1 2017-11-06 Jakub Jelinek <jakub@redhat.com>
3 PR tree-optimization/82838
4 * gimple-ssa-store-merging.c
5 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
6 on a separate gimple_seq which is then appended to seq.
8 2017-11-06 Jeff Law <law@redhat.com>
11 * config/i386/i386.c (PROBE_INTERVAL): Remove.
12 (get_probe_interval): New functions.
13 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
14 (ix86_adjust_stack_and_probe): Likewise.
15 (output_adjust_stack_and_probe): Likewise.
16 (ix86_emit_probe_stack_range): Likewise.
17 (ix86_expand_prologue): Likewise.
19 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
20 Alan Hayward <alan.hayward@arm.com>
21 David Sherwood <david.sherwood@arm.com>
23 PR tree-optimization/82816
24 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
25 if the modes of the two types are the same.
26 (convert_plusminus_to_widen): Likewise.
28 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
32 (usadv16qi): New define_expand.
33 (usadv8hi): New define_expand.
35 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
38 * ipa-inline-transform.c (update_noncloned_frequencies): Always
40 (inline_transform): Likewise.
41 * predict.c (counts_to_freqs): Remove useless conditional.
42 * profile-count.h (profile_count::apply_scale): Move sanity check.
43 * tree-inline.c (copy_bb): Always scale.
44 (copy_cfg_body): Likewise.
46 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
49 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
51 (*cmp_ite0): Add enabled_for_depr_it attribute.
52 (*cmp_ite1): Likewise.
54 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
56 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
59 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
61 * tree-vrp.c (vrp_int_const_binop): Return true on success and
62 return the value by pointer.
63 (extract_range_from_multiplicative_op_1): Update accordingly.
64 Return as soon as an operation fails.
66 2017-11-05 Tom de Vries <tom@codesourcery.com>
69 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
70 (DEF_SANITIZER_BUILTIN): ... here.
71 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
72 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
74 2017-11-05 Tom de Vries <tom@codesourcery.com>
77 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
79 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
80 ASM_OUTPUT_BEFORE_CASE_LABEL call.
81 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
83 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
84 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
85 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
87 2017-11-05 Tom de Vries <tom@codesourcery.com>
90 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
93 2017-11-04 Michael Clark <michaeljclark@mac.com>
95 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
96 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
98 (*addsi3_extended): Likewise.
99 (*addsi3_extended2): Likewise.
100 (<optab>si3): Likewise.
101 (<optab>di3): Likewise.
102 (<optab><mode>3): Likewise.
103 (<*optabe>si3_internal): Likewise.
104 (zero_extendqi<SUPERQI:mode>2): Likewise.
105 (*add<mode>hi3): Likewise.
106 (*xor<mode>hi3): Likewise.
107 (<optab>di3): Likewise.
108 (*<optab>si3_extend): Likewise.
109 (*sge<u>_<X:mode><GPR:mode>): Likewise.
110 (*slt<u>_<X:mode><GPR:mode>): Likewise.
111 (*sle<u>_<X:mode><GPR:mode>): Likewise.
113 2017-11-04 Andrew Waterman <andrew@sifive.com>
115 * config/riscv/riscv.c (riscv_option_override): Conditionally set
116 TARGET_STRICT_ALIGN based upon -mtune argument.
118 2017-11-04 Andrew Waterman <andrew@sifive.com>
120 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
122 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
124 config/i386/i386.c (choose_basereg): Use optional scratch
125 register and add assertion.
126 (x86_emit_outlined_ms2sysv_save): Use scratch register when
127 needed, and don't allocate stack.
128 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
129 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
130 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
132 2017-11-03 Jeff Law <law@redhat.com>
134 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
135 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
136 to probe at the start of a noreturn function.
138 2017-11-03 Jakub Jelinek <jakub@redhat.com>
140 PR tree-optimization/78821
141 * gimple-ssa-store-merging.c: Update the file comment.
142 (MAX_STORE_ALIAS_CHECKS): Define.
143 (struct store_operand_info): New type.
144 (store_operand_info::store_operand_info): New constructor.
145 (struct store_immediate_info): Add rhs_code and ops data members.
146 (store_immediate_info::store_immediate_info): Add rhscode, op0r
147 and op1r arguments to the ctor, initialize corresponding data members.
148 (struct merged_store_group): Add load_align_base and load_align
150 (merged_store_group::merged_store_group): Initialize them.
151 (merged_store_group::do_merge): Update them.
152 (merged_store_group::apply_stores): Pick the constant for
153 encode_tree_to_bitpos from one of the two operands, or skip
154 encode_tree_to_bitpos if neither operand is a constant.
155 (class pass_store_merging): Add process_store method decl. Remove
156 bool argument from terminate_all_aliasing_chains method decl.
157 (pass_store_merging::terminate_all_aliasing_chains): Remove
158 var_offset_p argument and corresponding handling.
159 (stmts_may_clobber_ref_p): New function.
160 (compatible_load_p): New function.
161 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
162 if there is overlap and rhs_code is not INTEGER_CST. For
163 non-overlapping stores terminate group if rhs is not mergeable.
164 (get_alias_type_for_stmts): Change first argument from
165 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
166 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
167 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
169 (get_location_for_stmts): Change first argument from
170 auto_vec<gimple *> & to vec<gimple *> &.
171 (struct split_store): Remove orig_stmts data member, add orig_stores.
172 (split_store::split_store): Create orig_stores rather than orig_stmts.
173 (find_constituent_stmts): Renamed to ...
174 (find_constituent_stores): ... this. Change second argument from
175 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
176 to info structures rather than the statements.
177 (split_group): Rename ALLOW_UNALIGNED argument to
178 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
179 it. Adjust find_constituent_stores caller.
180 (imm_store_chain_info::output_merged_store): Handle rhs_code other
181 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
182 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
183 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
184 (mem_valid_for_store_merging): New function.
185 (handled_load): New function.
186 (pass_store_merging::process_store): New method.
187 (pass_store_merging::execute): Use process_store method. Adjust
188 terminate_all_aliasing_chains caller.
190 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
192 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
193 Return true for more constants, symbols and label references.
194 (aarch64_valid_floating_const): Remove unused function.
196 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
199 * config/aarch64/aarch64.c (aarch64_layout_frame):
200 Undo forcing of LR at bottom of frame.
202 2017-11-03 Jeff Law <law@redhat.com>
205 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
206 for int_registers_saved.
208 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
209 extracted from tree-ssa-dom.c.
210 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
211 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
212 (record_equivalences_from_incoming_edge): Add additional argument
213 to single_pred_edge_ignoring_loop_edges call.
214 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
215 (uncprop_dom_walker::before_dom_children): Add additional argument
216 to single_pred_edge_ignoring_loop_edges call.
217 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
218 single_pred_edge_ignoring_loop_edges rather than open coding.
219 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
221 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
223 * match.pd (-(-A)): Rewrite.
225 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
227 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
228 (rs6000_emit_int_cmove): New declaration.
229 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
230 (rs6000_emit_sISEL): Delete.
231 (rs6000_emit_int_cmove): Make non-static.
232 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
233 instead of rs6000_emit_sISEL.
235 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
237 * asan.c (create_cond_insert_point): Maintain profile.
238 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
240 * basic-block.h (struct basic_block_def): Remove frequency.
241 (EDGE_FREQUENCY): Use to_frequency
242 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
244 (find_traces): Update to use to_frequency.
245 (find_traces_1_round): Likewise; use only IPA counts.
246 (bb_to_key): Likewise.
247 (connect_traces): Use IPA counts only.
248 (copy_bb_p): Update to use to_frequency.
249 (fix_up_crossing_landing_pad): Likewise.
250 (sanitize_hot_paths): Likewise.
251 * bt-load.c (basic_block_freq): Likewise.
252 * cfg.c (init_flow): Set count_max to uninitialized.
253 (check_bb_profile): Remove frequencies; check counts.
254 (dump_bb_info): Do not dump frequencies.
255 (update_bb_profile_for_threading): Update counts only.
256 (scale_bbs_frequencies_int): Likewise.
257 (MAX_SAFE_MULTIPLIER): Remove.
258 (scale_bbs_frequencies_gcov_type): Update counts only.
259 (scale_bbs_frequencies_profile_count): Update counts only.
260 (scale_bbs_frequencies): Update counts only.
261 * cfg.h (struct control_flow_graph): Add count-max.
262 (update_bb_profile_for_threading): Update prototype.
263 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
264 (find_many_sub_basic_blocks): Likewise.
265 * cfgcleanup.c (try_forward_edges): Likewise.
266 (try_crossjump_to_edge): Likewise.
267 * cfgexpand.c (expand_gimple_cond): Likewise.
268 (expand_gimple_tailcall): Likewise.
269 (construct_init_block): Likewise.
270 (construct_exit_block): Likewise.
271 * cfghooks.c (verify_flow_info): Check consistency of counts.
272 (dump_bb_for_graph): Do not dump frequencies.
273 (split_block_1): Do not update frequencies.
274 (split_edge): Do not update frequencies.
275 (make_forwarder_block): Do not update frequencies.
276 (duplicate_block): Do not update frequencies.
277 (account_profile_record): Do not update frequencies.
278 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
279 for global heuristics.
280 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
281 (expected_loop_iterations_unbounded): Use counts only.
282 * cfgloopmanip.c (scale_loop_profile): Simplify.
283 (create_empty_loop_on_edge): Simplify
285 (duplicate_loop_to_header_edge): Simplify
286 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
287 (update_br_prob_note): Take care of removing note when profile
289 (relink_block_chain): Do not dump frequency.
290 (rtl_account_profile_record): Use to_frequency.
291 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
292 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
293 (cgraph_update_edges_for_call_stmt_node): Likewise.
294 (cgraph_edge::verify_count_and_frequency): Update.
295 (cgraph_node::verify_node): Temporarily disable frequency verification.
296 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
298 (cgraph_edge::rebuild_edges): Convert to ipa counts.
299 * cgraphunit.c (init_lowered_empty_function): Do not initialize
301 (cgraph_node::expand_thunk): Update profile.
302 * except.c (dw2_build_landing_pads): Do not update frequency.
303 * final.c (compute_alignments): Use to_frequency.
304 (dump_basic_block_info): Do not dump frequency.
305 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
306 (dump_gimple_bb_header): Do not dump frequency.
307 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
309 * gimple-streamer-in.c (input_bb): Do not stream frequency.
310 * gimple-streamer-out.c (output_bb): Do not stream frequency.
311 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
312 (init_before_recovery): Do not update frequency.
313 (sched_create_recovery_edges): Do not update frequency.
314 * hsa-gen.c (convert_switch_statements): Do not update frequency.
315 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
316 (ipa_cp_c_finalize): Set max_count to uninitialized.
317 * ipa-fnsummary.c (get_minimal_bb): Use counts.
318 (param_change_prob): Use counts.
319 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
321 * ipa-split.c (consider_split): Use to_frequency.
322 (split_function): Use to_frequency.
323 * ira-build.c (loop_compare_func): Likewise.
324 (mark_loops_for_removal): Likewise.
325 (mark_all_loops_for_removal): Likewise.
326 * loop-doloop.c (doloop_modify): Do not update frequency.
327 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
329 * lto-streamer-in.c (input_function): Update count_max.
330 * omp-expand.c (expand_omp_taskreg): Update count_max.
331 * omp-simd-clone.c (simd_clone_adjust): Update profile.
332 * predict.c (maybe_hot_frequency_p): Use to_frequency.
333 (maybe_hot_count_p): Use ipa counts only.
334 (maybe_hot_bb_p): Simplify.
335 (maybe_hot_edge_p): Simplify.
336 (probably_never_executed): Do not take frequency argument.
337 (probably_never_executed_bb_p): Do not pass frequency.
338 (probably_never_executed_edge_p): Likewise.
339 (combine_predictions_for_bb): Check that profile is nonzero.
340 (propagate_freq): Do not set frequency.
341 (drop_profile): Simplify.
342 (counts_to_freqs): Simplify.
343 (expensive_function_p): Use to_frequency.
344 (propagate_unlikely_bbs_forward): Simplify.
345 (determine_unlikely_bbs): Simplify.
346 (estimate_bb_frequencies): Add hack to silence graphite issues.
347 (compute_function_frequency): Use ipa counts.
348 (pass_profile::execute): Update.
349 (rebuild_frequencies): Use counts only.
350 (force_edge_cold): Use counts only.
351 * profile-count.c (profile_count::dump): Dump new count types.
352 (profile_count::differs_from_p): Check compatiblity.
353 (profile_count::to_frequency): New function.
354 (profile_count::to_cgraph_frequency): New function.
355 * profile-count.h (struct function): Declare.
356 (enum profile_quality): Add profile_guessed_local and
357 profile_guessed_global0.
358 (class profile_proability): Decrease number of bits to 29;
359 update from_reg_br_prob_note and to_reg_br_prob_note.
360 (class profile_count: Update comment; decrease number of bits
361 to 61. Check compatibility.
362 (profile_count::compatible_p): New private member function.
363 (profile_count::ipa_p): New member function.
364 (profile_count::operator<): Handle global zero correctly.
365 (profile_count::operator>): Handle global zero correctly.
366 (profile_count::operator<=): Handle global zero correctly.
367 (profile_count::operator>=): Handle global zero correctly.
368 (profile_count::nonzero_p): New member function.
369 (profile_count::force_nonzero): New member function.
370 (profile_count::max): New member function.
371 (profile_count::apply_scale): Handle IPA scalling.
372 (profile_count::guessed_local): New member function.
373 (profile_count::global0): New member function.
374 (profile_count::ipa): New member function.
375 (profile_count::to_frequency): Declare.
376 (profile_count::to_cgraph_frequency): Declare.
377 * profile.c (OVERLAP_BASE): Delete.
378 (compute_frequency_overlap): Delete.
379 (compute_branch_probabilities): Do not use compute_frequency_overlap.
380 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
381 * sched-ebb.c (rank): Use counts only.
