30276 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
30277 (UNSPEC_IEEE128_CONVERT): Likewise.
30278 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
30279 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
30280 Use local variables for IBM extended format.
30281 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
30282 (fix_trunc<mode>si2_fprs): Likewise.
30283 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
30284 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
30285 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
30286 to know that we can now have integers of all sizes in vector
30288 (fix<uns>_<mode>di2_hw): Likewise.
30289 (float<uns>_<mode>si2_hw): Likewise.
30290 (fix_<mode>si2_hw): Likewise.
30291 (fixuns_<mode>si2_hw): Likewise.
30292 (float<uns>_<mode>di2_hw): Likewise.
30293 (float_<mode>di2_hw): Likewise.
30294 (float_<mode>si2_hw): Likewise.
30295 (floatuns_<mode>di2_hw): Likewise.
30296 (floatuns_<mode>si2_hw): Likewise.
30297 (xscvqp<su>wz_<mode>): Delete, no longer used.
30298 (xscvqp<su>dz_<mode>): Likewise.
30299 (xscv<su>dqp_<mode>): Likewise.
30300 (ieee128_mfvsrd_64bit): Likewise.
30301 (ieee128_mfvsrd_32bit): Likewise.
30302 (ieee128_mfvsrwz): Likewise.
30303 (ieee128_mtvsrw): Likewise.
30304 (ieee128_mtvsrd_64bit): Likewise.
30305 (ieee128_mtvsrd_32bit): Likewise.
30307 2017-01-31 Martin Liska <mliska@suse.cz>
30310 * ipa-prop.c (ipa_free_all_node_params): Call release method
30311 instead of ~sumbol_summary to not to trigger double times
30314 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
30316 PR tree-optimization/71691
30317 * bitmap.h (class auto_bitmap): New.
30318 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
30319 is_maybe_undefined instead of ssa_undefined_value_p.
30321 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30323 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
30324 __S390_ARCH_LEVEL__ to __ARCH__.
30326 2017-01-31 Jakub Jelinek <jakub@redhat.com>
30328 PR tree-optimization/79267
30329 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
30330 if should_remove_lhs_p is true.
30332 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
30335 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
30336 (add_alignment_attribute): New.
30337 (base_type_die): Add alignment attribute.
30338 (subrange_type_die): Likewise.
30339 (modified_type_die): Likewise.
30340 (gen_array_type_die): Likewise.
30341 (gen_descr_array_type_die: Likewise.
30342 (gen_enumeration_type_die): Likewise.
30343 (gen_subprogram_die): Likewise.
30344 (gen_variable_die): Likewise.
30345 (gen_field_die): Likewise.
30346 (gen_ptr_to_mbr_type_die): Likewise.
30347 (gen_struct_or_union_type_die): Likewise.
30348 (gen_subroutine_type_die): Likewise.
30349 (gen_typedef_die): Likewise.
30350 (base_type_cmp): Compare alignment attribute.
30352 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30355 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
30356 (setb_unsigned) New pattern for setb with CCUNS.
30357 * config/rs6000/rs6000.c (expand_block_compare): Use a different
30358 subfc./subfe sequence to avoid overflow problems. Generate a
30359 shorter sequence with cmpld/setb for power9.
30360 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
30361 for generating subfc. instruction.
30362 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
30363 now uses this instruction.
30365 2017-01-30 Ian Lance Taylor <iant@google.com>
30368 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
30369 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
30371 2017-01-30 Martin Sebor <msebor@redhat.com>
30373 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
30374 Move constant to the right of a relational operator.
30375 (get_mpfr_format_length, format_character, format_string): Ditto.
30376 (should_warn_p, maybe_warn): Same.
30378 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
30380 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
30383 * asan.c (get_translation_unit_decl): Remove function.
30384 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
30386 2017-01-30 Martin Liska <mliska@suse.cz>
30388 PR gcov-profile/79259
30389 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
30390 -fprofile-generate.
30392 2017-01-30 Martin Liska <mliska@suse.cz>
30395 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
30396 Initialize variables with NULL value.
30398 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
30401 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
30403 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
30405 2017-01-30 Richard Biener <rguenther@suse.de>
30407 PR tree-optimization/79276
30408 * tree-vrp.c (process_assert_insertions): Properly adjust common
30409 when removing a duplicate.
30411 2017-01-30 Richard Biener <rguenther@suse.de>
30413 PR tree-optimization/79256
30414 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
30415 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
30417 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
30419 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
30422 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
30423 ("*r<noxa>sbg_<mode>_sll_bitmask")
30424 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
30425 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
30426 Use contiguous_bitmask_nowrap_operand.
30428 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30431 * config/rs6000/altivec.h (vec_xl): Revise #define.
30432 (vec_xst): Likewise.
30434 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
30436 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
30438 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
30440 PR rtl-optimization/79194
30441 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
30442 traps before call to bypass_conditional_jumps.
30444 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
30446 PR tree-optimization/71374
30447 * lra-constraints.c (check_conflict_input_operands): New.
30448 (match_reload): Use it.
30450 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
30453 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
30454 account to calculate conflict_set.
30456 2017-01-27 Bin Cheng <bin.cheng@arm.com>
30458 PR rtl-optimization/78559
30459 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
30460 other_insn in combine.
30462 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
30464 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
30465 uint16_type_node for BT_UINT16.
30467 2017-01-27 David Malcolm <dmalcolm@redhat.com>
30469 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
30470 "RTL Tests" to menu.
30471 (GIMPLE Tests): New node.
30472 (RTL Tests): New node.
30474 2017-01-27 Richard Biener <rguenther@suse.de>
30476 PR tree-optimization/79245
30477 * tree-loop-distribution.c (distribute_loop): Apply cost
30478 modeling also to detected patterns.
30480 2017-01-27 Richard Biener <rguenther@suse.de>
30482 PR tree-optimization/71433
30483 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
30484 (compare_assert_loc): New function.
30485 (process_assert_insertions): Sort and optimize assert locations
30486 to remove duplicates and push down identical assertions on
30487 edges to their destination block.
30489 2017-01-27 Richard Biener <rguenther@suse.de>
30491 PR tree-optimization/79244
30492 * tree-vrp.c (remove_range_assertions): Forcefully propagate
30493 out SSA names even if abnormal.
30495 2017-01-27 Jakub Jelinek <jakub@redhat.com>
30497 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
30498 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
30499 instead of MPFR_RNDN.
30501 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
30504 * arm.c (arm_option_override): Don't call build_target_option_node
30505 until after doing all option overrides.
30506 (arm_valid_target_attribute_tree): Likewise.
30508 2017-01-27 Martin Liska <mliska@suse.cz>
30510 * doc/invoke.texi (-fprofile-arcs): Document profiling support
30511 for {cd}tors and C++ {cd}tors.
30513 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
30515 * config/s390/s390.md ("*setmem_long_and")
30516 ("*setmem_long_and_31z"): Use zero_extend instead of and.
30518 2017-01-26 Martin Sebor <msebor@redhat.com>
30520 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
30523 2017-01-26 Martin Sebor <msebor@redhat.com>
30525 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
30526 HAVE_DFmode before using XFmode or DFmode.
30527 (parse_directive): Avoid using the z length modifier to avoid
30528 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
30530 PR middle-end/78703
30531 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
30532 to accept adjustment as an array.
30533 (get_int_range): New function.
30534 (struct directive): Make width and prec arrays.
30535 (directive::set_width, directive::set_precision): Call get_int_range.
30536 (format_integer, format_floating): Handle width and precision ranges.
30537 (format_string, parse_directive): Same.
30539 2017-01-26 Jakub Jelinek <jakub@redhat.com>
30542 * dwarf2out.c (generate_skeleton_bottom_up): For children with
30543 comdat_type_p set, just clone them, but keep the children in the
30547 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
30548 which have direct callers with -fvar-tracking-assignments enabled
30550 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
30551 inside of type units.
30553 2017-01-26 Martin Sebor <msebor@redhat.com>
30555 PR middle-end/78703
30556 * gimple-ssa-sprintf.c (struct result_range): Add likely and
30558 (struct format_result): Replace number_chars, number_chars_min,
30559 and number_chars_max with a single member of struct result_range.
30561 (format_result::operator+=): Adjust.
30562 (struct fmtresult): Remove bounded. Handle likely and unlikely
30564 (fmtresult::adjust_for_width_or_precision): New function.
30565 (fmtresult:type_max_digits): New function.
30566 (bytes_remaining): Handle likely and unlikely counters.
30567 (min_bytes_remaining): Remove.
30568 (format_percent): Simplify.
30569 (format_integer, format_floating): Set likely and unlikely counters.
30570 (get_string_length, format_character, format_string): Same.
30571 (format_plain, should_warn_p): New function.
30572 (maybe_warn): Call should_warn_p. Update diagnostic messages
30573 and handle those for all directives, including plain strings.
30574 (format_directive): Handle likely and unlikely counters.
30575 Remove unnecessary quoting from diagnostics. Add an informational
30577 (add_bytes): Remove.
30578 (pass_sprintf_length::compute_format_length): Simplify.
30579 (try_substitute_return_value): Handle likely and unlikely counters.
30581 2017-01-26 Carl Love <cel@us.ibm.com>
30583 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
30584 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
30586 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
30589 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
30590 endianess for subregs into account.
30591 * lra-constraints.c (lra_constraints): Do risky transformations
30592 always on the first iteration.
30593 * lra-lives.c (check_pseudos_live_through_calls): Add arg
30594 last_call_used_reg_set.
30595 (process_bb_lives): Define and use last_call_used_reg_set.
30596 * lra.c (lra): Always continue after lra_constraints on the first
30599 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
30601 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
30603 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
30605 2017-01-26 Jakub Jelinek <jakub@redhat.com>
30607 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
30608 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
30609 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
30610 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
30611 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
30612 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
30613 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
30614 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
30615 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
30617 2017-01-26 Marek Polacek <polacek@redhat.com>
30620 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
30621 for the third operand.
30623 2017-01-26 Jakub Jelinek <jakub@redhat.com>
30625 PR middle-end/79236
30626 * omp-low.c (struct omp_context): Add simt_stmt field.
30627 (scan_omp_for): Return omp_context *.
30628 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
30629 context to the _simt_ SIMD stmt.
30630 (lower_omp_for): For combined SIMD with sibling _simt_
30631 SIMD, make sure to use the same decls in _looptemp_
30632 clauses as in the sibling.
30634 2017-01-26 David Sherwood <david.sherwood@arm.com>
30636 PR middle-end/79212
30637 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
30640 2017-01-26 Jakub Jelinek <jakub@redhat.com>
30643 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
30644 emit fld b; fld a; if possible.
30646 * brig-builtins.def: Update copyright years.
30647 * config/arm/arm_acle_builtins.def: Update copyright years.
30649 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
30652 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
30653 constraint instead of o for the stxsd instruction.
30655 2017-01-25 Carl Love <cel@us.ibm.com>
30657 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
30658 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
30660 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
30662 * doc/invoke.texi (C++ Dialect Options): Fix typo.
30664 2017-01-25 Richard Biener <rguenther@suse.de>
30666 PR tree-optimization/69264
30667 * target.def (vector_alignment_reachable): Improve documentation.
30668 * doc/tm.texi: Regenerate.
30669 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
30671 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
30672 earlier changes with respect to TYPE_USER_ALIGN.
30673 (vector_alignment_reachable_p): Likewise. Improve dumping.
30675 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30678 * config/arm/arm.md (xordi3): Force constant operand into a register
30681 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30683 * doc/invoke.texi (-fstore-merging): Correct default optimization
30684 levels at which it is enabled.
30685 (-O): Move -fstore-merging from list to...
30688 2017-01-25 Richard Biener <rguenther@suse.de>
30691 * omp-expand.c: Include debug.h.
30692 (expand_omp_taskreg): Make sure to generate early debug before
30693 outlining anything from a function.
30694 (expand_omp_target): Likewise.