382 * shrink-wrap.c (handle_simple_exit): Use counts only.
383 (try_shrink_wrapping): Use counts only.
384 (place_prologue_for_one_component): Use counts only.
385 * tracer.c (find_best_predecessor): Use to_frequency.
386 (find_trace): Use to_frequency.
387 (tail_duplicate): Use to_frequency.
388 * trans-mem.c (expand_transaction): Do not update frequency.
389 * tree-call-cdce.c: Do not update frequency.
390 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
391 (gimple_merge_blocks): Likewise.
392 (gimple_split_edge): Likewise.
393 (gimple_duplicate_sese_region): Likewise.
394 (gimple_duplicate_sese_tail): Likewise.
395 (move_sese_region_to_fn): Likewise.
396 (gimple_account_profile_record): Likewise.
397 (insert_cond_bb): Likewise.
398 * tree-complex.c (expand_complex_div_wide): Likewise.
399 * tree-eh.c (lower_resx): Update profile.
400 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
402 (initialize_cfun): Do not initialize frequencies
403 (freqs_to_counts): Delete.
404 (copy_cfg_body): Ignore count parameter.
406 (expand_call_inline): Update count_max.
407 (optimize_inline_calls): Update count_max.
408 (tree_function_versioning): Update count_max.
409 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
410 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
412 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
413 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
414 (try_peel_loop): Likewise.
415 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
417 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
418 (tree_transform_and_unroll_loop): Do not use frequencies
419 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
420 Use reliable prediction only.
421 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
422 * tree-ssa-sink.c (select_best_block): Use to_frequency.
423 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
425 * tree-ssa-threadupdate.c (create_block_for_threading): Do
427 (any_remaining_duplicated_blocks): Likewise.
428 (update_profile): Likewise.
429 (estimated_freqs_path): Delete.
430 (freqs_to_counts_path): Delete.
431 (clear_counts_path): Delete.
432 (ssa_fix_duplicate_block_edges): Likewise.
433 (duplicate_thread_path): Likewise.
434 * tree-switch-conversion.c (gen_inbound_check): Use counts.
435 * tree-tailcall.c (decrease_profile): Do not update frequency.
436 (eliminate_tail_call): Likewise.
437 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
438 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
439 (optimize_mask_stores): Likewise.
440 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
441 * ubsan.c (ubsan_expand_null_ifn): Update profile.
442 (ubsan_expand_ptr_ifn): Update profile.
443 * value-prof.c (gimple_ic): Simplify.
444 * value-prof.h (gimple_ic): Update prototype.
445 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
446 * ipa-inline.c (compute_uninlined_call_time): Be sure that
448 (want_inline_self_recursive_call_p): Likewise.
449 (resolve_noninline_speculation): Only cummulate defined counts.
450 (inline_small_functions): Use nonzero_p.
451 (ipa_inline): Do not access freed node.
453 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
455 * config/aarch64/aarch64.c (aarch64_override_options_internal):
456 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
458 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
460 * config/riscv/riscv.c (riscv_legitimize_move): Handle
461 non-legitimate address.
463 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
465 * config/rs6000/rs6000.md (*lt0_disi): Delete.
466 (*lt0_<mode>di, *lt0_<mode>si): New.
468 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
470 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
473 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
474 Jim Wilson <jim.wilson@linaro.org>
476 * config/aarch64/aarch64-cores.def (saphira): New CPU.
477 * config/aarch64/aarch64-tune.md: Regenerated.
478 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
479 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
481 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
483 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
484 (arc_expand_prologue): Restore blink for millicode.
485 * config/arc/linux.h (LINK_EH_SPEC): Defined.
487 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
490 * config/i386/i386.c (ix86_vector_duplicate_value): Use
491 gen_vec_duplicate after forcing the scalar into a register.
493 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
495 * combine (try_combine): Print the insns input to try_combine to the
498 2017-11-02 Steve Ellcey <sellcey@cavium.com>
501 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
502 Remove second argument from aarch64_process_target_attr call.
503 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
505 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
507 (aarch64_handle_attr_arch): Remove second argument.
508 (aarch64_handle_attr_cpu): Ditto.
509 (aarch64_handle_attr_tune): Ditto.
510 (aarch64_handle_attr_isa_flags): Ditto.
511 (aarch64_process_one_target_attr): Ditto.
512 (aarch64_process_target_attr): Ditto.
513 (aarch64_option_valid_attribute_p): Remove second argument.
514 on aarch64_process_target_attr call.
516 2017-11-02 David Malcolm <dmalcolm@redhat.com>
518 * diagnostic.c: Include "selftest-diagnostic.h".
519 (selftest::assert_location_text): New function.
520 (selftest::test_diagnostic_get_location_text): New function.
521 (selftest::diagnostic_c_tests): Call it.
523 2017-11-02 David Malcolm <dmalcolm@redhat.com>
525 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
526 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
527 (class selftest::test_diagnostic_context): Move to...
528 * selftest-diagnostic.c: New file.
529 * selftest-diagnostic.h: New file.
531 2017-11-02 James Bowman <james.bowman@ftdichip.com>
533 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
534 offset range for FT32B.
535 * config/ft32/ft32.h: option "mcompress" enables relaxation.
536 * config/ft32/ft32.md: Add TARGET_NOPM.
537 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
538 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
540 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
542 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
544 2017-11-02 Jeff Law <law@redhat.com>
546 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
547 virtual keyword on FINAL OVERRIDE members.
549 * tree-ssa-propagate.h (ssa_propagation_engine): Group
550 virtuals together. Add virtual destructor.
551 (substitute_and_fold_engine): Similarly.
553 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
555 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
557 2017-11-02 Richard Biener <rguenther@suse.de>
559 PR tree-optimization/82795
560 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
562 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
564 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
566 (gcc_GAS_CHECK_FEATURE): Remove.
567 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
568 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
569 * configure: Regenerate.
571 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
573 * config/arc/arc.c (hwloop_optimize): Account for empty
576 2017-11-02 Richard Biener <rguenther@suse.de>
579 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
580 Truncate ARRAY_REF index and element size.
582 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
584 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
586 2017-11-01 Jeff Law <law@redhat.com>
588 * tree-ssa-ccp.c (ccp_folder): New class derived from
589 substitute_and_fold_engine.
590 (ccp_folder::get_value): New member function.
591 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
592 (ccp_fold_stmt): Remove prototype.
593 (ccp_finalize): Call substitute_and_fold from the ccp_class.
594 * tree-ssa-copy.c (copy_folder): New class derived from
595 substitute_and_fold_engine.
596 (copy_folder::get_value): Renamed from get_value.
597 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
598 * tree-vrp.c (vrp_folder): New class derived from
599 substitute_and_fold_engine.
600 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
601 (vrp_folder::get_value): New member function.
602 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
603 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
604 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
605 provide a class interface to folder/substitute routines.
606 (ssa_prop_fold_stmt_fn): Remove typedef.
607 (ssa_prop_get_value_fn): Likewise.
608 (subsitute_and_fold): Remove prototype.
609 (replace_uses_in): Likewise.
610 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
611 Renamed from replace_uses_in. Call the virtual member function
612 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
613 (substitute_and_fold_dom_walker): Remove initialization of
614 data member entries for calbacks. Add substitute_and_fold_engine
615 member and initialize it.
616 (substitute_and_fold_dom_walker::before_dom_children0: Use the
617 member functions for get_value, replace_phi_args_in c
618 replace_uses_in, and fold_stmt calls.
619 (substitute_and_fold_engine::substitute_and_fold): Renamed from
620 substitute_and_fold. Remove assert. Update ctor call.
622 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
623 (ssa_prop_visit_phi_fn): Likewise.
624 (class ssa_propagation_engine): New class to provide an interface
626 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
628 (ssa_prop_visit_phi): Likewise.
629 (ssa_propagation_engine::simulate_stmt): Moved into class.
630 Call visit_phi/visit_stmt from the class rather than via
631 file scoped static variables.
632 (ssa_propagation_engine::simulate_block): Moved into class.
633 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
634 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
635 set file scoped statics for the visit_stmt/visit_phi callbacks.
636 * tree-complex.c (complex_propagate): New class derived from
637 ssa_propagation_engine.
638 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
639 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
640 (tree_lower_complex): Call ssa_propagate via the complex_propagate
642 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
643 ssa_propagation_engine.
644 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
645 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
646 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
647 * tree-ssa-copy.c (copy_prop): New class derived from
648 ssa_propagation_engine.
649 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
650 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
651 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
652 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
653 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
654 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
655 (execute_vrp): Call ssa_propagate from the vrp_prop class.
657 2017-11-01 Jakub Jelinek <jakub@redhat.com>
659 PR rtl-optimization/82778
660 PR rtl-optimization/82597
661 * compare-elim.c (struct comparison): Add in_a_setter field.
662 (find_comparison_dom_walker::before_dom_children): Remove killed
663 bitmap and df_simulate_find_defs call, instead walk the defs.
664 Compute last_setter and initialize in_a_setter. Merge definitions
665 with first initialization for a few variables.
666 (try_validate_parallel): Use insn_invalid_p instead of
667 recog_memoized. Return insn rather than just the pattern.
668 (try_merge_compare): Fix up comment. Don't uselessly test if
669 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
671 (execute_compare_elim_after_reload): Remove df_chain_add_problem
674 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
675 Alan Hayward <alan.hayward@arm.com>
676 David Sherwood <david.sherwood@arm.com>
678 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
679 aarch64_hard_regno_nregs to get the number of registers
682 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
683 Alan Hayward <alan.hayward@arm.com>
684 David Sherwood <david.sherwood@arm.com>
686 * config/aarch64/constraints.md (Upl): Rename to...
688 * config/aarch64/aarch64.md
689 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
692 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
693 Alan Hayward <alan.hayward@arm.com>
694 David Sherwood <david.sherwood@arm.com>
696 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
697 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
700 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
701 Alan Hayward <alan.hayward@arm.com>
702 David Sherwood <david.sherwood@arm.com>
704 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
705 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
706 (aarch64_evpc_dup): Generate rtl direcly, rather than using
708 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
710 * config/aarch64/iterators.md: Add a comment above the permute
711 unspecs to say that they are generated directly by
712 aarch64_expand_vec_perm_const.
713 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
715 2017-11-01 Nathan Sidwell <nathan@acm.org>
717 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
719 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
721 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
722 and medany code models, and describe what they do.
724 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
726 Revert accidental duplicate:
728 * combine.c (can_change_dest_mode): Reject changes in
729 REGMODE_NATURAL_SIZE.
731 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
733 PR rtl-optimization/64682
734 PR rtl-optimization/69567
735 PR rtl-optimization/69737
736 PR rtl-optimization/82683
737 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
738 register mentioned in the note, drop the note, unless it came from I3,
739 in which case it should go to I3 again.
741 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
743 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
744 and return false if not.
745 (clear_bytes_written_by, live_bytes_read): Update accordingly.
747 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
749 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
750 range is known to be empty.
752 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
753 Alan Hayward <alan.hayward@arm.com>
754 David Sherwood <david.sherwood@arm.com>
756 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
757 and CONST_VECTOR_NUNITS instead of computing the number of units from
758 the byte sizes of the vector and element.
759 (simplify_binary_operation_1): Likewise.
760 (simplify_const_binary_operation): Likewise.
761 (simplify_ternary_operation): Likewise.
763 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
764 Alan Hayward <alan.hayward@arm.com>
765 David Sherwood <david.sherwood@arm.com>
767 * var-tracking.c (INT_MEM_OFFSET): Replace with...
768 (int_mem_offset): ...this new function.
769 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
770 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
771 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
774 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
775 Alan Hayward <alan.hayward@arm.com>
776 David Sherwood <david.sherwood@arm.com>
778 * lower-subreg.c (interesting_mode_p): New function.
779 (compute_costs, find_decomposable_subregs, decompose_register)
780 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
781 (resolve_clobber, dump_choices): Use it.
783 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
784 Alan Hayward <alan.hayward@arm.com>
785 David Sherwood <david.sherwood@arm.com>
787 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
789 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
790 Alan Hayward <alan.hayward@arm.com>
791 David Sherwood <david.sherwood@arm.com>
793 * alias.c (find_base_value, find_base_term): Only process integer
794 truncations. Check the precision rather than the size.
796 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
797 Alan Hayward <alan.hayward@arm.com>
798 David Sherwood <david.sherwood@arm.com>
800 * machmode.h (is_narrower_int_mode): New function
801 * optabs.c (expand_float, expand_fix): Use it.
802 * dwarf2out.c (rotate_loc_descriptor): Likewise.
804 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
805 Alan Hayward <alan.hayward@arm.com>
806 David Sherwood <david.sherwood@arm.com>
808 * rtl.h (narrower_subreg_mode): New function.
809 * ira-color.c (update_costs_from_allocno): Use it.
811 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
812 Alan Hayward <alan.hayward@arm.com>
813 David Sherwood <david.sherwood@arm.com>
815 * optabs-query.h (convert_optab_p): New function, split out from...
816 (convert_optab_handler): ...here.
817 (widening_optab_handler): Delete.
818 (find_widening_optab_handler): Remove permit_non_widening parameter.
819 (find_widening_optab_handler_and_mode): Likewise. Provide an
820 override that operates on mode class wrappers.
821 * optabs-query.c (widening_optab_handler): Delete.
822 (find_widening_optab_handler_and_mode): Remove permit_non_widening
823 parameter. Assert that the two modes are the same class and that
824 the "from" mode is narrower than the "to" mode. Use
825 convert_optab_handler instead of widening_optab_handler.
826 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
827 instead of widening_optab_handler.
828 * expr.c (expand_expr_real_2): Update calls to
829 find_widening_optab_handler.
830 * optabs.c (expand_widen_pattern_expr): Likewise.
831 (expand_binop_directly): Take the insn_code as a parameter.
832 (expand_binop): Only call find_widening_optab_handler for
833 conversion optabs; use optab_handler otherwise. Update calls
834 to find_widening_optab_handler and expand_binop_directly.