30695 (grid_expand_target_grid_body): Likewise.
30697 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
30700 * asan.c (get_translation_unit_decl): New function.
30701 (asan_add_global): Extract modules file name from globals
30702 TRANSLATION_UNIT_DECL name.
30704 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
30707 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
30708 for long calls with APCS frame and VFP.
30710 2017-01-24 David Malcolm <dmalcolm@redhat.com>
30712 * cfg.c (original_copy_tables_initialized_p): New function.
30713 * cfg.h (original_copy_tables_initialized_p): New decl.
30714 * cfgrtl.c (relink_block_chain): Guard the call to
30715 free_original_copy_tables with a call to
30716 original_copy_tables_initialized_p.
30717 * cgraph.h (symtab_node::native_rtl_p): New decl.
30718 * cgraphunit.c (symtab_node::native_rtl_p): New function.
30719 (symtab_node::needed_p): Don't assert for early assembly output
30720 for __RTL functions.
30721 (cgraph_node::finalize_function): Set "force_output" for __RTL
30723 (cgraph_node::analyze): Bail out early for __RTL functions.
30724 (analyze_functions): Update assertion to support __RTL functions.
30725 (cgraph_node::expand): Bail out early for __RTL functions.
30726 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
30728 * function.h (struct function): Update comment for field
30730 * gimple-expr.c: Include "tree-pass.h".
30731 (gimple_has_body_p): Return false for __RTL functions.
30732 * Makefile.in (OBJS): Add run-rtl-passes.o.
30733 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
30735 (gcc::pass_manager::get_clean_slate): New accessor.
30736 * passes.c: Include "insn-addr.h".
30737 (should_skip_pass_p): Add logging. Update logic for running
30738 "expand" to be compatible with both __GIMPLE and __RTL. Guard
30739 property-provider override so it is only done for gimple passes.
30741 (skip_pass): New function.
30742 (execute_one_pass): Call skip_pass when skipping passes.
30743 * read-md.c (md_reader::read_char): Support filtering
30744 the input to a subset of line numbers.
30745 (md_reader::md_reader): Initialize fields
30746 m_first_line and m_last_line.
30747 (md_reader::read_file_fragment): New function.
30748 * read-md.h (md_reader::read_file_fragment): New decl.
30749 (md_reader::m_first_line): New field.
30750 (md_reader::m_last_line): New field.
30751 * read-rtl-function.c (function_reader::create_function): Only
30752 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
30753 curr_properties. Set DECL_INITIAL to a dummy block.
30754 (read_rtl_function_body_from_file_range): New function.
30755 * read-rtl-function.h (read_rtl_function_body_from_file_range):
30757 * run-rtl-passes.c: New file.
30758 * run-rtl-passes.h: New file.
30760 2017-01-24 Jeff Law <law@redhat.com>
30762 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
30765 2017-01-24 Bin Cheng <bin.cheng@arm.com>
30767 PR tree-optimization/79159
30768 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
30769 (record_nonwrapping_iv): Improve boundary using above function if no
30770 value range information.
30772 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
30773 Martin Jambor <mjambor@suse.cz>
30775 * brig-builtins.def: New file.
30776 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
30777 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
30778 (DEF_HSAIL_SAT_BUILTIN): Likewise.
30779 (DEF_HSAIL_INTR_BUILTIN): Likewise.
30780 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
30781 * builtin-types.def (BT_INT8): New.
30782 (BT_INT16): Likewise.
30783 (BT_UINT8): Likewise.
30784 (BT_UINT16): Likewise.
30785 (BT_FN_ULONG): Likewise.
30786 (BT_FN_UINT_INT): Likewise.
30787 (BT_FN_UINT_ULONG): Likewise.
30788 (BT_FN_UINT_LONG): Likewise.
30789 (BT_FN_UINT_PTR): Likewise.
30790 (BT_FN_ULONG_PTR): Likewise.
30791 (BT_FN_INT8_FLOAT): Likewise.
30792 (BT_FN_INT16_FLOAT): Likewise.
30793 (BT_FN_UINT32_FLOAT): Likewise.
30794 (BT_FN_UINT16_FLOAT): Likewise.
30795 (BT_FN_UINT8_FLOAT): Likewise.
30796 (BT_FN_UINT64_FLOAT): Likewise.
30797 (BT_FN_UINT16_UINT32): Likewise.
30798 (BT_FN_UINT32_UINT16): Likewise.
30799 (BT_FN_UINT16_UINT16_UINT16): Likewise.
30800 (BT_FN_INT_PTR_INT): Likewise.
30801 (BT_FN_UINT_PTR_UINT): Likewise.
30802 (BT_FN_LONG_PTR_LONG): Likewise.
30803 (BT_FN_ULONG_PTR_ULONG): Likewise.
30804 (BT_FN_VOID_UINT64_UINT64): Likewise.
30805 (BT_FN_UINT8_UINT8_UINT8): Likewise.
30806 (BT_FN_INT8_INT8_INT8): Likewise.
30807 (BT_FN_INT16_INT16_INT16): Likewise.
30808 (BT_FN_INT_INT_INT): Likewise.
30809 (BT_FN_UINT_FLOAT_UINT): Likewise.
30810 (BT_FN_FLOAT_UINT_UINT): Likewise.
30811 (BT_FN_ULONG_UINT_UINT): Likewise.
30812 (BT_FN_ULONG_UINT_PTR): Likewise.
30813 (BT_FN_ULONG_ULONG_ULONG): Likewise.
30814 (BT_FN_UINT_UINT_UINT): Likewise.
30815 (BT_FN_VOID_UINT_PTR): Likewise.
30816 (BT_FN_UINT_UINT_PTR: Likewise.
30817 (BT_FN_UINT32_UINT64_PTR): Likewise.
30818 (BT_FN_INT_INT_UINT_UINT): Likewise.
30819 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
30820 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
30821 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
30822 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
30823 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
30824 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
30825 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
30826 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
30827 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
30828 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
30829 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
30830 * doc/frontends.texi: List BRIG FE.
30831 * doc/install.texi (Testing): Add BRIG tesring requirements.
30832 * doc/invoke.texi (Overall Options): Mention BRIG.
30833 * doc/standards.texi (Standards): Doucment BRIG HSA version.
30835 2017-01-24 Richard Biener <rguenther@suse.de>
30837 PR translation/79208
30838 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
30840 2017-01-24 Martin Jambor <mjambor@suse.cz>
30843 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
30844 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
30845 and known_contexts.
30847 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
30849 PR middle-end/79123
30850 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
30851 casts from signed to unsigned really don't have a range.
30853 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
30855 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
30856 GMP_RNDx for compatiblity.
30858 2017-01-24 Martin Liska <mliska@suse.cz>
30861 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
30862 that would prevent us to call alloca with -1 as argument.
30864 2017-01-24 Jakub Jelinek <jakub@redhat.com>
30866 * dwarf2out.c (output_compilation_unit_header, output_file_names):
30867 Avoid -Wformat-security warning.
30869 2017-01-23 Andrew Pinski <apinski@cavium.com>
30871 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
30874 2017-01-23 Martin Sebor <msebor@redhat.com>
30876 PR middle-end/78703
30877 * gimple-ssa-sprintf.c (warn_level): New global.
30878 (format_integer): Use it here and throughout the rest of the file.
30879 Use the same switch to compute sign as base.
30880 (maybe_warn): New function.
30881 (format_directive): Factor out warnings into maybe_warn.
30882 Add debugging output. Use warn_level.
30883 (add_bytes): Use warn_level.
30884 (pass_sprintf_length::compute_format_length): Add debugging output.
30885 (try_substitute_return_value): Same.
30886 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
30888 PR middle-end/78703
30889 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
30890 (struct fmtresult, format_integer, format_floating): Adjust.
30891 (fmtresult::fmtresult): Set max correctly in two argument ctor.
30892 (get_string_length, format_string,format_directive): Same.
30893 (pass_sprintf_length::compute_format_length): Same.
30894 (try_substitute_return_value): Simplify slightly.
30896 PR middle-end/78703
30897 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
30898 (fmtresult::operator+=): Outlined.
30899 (struct fmtresult): Add ctors.
30900 (struct conversion_spec): Rename...
30901 (struct directive): ...to this. Add and remove data members.
30902 (directive::set_width, directive::set_precision): New functions.
30903 (format_percent): Use fmtresult ctor.
30904 (get_width_and_precision): Remove.
30905 (format_integer): Make naming changes. Avoid computing width and
30907 (format_floating): Same. Adjust indentation.
30908 (format_character, format_none): New functions.
30909 (format_string): Moved character handling to format_character.
30910 (format_directive): Remove arguments, change return type.
30911 (parse_directive): New function.
30912 (pass_sprintf_length::compute_format_length): Move directive
30913 parsing to parse_directive.
30915 2017-01-23 Jakub Jelinek <jakub@redhat.com>
30917 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
30918 (assign_assembler_name_if_needed): ... this.
30919 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
30920 (assign_assembler_name_if_needed): ... this.
30921 (free_lang_data_in_cgraph): Adjust callers.
30922 * cgraphunit.c (cgraph_node::analyze): Likewise.
30923 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
30926 2017-01-23 Richard Biener <rguenther@suse.de>
30928 PR tree-optimization/79088
30929 PR tree-optimization/79188
30930 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
30931 resetting loop bounds after last path deletion. Reset loop
30932 bounds of the target loop, make code match the comments.
30933 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
30934 Make sure loops need no fixups.
30936 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
30938 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
30939 exponent support with double type for first argument.
30940 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
30941 type returned by __builtin_vec_extract_sig,
30942 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
30943 functions from "vector int" to "vector unsigned int" or from
30944 "vector long long int" to "vector unsigned long long int".
30945 Changed type returned by __builtin_vec_extract_exp,
30946 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
30947 functions from "vector int" to "vector unsigned int" or from
30948 "vector long long int" to "vector unsigned long long int".
30949 Changed return type of __builtin_vec_test_data_class,
30950 __builtin_vec_test_data_class_sp, and
30951 __builtin_vec_test_data_class_dp from "vector int" to
30952 "vector bool int" or from "vector long long int" to "vector bool
30953 long long int" and changed second argument type from "unsigned
30954 int" to "int". Added new overloaded function forms "vector float
30955 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
30956 "vector float __builtin_vec_insert_exp_sp (vector float, vector
30957 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
30958 double, vector unsigned long long int)" and "vector double
30959 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
30960 long int)". Changed return type of
30961 __builtin_scalar_test_data_class and
30962 __builtin_scalar_test_data_class_sp and
30963 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
30964 int" and changed second argument from "unsigned int" to "int".
30965 Changed type returned by __builtin_scalar_test_neg,
30966 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
30967 from "int" to "bool int". Added new overloaded function form
30968 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
30969 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
30970 exponent double-precision with floating point first argument.
30971 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
30972 documentation of scalar_test_data_class, scalar_test_neg,
30973 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
30974 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
30975 vec_test_data_class built-in functions to reflect refinements in
30976 their type signatures.
30978 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
30980 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
30982 (aarch64_elf_asm_destructor): Likewise.
30984 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
30986 PR rtl-optimization/78634
30987 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
30988 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
30989 * ifcvt.c (noce_try_cmove): Add missing cost check.
30991 PR rtl-optimization/71724
30992 * combine.c (if_then_else_cond): Look for situations where it is
30993 beneficial to undo the work of one of the recursive calls.
30995 2017-01-23 Bin Cheng <bin.cheng@arm.com>
30997 PR tree-optimization/70754
30998 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
30999 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
31000 combined stmt before it if not NULL.
31001 (combine_chains): Process refs reversely and compute dominance point
31004 2017-01-23 Martin Liska <mliska@suse.cz>
31006 PR tree-optimization/79196
31007 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
31008 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
31010 (strlen_optimize_stmt): Call the renamed function.