835 Use convert_optab_handler instead of widening_optab_handler.
836 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
837 find_widening_optab_handler and use scalar_mode rather than
839 (convert_plusminus_to_widen): Likewise.
841 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
842 Alan Hayward <alan.hayward@arm.com>
843 David Sherwood <david.sherwood@arm.com>
845 * machmode.h (fixed_size_mode): New class.
846 * rtl.h (get_pool_mode): Return fixed_size_mode.
847 * gengtype.c (main): Add fixed_size_mode.
848 * target.def (get_raw_result_mode): Return a fixed_size_mode.
849 (get_raw_arg_mode): Likewise.
850 * doc/tm.texi: Regenerate.
851 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
852 * targhooks.c (default_get_reg_raw_mode): Likewise.
853 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
854 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
855 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
856 (msp430_get_raw_result_mode): Likewise.
857 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
858 * dbxout.c (dbxout_parms): Require fixed-size modes.
859 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
860 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
861 * omp-low.c (lower_oacc_reductions): Likewise.
862 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
863 (simplify_subreg): Update accordingly.
864 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
865 (force_const_mem): Update accordingly. Return NULL_RTX for modes
866 that aren't fixed-size.
867 (get_pool_mode): Return a fixed_size_mode.
868 (output_constant_pool_2): Take a fixed_size_mode.
870 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
871 Alan Hayward <alan.hayward@arm.com>
872 David Sherwood <david.sherwood@arm.com>
874 * doc/rtl.texi (vec_series): Document.
875 (const): Say that the operand can be a vec_series.
876 * rtl.def (VEC_SERIES): New rtx code.
877 * rtl.h (const_vec_series_p_1): Declare.
878 (const_vec_series_p): New function.
879 * emit-rtl.h (gen_const_vec_series): Declare.
880 (gen_vec_series): Likewise.
881 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
882 (gen_vec_series): Likewise.
883 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
884 * simplify-rtx.c (simplify_unary_operation): Handle negations
886 (simplify_binary_operation_series): New function.
887 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
888 (test_vector_ops_series): New function.
889 (test_vector_ops): Call it.
890 * config/powerpcspe/altivec.md (altivec_lvsl): Use
891 gen_const_vec_series.
892 (altivec_lvsr): Likewise.
893 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
895 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
896 Alan Hayward <alan.hayward@arm.com>
897 David Sherwood <david.sherwood@arm.com>
899 * doc/rtl.texi (const): Update description of address constants.
900 Say that vector constants are allowed too.
901 * common.md (E, F): Use CONSTANT_P instead of checking for
903 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
904 checking for CONST_VECTOR.
905 * expmed.c (make_tree): Use build_vector_from_val for a CONST
907 * expr.c (expand_expr_real_2): Check for vector modes instead
908 of checking for CONST_VECTOR.
909 * rtl.h (const_vec_p): New function.
910 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
911 (unwrap_const_vec_duplicate): Handle them here too.
913 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
914 David Malcolm <dmalcolm@redhat.com>
915 Alan Hayward <alan.hayward@arm.com>
916 David Sherwood <david.sherwood@arm.com>
918 * rtl.h (vec_duplicate_p): New function.
919 * selftest-rtl.c (assert_rtx_eq_at): New function.
920 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
921 (assert_rtx_eq_at): Declare.
922 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
923 * selftest-run-tests.c (selftest::run_tests): Call it.
924 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
925 (simplify_unary_operation_1): Recursively handle vector duplicates.
926 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
928 (simplify_subreg): Handle subregs of vector duplicates.
929 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
930 (selftest::simplify_rtx_c_tests): New functions.
932 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
933 Alan Hayward <alan.hayward@arm.com>
934 David Sherwood <david.sherwood@arm.com>
936 * emit-rtl.h (gen_const_vec_duplicate): Declare.
937 (gen_vec_duplicate): Likewise.
938 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
940 (gen_const_vector): ...here.
941 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
942 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
943 whose elements are all equal.
944 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
945 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
946 (simplify_relational_operation): Likewise.
947 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
949 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
950 (aarch64_expand_vector_init): Likewise.
951 * config/arm/arm.c (neon_vdup_constant): Likewise.
952 (neon_expand_vector_init): Likewise.
953 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
954 (arm_block_set_unaligned_vect): Likewise.
955 (arm_block_set_aligned_vect): Likewise.
956 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
957 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
958 (expand_vec_perm_even_odd_pack): Likewise.
959 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
960 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
961 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
962 gen_const_vec_duplicate.
963 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
964 * config/mips/mips.c (mips_gen_const_int_vector): Use
965 gen_const_vec_duplicate.
966 (mips_expand_vector_init): Use CONST0_RTX.
967 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
968 (define_split): Use gen_const_vec_duplicate.
969 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
970 (define_split): Use gen_const_vec_duplicate.
971 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
972 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
973 * config/spu/spu.c (spu_const): Likewise.
975 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
976 Alan Hayward <alan.hayward@arm.com>
977 David Sherwood <david.sherwood@arm.com>
979 * combine.c (can_change_dest_mode): Reject changes in
980 REGMODE_NATURAL_SIZE.
982 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
984 * configure.ac (--enable-libssp): New.
985 (gcc_cv_libc_provides_ssp): Check for explicit setting before
986 trying to determine target-specific default. Adjust indentation.
987 * configure: Regenerated.
988 * doc/install.texi (Configuration): Expand --disable-libssp
991 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
993 config/i386/i386.c (ix86_expand_epilogue): Correct stack
996 2017-10-31 Martin Jambor <mjambor@suse.cz>
999 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
1001 2017-10-31 David Malcolm <dmalcolm@redhat.com>
1003 * auto-profile.c (autofdo_source_profile::read): Use
1004 UNKNOWN_LOCATION rather than 0.
1005 * diagnostic-core.h (warning_at_rich_loc): Rename to...
1006 (warning_at): ...this overload.
1007 (warning_at_rich_loc_n): Rename to...
1008 (warning_n): ...this overload.
1009 (error_at_rich_loc): Rename to...
1010 (error_at): ...this overload.
1011 (pedwarn_at_rich_loc): Rename to...
1012 (pedwarn): ...this overload.
1013 (permerror_at_rich_loc): Rename to...
1014 (permerror): ...this overload.
1015 (inform_at_rich_loc): Rename to...
1016 (inform): ...this overload.
1017 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
1018 (diagnostic_n_impl_richloc): Rename to...
1019 (diagnostic_n_impl): ...this rich_location *-based decl.
1020 (inform_at_rich_loc): Rename to...
1021 (inform): ...this, and add an assertion.
1022 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
1023 (warning_at_rich_loc): Rename to...
1024 (warning_at): ...this, and add an assertion.
1025 (warning_at_rich_loc_n): Rename to...
1026 (warning_n): ...this, and add an assertion.
1027 (warning_n): Update location_t-based implementation for removal of
1028 location_t-based diagnostic_n_impl.
1029 (pedwarn_at_rich_loc): Rename to...
1030 (pedwarn): ...this, and add an assertion.
1031 (permerror_at_rich_loc): Rename to...
1032 (permerror): ...this, and add an assertion.
1033 (error_n): Update for removal of location_t-based diagnostic_n_impl.
1034 (error_at_rich_loc): Rename to...
1035 (error_at): ...this, and add an assertion.
1036 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
1037 (driver::do_spec_on_infiles): Likewise.
1038 * substring-locations.c (format_warning_va): Update for renaming
1039 of inform_at_rich_loc.
1041 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
1043 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
1044 _Float<N>X built-in functions so that the variant without the
1045 "__builtin_" prefix is only enabled for the GNU C and Objective C
1046 languages when they are in non-strict ANSI/ISO mode.
1047 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
1048 * target.def (floatn_builtin_p): Add a target hook to control
1049 whether _Float<N> and _Float<N>X built-in functions without the
1050 "__builtin_" prefix are enabled, and return true for C and
1051 Objective C in the default hook. Include langhooks.h in
1053 * targhooks.h (default_floatn_builtin_p): Likewise.
1054 * targhooks.c (default_floatn_builtin_p): Likewise.
1055 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
1056 floatn_builtin_p target hook.
1057 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
1059 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
1060 Eric Botcazou <ebotcazou@adacore.com>
1062 PR rtl-optimization/81803
1063 * lra-constraints.c (curr_insn_transform): Also reload the whole
1064 register for a strict subreg no wider than a word if this is for
1065 a WORD_REGISTER_OPERATIONS target.
1067 2017-10-31 Jason Merrill <jason@redhat.com>
1069 * gdbinit.in: Skip over inlines from timevar.h.
1071 2017-10-31 Martin Liska <mliska@suse.cz>
1073 * doc/gcov.texi: Document new option.
1074 * gcov.c (print_usage): Likewise print it.
1075 (process_args): Support the argument.
1076 (format_count): New function.
1077 (format_gcov): Use the function.
1079 2017-10-31 Martin Liska <mliska@suse.cz>
1081 * gcov.c (struct name_map): do not use typedef.
1082 Define operator== and operator<.
1083 (name_search): Remove.
1084 (name_sort): Remove.
1085 (main): Do not allocate names.
1086 (process_file): Add vertical space.
1087 (generate_results): Use std::find.
1088 (release_structures): Do not release memory.
1089 (find_source): Use std::find.
1091 2017-10-31 Martin Liska <mliska@suse.cz>
1093 * gcov.c (struct line_info): Remove it's typedef.
1094 (line_info::line_info): Add proper ctor.
1095 (line_info::has_block): Do not use a typedef.
1096 (struct source_info): Do not use typedef.
1097 (circuit): Likewise.
1098 (get_cycles_count): Likewise.
1099 (output_intermediate_file): Iterate via vector iterator.
1100 (add_line_counts): Use std::vector methods.
1101 (accumulate_line_counts): Likewise.
1102 (output_lines): Likewise.
1104 2017-10-31 Martin Liska <mliska@suse.cz>
1106 * gcov.c (struct source_info): Remove typedef.
1107 (source_info::source_info): Add proper ctor.
1108 (accumulate_line_counts): Use struct, not it's typedef.
1109 (output_gcov_file): Likewise.
1110 (output_lines): Likewise.
1111 (main): Do not allocate an array.
1112 (output_intermediate_file): Use size of vector container.
1113 (process_file): Resize the vector.
1114 (generate_results): Do not preallocate, use newly added vector
1116 (release_structures): Do not release sources.
1117 (find_source): Use vector methods.
1118 (add_line_counts): Do not use typedef.
1120 2017-10-31 Martin Liska <mliska@suse.cz>
1122 * doc/gcov.texi: Document that.
1123 * gcov.c (add_line_counts): Mark lines with a non-executed
1125 (output_line_beginning): Handle such lines.
1126 (output_lines): Pass new argument.
1127 (output_intermediate_file): Print it in intermediate format.
1129 2017-10-31 Martin Liska <mliska@suse.cz>
1131 * color-macros.h: New file.
1132 * diagnostic-color.c: Factor out color related to macros to
1134 * doc/gcov.texi: Document -k option.
1135 * gcov.c (INCLUDE_STRING): Include string.h.
1136 (print_usage): Add -k option.
1137 (process_args): Parse it.
1138 (pad_count_string): New function.
1139 (output_line_beginning): Likewise.
1140 (DEFAULT_LINE_START): New macro.
1141 (output_lines): Support color output.
1143 2017-10-31 Martin Liska <mliska@suse.cz>
1145 PR gcov-profile/82633
1146 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
1147 their interaction with GCOV infrastructure.
1148 * configure.ac: Add -fkeep-{inline,static}-functions to
1150 * configure: Regenerate.
1152 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
1155 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
1157 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
1160 * config/rs6000/rs6000.md (allocate_stack): Force update interval
1161 into a register if it does not fit into an immediate offset field.
1163 2017-10-31 Olivier Hainque <hainque@adacore.com>
1165 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
1167 2017-10-31 Julia Koval <julia.koval@intel.com>
1169 * config.gcc: Add gfniintrin.h.
1170 * config/i386/gfniintrin.h: New.
1171 * config/i386/i386-builtin-types.def
1172 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1173 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1174 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1175 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1176 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1177 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
1178 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
1179 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
1180 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
1181 V64QI_FTYPE_V64QI_V64QI_INT): New types.
1182 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1183 * config/i386/immintrin.h: Include gfniintrin.h.
1184 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
1186 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
1188 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
1190 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1192 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
1193 (arm_ashldi3_1bit): Remove pattern.
1194 (ashrdi3): Remove shift by 1 expansion.
1195 (arm_ashrdi3_1bit): Remove pattern.
1196 (lshrdi3): Remove shift by 1 expansion.
1197 (arm_lshrdi3_1bit): Remove pattern.
1198 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
1199 cost of ashldi3 by 1.
1200 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
1201 (<shift>di3_neon): Likewise.
1203 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
1205 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
1206 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
1207 and (*aarch64_simd_mov<VQ:mode>).
1208 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
1209 pattern alternative.
1210 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
1211 attributes to match pattern alternative.
1213 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
1215 * config.gcc (powerpc*-*-*): Add emmintrin.h.
1216 * config/rs6000/emmintrin.h: New file.
1217 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
1219 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1221 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
1222 * (movdi_vfp_cortexa8): Remove pattern.
1224 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1226 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
1228 (Specific, alpha*-dec-osf5.1): Remove.
1229 (Specific, mips-sgi-irix5): Remove.
1230 (Specific, mips-sgi-irix6): Remove.
1232 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1235 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
1236 arguments to clear_bit_region_be.
1238 2017-10-30 Jim Wilson <wilson@tuliptree.org>
1240 * gimplify.c: Include memmodel.h.
1242 2017-10-30 Martin Jambor <mjambor@suse.cz>
1244 * omp-grid.c (grid_attempt_target_gridification): Also insert a
1245 condition whether loop should be executed at all.
1247 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
1249 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1250 gimple folding of vec_madd() intrinsics.
1251 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
1252 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
1253 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
1255 2017-10-30 Richard Biener <rguenther@suse.de>
1257 PR tree-optimization/82762
1259 2017-10-23 Richard Biener <rguenther@suse.de>
1261 PR tree-optimization/82129
1263 2017-08-01 Richard Biener <rguenther@suse.de>
1265 PR tree-optimization/81181
1266 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
1267 (compute_antic): ... end of iteration here.
1269 2017-10-30 Joseph Myers <joseph@codesourcery.com>
1271 * doc/invoke.texi (C Dialect Options): Document -std=c17,
1272 -std=iso9899:2017 and -std=gnu17.
1273 * doc/standards.texi (C Language): Document C17 support.
1274 * doc/cpp.texi (Overview): Mention -std=c17.
1275 (Standard Predefined Macros): Document C11 and C17 values of
1276 __STDC_VERSION__. Do not refer to C99 support as incomplete.
1277 * doc/extend.texi (Inline): Do not list individual options for
1278 standards newer than C99.
1279 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
1281 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
1284 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
1286 * asan.c (asan_finish_file): Align asan globals array by shadow
1289 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1292 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
1293 (struct store_immediate_info): Add bitregion_start and bitregion_end
1295 (store_immediate_info::store_immediate_info): Add brs and bre
1296 arguments and initialize bitregion_{start,end} from those.
1297 (struct merged_store_group): Add bitregion_start, bitregion_end,
1298 align_base and mask fields. Drop unnecessary struct keyword from
1299 struct store_immediate_info. Add do_merge method.
1300 (clear_bit_region_be): Use memset instead of loop storing zeros.
1301 (merged_store_group::do_merge): New method.
1302 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
1303 stores as long as the surrounding bitregions have no gaps.
1304 (merged_store_group::merge_overlapping): Use do_merge.
1305 (merged_store_group::apply_stores): Test that bitregion_{start,end}
1306 is byte aligned, rather than requiring that start and width are
1307 byte aligned. Drop unnecessary struct keyword from
1308 struct store_immediate_info. Allocate and populate also mask array.
1309 Make start of the arrays relative to bitregion_start rather than
1310 start and size them according to bitregion_{end,start} difference.
1311 (struct imm_store_chain_info): Drop unnecessary struct keyword from
1312 struct store_immediate_info.
1313 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
1314 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
1315 struct keyword from struct store_immediate_info.
1316 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
1317 between stores as long as the surrounding bitregions have no gaps.
1319 (struct split_store): Add orig non-static data member.
1320 (split_store::split_store): Initialize orig to false.
1321 (find_constituent_stmts): Return store_immediate_info *, non-NULL
1322 if there is exactly a single original stmt. Change stmts argument
1323 to pointer from reference, if NULL, don't push anything to it. Add
1324 first argument, use it to optimize skipping over orig stmts that
1325 are known to be before bitpos already. Simplify.
1326 (split_group): Return unsigned int count how many stores are or
1327 would be needed rather than a bool. Add allow_unaligned argument.
1328 Change split_stores argument from reference to pointer, if NULL,
1329 only do a dry run computing how many stores would be produced.
1330 Rewritten algorithm to use both alignment and misalign if
1331 !allow_unaligned and handle bitfield stores with gaps.
1332 (imm_store_chain_info::output_merged_store): Set start_byte_pos
1333 from bitregion_start instead of start. Compute allow_unaligned
1334 here, if true, do 2 split_group dry runs to compute which one
1335 produces fewer stores and prefer aligned if equal. Punt if
1336 new count is bigger or equal than original before emitting any
1337 statements, rather than during that. Remove no longer needed
1338 new_ssa_names tracking. Replace num_stmts with
1339 split_stores.length (). Use 32-bit stack allocated entries
1340 in split_stores auto_vec. Try to reuse original store lhs/rhs1
1341 if possible. Handle bitfields with gaps.
1342 (pass_store_merging::execute): Ignore bitsize == 0 stores.
1343 Compute bitregion_{start,end} for the stores and construct
1344 store_immediate_info with that. Formatting fixes.
1346 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
1349 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
1350 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
1352 2017-10-29 Jim Wilson <wilson@tuliptree.org>
1354 * gimplify.c: Include tm_p.h.
1356 * common.opt (gcoff): Re-add as ignored option.
1357 (gcoff1, gcoff2, gcoff3): Likewise.
1359 * Makefile.in (OBJS): Delete sdbout.o.
1360 (GTFILES): Delete $(srcdir)/sdbout.c.
1361 * debug.h: Delete sdb_debug_hooks.
1362 * final.c: Delete sdbout.h include.
1363 (final_scan_insn): Delete SDB_DEBUG check.
1364 (rest_of_clean_state): Likewise.
1365 * output.h: Delete sdb_begin_function_line.
1368 * toplev.c: Delete sdbout.h include.
1369 (process_options): Delete SDB_DEBUG check.
1370 * tree-core.h (tree_type_common): Delete pointer field of
1372 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
1373 TYPE_SYMTAB_POINTER.
1374 * tree.h (TYPE_SYMTAB_POINTER): Delete.
1375 (TYPE_SYMTAB_IS_POINTER): Delete.
1376 (TYPE_SYMTAB_IS_DIE): Renumber.
1377 * xcoffout.c: Refer to former sdbout.c file.
1378 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
1380 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
1381 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
1382 (Debugging Options): Delete -gcoff.
1383 (-gstabs): Delete SDB reference.
1385 (-gcoff@var{level}): Delete.
1386 * doc/passes.texi (Debugging information output): Delete SDB and
1387 sdbout.c references.
1388 * doc/tm.texi: Regenerate.
1389 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
1390 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
1392 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
1393 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
1394 (SDB_DEBUGGING_INFO): Delete.
1395 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
1396 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
1397 * target.def (output_source_filename): Delete COFF reference.
1399 * common.opt (gcoff): Delete.
1400 (gxcoff+): Update Negative chain.
1401 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
1403 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
1404 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
1405 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
1407 (expand_function_start): Change sdb reference to past tense.
1408 (expand_function_end): Change sdb reference to past tense.
1409 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
1410 * opts.c (debug_type_names): Delete coff entry.
1411 (common_handle_option): Delete OPT_gcoff case.
1412 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
1414 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
1415 * config/cris/cris.h: Delete SDB reference in comment.
1416 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
1417 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
1418 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
1419 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
1421 (ix86_expand_prologue): Likewise.
1422 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
1423 * config/ia64/ia64.h: Likewise.
1424 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
1425 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
1427 * config/mmix/mmix.h: Likewise.
1428 * config/nds32/nds32.c: Likewise.
1429 * config/stormy/storym16.h: Likewise.
1430 * config/visium/visium.h: Likewise.
1431 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
1433 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
1435 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
1436 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
1437 FRAME_POINTER_REGNUM point at high end of local var area.
1439 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
1441 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
1442 Move comment around. Do not reset best_edge for a copiable
1443 destination if the copy would cause a partition change.
1444 (better_edge_p): Remove redundant check.
1446 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
1448 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
1450 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
1452 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
1453 for math functions that have _Float<N> and _Float<N>X variants.
1454 (mathfn_built_in_2): Add support for math functions that have
1455 _Float<N> and _Float<N>X variants.
1456 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
1457 (expand_builtin_mathfn_ternary): Add support for fma with
1458 _Float<N> and _Float<N>X variants.
1459 (expand_builtin): Likewise.
1460 (fold_builtin_3): Likewise.
1461 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
1462 create math function _Float<N> and _Float<N>X variants as external
1464 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
1465 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
1466 the __builtin_ prefix and if not strict ansi, without the prefix.
1467 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
1468 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
1469 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
1470 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
1471 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
1472 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
1473 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
1474 function signatures for fma _Float<N> and _Float<N>X variants.
1475 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
1476 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
1477 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
1478 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
1479 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
1480 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
1481 * gencfn-macros.c (print_case_cfn): Add support for math functions
1482 that have _Float<N> and _Float<N>X variants.
1483 (print_define_operator_list): Likewise.
1484 (fltfn_suffixes): Likewise.
1486 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
1487 for math functions that have _Float<N> and _Float<N>X variants.
1488 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
1489 and _Float<N>X variants.
1490 (COPYSIGN): Likewise.
1493 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
1494 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
1496 (integer_valued_read_call_p): Likewise.
1497 * fold-const-call.c (fold_const_call_ss): Likewise.
1498 (fold_const_call_sss): Add support for copysign, fmin, and fmax
1499 _Float<N> and _Float<N>X variants.
1500 (fold_const_call_ssss): Add support for fma _Float<N> and
1501 _Float<N>X variants.
1502 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
1503 support for copysign and fma _Float<N> and _Float<N>X variants.
1504 (backprop::process_builtin_call_use): Likewise.
1505 * tree-call-cdce.c (can_test_argument_range); Add support for
1506 sqrt _Float<N> and _Float<N>X variants.
1507 (edom_only_function): Likewise.
1508 (get_no_error_domain): Likewise.
1509 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
1510 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
1511 copysign _Float<N> and _Float<N>X variants.
1512 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
1513 handled by machine independent code.
1514 (FMAF128): Likewise.
1515 * doc/cpp.texi (Common Predefined Macros): Document defining
1516 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
1517 fma _Float<N> and _Float<N>X variants.
1519 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
1522 * config/i386/i386-modes.def (CCFPU): Remove definition.
1523 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
1524 (ix86_cc_modes_compatible): Ditto.
1525 (ix86_expand_carry_flag_compare): Ditto.
1526 (ix86_expand_int_movcc): Ditto.
1527 (ix86_expand_int_addcc): Ditto.
1528 (ix86_reverse_condition): Ditto.
1529 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
1530 Return true/false for unordered/ordered fp comparisons.
1531 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
1532 (ix86_prepare_fp_compare_args): Update for rename.
1533 (ix86_expand_fp_compare): Update for rename. Generate unordered
1534 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
1535 (ix86_expand_sse_compare_and_jump): Ditto.
1536 * config/i386/predicates.md (fcmov_comparison_operator):
1537 Remove CCFPU mode handling.
1538 (ix86_comparison_operator): Ditto.
1539 (ix86_carry_flag_operator): Ditto.
1540 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
1541 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
1542 (*cmpu<mode>_cc_i387): Ditto.
1543 (FPCMP): Remove mode iterator.
1544 (unord): Remove mode attribute.
1545 (unord_subst): New define_subst transformation
1546 (unord): New define_subst attribute.
1548 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
1549 (*cmpi<unord>xf_i387): Ditto.
1550 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
1551 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
1552 using unord_subst transformation.
1553 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
1554 (round_saeonly): Also handle CCFP mode.
1555 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
1556 Remove UNSPEC_SAHF unspec handling.
1558 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
1560 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
1562 2017-10-27 Jeff Law <law@redhat.com>
1564 * gimple-ssa-sprintf.c: Include domwalk.h.
1565 (class sprintf_dom_walker): New class, derived from dom_walker.
1566 (sprintf_dom_walker::before_dom_children): New function.
1567 (struct call_info): Moved into sprintf_dom_walker class
1568 (compute_formath_length, handle_gimple_call): Likewise.
1569 (sprintf_length::execute): Call the dominator walker rather
1570 than walking the statements.
1572 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
1573 gimple statement locations.
1574 (check_array_bounds): Corresponding changes. Get the statement's
1575 location directly from wi->stmt.
1577 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
1580 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
1582 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
1584 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
1585 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
1587 2017-10-27 Jakub Jelinek <jakub@redhat.com>
1590 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
1591 * config/i386/i386.c (maybe_get_pool_constant): Removed.
1592 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
1593 maybe_get_pool_constant.
1594 * config/i386/predicates.md (zero_extended_scalar_load_operand):
1597 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1599 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
1600 2.26 caveat. Update gas and gld versions.
1601 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
1604 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1606 * cgraph.h (set_malloc_flag): Declare.
1607 * cgraph.c (set_malloc_flag_1): New function.
1608 (set_malloc_flag): Likewise.
1609 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
1610 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
1612 (read_ipa_call_summary): Add support for reading is_return_callee.
1613 (write_ipa_call_summary): Stream is_return_callee.
1614 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
1615 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
1616 ipa-prop.h, ipa-fnsummary.h.
1617 (pure_const_names): Change to static.
1618 (malloc_state_e): Define.
1619 (malloc_state_names): Define.
1620 (funct_state_d): Add field malloc_state.
1621 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
1622 (check_retval_uses): New function.
1623 (malloc_candidate_p): Likewise.
1624 (analyze_function): Add support for malloc attribute.
1625 (pure_const_write_summary): Stream malloc_state.
1626 (pure_const_read_summary): Add support for reading malloc_state.
1627 (dump_malloc_lattice): New function.
1628 (propagate_malloc): New function.
1629 (warn_function_malloc): New function.
1630 (ipa_pure_const::execute): Call propagate_malloc and
1631 ipa_free_fn_summary.
1632 (pass_local_pure_const::execute): Add support for malloc attribute.
1633 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
1634 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
1636 2017-10-27 Martin Liska <mliska@suse.cz>
1638 PR gcov-profile/82457
1639 * doc/invoke.texi: Document that one needs a non-strict ISO mode
1640 for fork-like functions to be properly instrumented.
1642 2017-10-27 Richard Biener <rguenther@suse.de>
1645 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
1646 info when we redirected EH.
1648 2017-10-26 Michael Collison <michael.collison@arm.com>
1650 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
1652 (<optab>_trunchf<GPI:mode>2: New pattern.
1653 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
1654 * config/aarch64/iterators.md (wv): New mode attribute.
1655 (vf, VF): New mode attributes.
1656 (vgp, VGP): New mode attributes.
1657 (s): Update attribute with SImode and DImode prefixes.
1659 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
1661 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
1662 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
1663 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
1664 (nios2_option_overide): Initialize it. Don't allow R0-relative
1665 addressing with PIC.
1666 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
1667 (nios2_symbolic_constant_p): Likewise.
1668 (nios2_legitimate_address_p): Likewise.
1669 (nios2_r0rel_section_name_p): New.