31012 2017-01-23 Michael Matz <matz@suse.de>
31014 PR tree-optimization/78384
31015 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
31017 2017-01-23 Richard Biener <rguenther@suse.de>
31019 PR tree-optimization/79186
31020 * tree-vrp.c (register_new_assert_for): Make sure we've seen
31021 both incoming edges before moving an assert.
31023 2017-01-23 Martin Jambor <mjambor@suse.cz>
31025 * ipa-prop.c (load_from_param_1): Removed.
31026 (load_from_unmodified_param): Bits from load_from_param_1 put back
31028 (load_from_param): Removed.
31029 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
31030 with stmt. Reverted back to use of load_from_unmodified_param.
31032 2017-01-23 Martin Jambor <mjambor@suse.cz>
31035 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
31036 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
31037 field a pointer to garbage collected vector, mark lattices and
31038 ipcp_orig_node with GTY((skip)).
31039 (ipa_get_param_count): Adjust to descriptors being a pointer.
31040 (ipa_get_param): Likewise.
31041 (ipa_get_type): Likewise.
31042 (ipa_get_param_move_cost): Likewise.
31043 (ipa_set_param_used): Likewise.
31044 (ipa_get_controlled_uses): Likewise.
31045 (ipa_set_controlled_uses): Likewise.
31046 (ipa_is_param_used): Likewise.
31047 (ipa_node_params_t): Move into garbage collector. New methods insert
31049 (ipa_node_params_sum): Annotate wth GTY(()).
31050 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
31052 (ipa_load_from_parm_agg): Adjust declaration.
31053 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
31054 * ipa-profile.c (ipa_profile): Likewise.
31055 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
31056 (ipa_populate_param_decls): Make descriptors parameter garbage
31058 (ipa_dump_param): Adjust to descriptors being a pointer.
31059 (ipa_alloc_node_params): Likewise.
31060 (ipa_initialize_node_params): Likewise.
31061 (load_from_param_1): Make descriptors parameter garbage collected.
31062 (load_from_unmodified_param): Likewise.
31063 (load_from_param): Likewise.
31064 (ipa_load_from_parm_agg): Likewise.
31065 (ipa_node_params::~ipa_node_params): Removed.
31066 (ipa_free_all_node_params): Remove call to delete operator.
31067 (ipa_node_params_t::insert): New.
31068 (ipa_node_params_t::remove): Likewise.
31069 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
31070 copy known_csts and known_contexts vectors.
31071 (ipa_read_node_info): Adjust to descriptors being a pointer.
31072 (ipcp_modif_dom_walker): Make m_descriptors field garbage
31074 (ipcp_transform_function): Make descriptors variable garbage
31077 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
31079 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
31080 * config/i386/avx512dqintrin.h: Ditto.
31081 * config/i386/avx512fintrin.h: Ditto.
31082 * config/i386/i386.c: Handle new builtins.
31083 * config/i386/i386-builtin.def: Add new builtins.
31084 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
31085 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
31087 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31088 Martin Liska <mliska@suse.cz>
31090 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
31091 * asan.c (asan_expand_poison_ifn): Support stores and use
31092 appropriate ASAN report function.
31093 * internal-fn.c (expand_ASAN_POISON_USE): New function.
31094 * internal-fn.def (ASAN_POISON_USE): Declare.
31095 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
31096 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
31097 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
31098 ASAN_POISON calls w/o LHS.
31099 * tree-ssa.c (execute_update_addresses_taken): Create clobber
31100 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
31101 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
31102 * gimplify.c (asan_poison_variables): Add attribute
31103 use_after_scope_memory to variables that really needs to live
31105 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
31106 having the attribute.
31108 2017-01-23 Martin Liska <mliska@suse.cz>
31110 * asan.c (create_asan_shadow_var): New function.
31111 (asan_expand_poison_ifn): Likewise.
31112 * asan.h (asan_expand_poison_ifn): New declaration.
31113 * internal-fn.c (expand_ASAN_POISON): Likewise.
31114 * internal-fn.def (ASAN_POISON): New builtin.
31115 * sanopt.c (pass_sanopt::execute): Expand
31116 asan_expand_poison_ifn.
31117 * tree-inline.c (copy_decl_for_dup_finish): Make function
31119 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
31120 * tree-ssa.c (is_asan_mark_p): New function.
31121 (execute_update_addresses_taken): Rewrite local variables
31122 (identified just by use-after-scope as addressable) into SSA.
31124 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31126 * doc/install.texi (Specific): opensource.apple.com uses https
31127 now. Remove trailing slash.
31129 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31131 * README.Portability: Remove note on an Irix compatibility issue.
31133 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
31135 * gcov.c (INCLUDE_ALGORITHM): Define.
31136 (INCLUDE_VECTOR): Define.
31137 No longer include <vector> and <algorithm> directly.
31139 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
31141 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
31143 * doc/invoke.texi (Code Gen Options): Ditto.
31145 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
31148 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
31150 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
31152 rtl-optimization/79125
31153 * cprop.c (local_cprop_pass): Handle cases where we make an
31154 unconditional trap.
31156 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
31160 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
31161 read from, for big endian.
31163 2017-01-20 Jiong Wang <jiong.wang@arm.com>
31165 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
31166 register pauth builtins for LP64 only.
31168 2017-01-20 Marek Polacek <polacek@redhat.com>
31171 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
31174 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
31176 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
31178 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
31179 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
31180 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
31182 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31185 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
31186 in big-endian mode when they are not a single duplicated value.
31188 2017-01-20 Richard Biener <rguenther@suse.de>
31190 * BASE-VER: Bump to 7.0.1.
31192 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
31194 * omp-low.c (omplow_simd_context): New struct. Use it...
31195 (lower_rec_simd_input_clauses): ...here and...
31196 (lower_rec_input_clauses): ...here to hold common data. Adjust all
31197 references to idx, lane, max_vf, is_simt.
31199 2017-01-20 Graham Markall <graham.markall@embecosm.com>
31201 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
31204 2017-01-20 Martin Jambor <mjambor@suse.cz>
31206 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
31207 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
31209 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
31210 (GTFILES): Rename hsa.c to hsa-common.c.
31211 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
31212 * hsa-dump.c: Likewise.
31213 * hsa-gen.c: Likewise.
31214 * hsa-regalloc.c: Likewise.
31215 * ipa-hsa.c: Likewise.
31216 * omp-expand.c: Likewise.
31217 * omp-low.c: Likewise.
31218 * toplev.c: Likewise.
31220 2017-01-20 Marek Polacek <polacek@redhat.com>
31223 * doc/invoke.texi: Document -Wduplicated-branches.
31224 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
31225 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
31226 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
31227 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
31228 return 0 only when not OEP_LEXICOGRAPHIC.
31229 (fold_build_cleanup_point_expr): Use the expression
31230 location when building CLEANUP_POINT_EXPR.
31231 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
31232 * tree.c (add_expr): Handle error_mark_node.
31234 2017-01-20 Martin Liska <mliska@suse.cz>
31237 * tree-profile.c (init_ic_make_global_vars): Do not call
31239 (gimple_init_gcov_profiler): Likewise.
31241 2017-01-20 Martin Liska <mliska@suse.cz>
31244 * cgraph.h (maybe_create_reference): Remove argument and
31246 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
31248 * ipa-cp.c (create_specialized_node): Likewise.
31249 * symtab.c (symtab_node::maybe_create_reference): Handle
31250 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
31252 2017-01-20 Martin Liska <mliska@suse.cz>
31254 * read-rtl-function.c (function_reader::create_function): Use
31255 build_decl instread of build_decl_stat.
31257 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
31259 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
31260 * config/i386/avx512dqintrin.h: Ditto.
31261 * config/i386/avx512fintrin.h: Ditto.
31262 * config/i386/i386-builtin-types.def: Add new types.
31263 * config/i386/i386.c: Handle new types.
31264 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
31265 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
31266 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
31267 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
31268 (__builtin_ia32_kshiftridi): New.
31269 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
31271 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
31275 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
31276 define to rs6000_init_stack_protect_guard.
31277 (rs6000_init_stack_protect_guard): New function.
31279 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
31280 Yunqiang Su <yunqiang.su@imgtec.com>
31282 * config.gcc (supported_defaults): Add madd4.
31283 (with_madd4): Add validation.
31284 (all_defaults): Add madd4.
31285 * config/mips/mips.opt (mmadd4): New option.
31286 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
31288 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
31290 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
31291 (ISA_HAS_FUSED_MADD4): Likewise.
31292 * doc/invoke.texi (-mmadd4): Document the new option.
31293 * doc/install.texi (--with-madd4): Document the new option.
31295 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31297 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
31298 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
31299 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
31300 (aarch64_init_pauth_hint_builtins): New.
31301 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
31302 (aarch64_expand_builtin): Expand new builtins.
31304 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31306 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
31307 * combine-stack-adj.c (no_unhandled_cfa): Handle
31308 REG_CFA_TOGGLE_RA_MANGLE.
31309 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
31310 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
31311 info for return address signing.
31312 (aarch64_expand_epilogue): Likewise.
31314 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31316 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
31317 * config/aarch64/aarch64-protos.h
31318 (aarch64_return_address_signing_enabled): New declaration.
31319 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
31321 (aarch64_expand_prologue): Sign return address before it's pushed onto
31323 (aarch64_expand_epilogue): Authenticate return address fetched from
31325 (aarch64_override_options): Sanity check for ILP32 and ISA level.
31326 (aarch64_attributes): New function attributes for "sign-return-address".
31327 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
31328 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
31329 ("*do_return"): Generate combined instructions according to key index.
31330 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
31331 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
31333 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
31334 * config/aarch64/aarch64.opt (msign-return-address=): New.
31335 * doc/extend.texi (AArch64 Function Attributes): Documents
31336 "sign-return-address=".
31337 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
31339 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
31341 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
31342 overall option summary.
31344 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31346 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
31347 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
31348 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
31349 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
31351 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
31353 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
31354 -mpower9-minmax by default for -mcpu=power9.
31355 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
31356 128-bit floating point.
31358 2017-01-20 Alan Modra <amodra@gmail.com>
31360 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
31361 optimizing for size.
31363 2017-01-20 Alan Modra <amodra@gmail.com>
31366 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
31367 for strcmp and strncmp from corresponding builtin decl.
31369 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
31371 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
31372 instead of i386/rtems-64.h.
31373 * config/i386/rtems-64.h: Remove.
31375 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
31379 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
31381 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
31383 2017-01-19 Tamar Christina <tamar.christina@arm.com>
31385 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
31386 Change int to HOST_WIDE_INT.
31387 * config/aarch64/aarch64-protos.h
31388 (aarch64_simd_gen_const_vector_dup): Likewise.
31389 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
31391 2017-01-19 David Malcolm <dmalcolm@redhat.com>
31393 * langhooks-def.h (lhd_type_for_size): New decl.
31394 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
31395 * langhooks.c (lhd_type_for_size): New function, taken from
31398 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
31400 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
31401 define_bypass for CR latency.
31402 (power9-cracked-alu): Update bypass latency and remove power9-branch.
31403 (power9-alu2): Add define_bypass for CR latency.
31405 (power9-mul): Update insn latency.
31406 (power9-mul-compare): Update insn latency, bypass latency and remove
31409 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31411 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
31413 * config/aarch64/aarch64.md
31414 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
31415 aarch64_nopcrelative_literal_loads.
31416 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
31418 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
31420 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
31421 TARGET_LOONGSON_3A.
31422 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
31424 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
31427 * config.gcc (supported_defaults): Add lxc1-sxc1.
31428 (with_lxc1_sxc1): Add validation.
31429 (all_defaults): Add lxc1-sxc1.
31430 * config/mips/mips.opt (mlxc1-sxc1): New option.
31431 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
31433 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
31434 __mips_no_lxc1_sxc1.
31435 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
31436 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
31437 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
31439 2017-01-19 Richard Biener <rguenther@suse.de>
31441 PR tree-optimization/72488
31442 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
31443 sure to restore SSA info.