1670 (nios2_symbol_ref_in_r0rel_data_p): New.
1671 (nios2_emit_move_sequence): Handle r0rel_constant_p.
1672 (r0rel_constant_p): New.
1673 (nios2_print_operand_address): Handle r0rel_constant_p.
1674 (nios2_cdx_narrow_form_p): Likewise.
1675 * config/nios2/nios2.opt (mr0rel-sec=): New option.
1676 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
1677 (Nios II Options): Document -mr0rel-sec.
1679 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
1681 * config/nios2/nios2.c: Include xregex.h.
1682 (nios2_gprel_sec_regex): New.
1683 (nios2_option_overide): Initialize it. Don't allow GP-relative
1684 addressing with PIC.
1685 (nios2_small_section_name_p): Check for regex match.
1686 * config/nios2/nios2.opt (mgprel-sec=): New option.
1687 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
1688 (Nios II Options): Document -mgprel-sec.
1690 2017-10-26 Jim Wilson <wilson@tuliptree.org>
1692 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
1694 2017-10-26 Tom de Vries <tom@codesourcery.com>
1696 PR tree-optimization/82707
1697 * gimple.c (gimple_copy): Fix unsharing of
1698 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
1700 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
1702 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
1703 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
1704 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
1705 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
1706 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
1707 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
1708 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
1709 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
1710 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
1711 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
1712 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
1713 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
1714 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
1715 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
1716 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
1717 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
1718 _mm512_mask_cmpunord_ps_mask): New intrinsics.
1720 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
1722 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
1724 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
1725 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
1726 warning to rs6000.c. Remove the Undocumented flag, since it has
1728 (-mabi=ibmlongdouble): Likewise.
1729 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
1730 already set, set the default format for long double.
1731 (rs6000_debug_reg_global): Print whether long double is IBM or
1733 (rs6000_option_override_internal): Rework setting long double
1734 format. Only warn if the user is changing the long double default
1735 and they did not use -Wno-psabi.
1736 * doc/invoke.texi (PowerPC options): Update the documentation for
1737 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
1739 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
1740 Alan Hayward <alan.hayward@arm.com>
1741 David Sherwood <david.sherwood@arm.com>
1743 * rtl.h (wider_subreg_mode): New function.
1744 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
1745 rather than an unsigned int *.
1746 * ira-color.c (regno_max_ref_width): Replace with...
1747 (regno_max_ref_mode): ...this new variable.
1748 (coalesced_pseudo_reg_slot_compare): Update accordingly.
1749 Use wider_subreg_mode.
1750 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
1751 rather than an unsigned int *.
1752 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
1753 (process_alt_operands): Likewise.
1754 (invariant_p): Likewise.
1755 * lra-spills.c (assign_mem_slot): Likewise.
1756 (add_pseudo_to_slot): Likewise.
1757 * lra.c (collect_non_operand_hard_regs): Likewise.
1758 (add_regs_to_insn_regno_info): Likewise.
1759 * reload1.c (regno_max_ref_width): Replace with...
1760 (regno_max_ref_mode): ...this new variable.
1761 (reload): Update accordingly. Update call to
1762 ira_sort_regnos_for_alter_reg.
1763 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
1764 (init_eliminable_invariants): Update to use regno_max_ref_mode.
1765 (scan_paradoxical_subregs): Likewise.
1767 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
1769 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
1770 (aarch64_frame): Add emit_frame_chain boolean.
1771 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1772 Move eh_return case to aarch64_layout_frame.
1773 (aarch64_layout_frame): Initialize emit_frame_chain.
1774 (aarch64_expand_prologue): Use emit_frame_chain.
1776 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
1778 * config/aarch64/aarch64.c (aarch64_layout_frame):
1779 Ensure LR is always stored at the bottom of the callee-saves.
1780 Remove rarely used frame layout which saves callee-saves at top of
1781 frame, so the store of LR can be used as a valid probe in all cases.
1783 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
1785 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
1786 Improve unaligned TImode/TFmode base/offset split.
1788 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
1789 Alan Hayward <alan.hayward@arm.com>
1790 David Sherwood <david.sherwood@arm.com>
1792 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
1793 * combine.c (find_single_use_1): Likewise.
1794 (expand_field_assignment): Likewise.
1795 (move_deaths): Likewise.
1796 * lra-constraints.c (simplify_operand_subreg): Likewise.
1797 (curr_insn_transform): Likewise.
1798 * lra.c (collect_non_operand_hard_regs): Likewise.
1799 (add_regs_to_insn_regno_info): Likewise.
1800 * rtlanal.c (reg_referenced_p): Likewise.
1801 (covers_regno_no_parallel_p): Likewise.
1803 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
1805 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
1806 Don't print any bits outside the precision of the value.
1807 * wide-int.cc (test_printing): Add some new tests.
1809 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1811 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
1812 supports -xbrace_comment option.
1813 * configure: Regenerate.
1814 * config.in: Regenerate.
1815 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
1816 (ASM_CPU_SPEC): Use it.
1818 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
1820 * target.def (static_rtx_alignment): New hook.
1821 * targhooks.h (default_static_rtx_alignment): Declare.
1822 * targhooks.c (default_static_rtx_alignment): New function.
1823 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
1824 * doc/tm.texi: Regenerate.
1825 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
1826 instead of targetm.constant_alignment. Remove call to
1828 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
1829 (cris_preferred_mininum_alignment): New function, split out from...
1830 (cris_constant_alignment): ...here.
1831 (cris_static_rtx_alignment): New function.
1832 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
1834 (ix86_constant_alignment): ...here.
1835 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
1836 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
1837 (mmix_static_rtx_alignment): New function.
1838 * config/spu/spu.c (spu_static_rtx_alignment): New function.
1839 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
1841 2017-10-26 Tamar Christina <tamar.christina@arm.com>
1844 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
1845 Add flag_trapping_math and flag_fp_int_builtin_inexact.
1847 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
1849 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
1850 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
1851 mark as a sign-extending load.
1852 (local_pic_load_u): Define.
1854 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
1857 * fold-const.c (operand_equal_for_comparison_p): Also return true
1858 if ARG0 is a simple variant of ARG1 with narrower precision.
1859 (fold_ternary_loc): Always pass unstripped operands to the predicate.
1861 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
1863 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
1865 * i386.h (processor_costs): Add gather_static, gather_per_elt,
1866 scatter_static, scatter_per_elt.
1867 * x86-tune-costs.h: Add new cost entries.
1869 2017-10-25 Richard Biener <rguenther@suse.de>
1871 * tree-ssa-sccvn.h (vn_eliminate): Declare.
1872 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
1873 class pass_fre): Move to ...
1874 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
1875 class pass_fre): ... here and adjust for statistics.
1877 2017-10-25 Jakub Jelinek <jakub@redhat.com>
1880 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
1881 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
1882 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
1883 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
1886 2017-10-25 Richard Biener <rguenther@suse.de>
1888 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
1889 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
1890 eliminate_push_avail, eliminate_insert): Move inside...
1891 (class eliminate_dom_walker): ... this class in preparation
1893 (fini_eliminate): Remove by merging with ...
1894 (eliminate): ... this function. Adjust for class changes.
1895 (pass_pre::execute): Remove fini_eliminate call.
1896 (pass_fre::execute): Likewise.
1898 2017-10-24 Jakub Jelinek <jakub@redhat.com>
1901 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
1902 (VPERMI2, VPERMI2I): New mode iterators.
1903 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
1904 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
1906 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
1907 mode iterator. Remove 3 old define_insn patterns.
1908 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
1909 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
1910 VPERMI2 mode iterator, remove the other two expanders.
1911 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
1912 to use VPERMI2 mode iterator, add another alternative for vpermi2*
1913 instructions, remove the other two patterns.
1914 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
1915 mode iterator, remove the other two patterns.
1916 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
1917 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
1918 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
1919 and adjust argument order accordingly.
1920 (ix86_expand_vec_perm): Adjust caller.
1921 (expand_vec_perm_1): Likewise.
1922 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
1923 (expand_vec_perm_vpermt2_vpshub2): ... this.
1924 (ix86_expand_vec_perm_const_1): Adjust caller.
1925 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
1928 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
1929 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
1930 (vec_shl_<mode>): Remove unused expander.
1931 (avx512bw_<shift_insn><mode>3): New define_insn.
1932 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
1933 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
1935 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
1938 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
1941 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
1943 PR rtl-optimization/82396
1944 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
1945 (autopref_multipass_init): Simplify initialization.
1946 (autopref_rank_data): Simplify sort order.
1947 * gcc/sched-int.h (autopref_multipass_data_): Remove
1948 multi_mem_insn_p, min_offset and max_offset.
1950 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
1953 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1954 Check special value of flag_omit_frame_pointer.
1955 (aarch64_can_eliminate): Likewise.
1956 (aarch64_override_options_after_change_1): Simplify handling of
1957 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
1959 2017-10-24 Richard Biener <rguenther@suse.de>
1961 PR tree-optimization/82697
1962 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
1963 zero for conditional load and unconditional store.
1965 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
1967 * doc/install.texi: Document bootstrap-cet.
1969 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
1972 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
1973 ENDBR instruction at function entrance if function is only
1976 2017-10-24 Jakub Jelinek <jakub@redhat.com>
1979 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
1980 patterns to better describe from which operation the CF is computed.
1981 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
1982 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
1983 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
1984 is 0, use _0 suffixed expanders instead of emitting a comparison
1987 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
1989 * config/i386/i386.md(*movsf_internal, *movdf_internal):
1990 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
1992 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
1995 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
1996 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
1997 * loop-iv.c (iv_get_reaching_def): Likewise.
1998 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
1999 variable is promoted and the partition contains undefined values.
2001 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2003 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
2005 (nios2_address_cost): Define.
2006 (nios2_legitimize_address): Recognize (exp + constant) directly.
2007 (TARGET_ADDRESS_COST): Define.
2009 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2011 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
2012 (nios2_symbolic_memory_operand_p): Declare.
2013 (nios2_split_large_constant): Declare.
2014 (nios2_split_symbolic_memory_operand): Declare.
2015 * config/nios2/nios2.c: Adjust includes.
2016 (nios2_symbolic_constant_allowed): New.
2017 (nios2_symbolic_constant_p): New.
2018 (nios2_plus_symbolic_constant_p): New.
2019 (nios2_valid_addr_expr_p): Recognize addresses involving
2021 (nios2_legitimate_address_p): Likewise, also LO_SUM.
2022 (nios2_symbolic_memory_operand_p): New.
2023 (nios2_large_constant_p): New.
2024 (nios2_split_large_constant): New.
2025 (nios2_split_plus_large_constant): New.
2026 (nios2_split_symbolic_memory_operand): New.
2027 (nios2_legitimize_address): Code refactoring. Handle addresses
2028 involving symbolic constants.
2029 (nios2_emit_move_sequence): Likewise.
2030 (nios2_print_operand): Improve error output.
2031 (nios2_print_operand_address): Handle LO_SUM.
2032 (nios2_cdx_narrow_form_p): Likewise.
2033 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
2034 operands involving symbolic constants.
2035 (movhi_internal, movsi_internal): Likewise.
2036 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
2037 (extendhisi2, extendqi<mode>2): Likewise.
2039 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2041 * tree-pass.h (PROP_rtl_split_insns): Define.
2042 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
2044 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2046 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
2048 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2051 * target.def (const_not_ok_for_debug_p): Default to
2052 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
2053 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
2054 * targhooks.c (default_const_not_ok_for_debug_p): New function.
2055 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
2056 which targetm.const_not_ok_for_debug_p returned true.
2057 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
2059 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
2061 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
2062 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
2063 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
2065 (ix86_const_not_ok_for_debug_p): New function.
2066 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
2067 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
2069 2017-10-23 David Malcolm <dmalcolm@redhat.com>
2072 * system.h: Conditionally include "unique-ptr.h" if
2073 INCLUDE_UNIQUE_PTR is defined.
2074 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
2075 of defining INCLUDE_UNIQUE_PTR before including "system.h".
2077 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
2079 * config/rl78/rl78.md: New define_expand "subdi3".
2081 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
2084 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
2085 DF_REF_INSN if DF_REF_INSN_INFO is false.
2087 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
2089 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
2090 xmm_move instead of sse_move.
2091 (sse_store_index): New function.
2092 (ix86_register_move_cost): Be more sensible about mismatch stall;
2093 model AVX moves correctly; make difference between sse->integer and
2095 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
2096 moves; make difference between SSE and AVX.
2097 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
2098 and zmm_move. Increase size of sse load and store tables;
2099 add unaligned load and store tables; add ssemmx_to_integer.
2100 * x86-tune-costs.h: Update all entries according to real
2101 move latencies from Agner Fog's manual and chip documentation.
2103 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2106 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
2107 * config/i386/constraints.md (Wf): New constraint.
2108 * config/i386/i386.md (UNSPEC_SBB): New unspec.
2109 (cmp<dwi>_doubleword): Removed.
2110 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
2111 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
2112 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
2113 expand with cmp<dwi>_doubleword. For LTU and GEU use
2114 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
2116 * common.opt (gcolumn-info): Enable by default.
2117 * doc/invoke.texi (gcolumn-info): Document new default.
2119 2017-10-23 Richard Biener <rguenther@suse.de>
2121 PR tree-optimization/82672
2122 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
2123 Fold the stmt if we propagated into it.
2125 2017-10-23 Richard Biener <rguenther@suse.de>
2127 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
2128 (bitmap_remove_expr_from_set): ... this. All callers call this
2129 for non-constant values.
2130 (bitmap_set_subtract): Rename to...
2131 (bitmap_set_subtract_expressions): ... this. Adjust and
2133 (bitmap_set_contains_value): Remove superfluous check.
2134 (bitmap_set_replace_value): Inline into single caller ...
2135 (bitmap_value_replace_in_set): ... here and simplify.
2136 (dependent_clean): Merge into ...
2137 (clean): ... this using an overload. Adjust.
2138 (prune_clobbered_mems): Adjust.
2139 (compute_antic_aux): Likewise.
2140 (compute_partial_antic_aux): Likewise.
2142 2017-10-23 Richard Biener <rguenther@suse.de>
2144 PR tree-optimization/82129
2146 2017-08-01 Richard Biener <rguenther@suse.de>
2148 PR tree-optimization/81181
2149 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2150 (compute_antic): ... end of iteration here.
2152 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2154 * target.def (starting_frame_offset): New hook.
2155 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
2156 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
2157 * doc/tm.texi.in: Regenerate.
2158 * hooks.h (hook_hwi_void_0): Declare.
2159 * hooks.c (hook_hwi_void_0): New function.
2160 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
2161 STARTING_FRAME_OFFSET.
2162 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
2163 * reload1.c (reload): Likewise.
2164 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
2165 instead of STARTING_FRAME_OFFSET.
2166 * function.c (try_fit_stack_local): Likewise.
2167 (assign_stack_local_1): Likewise
2168 (instantiate_virtual_regs): Likewise.
2169 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
2170 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
2171 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
2172 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
2173 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
2174 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
2175 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
2176 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
2177 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
2178 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
2179 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
2180 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
2181 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
2182 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
2183 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
2184 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
2185 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
2186 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
2187 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
2188 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
2189 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
2190 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
2191 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
2192 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
2193 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
2194 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
2195 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
2196 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
2197 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
2198 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
2199 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
2200 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
2201 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
2202 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
2203 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
2204 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
2205 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
2206 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
2207 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
2208 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
2209 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
2210 * config/avr/avr.c (avr_starting_frame_offset): Make static and
2211 return a HOST_WIDE_INT.
2212 (avr_builtin_setjmp_frame_value): Use it instead of
2213 STARTING_FRAME_OFFSET.
2214 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2215 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
2216 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
2218 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2219 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
2220 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
2221 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2222 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
2223 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
2224 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2225 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
2226 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
2227 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2228 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
2229 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
2231 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2232 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
2233 * config/mips/mips.c (mips_compute_frame_info): Refer to
2234 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2235 (mips_starting_frame_offset): New function.
2236 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2237 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
2238 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
2239 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
2240 and return a HOST_WIDE_INT.
2241 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2242 (mmix_initial_elimination_offset): Refer to
2243 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2244 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
2245 * config/pa/pa.c (pa_starting_frame_offset): New function.
2246 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
2247 (pa_expand_prologue): Likewise.
2248 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2249 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
2250 !FRAME_GROWS_DOWNWARD handling to...
2251 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2252 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
2253 !FRAME_GROWS_DOWNWARD handling to...
2254 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2255 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
2256 !FRAME_GROWS_DOWNWARD handling to...
2257 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2258 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
2260 (rs6000_starting_frame_offset): New function.
2261 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
2262 !FRAME_GROWS_DOWNWARD handling to...
2263 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2264 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
2265 !FRAME_GROWS_DOWNWARD handling to...
2266 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2267 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
2268 !FRAME_GROWS_DOWNWARD handling to...
2269 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2270 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
2271 (rs6000_starting_frame_offset): New function.
2272 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
2273 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
2274 * config/vax/vax.c (vax_starting_frame_offset): New function.
2275 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
2276 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2277 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
2278 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
2279 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2280 * system.h (STARTING_FRAME_OFFSET): Poison.
2282 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2284 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
2285 SCALAR_TYPE_MODE instead of TYPE_MODE.
2287 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2288 Alan Hayward <alan.hayward@arm.com>
2289 David Sherwood <david.sherwood@arm.com>
2291 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
2293 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2294 Alan Hayward <alan.hayward@arm.com>
2295 David Sherwood <david.sherwood@arm.com>
2297 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
2299 2017-10-23 Richard Biener <rguenther@suse.de>
2301 PR tree-optimization/82129
2302 * tree-ssa-pre.c (bitmap_set_and): Remove.
2303 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
2304 canonicalizing expressions in the set to those with lowest
2305 ID rather than taking that from the first edge.
2307 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2309 * combine.c (rtx_equal_for_field_assignment_p): Use
2310 byte_lowpart_offset.
2312 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2313 Alan Hayward <alan.hayward@arm.com>
2314 David Sherwood <david.sherwood@arm.com>
2316 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
2317 to a SUBREG_PROMOTED_VAR.
2319 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2320 Alan Hayward <alan.hayward@arm.com>
2321 David Sherwood <david.sherwood@arm.com>
2323 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
2324 (expand_debug_source_expr): Likewise.
2325 * combine.c (combine_simplify_rtx): Likewise.
2326 * cse.c (fold_rtx): Likewise.
2327 * optabs.c (expand_float): Likewise.
2328 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2329 (simplify_binary_operation_1): Likewise.
2331 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2332 Alan Hayward <alan.hayward@arm.com>
2333 David Sherwood <david.sherwood@arm.com>
2335 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
2336 (record_promoted_value): Likewise.
2337 * expr.c (expand_expr_real_2): Likewise.
2338 * ree.c (update_reg_equal_equiv_notes): Likewise.
2339 (combine_set_extension): Likewise.
2340 * rtlanal.c (low_bitmask_len): Likewise.
2341 * simplify-rtx.c (neg_const_int): Likewise.
2342 (simplify_binary_operation_1): Likewise.
2344 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2345 Alan Hayward <alan.hayward@arm.com>
2346 David Sherwood <david.sherwood@arm.com>
2348 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
2349 * regcprop.c (maybe_mode_change): Likewise.
2350 * reload1.c (alter_reg): Likewise.
2352 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2354 * inchash.h (inchash::hash::add_wide_int): New function.
2355 * lto-streamer-out.c (hash_tree): Use it.
2357 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2359 * inchash.h (inchash::hash::add_wide_int): Rename to...
2360 (inchash::hash::add_hwi): ...this.
2361 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
2362 (polymorphic_call_target_hasher::hash): Likewise.
2363 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
2364 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
2365 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
2366 * lto-streamer-out.c (hash_tree): Likewise.
2367 * optc-save-gen.awk: Likewise.
2368 * tree.c (add_expr): Likewise.
2370 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
2373 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
2374 for ordered inequality comparisons even with TARGET_IEEE_FP.
2376 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
2379 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
2380 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
2381 Expand with cmp<dwi>_doubleword.
2383 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2385 * extend.texi: Add x86 specific to 'nocf_check' attribute.
2386 List CET intrinsics.
2387 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
2388 specific to -fcf-protection option.
2390 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2392 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
2393 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
2394 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
2395 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
2396 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
2397 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
2398 (extra_objs): Add cet.o for Linux/x86 targets.
2399 (tmake_file): Add i386/t-cet for Linux/x86 targets.
2400 * config/i386/cet.c: New file.
2401 * config/i386/cetintrin.h: Likewise.
2402 * config/i386/t-cet: Likewise.
2403 * config/i386/cpuid.h (bit_SHSTK): New.
2404 (bit_IBT): Likewise.
2405 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
2406 pass IBT and SHSTK bits.
2407 * config/i386/i386-builtin-types.def
2408 (VOID_FTYPE_UNSIGNED_PVOID): New.
2409 (VOID_FTYPE_UINT64_PVOID): Likewise.
2410 * config/i386/i386-builtin.def: Add CET intrinsics.
2411 * config/i386/i386-c.c (ix86_target_macros_internal): Add
2412 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
2413 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
2414 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
2416 * config/i386/i386.c (rest_of_insert_endbranch): New.
2417 (pass_data_insert_endbranch): Likewise.
2418 (pass_insert_endbranch): Likewise.
2419 (make_pass_insert_endbranch): Likewise.
2420 (ix86_notrack_prefixed_insn_p): Likewise.
2421 (ix86_target_string): Add -mibt, -mshstk flags.
2422 (ix86_option_override_internal): Add flag_cf_protection
2424 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
2425 (ix86_print_operand): Add 'notrack' prefix output.
2426 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
2427 (ix86_expand_builtin): Expand CET intrinsics.
2428 (x86_output_mi_thunk): Add 'endbranch' instruction.
2429 * config/i386/i386.h (TARGET_IBT): New.
2430 (TARGET_IBT_P): Likewise.
2431 (TARGET_SHSTK): Likewise.
2432 (TARGET_SHSTK_P): Likewise.
2433 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
2434 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
2435 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
2436 (builtin_setjmp_setup): New pattern.
2437 (builtin_longjmp): Likewise.
2438 (rdssp<mode>): Likewise.
2439 (incssp<mode>): Likewise.
2440 (saveprevssp): Likewise.
2441 (rstorssp): Likewise.
2442 (wrss<mode>): Likewise.
2443 (wruss<mode>): Likewise.
2444 (setssbsy): Likewise.
2445 (clrssbsy): Likewise.
2446 (nop_endbr): Likewise.
2447 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
2449 * config/i386/immintrin.h: Include <cetintrin.h>.
2450 * config/i386/linux-common.h
2451 (file_end_indicate_exec_stack_and_cet): New prototype.
2452 (TARGET_ASM_FILE_END): New.
2454 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
2456 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
2457 latencies instead of having separate table; make difference between
2458 integer and float costs.
2459 * i386.h (processor_costs): Remove scalar_stmt_cost,
2460 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
2461 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
2463 * x86-tune-costs.h: Remove entries which has been removed in
2464 procesor_costs from all tables; make cond_taken_branch_cost
2465 and cond_not_taken_branch_cost COST_N_INSNS based.
2467 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
2469 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
2471 2017-10-20 Jakub Jelinek <jakub@redhat.com>
2473 * config/i386/i386.md (isa): Remove fma_avx512f.
2474 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
2475 <avx512>_fmadd_<mode>_mask3<round_name>,
2476 <avx512>_fmsub_<mode>_mask<round_name>,
2477 <avx512>_fmsub_<mode>_mask3<round_name>,
2478 <avx512>_fnmadd_<mode>_mask<round_name>,
2479 <avx512>_fnmadd_<mode>_mask3<round_name>,
2480 <avx512>_fnmsub_<mode>_mask<round_name>,
2481 <avx512>_fnmsub_<mode>_mask3<round_name>,
2482 <avx512>_fmaddsub_<mode>_mask<round_name>,
2483 <avx512>_fmaddsub_<mode>_mask3<round_name>,
2484 <avx512>_fmsubadd_<mode>_mask<round_name>,
2485 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
2486 (*vec_widen_umult_even_v16si<mask_name>,
2487 *vec_widen_smult_even_v16si<mask_name>): Likewise.
2488 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
2490 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2492 * extend.texi: Add 'nocf_check' documentation.
2493 * gimple.texi: Add second parameter to
2494 gimple_build_call_from_tree.
2495 * invoke.texi: Add -fcf-protection documentation.
2496 * rtl.texi: Add REG_CALL_NOTRACK documenation.
2498 2017-10-20 Richard Biener <rguenther@suse.de>
2500 PR tree-optimization/82473
2501 * tree-vect-loop.c (vectorizable_reduction): Properly get at
2502 the largest input type.
2504 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2506 * c-attribs.c (handle_nocf_check_attribute): New function.
2507 (c_common_attribute_table): Add 'nocf_check' handling.
2508 * gimple-parser.c: Add second argument NULL to
2509 gimple_build_call_from_tree.
2510 * attrib.c (comp_type_attributes): Check nocf_check attribute.
2511 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
2513 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
2514 * common.opt: Add fcf-protection flag.
2515 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
2516 * flag-types.h: Add enum cf_protection_level.
2517 * gimple.c (gimple_build_call_from_tree): Add second parameter.
2518 Add 'nocf_check' attribute propagation to gimple call.
2519 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
2520 (gimple_build_call_from_tree): Update prototype.
2521 (gimple_call_nocf_check_p): New function.
2522 (gimple_call_set_nocf_check): Likewise.
2523 * gimplify.c: Add second argument to gimple_build_call_from_tree.
2524 * ipa-icf.c: Add nocf_check attribute in statement hash.
2525 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
2526 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
2527 * toplev.c (process_options): Add flag_cf_protection handling.
2529 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
2531 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
2533 2017-10-20 Richard Biener <rguenther@suse.de>
2535 PR tree-optimization/82603
2536 * tree-if-conv.c (predicate_mem_writes): Make sure to only
2537 remove false predicated stores.
2539 2017-10-20 Richard Biener <rguenther@suse.de>
2541 * graphite-isl-ast-to-gimple.c
2542 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
2543 Remove return value and simplify, dump copied stmt after lhs
2545 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
2546 Reduce dump verbosity.
2547 (gsi_insert_earliest): Likewise.
2548 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
2549 * graphite.c (print_global_statistics): Adjust dumping.
2550 (print_graphite_scop_statistics): Likewise.
2551 (print_graphite_statistics): Do not dump loops here.
2552 (graphite_transform_loops): But here.
2554 2017-10-20 Nicolas Roche <roche@adacore.com>
2556 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
2557 * configure: Regenerate.
2559 2017-10-20 Jakub Jelinek <jakub@redhat.com>
2562 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
2563 functions when optimizing replace GIMPLE_RETURN stmts with
2564 calls to __builtin_unreachable ().
2567 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
2568 for -fsanitize=thread link of executables.
2569 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
2570 link of executables.
2573 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
2575 (<shift_insn><mode>3<mask_name>): Change the last of the 3
2576 define_insns for logical vector shifts to use VI248_AVX512BW
2577 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
2578 condition, useless isa and prefix attributes. Change the first
2579 2 of these define_insns to ...
2580 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
2581 define_insn for avx512vl.
2582 (<shift_insn><mode>3): ... and this, new define_insn without
2583 masking for non-avx512vl.
2586 * config/i386/sse.md (*andnot<mode>3,
2587 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
2588 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
2589 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
2590 not applied use empty suffix even for TARGET_AVX512VL.
2591 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
2592 is applied, supply evex,evex or evex,evex,evex instead of just
2595 2017-10-20 Julia Koval <julia.koval@intel.com>
2597 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
2598 (OPTION_MASK_ISA_GFNI_UNSET): New.