31444 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
31446 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
31448 PR rtl-optimization/79121
31449 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
31450 of the inner type when shifting an extended value.
31452 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
31455 * symtab.c (symtab_node::equal_address_to): Fix comparing of
31456 interposable aliases.
31458 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
31461 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
31462 Use the evmergelohi instruction.
31463 (mov_si<mode>_e500_subreg4_2_le): Likewise.
31464 (mov_sitf_e500_subreg8_2_be): Likewise.
31465 (mov_sitf_e500_subreg12_2_le): Likewise.
31466 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
31467 (mov_si<mode>_e500_subreg4_2_be): Likewise.
31468 (mov_sitf_e500_subreg8_2_le): Likewise.
31469 (mov_sitf_e500_subreg12_2_be): Likewise.
31471 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31473 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
31474 attribute from vecsimple to vecperm.
31475 (altivec_vbpermq2): Likewise.
31477 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31480 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
31482 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31483 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
31484 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
31485 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
31486 case where N arg is SIZE_MAX.
31487 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
31488 (cmpstrsi): Add pattern.
31490 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
31492 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
31493 __builtin_vec_revb builtins.
31494 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
31495 built-in functions to support generation of the ISA 3.0 XXBR<x>
31496 vector byte reverse instructions.
31497 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
31498 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
31499 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
31500 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
31501 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
31502 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
31503 (P9V_BUILTIN_VEC_REVB): Likewise.
31504 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
31505 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
31506 (p9_xxbrq_v16qi): Likewise.
31507 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
31508 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
31509 (p9_xxbrh_v8hi): Likewise.
31510 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
31511 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
31512 vec_revb built-in functions.
31514 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
31516 PR rtl-optimization/78952
31517 * config/i386/i386.md (any_extract): New code iterator.
31518 (*insvqi_2): Use any_extract for source operand.
31519 (*insvqi_3): Use any_shiftrt for source operand.
31521 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
31523 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
31525 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
31527 2017-01-18 Matthias Klose <doko@ubuntu.com>
31529 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
31531 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31533 * config/rs6000/altivec.h (vec_bperm): Change #define.
31534 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
31535 (altivec_vbpermq2): New define_insn.
31536 (altivec_vbpermd): Likewise.
31537 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
31538 function interface.
31539 (VBPERMD): Likewise.
31540 (VBPERM): New polymorphic function interface.
31541 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
31542 Add entries for P9V_BUILTIN_VEC_VBPERM.
31543 * doc/extend.texi: Add interfaces for vec_bperm.
31545 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31547 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
31548 first letter of error messages.
31549 (s390_resolve_overloaded_builtin): Likewise.
31550 * config/s390/s390.c (s390_expand_builtin): Likewise.
31551 (s390_invalid_arg_for_unprototyped_fn): Likewise.
31552 (s390_valid_target_attribute_inner_p): Likewise.
31553 * config/s390/s390.md ("tabort"): Likewise.
31555 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
31557 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
31558 (ISA_AVOID_DIV_HILO): New macro.
31559 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
31560 (ISA_HAS_DDIV): Likewise.
31562 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
31564 * doc/invoke.texi (fabi-version): Correct number of occurrences.
31566 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
31568 * doc/invoke.texi (fabi-version): Spelling fix.
31570 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
31573 * doc/invoke.texi (fabi-version): Mention mangling fix for
31576 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
31579 * doc/invoke.texi (fabi-version): Document discriminator mangling.
31581 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
31584 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
31585 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
31587 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
31589 (stack_protect_test): Ditto.
31590 * config/rs6000/rs6000.opt (mstack-protector-guard=,
31591 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
31593 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
31594 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
31595 -mstack-protector-guard-offset=.
31596 (RS/6000 and PowerPC Options): Ditto.
31598 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
31600 * config/i386/i386.h (MASK_CLASS_P): New define.
31601 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
31602 there are no registers from different register sets also when
31603 mask registers are used. Update function comment.
31604 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
31605 to (*k/*r) and (*k/*km) alternatives.
31607 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
31609 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
31610 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
31611 (EH_RETURN_HANDLER_RTX): New define.
31612 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
31613 Force frame pointer in EH return functions.
31614 (aarch64_expand_epilogue): Add barrier for eh_return.
31615 (aarch64_final_eh_return_addr): Remove.
31616 (aarch64_eh_return_handler_rtx): New function.
31617 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
31619 (aarch64_eh_return_handler_rtx): New prototype.
31621 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31623 * config/rs6000/altivec.h (vec_rlmi): New #define.
31624 (vec_vrlnm): Likewise.
31625 (vec_rlnm): Likewise.
31626 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
31627 (UNSPEC_VRLNM): Likewise.
31628 (VIlong): New mode iterator.
31629 (altivec_vrl<VI_char>mi): New define_insn.
31630 (altivec_vrl<VI_char>nm): Likewise.
31631 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
31633 (VRLDNM): Likewise.
31634 (RLNM): New polymorphic function entry.
31635 (VRLWMI): New monomorphic function entry.
31636 (VRLDMI): Likewise.
31637 (RLMI): New polymorphic function entry.
31638 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
31639 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
31640 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
31643 2017-01-17 Jakub Jelinek <jakub@redhat.com>
31646 * dwarf2out.c (field_byte_offset): Restore the
31647 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
31648 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
31649 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
31652 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
31655 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
31657 2017-01-17 Jakub Jelinek <jakub@redhat.com>
31660 * dwarf2out.c (add_data_member_location_attribute): For constant
31661 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
31662 instead of DW_AT_data_member_location, DW_AT_bit_offset and
31663 DW_AT_byte_size attributes.
31665 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
31667 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
31668 after forcing to constant memory when the code model is medium.
31670 2017-01-17 Julia Koval <julia.koval@intel.com>
31673 * config/i386/avx512fintrin.h
31674 (_mm512_i32gather_ps): Change __addr type to void const*.
31675 (_mm512_mask_i32gather_ps): Ditto.
31676 (_mm512_i32gather_pd): Ditto.
31677 (_mm512_mask_i32gather_pd): Ditto.
31678 (_mm512_i64gather_ps): Ditto.
31679 (_mm512_mask_i64gather_ps): Ditto.
31680 (_mm512_i64gather_pd): Ditto.
31681 (_mm512_mask_i64gather_pd): Ditto.
31682 (_mm512_i32gather_epi32): Ditto.
31683 (_mm512_mask_i32gather_epi32): Ditto.
31684 (_mm512_i32gather_epi64): Ditto.
31685 (_mm512_mask_i32gather_epi64): Ditto.
31686 (_mm512_i64gather_epi32): Ditto.
31687 (_mm512_mask_i64gather_epi32): Ditto.
31688 (_mm512_i64gather_epi64): Ditto.
31689 (_mm512_mask_i64gather_epi64): Ditto.
31690 (_mm512_i32scatter_ps): Change __addr type to void*.
31691 (_mm512_mask_i32scatter_ps): Ditto.
31692 (_mm512_i32scatter_pd): Ditto.
31693 (_mm512_mask_i32scatter_pd): Ditto.
31694 (_mm512_i64scatter_ps): Ditto.
31695 (_mm512_mask_i64scatter_ps): Ditto.
31696 (_mm512_i64scatter_pd): Ditto.
31697 (_mm512_mask_i64scatter_pd): Ditto.
31698 (_mm512_i32scatter_epi32): Ditto.
31699 (_mm512_mask_i32scatter_epi32): Ditto.
31700 (_mm512_i32scatter_epi64): Ditto.
31701 (_mm512_mask_i32scatter_epi64): Ditto.
31702 (_mm512_i64scatter_epi32): Ditto.
31703 (_mm512_mask_i64scatter_epi32): Ditto.
31704 (_mm512_i64scatter_epi64): Ditto.
31705 (_mm512_mask_i64scatter_epi64): Ditto.
31706 * config/i386/avx512pfintrin.h
31707 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
31708 (_mm512_mask_prefetch_i32gather_ps): Ditto.
31709 (_mm512_mask_prefetch_i64gather_pd): Ditto.
31710 (_mm512_mask_prefetch_i64gather_ps): Ditto.
31711 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
31712 (_mm512_prefetch_i32scatter_ps): Ditto.
31713 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
31714 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
31715 (_mm512_prefetch_i64scatter_pd): Ditto.
31716 (_mm512_prefetch_i64scatter_ps): Ditto.
31717 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
31718 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
31719 * config/i386/avx512vlintrin.h
31720 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
31721 (_mm_mmask_i32gather_ps): Ditto.
31722 (_mm256_mmask_i32gather_pd): Ditto.
31723 (_mm_mmask_i32gather_pd): Ditto.
31724 (_mm256_mmask_i64gather_ps): Ditto.
31725 (_mm_mmask_i64gather_ps): Ditto.
31726 (_mm256_mmask_i64gather_pd): Ditto.
31727 (_mm_mmask_i64gather_pd): Ditto.
31728 (_mm256_mmask_i32gather_epi32): Ditto.
31729 (_mm_mmask_i32gather_epi32): Ditto.
31730 (_mm256_mmask_i32gather_epi64): Ditto.
31731 (_mm_mmask_i32gather_epi64): Ditto.
31732 (_mm256_mmask_i64gather_epi32): Ditto.
31733 (_mm_mmask_i64gather_epi32): Ditto.
31734 (_mm256_mmask_i64gather_epi64): Ditto.
31735 (_mm_mmask_i64gather_epi64): Ditto.
31736 (_mm256_i32scatter_ps): Change __addr type to void*.
31737 (_mm256_mask_i32scatter_ps): Ditto.
31738 (_mm_i32scatter_ps): Ditto.
31739 (_mm_mask_i32scatter_ps): Ditto.
31740 (_mm256_i32scatter_pd): Ditto.
31741 (_mm256_mask_i32scatter_pd): Ditto.
31742 (_mm_i32scatter_pd): Ditto.
31743 (_mm_mask_i32scatter_pd): Ditto.
31744 (_mm256_i64scatter_ps): Ditto.
31745 (_mm256_mask_i64scatter_ps): Ditto.
31746 (_mm_i64scatter_ps): Ditto.
31747 (_mm_mask_i64scatter_ps): Ditto.
31748 (_mm256_i64scatter_pd): Ditto.
31749 (_mm256_mask_i64scatter_pd): Ditto.
31750 (_mm_i64scatter_pd): Ditto.
31751 (_mm_mask_i64scatter_pd): Ditto.
31752 (_mm256_i32scatter_epi32): Ditto.
31753 (_mm256_mask_i32scatter_epi32): Ditto.
31754 (_mm_i32scatter_epi32): Ditto.
31755 (_mm_mask_i32scatter_epi32): Ditto.
31756 (_mm256_i32scatter_epi64): Ditto.
31757 (_mm256_mask_i32scatter_epi64): Ditto.
31758 (_mm_i32scatter_epi64): Ditto.
31759 (_mm_mask_i32scatter_epi64): Ditto.
31760 (_mm256_i64scatter_epi32): Ditto.
31761 (_mm256_mask_i64scatter_epi32): Ditto.
31762 (_mm_i64scatter_epi32): Ditto.
31763 (_mm_mask_i64scatter_epi32): Ditto.
31764 (_mm256_i64scatter_epi64): Ditto.
31765 (_mm256_mask_i64scatter_epi64): Ditto.
31766 (_mm_i64scatter_epi64): Ditto.
31767 (_mm_mask_i64scatter_epi64): Ditto.
31768 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
31769 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
31770 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
31771 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
31772 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
31773 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
31774 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
31775 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
31776 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
31777 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
31778 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
31779 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
31780 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
31781 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
31782 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
31783 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
31784 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
31785 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
31786 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
31787 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
31788 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
31789 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
31790 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
31791 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
31792 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
31793 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
31794 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
31795 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
31796 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
31797 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
31798 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
31799 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
31800 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
31801 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
31802 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
31803 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
31804 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
31805 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
31806 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
31807 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
31808 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
31809 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
31810 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
31811 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
31812 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
31813 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
31814 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
31815 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
31816 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
31817 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
31818 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
31819 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
31820 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
31821 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
31822 definitions accordingly.
31824 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
31825 Kuan-Lin Chen <kuanlinchentw@gmail.com>
31828 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
31831 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
31834 * ira-conflicts.c (ira_build_conflicts): Update total conflict
31835 hard regs for inner regno.
31837 2017-01-17 Martin Liska <mliska@suse.cz>
31840 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
31841 assumption and add comment.
31843 2017-01-17 Nathan Sidwell <nathan@acm.org>
31845 * ipa-visibility.c (localize_node): New function, broken out of ...
31846 (function_and_variable_visibility): ... here. Call it.
31848 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
31850 PR middle-end/77445
31851 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
31852 correctly set frequency of oudgoing edge.
31853 (duplicate_thread_path): Fix profile updating.
31855 2017-01-17 Jakub Jelinek <jakub@redhat.com>
31858 * configure.ac: Add GCC_BASE_VER.
31859 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
31860 version from BASE-VER file.
31861 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
31862 (gcc.o): Depend on $(BASEVER).
31863 * common.opt (dumpfullversion): New option.
31864 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
31865 * doc/invoke.texi: Document -dumpfullversion.
31866 * doc/install.texi: Document --with-gcc-major-version-only.
31867 * configure: Regenerated.
31869 2017-01-17 Richard Biener <rguenther@suse.de>
31871 PR tree-optimization/71433
31872 * tree-vrp.c (register_new_assert_for): Merge same asserts
31873 on all incoming edges.
31874 (process_assert_insertions_for): Handle insertions at the
31877 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
31879 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
31880 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
31882 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
31885 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
31888 2017-01-17 Alan Modra <amodra@gmail.com>
31891 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
31892 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
31893 symbolic stack limit when pic.
31895 2017-01-16 Martin Sebor <msebor@redhat.com>
31897 PR tree-optimization/78608
31898 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
31900 2017-01-16 Jeff Law <law@redhat.com>
31903 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
31904 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
31905 for several include directories that may be relative to sysroot.
31906 * config/i386/x-mingw32 (gplus_includedir): Define.
31907 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
31908 (native_system_includedir): Likewise.
31909 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
31910 override if TARGET_SYSTEM_ROOT is defined.
31911 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
31913 PR tree-optimization/79090
31914 PR tree-optimization/33562
31915 PR tree-optimization/61912
31916 PR tree-optimization/77485
31917 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
31918 and computed trims into the dump file.
31920 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
31922 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
31924 2017-01-16 Jakub Jelinek <jakub@redhat.com>
31927 * gimplify.c (gimplify_init_constructor): If want_value and
31928 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
31932 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
31933 sequence. Formatting fixes.
31934 (doloop_optimize): Formatting fixes.
31937 * gcc.c (debug_level_greater_than_spec_func): New function.
31938 (static_spec_functions): Add debug-level-gt spec function.
31939 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
31941 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
31942 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
31943 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
31944 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
31945 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
31946 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
31948 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
31950 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
31951 QImode fixups to general and mask registers only.
31953 2017-01-16 Carl Love <cel@us.ibm.com>
31955 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
31956 for built-in functions
31957 vector signed char vec_nabs (vector signed char)
31958 vector signed short vec_nabs (vector signed short)
31959 vector signed int vec_nabs (vector signed int)
31960 vector signed long long vec_nabs (vector signed long long)
31961 vector float vec_nabs (vector float)
31962 vector double vec_nabs (vector double)
31963 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
31965 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
31966 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
31967 * doc/extend.texi: Update the documentation file for the new built-in
31970 2017-01-16 Martin Sebor <msebor@redhat.com>
31972 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
31975 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31977 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
31978 UNSPEC_VSX__XXSPLTD to require special splat handling.
31980 2017-01-16 David Malcolm <dmalcolm@redhat.com>
31983 * system.h: Poison strndup.
31985 2017-01-16 Alan Modra <amodra@gmail.com>
31988 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
31991 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
31993 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
31995 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
31997 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
31998 call recog here. Assert that INSN_CODE (insn) is non-negative.
32000 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32003 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
32005 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
32006 in the currently scheduled RTL fragment.
32008 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32010 PR rtl-optimization/78751
32011 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
32014 2017-01-14 Jeff Law <law@redhat.com>
32016 PR tree-optimization/79090
32017 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
32018 variable length stores.
32019 (compute_trims): Delete dead assignment to *trim_tail.
32020 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
32023 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
32025 PR rtl-optimization/78626
32026 PR rtl-optimization/78727
32027 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
32028 of a block, and split such blocks after everything else is finished.
32030 2017-01-14 Alan Modra <amodra@gmail.com>
32033 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
32034 target legitimate_combined_insn.
32035 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
32036 (rs6000_legitimate_combined_insn): New function.
32037 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
32039 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
32040 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
32041 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
32043 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
32045 * doc/frontends.texi (G++ and GCC): Remove references to Java.
32047 2017-01-13 Jeff Law <law@redhat.com>
32049 PR tree-optimization/33562
32050 PR tree-optimization/61912
32051 PR tree-optimization/77485
32052 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
32054 (delete_dead_assignment): Likewise.
32055 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
32056 statement to delete_dead_call and delete_dead_assignment.
32058 2017-01-13 David Malcolm <dmalcolm@redhat.com>
32061 * substring-locations.c (format_warning_va): Strengthen case 1 so
32062 that both endpoints of the substring must be within the format
32063 range for just the substring to be printed.
32065 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
32067 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
32068 * config/i386/i386.c (ix86_target_string): Add missing options
32069 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
32070 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
32071 flags_other and ix86_target_other to flags2_other. Display unknown
32073 (ix86_valid_target_attribute_inner_p): Add missing options and
32074 reorder options by implied ISAs, as in ix86_target_string.
32076 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32078 * hash-table.h (hash_table::too_empty_p): New function.
32079 (hash_table::expand): Use it.
32080 (hash_table::traverse): Likewise.
32081 (hash_table::empty_slot): Use sizeof (value_type) instead of
32082 sizeof (PTR) to convert bytes to elements. Shrink the table
32083 if the current size is excessive for the current number of
32086 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32088 * ira-costs.c (record_reg_classes): Break from the inner loop
32089 early once alt_fail is known to be true. Update outer loop
32090 handling accordingly.
32092 2017-01-13 Jeff Law <law@redhat.com>
32094 * tree-ssa-dse.c (decrement_count): New function.
32095 (increment_start_addr, maybe_trim_memstar_call): Likewise.
32096 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
32097 when we know the partially dead statement is a mem* function.
32099 PR tree-optimization/61912
32100 PR tree-optimization/77485
32101 * tree-ssa-dse.c: Include expr.h.
32102 (maybe_trim_constructor_store): New function.
32103 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
32105 PR tree-optimization/33562
32106 PR tree-optimization/61912
32107 PR tree-optimization/77485
32108 * doc/invoke.texi: Document new dse-max-object-size param.
32109 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
32110 * tree-ssa-dse.c: Include params.h.
32111 (dse_store_status): New enum.
32112 (initialize_ao_ref_for_dse): New, partially extracted from
32114 (valid_ao_ref_for_dse, normalize_ref): New.
32115 (setup_live_bytes_from_ref, compute_trims): Likewise.
32116 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
32117 (maybe_trim_partially_dead_store): Likewise.
32118 (maybe_trim_complex_store): Likewise.
32119 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
32120 Track what bytes live from the original store. Return tri-state
32121 for dead, partially dead or live.
32122 (dse_dom_walker): Add constructor, destructor and new private members.
32123 (delete_dead_call, delete_dead_assignment): New extracted from
32125 (dse_optimize_stmt): Make a member of dse_dom_walker.
32126 Use initialize_ao_ref_for_dse.
32128 PR tree-optimization/33562
32129 PR tree-optimization/61912
32130 PR tree-optimization/77485
32131 * sbitmap.h (bitmap_count_bits): Prototype.
32132 (bitmap_clear_range, bitmap_set_range): Likewise.
32133 * sbitmap.c (bitmap_clear_range): New function.
32134 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
32136 2017-01-13 Martin Liska <mliska@suse.cz>
32139 * function.c (set_cfun): Add new argument force.
32140 * function.h (set_cfun): Likewise.
32141 * ipa-inline-transform.c (inline_call): Use the function when
32142 strict alising from is dropped for function we inline to.
32144 2017-01-13 Richard Biener <rguenther@suse.de>
32146 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
32147 for dumping GIMPLE INTEGER_CSTs.
32149 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32151 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
32152 to 201112L since C++17.
32154 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
32157 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
32158 if -fsanitize=kernel-address is present.
32160 2017-01-13 Richard Biener <rguenther@suse.de>
32162 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
32163 as _Literal ( type ) number in case usual suffixes do not
32164 preserve all information.
32166 2017-01-13 Richard Biener <rguenther@suse.de>
32168 PR tree-optimization/77283
32169 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
32170 and ssa-iterators.h.
32171 (is_feasible_trace): Implement a cost model based on joiner
32174 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
32177 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
32178 char or short to __float128/_Float128 directly.
32180 2017-01-12 Martin Sebor <msebor@redhat.com>
32182 to -Wformat-overflow.
32183 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
32184 (min_bytes_remaining): Same.
32185 (get_string_length): Same.
32186 (format_string): Same.
32187 (format_directive): Same.
32189 (pass_sprintf_length::handle_gimple_call): Same.
32191 2017-01-12 Jakub Jelinek <jakub@redhat.com>
32193 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
32194 info.nowrite calls with no lhs that can't throw. Return bool
32195 whether gsi_remove has been called or not.
32196 (pass_sprintf_length::handle_gimple_call): Return bool whether
32197 try_substitute_return_value called gsi_remove. Formatting fix.
32198 (pass_sprintf_length::execute): Don't use gsi_remove if
32199 handle_gimple_call returned true.
32202 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
32203 be removed due to side-effects, don't remove following barrier nor
32204 turn the successor edge into fallthru edge.
32206 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32209 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
32210 element-reversing loads and stores as not swappable.
32212 2017-01-12 Nathan Sidwell <nathan@acm.org>
32213 Nicolai Stange <nicstange@gmail.com>
32215 * combine.c (try_combine): Don't ignore result of overlap checking
32216 loop. Combine overlap & asm check into single loop.
32218 2017-01-12 Richard Biener <rguenther@suse.de>
32220 * tree-pretty-print.c (dump_generic_node): Provide -gimple
32221 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
32223 2017-01-12 Richard Biener <rguenther@suse.de>
32225 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
32226 and TS_TARGET_OPTION directly derive from TS_BASE.
32227 * tree-core.h (tree_optimization_option): Derive from tree_base.
32228 (tree_target_option): Likewise.
32230 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
32232 * config/i386/i386.c (memory_address_length): Increase len
32233 only when rip_relative_addr_p returns false.
32235 2017-01-11 Julia Koval <julia.koval@intel.com>
32237 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
32238 (OPTION_MASK_ISA_SGX_SET): New.
32239 (ix86_handle_option): Handle OPT_msgx.
32240 * config.gcc: Added sgxintrin.h.
32241 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
32242 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
32243 * config/i386/i386.c (ix86_target_string): Add -msgx.
32245 (ix86_option_override_internal): Handle new options.
32246 (ix86_valid_target_attribute_inner_p): Add sgx.
32247 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
32248 * config/i386/i386.opt: Add msgx.
32249 * config/i386/sgxintrin.h: New file.
32250 * config/i386/x86intrin.h: Add sgxintrin.h.
32252 2017-01-11 Jakub Jelinek <jakub@redhat.com>
32255 * fold-const.c (maybe_nonzero_address): Return 1 for function
32257 (tree_single_nonzero_warnv_p): Don't handle function local objects
32261 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
32264 2017-01-11 David Malcolm <dmalcolm@redhat.com>
32267 * opts.c: Include "spellcheck.h"
32268 (struct string_fragment): New struct.