2599 (ix86_handle_option): Handle OPT_mgfni.
2600 * config/i386/cpuid.h (bit_GFNI): New.
2601 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
2602 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
2603 * config/i386/i386.c (ix86_target_string): Add -mgfni.
2604 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
2605 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
2606 * config/i386/i386.opt: Add mgfni.
2608 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
2610 * config/msp430/msp430.c (msp430_option_override): Disable
2611 -fdelete-null-pointer-checks.
2612 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
2614 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
2616 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
2617 of x87 and SSE instructions.
2619 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
2621 * asan.c (create_cond_insert_point): Do not update edge count.
2622 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
2623 (afdo_propagate_circuit): Likewise.
2624 (afdo_calculate_branch_prob): Likewise.
2625 (afdo_annotate_cfg): Likewise.
2626 * basic-block.h (struct edge_def): Remove count.
2627 (edge_def::count): New accessor.
2628 * bb-reorder.c (rotate_loop): Update.
2629 (find_traces_1_round): Update.
2630 (connect_traces): Update.
2631 (sanitize_hot_paths): Update.
2632 * cfg.c (unchecked_make_edge): Update.
2633 (make_single_succ_edge): Update.
2634 (check_bb_profile): Update.
2635 (dump_edge_info): Update.
2636 (update_bb_profile_for_threading): Update.
2637 (scale_bbs_frequencies_int): Update.
2638 (scale_bbs_frequencies_gcov_type): Update.
2639 (scale_bbs_frequencies_profile_count): Update.
2640 (scale_bbs_frequencies): Update.
2641 * cfganal.c (connect_infinite_loops_to_exit): Update.
2642 * cfgbuild.c (compute_outgoing_frequencies): Update.
2643 (find_many_sub_basic_blocks): Update.
2644 * cfgcleanup.c (try_forward_edges): Update.
2645 (try_crossjump_to_edge): Update
2646 * cfgexpand.c (expand_gimple_cond): Update
2647 (expand_gimple_tailcall): Update
2648 (construct_exit_block): Update
2649 * cfghooks.c (verify_flow_info): Update
2650 (redirect_edge_succ_nodup): Update
2651 (split_edge): Update
2652 (make_forwarder_block): Update
2653 (duplicate_block): Update
2654 (account_profile_record): Update
2655 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
2656 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
2657 * cfgloopmanip.c (scale_loop_profile): Update.
2659 (lv_adjust_loop_entry_edge): Update.
2660 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
2661 (force_nonfallthru_and_redirect): Update.
2662 (purge_dead_edges): Update.
2663 (rtl_flow_call_edges_add): Update.
2664 * cgraphunit.c (init_lowered_empty_function): Update.
2665 (cgraph_node::expand_thunk): Update.
2666 * gimple-pretty-print.c (dump_probability): Update.
2667 (dump_edge_probability): Update.
2668 * gimple-ssa-isolate-paths.c (isolate_path): Update.
2669 * haifa-sched.c (sched_create_recovery_edges): Update.
2670 * hsa-gen.c (convert_switch_statements): Update.
2671 * ifcvt.c (dead_or_predicable): Update.
2672 * ipa-inline-transform.c (inline_transform): Update.
2673 * ipa-split.c (split_function): Update.
2674 * ipa-utils.c (ipa_merge_profiles): Update.
2675 * loop-doloop.c (add_test): Update.
2676 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
2677 * lto-streamer-in.c (input_cfg): Update.
2678 (input_function): Update.
2679 * lto-streamer-out.c (output_cfg): Update.
2680 * modulo-sched.c (sms_schedule): Update.
2681 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
2682 * predict.c (maybe_hot_edge_p): Update.
2683 (unlikely_executed_edge_p): Update.
2684 (probably_never_executed_edge_p): Update.
2685 (dump_prediction): Update.
2686 (drop_profile): Update.
2687 (propagate_unlikely_bbs_forward): Update.
2688 (determine_unlikely_bbs): Update.
2689 (force_edge_cold): Update.
2690 * profile.c (compute_branch_probabilities): Update.
2691 * reg-stack.c (better_edge): Update.
2692 * shrink-wrap.c (handle_simple_exit): Update.
2693 * tracer.c (better_p): Update.
2694 * trans-mem.c (expand_transaction): Update.
2695 (split_bb_make_tm_edge): Update.
2696 * tree-call-cdce.c: Update.
2697 * tree-cfg.c (gimple_find_sub_bbs): Update.
2698 (gimple_split_edge): Update.
2699 (gimple_duplicate_sese_region): Update.
2700 (gimple_duplicate_sese_tail): Update.
2701 (gimple_flow_call_edges_add): Update.
2702 (insert_cond_bb): Update.
2703 (execute_fixup_cfg): Update.
2704 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
2705 * tree-complex.c (expand_complex_div_wide): Update.
2706 * tree-eh.c (lower_resx): Update.
2707 (unsplit_eh): Update.
2708 (cleanup_empty_eh_move_lp): Update.
2709 * tree-inline.c (copy_edges_for_bb): Update.
2710 (freqs_to_counts): Update.
2711 (copy_cfg_body): Update.
2712 * tree-ssa-dce.c (remove_dead_stmt): Update.
2713 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
2714 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
2715 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
2716 (unloop_loops): Update.
2717 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
2718 * tree-ssa-loop-split.c (connect_loops): Update.
2719 (split_loop): Update.
2720 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
2721 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
2722 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
2723 * tree-ssa-reassoc.c (branch_fixup): Update.
2724 * tree-ssa-tail-merge.c (replace_block_by): Update.
2725 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
2726 (compute_path_counts): Update.
2727 (update_profile): Update.
2728 (recompute_probabilities): Update.
2729 (update_joiner_offpath_counts): Update.
2730 (estimated_freqs_path): Update.
2731 (freqs_to_counts_path): Update.
2732 (clear_counts_path): Update.
2733 (ssa_fix_duplicate_block_edges): Update.
2734 (duplicate_thread_path): Update.
2735 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
2736 (case_bit_test_cmp): Update.
2737 (collect_switch_conv_info): Update.
2738 (gen_inbound_check): Update.
2739 (do_jump_if_equal): Update.
2740 (emit_cmp_and_jump_insns): Update.
2741 * tree-tailcall.c (decrease_profile): Update.
2742 (eliminate_tail_call): Update.
2743 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
2744 (vect_do_peeling): Update.
2745 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
2746 * ubsan.c (ubsan_expand_null_ifn): Update.
2747 (ubsan_expand_ptr_ifn): Update.
2748 * value-prof.c (gimple_divmod_fixed_value): Update.
2749 (gimple_mod_pow2): Update.
2750 (gimple_mod_subtract): Update.
2751 (gimple_ic): Update.
2752 (gimple_stringop_fixed_value): Update.
2754 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
2757 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
2759 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
2761 PR rtl-optimization/82395
2762 * ira-color.c (allocno_priority_compare_func): Fix comparison step
2763 based on non_spilled_static_chain_regno_p.
2765 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
2767 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
2768 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
2769 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
2771 2017-10-19 Martin Sebor <msebor@redhat.com>
2773 PR tree-optimization/82596
2774 * tree.c (array_at_struct_end_p): Handle STRING_CST.
2776 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
2778 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
2779 (get_mem_refs_of_builtin_call): Likewise.
2780 * builtins.c (expand_builtin_apply): Adjust call to
2781 allocate_dynamic_stack_space.
2782 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
2783 the third argument to allocate_dynamic_stack_space, otherwise -1.
2784 (expand_builtin): Deal with all alloca variants.
2785 (is_inexpensive_builtin): Likewise.
2786 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
2787 * calls.c (special_function_p): Deal with all alloca variants.
2788 (initialize_argument_information): Adjust call to
2789 allocate_dynamic_stack_space.
2790 (expand_call): Likewise.
2791 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
2792 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
2793 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
2794 use it for the stack usage computation.
2795 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
2796 * function.c (gimplify_parameters): Call build_alloca_call_expr.
2797 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
2798 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
2799 (in_loop_p): Remove first argument and useless check.
2800 (pass_walloca::execute): Remove useless test and adjust call to above.
2801 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
2802 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
2803 (gimplify_call_expr): Deal with all alloca variants.
2804 * hsa-gen.c (gen_hsa_alloca): Likewise.
2805 (gen_hsa_insns_for_call): Likewise.
2806 * ipa-pure-const.c (special_builtin_state): Likewise.
2807 * tree-chkp.c (chkp_build_returned_bound): Likewise.
2808 * tree-object-size.c (alloc_object_size): Likewise.
2809 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
2810 (call_may_clobber_ref_p_1): Likewise.
2811 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
2812 (ccp_fold_stmt): Likewise.
2813 (optimize_stack_restore): Likewise.
2814 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
2815 (mark_all_reaching_defs_necessary_1): Likewise.
2816 (propagate_necessity): Likewise.
2817 (eliminate_unnecessary_stmts): Likewise.
2818 * tree.c (build_common_builtin_nodes): Build
2819 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
2820 (build_alloca_call_expr): New function.
2821 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
2822 (CASE_BUILT_IN_ALLOCA): Likewise.
2823 (build_alloca_call_expr): Declare.
2824 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
2826 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
2829 * dwarf2out.c (new_die_raw): New static inline function.
2830 (new_die): Use it to create the DIE.
2831 (add_AT_external_die_ref): Likewise.
2832 (clone_die): Likewise.
2833 (clone_as_declaration): Likewise.
2834 (dwarf2out_vms_debug_main_pointer): Likewise.
2835 (base_type_die): Likewise. Remove early return for corner cases.
2836 Do not call add_pubtype on the DIE here.
2837 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
2838 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
2839 typedefs for base types with DW_AT_endianity. Make sure a DIE with
2840 native order exists for base types, attach the DIE manually and call
2841 add_pubtype on it. Do not equate a reverse order DIE to the type.
2843 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
2845 * config/arm/arm.c (align_ok_ldrd_strd): New function.
2846 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
2848 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
2850 2017-10-19 Jakub Jelinek <jakub@redhat.com>
2852 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
2853 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
2854 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
2855 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
2856 * opts.c (sanitizer_opts): Add builtin.
2857 * ubsan.c (instrument_builtin): New function.
2858 (pass_ubsan::execute): Call it.
2859 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
2860 * doc/invoke.texi: Document -fsanitize=builtin.
2862 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
2863 builtins, store max (log2 (align), 0) into uchar field instead of
2864 align into uptr field.
2865 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
2866 store uchar 0 field instead of uptr 0 field.
2867 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
2868 instead of passing one address of struct with 2 locations pass
2869 two addresses of structs with 1 location each.
2870 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
2871 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
2872 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
2873 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
2874 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
2875 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
2876 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
2877 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
2879 2017-10-19 Martin Liska <mliska@suse.cz>
2882 * file-find.c (remove_prefix): Remove.
2883 * file-find.h (remove_prefix): Likewise.
2884 * gcc-ar.c: Remove smartness of lookup.
2886 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
2888 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
2889 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
2890 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
2892 2017-10-19 Jakub Jelinek <jakub@redhat.com>
2895 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
2896 (setcc + and to xor + setcc): New peephole2.
2898 2017-10-19 Tom de Vries <tom@codesourcery.com>
2900 * doc/sourcebuild.texi (Test Directives, Variants of
2901 dg-require-support): Add dg-require-stack-size.
2903 2017-10-19 Martin Liska <mliska@suse.cz>
2906 * gimplify.c (gimplify_decl_expr): Do not instrument variables
2907 that have a large alignment.
2908 (gimplify_target_expr): Likewise.
2910 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
2912 PR rtl-optimization/82602
2913 * ira.c (rtx_moveable_p): Return false for volatile asm.
2915 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
2918 * config/i386/i386-modes.def (CCGZ): New CC mode.
2919 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
2920 * config/i386/predicates.md (ix86_comparison_operator):
2922 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
2923 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
2924 with double-word subtraction.
2925 (put_condition_code): Handle CCGZmode.
2927 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
2929 * wide-int.cc (debug (const wide_int &)): New.
2930 (debug (const wide_int *)): New.
2931 (debug (const widest_int &)): New.
2932 (debug (const widest_int *)): New.
2934 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
2937 * lra-constraints.c (curr_insn_transform): Use non-input operand
2938 instead of output one for matched reload.
2940 2017-10-18 Bin Cheng <bin.cheng@arm.com>
2942 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
2943 (tree-ssa-loop-ivopts.h): New header file.
2944 (struct builtin_info): New fields.
2945 (classify_builtin_1): Compute and record base and offset parts for
2946 memset builtin partition by calling strip_offset.
2947 (offset_cmp, fuse_memset_builtins): New functions.
2948 (finalize_partitions): Fuse adjacent memset partitions by calling
2950 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
2951 Expose the interface.
2952 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
2954 2017-10-18 Bin Cheng <bin.cheng@arm.com>
2956 PR tree-optimization/82574
2957 * tree-loop-distribution.c (find_single_drs): New parameter. Check
2958 that data reference must be executed exactly once per iteration
2959 against the outermost loop in nest.
2960 (classify_partition): Update call to above function.
2962 2017-10-18 Richard Biener <rguenther@suse.de>
2964 PR tree-optimization/82591
2965 * graphite.c (graphite_transform_loops): Move code gen message
2967 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
2968 Here. Handle scop_to_isl_ast failing.
2969 (scop_to_isl_ast): Limit the number of ISL operations.
2971 2017-10-18 Richard Biener <rguenther@suse.de>
2973 * graphite-isl-ast-to-gimple.c
2974 (translate_isl_ast_to_gimple::set_rename): Simplify.
2975 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
2976 (graphite_copy_stmts_from_block): ... here.
2977 (copy_bb_and_scalar_dependences): Simplify.
2978 (add_parameters_to_ivs_params): Canonicalize.
2979 (generate_entry_out_of_ssa_copies): Simplify.
2980 * graphite-sese-to-poly.c (extract_affine_name): Simplify
2981 by passing in ISL dimension.