32269 (struct edit_distance_traits<const string_fragment &>): New
32271 (get_closest_sanitizer_option): New function.
32272 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
32274 2017-01-11 Jakub Jelinek <jakub@redhat.com>
32276 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
32278 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
32279 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
32280 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
32281 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
32282 for initial die_offset if dwarf_split_debug_info.
32283 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
32284 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
32286 (output_skeleton_debug_sections): Formatting fix. Use
32287 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
32288 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
32290 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
32292 * config/arm/cortex-a53.md: Add bypasses for
32293 cortex_a53_r2f_cvt.
32294 (cortex_a53_r2f): Only use for transfers.
32295 (cortex_a53_f2r): Likewise.
32296 (cortex_a53_r2f_cvt): Add reservation for conversions.
32297 (cortex_a53_f2r_cvt): Likewise.
32299 2017-01-11 Tamar Christina <tamar.christina@arm.com>
32301 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
32302 to all inlined functions, change static to extern.
32304 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
32307 * config/arm/arm.c (legitimize_pic_address): Handle reference to
32309 (arm_assemble_integer): Likewise.
32311 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
32313 * config.gcc: Use new awk script to check CPU, FPU and architecture
32314 parameters for --with-... options.
32315 * config/arm/parsecpu.awk: New file
32316 * config/arm/arm-cpus.in: New file.
32317 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
32319 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
32321 * config/arm/t-arm: Update dependency rules.
32322 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
32323 of processing .def files.
32324 * config/arm/genopt.sh: Deleted.
32325 * config/arm/gentune.sh: Deleted.
32326 * config/arm/arm-cores.def: Deleted.
32327 * config/arm/arm-arches.def: Deleted.
32328 * config/arm/arm-fpus.def: Deleted.
32329 * config/arm/arm-tune.md: Regenerated.
32330 * config/arm/arm-tables.opt: Regenerated.
32331 * config/arm/arm-cpu.h: New generated file.
32332 * config/arm/arm-cpu-data.h: New generated file.
32333 * config/arm/arm-cpu-cdata.h: New generated file.
32335 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
32338 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
32340 (input_varpool_node): Unpack dynamically_initialized bit.
32342 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
32344 PR rtl-optimization/79032
32345 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
32346 the alignment of the adjusted memory reference against that of MODE,
32347 instead of the alignment of the original memory reference.
32349 2017-01-11 Martin Jambor <mjambor@suse.cz>
32351 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
32353 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
32354 decorated functions.
32356 2017-01-11 Richard Biener <rguenther@suse.de>
32358 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
32359 set range/nonnull info for PHI results. Do not set it on
32360 stmts marked for removal.
32362 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
32364 * expr.c (store_field): In the bitfield case, fetch the return value
32365 from the registers before applying a single big-endian adjustment.
32366 Always do a final load for a BLKmode value not larger than a word.
32368 2017-01-10 David Malcolm <dmalcolm@redhat.com>
32371 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
32372 that we correctly handle column numbers greater than
32373 LINE_MAP_MAX_COLUMN_NUMBER.
32375 2017-01-10 Martin Sebor <msebor@redhat.com>
32377 PR middle-end/78245
32378 * gimple-ssa-sprintf.c (get_destination_size): Call
32379 {init,fini}object_sizes.
32380 * tree-object-size.c (addr_object_size): Adjust.
32381 (pass_through_call): Adjust.
32382 (pass_object_sizes::execute): Adjust.
32383 * tree-object-size.h (fini_object_sizes): Declare.
32385 2017-01-10 Martin Sebor <msebor@redhat.com>
32387 PR tree-optimization/78775
32388 * builtins.c (get_size_range): Move...
32389 * calls.c: ...to here.
32390 (alloc_max_size): Accept zero argument.
32391 (operand_signed_p): Remove.
32392 (maybe_warn_alloc_args_overflow): Call get_size_range.
32393 * calls.h (get_size_range): Declare.
32395 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
32397 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
32398 from TI's devices.csv file as of September 2016.
32399 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
32401 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
32403 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
32404 * doc/invoke.texi: Likewise.
32405 * doc/md.texi: Likewise.
32406 * doc/objc.texi: Likewise.
32408 2017-01-10 Joshua Conner <joshconner@google.com>
32410 * config/arm/fuchsia-elf.h: New file.
32411 * config/fuchsia.h: New file.
32412 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
32413 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
32415 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
32417 2016-01-10 Richard Biener <rguenther@suse.de>
32419 PR tree-optimization/79034
32420 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
32421 Propagate out degenerate PHIs in the joiner.
32423 2017-01-10 Martin Liska <mliska@suse.cz>
32425 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
32426 (sort_congruence_classes_by_decl_uid): Likewise.
32427 (sort_congruence_class_groups_by_decl_uid): Likewise.
32428 (sem_item_optimizer::merge_classes): Sort class, groups in these
32429 classes and members in the groups by DECL_UID of declarations.
32430 This would make merge operations stable.
32432 2017-01-10 Martin Liska <mliska@suse.cz>
32434 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
32435 usage of m_classes_vec.
32436 (sem_item_optimizer::~sem_item_optimizer): Likewise.
32437 (sem_item_optimizer::get_group_by_hash): Likewise.
32438 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
32439 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
32440 (sem_item_optimizer::verify_classes): Likewise.
32441 (sem_item_optimizer::process_cong_reduction): Likewise.
32442 (sem_item_optimizer::dump_cong_classes): Likewise.
32443 (sem_item_optimizer::merge_classes): Likewise.
32444 * ipa-icf.h (congruence_class_hash): Rename from
32445 congruence_class_group_hash. Remove declaration of m_classes_vec.
32447 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
32449 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
32450 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
32451 * config.gcc: Add avx512vpopcntdqintrin.h.
32452 * config/i386/avx512vpopcntdqintrin.h: New.
32453 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
32454 * config/i386/i386-builtin-types.def: Add new types.
32455 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
32456 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
32457 __builtin_ia32_vpopcountq_v8di_mask): New.
32458 * config/i386/i386-c.c (ix86_target_macros_internal): Define
32459 __AVX512VPOPCNTDQ__.
32460 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
32461 (PTA_AVX512VPOPCNTDQ): Define.
32462 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
32463 TARGET_AVX512VPOPCNTDQ_P): Define.
32464 * config/i386/i386.opt: Add mavx512vpopcntdq.
32465 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
32466 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
32468 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32470 PR middle-end/77484
32471 * predict.def (PRED_CALL): Set to 67.
32473 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
32475 * expr.c (store_field): In the bitfield case, if the value comes from
32476 a function call and is of an aggregate type returned in registers, do
32477 not modify the field mode; extract the value in all cases if the mode
32478 is BLKmode and the size is not larger than a word.
32480 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
32483 * config/i386/cpuid.h: Fix undefined behavior.
32485 2017-01-04 Jeff Law <law@redhat.com>
32487 PR tree-optimization/79007
32488 PR tree-optimization/67955
32489 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
32490 conservative for pt.null when flag_non_call_exceptions is on.
32492 2017-01-09 Jakub Jelinek <jakub@redhat.com>
32494 PR translation/79019
32495 PR translation/79020
32496 * params.def (PARAM_INLINE_MIN_SPEEDUP,
32497 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
32498 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
32500 * config/avr/avr.opt (maccumulate-args): Likewise.
32501 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
32502 * common.opt (freport-bug): Likewise.
32503 * cif-code.def (CIF_FINAL_ERROR): Likewise.
32504 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
32505 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
32506 translatable string.
32507 * config/i386/i386.c (function_value_32): Likewise.
32508 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
32509 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
32511 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
32512 * common/config/msp430/msp430-common.c (msp430_handle_option):
32514 * symtab.c (symtab_node::verify_base): Likewise.
32515 * opts.c (set_debug_level): Likewise.
32516 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
32517 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
32518 missing whitespace to translatable strings.
32519 * config/avr/avr.md (bswapsi2): Fix typo in comment.
32520 * config/sh/superh.h: Likewise.
32521 * config/i386/xopintrin.h: Likewise.
32522 * config/i386/znver1.md: Likewise.
32523 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
32524 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
32525 * double-int.h (struct double_int): Likewise.
32526 * double-int.c (div_and_round_double): Likewise.
32527 * wide-int.cc: Likewise.
32528 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
32529 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
32530 * cfgcleanup.c (crossjumps_occured): Renamed to ...
32531 (crossjumps_occurred): ... this.
32532 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
32535 PR tree-optimization/78899
32536 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
32537 returning bool return struct loop *, NULL for failure and the new
32539 (versionable_outer_loop_p): Don't version outer loop if it has
32540 dont_vectorized bit set.
32541 (tree_if_conversion): When versioning outer loop, ensure
32542 tree_if_conversion is performed also on the inner loop of the
32543 non-vectorizable outer loop copy.
32544 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
32545 LOOP_VECTORIZED in inner loop of the scalar outer loop and
32546 prevent vectorization of it.
32547 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
32548 the outer loop vectorization of the non-scalar version is attempted
32549 before vectorization of the inner loop in scalar version. If
32550 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
32551 vectorization of its inner loop.
32552 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
32553 has 2 inner loops, rename also on edges from bb whose single pred
32554 is outer_loop->header. Fix typo in function comment.
32556 2017-01-09 Martin Sebor <msebor@redhat.com>
32559 * asan.c (asan_emit_stack_protection): Increase local buffer size
32560 to avoid snprintf truncation warning.
32562 2017-01-09 Andrew Pinski <apinski@cavium.com>
32564 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
32565 to reference thunderx2t99 for the tuning structure
32566 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
32568 (thunderx2t99_extra_costs): This.
32569 * config/aarch64/aarch64-tune.md: Regenerate.
32570 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
32571 (vulcan_addrcost_table): This.
32572 (vulcan_regmove_cost): Rename to ...
32573 (thunderx2t99_regmove_cost): This.
32574 (vulcan_vector_cost): Rename to ...
32575 (thunderx2t99_vector_cost): this.
32576 (vulcan_branch_cost): Rename to ...
32577 (thunderx2t99_branch_cost): This.
32578 (vulcan_tunings): Rename to ...
32579 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
32580 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
32582 2017-01-09 Martin Jambor <mjambor@suse.cz>
32586 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
32587 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
32588 (propagate_vr_accross_jump_function): Use the above function for all
32589 value range computations for pass-through jump functions and type
32590 converasion from explicit value range values.
32591 (ipcp_propagate_stage): Do not attempt to deduce types of formal
32592 parameters from TYPE_ARG_TYPES.
32593 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
32594 (ipa_write_node_info): Stream type of the actual argument.
32595 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
32597 2017-01-09 Martin Liska <mliska@suse.cz>
32600 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
32601 (lookup_compiler): Do not show error message with have_E.
32603 2017-01-09 Jakub Jelinek <jakub@redhat.com>
32605 PR tree-optimization/78938
32606 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
32607 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
32608 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
32609 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
32612 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32614 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
32617 2017-01-09 Richard Biener <rguenther@suse.de>
32619 PR tree-optimization/78997
32620 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
32621 name condition properly.
32623 2017-01-09 Richard Biener <rguenther@suse.de>
32626 * dwarf2out.c (is_cxx): New overload with context.
32627 (is_naming_typedef_decl): Use it.
32629 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
32631 * invoke.texi (Option Summary): Correct spacing in option lists
32632 and add line breaks to fix over-long lines.
32634 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
32636 PR middle-end/17660
32638 * extend.texi (Common Variable Attributes): Add xref to GCC
32639 Internals manual to explain mode attribute keywords.
32641 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
32644 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
32645 and Preprocessor Options.
32646 (Options for Linking): Document -pthread here....
32647 (RS/6000 and PowerPC Options): ...not here.
32648 (Solaris 2 Options): ...or here.
32649 * doc/cppopts.texi: Document -pthread.
32651 2017-01-08 Martin Sebor <msebor@redhat.com>
32653 PR middle-end/77708
32654 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
32655 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
32656 New member functions.
32657 (format_directive): Used them.
32659 (pass_sprintf_length::handle_gimple_call): Same.
32660 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
32661 to avoid truncation for any argument.
32662 (extract_affine_mul): Same.
32663 * tree.c (get_file_function_name): Same.
32665 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32667 PR middle-end/77484
32668 * predict.def (PRED_INDIR_CALL): Set to 86.
32670 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
32672 PR preprocessor/54124
32673 * doc/cppopts.texi: Reformat -d subtable to list the full name
32674 of the options. Add cross-reference to the docs for the general
32675 compiler -d options.
32676 * doc/invoke.texi (Developer Options): Add cross-reference to the
32677 preprocessor-specific -d option documentation.
32679 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
32681 PR preprocessor/13498
32682 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
32683 redudant material, and reflect new command-line options.
32684 (System Headers): Likewise.
32686 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
32688 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
32689 -isystem, and -idirafter. Copy-edit.
32690 * doc/cppopts.texi: Copy-edit. Remove contradiction about
32691 default for -ftrack-macro-expansion. Delete obsolete and
32692 badly-formatted implementation details about -fdebug-cpp output.
32693 * doc/cppwarnopts.texi: Copy-edit.
32695 2017-01-07 David Malcolm <dmalcolm@redhat.com>
32698 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
32699 that the transition from a max line width >= 1<<10 to narrower
32700 lines works correctly.
32702 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
32704 * doc/options.texi (PerFunction): New.
32705 * opt-functions.awk (switch_flags): Map both Optimization and
32706 PerFunction to CL_OPTIMIZATION.
32707 * opth-gen.awk: Test for PerFunction flag along with
32709 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
32710 it only when the latter is present. Skip those that don't in
32711 the hash function generator.
32712 * common.opt (fvar-tracking): Mark as PerFunction instead of
32714 (fvar-tracking-assignments): Likewise.
32715 (fvar-tracking-assignments-toggle): Likewise.
32716 (fvar-tracking-uninit): Likewise.
32718 2017-01-07 Jakub Jelinek <jakub@redhat.com>
32720 PR translation/79018
32721 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
32724 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
32727 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
32728 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
32729 TARGET_LONG_JUMP_TABLE_OFFSETS.
32730 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
32731 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
32732 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
32733 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
32734 * config/m68k/m68k.md (tablejump expander): Likewise.
32735 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
32736 TARGET_LONG_JUMP_TABLE_OFFSETS.
32737 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
32738 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
32740 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
32741 David Holsgrove <david.holsgrove@xilinx.com>
32743 * common/config/microblaze/microblaze-common.c
32744 (TARGET_EXCEPT_UNWIND_INFO): Remove.
32745 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
32747 * config/microblaze/microblaze.c (microblaze_must_save_register)
32748 (microblaze_expand_epilogue, microblaze_return_addr): Handle
32750 (microblaze_eh_return): New function.
32751 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
32752 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
32753 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
32754 * config/microblaze/microblaze.md (eh_return): New pattern.
32756 2017-01-06 Jakub Jelinek <jakub@redhat.com>
32758 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
32759 GCC_DIAGNOSTIC_STRINGIFY): Define.
32761 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
32763 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32765 * config/arm/arm.md (<mcrr>): New.
32767 * config/arm/arm.c (arm_arch5te): New.
32768 (arm_option_override): Set arm_arch5te.
32769 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
32771 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
32772 (arm_mcrr_qualifiers): ... this. New.
32773 (MRRC_QUALIFIERS): Define to...
32774 (arm_mrrc_qualifiers): ... this. New.
32775 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
32777 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
32778 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
32779 (MRRCI, mrrc, MRRC): New.
32780 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
32781 VUNSPEC_MRRC2): New.
32783 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32785 * config/arm/arm.md (<mcr>): New.
32787 * config/arm/arm.c (arm_coproc_builtin_available): Add
32788 support for mcr, mrc, mcr2 and mrc2.
32789 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
32790 (arm_mcr_qualifiers): ... this. New.
32791 (MRC_QUALIFIERS): Define to ...
32792 (arm_mrc_qualifiers): ... this. New.
32793 (MCR_QUALIFIERS): Define to ...
32794 (arm_mcr_qualifiers): ... this. New.
32795 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
32797 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
32798 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
32799 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
32800 VUNSPEC_MRC2): New.
32802 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32804 * config/arm/arm.md (*ldc): New.
32808 * config/arm/arm.c (arm_coproc_builtin_available): Add
32809 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
32810 (arm_coproc_ldc_stc_legitimate_address): New.
32811 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
32812 'qualifier_const_pointer'.
32813 (LDC_QUALIFIERS): Define to...
32814 (arm_ldc_qualifiers): ... this. New.
32815 (STC_QUALIFIERS): Define to...
32816 (arm_stc_qualifiers): ... this. New.
32817 * config/arm/arm-protos.h
32818 (arm_coproc_ldc_stc_legitimate_address): New.
32819 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
32820 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
32821 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
32822 stc2, stcl, stc2l): New.
32823 * config/arm/constraints.md (Uz): New.
32824 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
32825 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
32826 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
32827 VUNSPEC_STC2L): New.
32829 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32831 * config/arm/arm.md (<cdp>): New.
32832 * config/arm/arm.c (neon_const_bounds): Rename this ...
32833 (arm_const_bounds): ... this.
32834 (arm_coproc_builtin_available): New.
32835 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
32836 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
32837 (CDP_QUALIFIERS): Define to...
32838 (arm_cdp_qualifiers): ... this. New.
32840 (arm_expand_builtin_args): Add case for 6 arguments.
32841 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
32842 (arm_const_bounds): ... this.
32843 (arm_coproc_builtin_available): New.
32844 * config/arm/arm_acle.h (__arm_cdp): New.
32846 * config/arm/arm_acle_builtins.def (cdp): New.
32848 * config/arm/iterators.md (CDPI,CDP,cdp): New.
32849 * config/arm/neon.md: Rename all 'neon_const_bounds' to
32850 'arm_const_bounds'.
32851 * config/arm/types.md (coproc): New.
32852 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
32853 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
32854 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
32855 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
32857 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32859 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
32860 (UBINOP_QUALIFIERS): New.
32862 (acle_builtin_data): New. Change comment.
32863 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
32864 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
32865 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
32866 arm_acle_builtins.def.
32867 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
32868 (arm_init_acle_builtins): New.
32869 (CRC32_BUILTIN): Remove.
32870 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
32871 crc32cb, crc32ch and crc32cw.
32872 (arm_init_crc32_builtins): Remove.
32873 (arm_init_builtins): Use arm_init_acle_builtins rather
32874 than arm_init_crc32_builtins.
32875 (arm_expand_acle_builtin): New.
32876 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
32877 * config/arm/arm_acle_builtins.def: New.
32879 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32881 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
32882 (arm_builtin_datum): ... this.
32883 (arm_init_neon_builtin): Rename to ...
32884 (arm_init_builtin): ... this. Add a new parameters PREFIX
32885 and USE_SIG_IN_NAME.
32886 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
32887 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
32888 'arm_builtin_datum'.
32889 (arm_init_vfp_builtins): Likewise.
32890 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
32891 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
32892 (arm_expand_neon_args): Rename to ...
32893 (arm_expand_builtin_args): ... this. Rename builtin_arg
32894 enum values and differentiate between ARG_BUILTIN_MEMORY
32895 and ARG_BUILTIN_NEON_MEMORY.
32896 (arm_expand_neon_builtin_1): Rename to ...
32897 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
32898 values, arm_expand_builtin_args and add bool parameter NEON.
32899 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
32900 (arm_expand_vfp_builtin): Likewise.
32901 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
32903 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32905 PR middle-end/77484
32906 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
32907 * predict.c (tree_estimate_probability_bb): Reverse direction of
32908 polymorphic call predictor.
32910 2017-01-06 David Malcolm <dmalcolm@redhat.com>
32912 * passes.c (execute_one_pass): Split out pass-skipping logic into...
32913 (determine_pass_name_match): ...this new function and...
32914 (should_skip_pass_p): ...this new function.
32916 2017-01-06 Nathan Sidwell <nathan@acm.org>
32918 * ipa-visibility.c (function_and_variable_visibility): Reformat
32919 comments and long lines. Remove extrneous if.
32920 * symtab.c (symtab_node::make_decl_local): Fix code format.
32921 (symtab_node::set_section_for_node): Fix comment typo.
32923 2017-01-06 Martin Liska <mliska@suse.cz>
32926 * lra-constraints.c: Rename invariant to lra_invariant.
32927 * predict.c (set_even_probabilities): Initialize e to NULL.
32929 2017-01-05 Martin Sebor <msebor@redhat.com>
32931 PR tree-optimization/78910
32932 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
32933 (format_integer): Correct off-by-one error in the handling
32934 of precision with negative numbers in signed conversions..
32936 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
32938 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
32940 2017-01-05 Jakub Jelinek <jakub@redhat.com>
32942 PR tree-optimization/71016
32943 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
32944 factor_out_conditional_conversion. Formatting fix.
32945 (factor_out_conditional_conversion): Add cond_stmt argument.
32946 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
32947 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
32950 2017-01-05 David Malcolm <dmalcolm@redhat.com>
32952 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
32953 read-rtl-function.o, and selftest-rtl.o.
32954 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
32955 (selftest::aarch64_test_loading_full_dump): New function.
32956 (selftest::aarch64_run_selftests): New function.
32957 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
32958 selftest::aarch64_run_selftests.
32959 * config/i386/i386.c
32960 (selftest::ix86_test_loading_dump_fragment_1): New function.
32961 (selftest::ix86_test_loading_call_insn): New function.
32962 (selftest::ix86_test_loading_full_dump): New function.
32963 (selftest::ix86_test_loading_unspec): New function.
32964 (selftest::ix86_run_selftests): Call the new functions.
32965 * emit-rtl.c (maybe_set_max_label_num): New function.
32966 * emit-rtl.h (maybe_set_max_label_num): New decl.
32967 * function.c (instantiate_decls): Guard call to
32968 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
32969 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
32971 * gensupport.c (gen_reader::gen_reader): Pass "false"
32972 for new "compact" param of rtx_reader.
32973 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
32974 rather than an empty string for NULL strings.
32975 * read-md.c: Potentially include config.h rather than bconfig.h.
32976 Wrap include of errors.h with #ifdef GENERATOR_FILE.
32977 (have_error): New global, copied from errors.c.
32978 (md_reader::read_name): Rename to...
32979 (md_reader::read_name_1): ...this, adding "out_loc" param,
32980 and converting "missing name or number" to returning false, rather
32982 (md_reader::read_name): Reimplement in terms of read_name_1.
32983 (md_reader::read_name_or_nil): New function.
32984 (md_reader::read_string): Handle "(nil)" by returning NULL.
32985 (md_reader::md_reader): Add new param "compact".
32986 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
32987 (md_reader::read_file): New method.
32988 * read-md.h (md_reader::md_reader): Add new param "compact".
32989 (md_reader::read_file): New method.
32990 (md_reader::is_compact): New accessor.
32991 (md_reader::read_name): Convert return type from void to file_location.
32992 (md_reader::read_name_or_nil): New decl.
32993 (md_reader::read_name_1): New decl.
32994 (md_reader::m_compact): New field.
32995 (noop_reader::noop_reader): Pass "false" for new "compact" param
32997 (rtx_reader::rtx_reader): Add new "compact" param.
32998 (rtx_reader::read_rtx_operand): Make virtual and convert return
32999 type from void to rtx.
33000 (rtx_reader::read_until): New decl.
33001 (rtx_reader::handle_any_trailing_information): New virtual function.
33002 (rtx_reader::postprocess): New virtual function.
33003 (rtx_reader::finalize_string): New virtual function.
33004 (rtx_reader::m_in_call_function_usage): New field.
33005 (rtx_reader::m_reuse_rtx_by_id): New field.
33006 * read-rtl-function.c: New file.
33007 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
33008 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
33009 (selftest::verify_three_block_rtl_cfg): New decl.
33010 * read-rtl-function.h: New file.
33011 * read-rtl.c: Potentially include config.h rather than bconfig.h.
33012 For host, include function.h, memmodel.h, and emit-rtl.h.
33013 (one_time_initialization): New function.
33014 (struct compact_insn_name): New struct.
33015 (compact_insn_names): New array.
33016 (find_code): Handle insn codes in compact dumps.
33017 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
33018 (bind_subst_iter_and_attr): Likewise.
33019 (add_condition_to_string): Likewise.
33020 (add_condition_to_rtx): Likewise.
33021 (apply_attribute_uses): Likewise.
33022 (add_current_iterators): Likewise.
33023 (apply_iterators): Likewise.
33024 (initialize_iterators): Guard usage of apply_subst_iterator with
33025 #ifdef GENERATOR_FILE.
33026 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
33027 (md_reader::read_mapping): Likewise.
33028 (add_define_attr_for_define_subst): Likewise.
33029 (add_define_subst_attr): Likewise.
33030 (read_subst_mapping): Likewise.
33031 (check_code_iterator): Likewise.
33032 (rtx_reader::read_rtx): Likewise. Move one-time initialization
33034 (one_time_initialization): New function.
33035 (rtx_reader::read_until): New method.
33036 (read_flags): New function.
33037 (parse_reg_note_name): New function.
33038 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
33039 Handle reuse_rtx ids.
33040 Wrap iterator lookup within #ifdef GENERATOR_FILE.
33041 Add parsing support for RTL dumps, mirroring the special-cases in
33042 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
33043 values, and calling handle_any_trailing_information.
33044 (rtx_reader::read_rtx_operand): Convert return type from void
33045 to rtx, returning return_rtx. Handle case 'e'. Call
33046 finalize_string on XSTR and XTMPL fields.
33047 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
33048 "(nil)" values were omitted. Call the postprocess vfunc on the
33050 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
33051 class ctor. Initialize m_in_call_function_usage. Call
33052 one_time_initialization.
33053 * rtl-tests.c (selftest::test_uncond_jump): Call
33054 set_new_first_and_last_insn.
33055 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
33056 * selftest-rtl.c: New file.
33057 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
33058 (selftest::get_insn_by_uid): New decl.
33059 * selftest-run-tests.c (selftest::run_tests): Call
33060 read_rtl_function_c_tests.
33061 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
33062 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
33065 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
33067 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
33068 operands in a special way. Assert that pos+len <= mode precision.
33070 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33072 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
33073 3 argument Alias with unlimited for the negative form.
33074 (fno-vect-cost-model): Removed.
33076 2017-01-05 Martin Liska <mliska@suse.cz>
33078 * hsa-gen.c (gen_hsa_divmod): New function.
33079 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
33081 2017-01-05 Martin Liska <mliska@suse.cz>
33084 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
33087 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33089 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
33090 small constant length operands.
33092 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33094 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
33095 between loop iterations.
33097 2017-01-05 Martin Liska <mliska@suse.cz>
33100 * gimplify.c (gimplify_decl_expr): Compare to
33101 asan_poisoned_variables instread of checking flags.
33102 (gimplify_target_expr): Likewise.
33103 (gimplify_expr): Likewise.
33104 (gimplify_function_tree): Conditionally initialize
33105 asan_poisoned_variables.
33107 2017-01-04 Jeff Law <law@redhat.com>
33109 PR tree-optimizatin/78812
33110 * rtl.h (contains_mem_rtx_p): Prototype.
33111 * ifcvt.c (containts_mem_rtx_p): Move from here to...
33112 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
33113 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
33114 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
33115 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
33117 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33119 * input.c (assert_char_at_range): Default-initialize actual_range.
33121 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33123 * df-scan.c (df_ref_create_structure): Make regno unsigned,
33124 to match the caller.
33126 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33128 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
33129 insns after final jump in test to emit dummy move.
33131 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33133 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
33134 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
33136 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33138 * multiple_target.c (create_dispatcher_calls): Init e_next.
33139 * tree-ssa-loop-split.c (split_loop): Init border.
33140 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
33143 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
33148 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
33149 (altivec_register_operand): Do not return true if the operand
33150 contains a SUBREG mixing SImode and SFmode.
33151 (vsx_register_operand): Likewise.
33152 (vsx_reg_sfsubreg_ok): New predicate.
33153 (vfloat_operand): Do not return true if the operand contains a
33154 SUBREG mixing SImode and SFmode.
33155 (vint_operand): Likewise.
33156 (vlogical_operand): Likewise.
33157 (gpc_reg_operand): Likewise.
33158 (int_reg_operand): Likewise.
33159 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
33160 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
33161 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
33163 (rs6000_emit_move_si_sf_subreg): New helper function.
33164 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
33165 fixup SUBREGs involving SImode and SFmode.
33166 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
33167 numbers for the new peephole2 optimization.
33168 (peephole2 for SFmode unions): New peephole2 to optimize cases in
33169 the GLIBC math library that do AND/IOR/XOR operations on single
33170 precision floating point.
33171 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
33172 target macros to say whether we need to avoid SUBREGs mixing
33174 (TARGET_ALLOW_SF_SUBREG): Likewise.
33175 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
33176 (UNSPEC_SI_FROM_SF): Likewise.
33177 (iorxor): Change spacing.
33178 (and_ior_xor): New iterator for AND, IOR, and XOR.
33179 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
33180 (movdi_from_sf_zero_ext): Likewise.
33181 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
33182 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
33183 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
33184 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
33185 (fms<mode>4): Likewise.
33186 (fnma<mode>4): Likewise.
33187 (fnms<mode>4): Likewise.
33188 (nfma<mode>4): Likewise.
33189 (nfms<mode>4): Likewise.
33191 2017-01-04 Marek Polacek <polacek@redhat.com>
33194 * doc/invoke.texi: Document -Wpointer-compare.
33196 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33198 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
33201 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
33202 descriptions for -gdwarf-5 and emit them as uleb128 instead of
33205 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
33208 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
33209 documentation of the powerpc_popcntb_ok attribute.
33210 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
33211 code to issue warning messages if a requested CPU configuration is
33212 not supported by the binary (assembler and loader) toolchain.
33213 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
33214 made to define a built-in function that has been disabled.
33215 (paired_init_builtins): Add assertion to prevent ICE if attempt is
33216 made to define a built-in function that has been disabled.
33217 (altivec_init_builtins): Add comment explaining why definition
33218 of the DST built-in functions is not preceded by an assertion
33219 check. Add assertions to prevent ICE if attempts are made to
33220 define an altivec predicate or an abs* built-in function that has
33222 (htm_init_builtins): Add comment explaining why definition of the
33223 htm built-in functions is not preceded by an assertion check.
33225 2017-01-04 Jeff Law <law@redhat.com>
33227 PR tree-optimizatin/67955
33228 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
33229 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
33230 the points-to solution does not include pt_null. Use DECL_PT_UID
33233 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
33235 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
33236 Use gen_int_mode instead of gen_lopwart for const_int operands.
33238 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33240 PR tree-optimization/71563
33241 * match.pd: Simplify X << Y into X if Y is known to be 0 or
33242 out of range value - has low bits known to be zero.
33244 2017-01-04 Alan Modra <amodra@gmail.com>
33246 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
33247 * configure: Regenerate.
33248 * config.in: Regenerate.
33250 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33253 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
33254 a substring of the message, but strcmp with the whole message. Ifdef
33255 ENABLE_NLS, translate the message first using dgettext.
33257 2017-01-03 Jeff Law <law@redhat.com>
33259 PR tree-optimizatin/78856
33260 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
33261 (mark_threaded_blocks): Remove code to truncate thread paths that
33262 cross multiple loop headers. Instead invalidate the cached loop
33263 iteration information and handle case of a thread path walking
33264 into an irreducible region.
33266 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
33269 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
33270 assertions. Add support for doing the signbit if the IEEE 128-bit
33271 floating point value is in a GPR.
33272 * config/rs6000/rs6000.md (Fsignbit): Delete.
33273 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
33274 Update the length attribute if the value is in a GPR.
33275 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
33276 the sign or zero extension instruction, since the value is always 0/1.
33277 (signbit<mode>2_dm2): Delete using <Fsignbit>.
33280 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
33281 extracting SImode to a GPR register so that we can generate a
33282 store, limit the vector to be in a traditional Altivec register
33283 for the vextuwrx instruction.
33285 2017-01-03 Ian Lance Taylor <iant@google.com>
33287 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
33289 2017-01-03 Martin Sebor <msebor@redhat.com>
33291 PR tree-optimization/78696
33292 * gimple-ssa-sprintf.c (format_floating): Correct handling of
33293 precision. Use MPFR for %f for greater fidelity. Correct handling
33295 (pass_sprintf_length::compute_format_length): Set width and precision
33296 specified by asrerisk to void_node for vararg functions.
33297 (try_substitute_return_value): Adjust dump output.
33299 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
33301 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
33303 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
33305 * doc/invoke.texi (SPARC options): Document -mlra as the default.
33306 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
33307 -mlra/-mno-lra was passed to the compiler.
33309 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
33311 PR rtl-optimization/65618
33312 * emit-rtl.c (try_split): Move initialization of "before" and
33313 "after" to just before the call to emit_insn_after_setloc.
33315 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
33317 * doc/md.texi (Standard Names): Remove reference to Java frontend.
33319 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
33321 * dwarf2out.c (gen_enumeration_type_die): When
33322 -gno-strict-dwarf, add a DW_AT_encoding attribute.
33324 2017-01-03 Jakub Jelinek <jakub@redhat.com>
33326 PR tree-optimization/78965
33327 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
33328 Change first argument from const call_info & to call_info &. For %n
33329 set info.nowrite to false.
33331 PR middle-end/78901
33332 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
33333 possibly throwing calls.
33335 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
33336 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
33337 and fns handling, rather than in a separate case SSA_NAME.
33339 2017-01-02 Jeff Law <law@redhat.com>
33341 * config/darwin-driver.c (darwin_driver_init): Const-correctness
33342 fixes for first_period and second_period variables.
33344 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
33347 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
33348 (*insvqi_1): New insn pattern.
33349 (*insvqi_1_mem_rex64): Ditto.
33350 (*insvqi_2): Ditto.
33351 (*insvqi_3): Rename from *insvqi.
33353 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
33355 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
33357 * doc/cfg.texi (Edges): Remove reference to Java.
33358 (Maintaining the CFG): Ditto.
33360 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33362 PR middle-end/77674
33363 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
33364 transparent aliases.
33366 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33368 PR middle-end/77484
33369 * predict.def (PRED_CALL): Update hitrate.
33370 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
33371 * predict.c (tree_estimate_probability_bb): Split CALL predictor
33372 into direct/indirect/polymorphic variants.
33374 2017-01-01 Jakub Jelinek <jakub@redhat.com>
33376 Update copyright years.
33378 * gcc.c (process_command): Update copyright notice dates.
33379 * gcov-dump.c (print_version): Ditto.
33380 * gcov.c (print_version): Ditto.
33381 * gcov-tool.c (print_version): Ditto.
33382 * gengtype.c (create_file): Ditto.
33383 * doc/cpp.texi: Bump @copying's copyright year.
33384 * doc/cppinternals.texi: Ditto.
33385 * doc/gcc.texi: Ditto.
33386 * doc/gccint.texi: Ditto.
33387 * doc/gcov.texi: Ditto.
33388 * doc/install.texi: Ditto.
33389 * doc/invoke.texi: Ditto.
33391 Copyright (C) 2017 Free Software Foundation, Inc.
33393 Copying and distribution of this file, with or without modification,
33394 are permitted in any medium without royalty provided the copyright
33395 notice and this notice are preserved.