2982 (parameter_index_in_region_1): Rename to ...
2983 (parameter_index_in_region): ... this.
2984 (extract_affine): Adjust assert, pass down parameter index.
2985 (add_param_constraints): Use range-info when available.
2986 (build_scop_context): Adjust.
2987 * sese.c (new_sese_info): Adjust.
2988 (free_sese_info): Likewise.
2989 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
2990 Remove unused typedefs.
2991 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
2993 2017-10-18 Martin Liska <mliska@suse.cz>
2995 * combine.c (simplify_compare_const): Add gcc_fallthrough.
2997 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
2999 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
3000 (s390_sched_init): Do not reset s390_sched_state if we entered the
3001 current basic block via a fallthru edge and all others are unlikely.
3003 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
3005 * config/s390/s390.c (NUM_SIDES): New variable.
3006 (LONGRUNNING_THRESHOLD): New variable.
3007 (LATENCY_FACTOR): New variable.
3008 (s390_sched_score): Decrease score for long-running instructions on
3010 (s390_sched_variable_issue): Perform bookkeeping for long-running
3013 2017-10-18 Richard Biener <rguenther@suse.de>
3015 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
3016 Simplify with removal of the parameter rename map.
3017 (set_rename): Likewise.
3018 (should_copy_to_new_region): Likewise.
3019 (graphite_copy_stmts_from_block): Likewise.
3020 (copy_bb_and_scalar_dependences): Remove initialization of
3021 unused copied_bb_map.
3023 (copy_internal_parameters): Likewise.
3024 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
3025 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
3026 Use INTEGRAL_TYPE_P.
3027 (parameter_index_in_region_1): Rename to ...
3028 (assign_parameter_index_in_region): ... this. Assert we have
3029 a parameter we handle.
3030 (scan_tree_for_params): Adjust.
3031 * sese.h (parameter_rename_map_t): Remove.
3032 (struct sese_info_t): Remove unused parameter_rename_map and
3033 copied_bb_map members.
3034 * sese.c (new_sese_info): Adjust.
3035 (free_sese_info): Likewise.
3037 2017-10-18 Martin Liska <mliska@suse.cz>
3040 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
3041 on an abnormal edge.
3043 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
3045 * doc/invoke.texi (ffunction-sections and fdata-sections):
3048 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3050 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
3051 the use statement can throw internally.
3053 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3055 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
3056 any RTX present on the RHS of a SET.
3057 * compare-elim.c (try_eliminate_compare): Restore comment.
3059 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3061 * langhooks.h (struct lang_hooks): Document that tree_size langhook
3062 may be also called on tcc_type nodes.
3063 * langhooks.c (lhd_tree_size): Likewise.
3065 2017-10-17 David Malcolm <dmalcolm@redhat.com>
3067 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
3068 format_warning_at_substring.
3069 (maybe_warn): Convert source_range * param to a location_t. Pass
3070 UNKNOWN_LOCATION rather than NULL to fmtwarn.
3071 (format_directive): Remove code to extract source_ranges and
3072 source_range * in favor of just a location_t.
3073 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
3075 * substring-locations.c (format_warning_va): Convert
3076 source_range * param to a location_t.
3077 (format_warning_at_substring): Likewise.
3078 * substring-locations.h (format_warning_va): Likewise.
3079 (format_warning_at_substring): Likewise.
3081 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
3083 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
3085 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
3086 and scatter/gather ops.
3088 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
3089 vec_gather_load and vec_scatter_store.
3090 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
3091 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
3092 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
3093 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
3094 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
3095 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
3097 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
3099 * reg-stack.c (compare_for_stack_reg): Add bool argument.
3100 Detect FTST instruction and handle its register pops. Only pop
3101 second operand if can_pop_second_op is true.
3102 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
3103 set can_pop_second_op to false in the compare_for_stack_reg call.
3105 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
3106 output_fp_compare for stack register operands.
3107 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
3108 instructions here. Do not emit stack register pops here. Assert
3109 that FCOMPP pops next to top stack register. Rewrite function.
3111 2017-10-17 Nathan Sidwell <nathan@acm.org>
3114 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
3115 use DECL_ASSEMBLER_NAME_RAW.
3118 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
3121 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
3122 Wilco Dijkstra <wilco.dijkstra@arm.com>
3124 * builtins.c (expand_builtin_update_setjmp_buf): Add a
3125 converstion to Pmode from the buf_addr.
3127 2017-10-17 Richard Biener <rguenther@suse.de>
3129 * graphite-dependences.c (scop_get_reads_and_writes): Change
3130 output parameters to references.
3132 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
3134 PR 71026/tree-optimization
3135 * fold-const.c (distribute_real_division): Removed.
3136 (fold_binary_loc): Remove calls to distribute_real_divison.
3138 2017-10-17 Richard Biener <rguenther@suse.de>
3140 * graphite-scop-detection.c
3141 (scop_detection::stmt_has_simple_data_refs_p): Always use
3142 the full nest as region.
3143 (try_generate_gimple_bb): Likewise.
3144 * sese.c (scalar_evolution_in_region): Simplify now that
3145 SCEV can handle instantiation in regions.
3146 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
3147 in the non-loop part of a function if requested.
3149 2017-10-17 Richard Biener <rguenther@suse.de>
3151 PR tree-optimization/82563
3152 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
3154 (graphite_regenerate_ast_isl): Call it.
3155 * graphite-scop-detection.c (build_scops): Remove entry edge split.
3157 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3159 PR tree-optimization/82549
3160 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
3161 Formatting fixes. Instead of calling make_bit_field_ref with negative
3164 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
3166 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
3167 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
3168 _mm_maskz_reduce_ss): New.
3169 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
3170 __builtin_ia32_reducess_mask): Ditto..
3171 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
3172 * config/i386/sse.md (reduces<mode>): Renamed to ...
3173 (reduces<mode><mask_scalar_name>): ... this.
3174 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
3176 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
3177 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
3179 2017-10-16 David Malcolm <dmalcolm@redhat.com>
3181 * Makefile.in (OBJS): Add unique-ptr-tests.o.
3182 * selftest-run-tests.c (selftest::run_tests): Call
3183 selftest::unique_ptr_tests_cc_tests.
3184 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
3185 * unique-ptr-tests.cc: New file.
3187 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
3190 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3192 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3193 (make_hard_regno_born, make_hard_regno_dead): Update
3194 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
3196 2017-10-16 Jeff Law <law@redhat.com>
3198 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
3200 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
3202 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
3204 2017-10-16 Olivier Hainque <hainque@adacore.com>
3206 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
3207 with_cpu if we were configured for an e500v2 target cpu name.
3209 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
3211 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
3212 * doc/invoke.texi: Document +nodsp as a valid extension for
3215 2017-10-16 Martin Liska <mliska@suse.cz>
3217 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
3218 (test_set_range): Likewise.
3219 (test_range_functions): Rename to ...
3220 (test_bit_in_range): ... this.
3221 (sbitmap_c_tests): Add new test.
3223 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3225 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
3227 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
3228 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
3230 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3232 * config/aarch64/aarch64-builtins.c
3233 (aarch64_types_quadopu_lane_qualifiers): New.
3234 (TYPES_QUADOPU_LANE): New.
3235 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
3236 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
3237 (aarch64_<sur>dot_laneq<vsi2qi>): New.
3238 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
3239 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
3240 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
3241 (Vdottype, DOTPROD): New.
3242 (sur): Add SDOT and UDOT.
3244 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3246 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
3247 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
3248 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
3250 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
3251 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
3252 Enable TARGET_DOTPROD.
3253 (cortex-a75.cortex-a55): Likewise.
3254 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
3256 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3258 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
3259 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
3261 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
3263 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
3264 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
3265 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
3266 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
3267 * config/arm/types.md (neon_dot, neon_dot_q): New.
3268 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
3270 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3272 * config/arm/arm.h (TARGET_DOTPROD): New.
3273 * config/arm/arm.c (arm_arch_dotprod): New.
3274 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
3275 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
3276 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
3277 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
3278 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
3279 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
3280 * doc/invoke.texi (armv8.2-a): Document dotprod
3282 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3284 * i386.c (ix86_vec_cost): New function.
3285 (ix86_rtx_costs): Handle vector operations better.
3286 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
3287 * x86-tune-costs.h: Add new costs to all tables.
3289 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3291 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
3293 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
3294 divsd, sqrtss and sqrtsd
3295 * x86-tune-costs.h: Add new entries to all costs.
3296 (znver1_cost): Fix to match real instruction latencies.
3298 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3299 Michael Collison <michael.collison@arm.com>
3301 * compare-elim.c: Include emit-rtl.h.
3302 (can_merge_compare_into_arith): New function.
3303 (try_validate_parallel): Likewise.
3304 (try_merge_compare): Likewise.
3305 (try_eliminate_compare): Call the above when no previous clobber
3307 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
3310 2017-10-14 Jakub Jelinek <jakub@redhat.com>
3314 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
3315 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
3319 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
3320 to be any operand_equal_p operands. For & (B - 1) require
3321 B to be power of 2. Recognize
3322 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
3324 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3327 * optabs.c (expand_memory_blockage): Fix call of
3328 targetm.have_memory_blockage.
3330 2017-10-14 Jakub Jelinek <jakub@redhat.com>
3333 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
3334 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
3335 objects to extra_objs instead of overwriting it.
3337 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3339 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
3340 Use any_fp_register_operand as operand[3] predicate. Simplify
3341 equality test for operands[2] and operands[4] memory location.
3342 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
3343 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
3344 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
3345 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
3346 any_fp_register_operand as operand[1] predicate. Simplify
3347 equality test for operands[0] and operands[3] memory location.
3348 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
3349 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
3350 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
3352 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3354 * target-insns.def: Add memory_blockage.
3355 * optabs.c (expand_memory_blockage): New function.
3356 (expand_asm_memory_barrier): Rename ...
3357 (expand_asm_memory_blockage): ... to this.
3358 (expand_mem_thread_fence): Call expand_memory_blockage
3359 instead of expand_asm_memory_barrier.
3360 (expand_mem_singnal_fence): Ditto.
3361 (expand_atomic_load): Ditto.
3362 (expand_atomic_store): Ditto.
3363 * doc/md.texi (Standard Pattern Names For Generation):
3364 Document memory_blockage instruction pattern.
3366 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
3368 * config/rl78/rl78.c (rl78_emit_libcall): New function.
3369 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
3370 * config/rl78/rl78.md: New define_expand "adddi3".
3372 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
3374 * cfghooks.c (verify_flow_info): Disable check that all probabilities
3377 2017-10-13 Jeff Law <law@redhat.com>
3379 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
3381 2017-10-13 Jakub Jelinek <jakub@redhat.com>
3384 * internal-fn.c (expand_mul_overflow): If both operands have
3385 the same highpart of -1 or 0 and the topmost bit of lowpart
3386 is different, overflow is if res <= 0 rather than res < 0.
3388 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
3390 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
3391 TARGET_P9_VECTOR code for unaligned_load case.
3393 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
3395 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
3397 2017-10-13 Nathan Sidwell <nathan@acm.org>
3399 * tree-core.h (tree_contains_struct): Make bool.
3400 * tree.c (tree_contains_struct): Likewise.
3401 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
3402 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
3403 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
3404 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
3405 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
3407 2017-10-13 Richard Biener <rguenther@suse.de>
3409 * graphite-isl-ast-to-gimple.c
3410 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
3411 parameters and dominance check.
3412 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
3413 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
3414 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
3415 Do not update SSA form here or do intermediate IL verification.
3416 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
3417 (graphite_initialize): Remove check on the number of loops in
3418 the function and inline into graphite_transform_loops.
3419 (graphite_finalize): Inline into graphite_transform_loops.
3420 (graphite_transform_loops): Perform SSA update and IL verification
3422 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
3424 2017-10-13 Richard Biener <rguenther@suse.de>
3426 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
3427 graphite_expression_type_precision): Avoid global constructor
3429 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
3430 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
3431 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
3432 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
3433 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
3434 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
3436 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
3439 * config/i386/i386.h (ix86_red_zone_size): New.
3440 * config/i386/i386.md (push peephole2s): Replace
3441 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
3443 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3444 Alan Hayward <alan.hayward@arm.com>
3445 David Sherwood <david.sherwood@arm.com>
3447 * combine.c (can_change_dest_mode): Reject changes in
3448 REGMODE_NATURAL_SIZE.
3450 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3451 Alan Hayward <alan.hayward@arm.com>
3452 David Sherwood <david.sherwood@arm.com>
3454 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
3455 (expand_debug_source_expr): Likewise.
3456 * combine.c (combine_simplify_rtx): Likewise.
3457 * cse.c (fold_rtx): Likewise.
3458 * fwprop.c (canonicalize_address): Likewise.
3459 * targhooks.c (default_shift_truncation_mask): Likewise.
3461 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3462 Alan Hayward <alan.hayward@arm.com>
3463 David Sherwood <david.sherwood@arm.com>
3465 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
3466 (widened_mode): Likewise.
3467 (expand_unop): Likewise.
3468 * ree.c (transform_ifelse): Likewise.
3469 (merge_def_and_ext): Likewise.
3470 (combine_reaching_defs): Likewise.
3471 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3473 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3474 Alan Hayward <alan.hayward@arm.com>
3475 David Sherwood <david.sherwood@arm.com>
3477 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
3478 * combine.c (gen_lowpart_for_combine): Likewise.
3479 * dwarf2out.c (rtl_for_decl_location): Likewise.
3480 * final.c (alter_subreg): Likewise.
3481 * rtlhooks.c (gen_lowpart_general): Likewise.
3482 (gen_lowpart_if_possible): Likewise.
3484 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3485 Alan Hayward <alan.hayward@arm.com>
3486 David Sherwood <david.sherwood@arm.com>
3488 * calls.c (expand_call): Use subreg_lowpart_offset.
3489 * cse.c (cse_insn): Likewise.
3490 * regcprop.c (copy_value): Likewise.
3491 (copyprop_hardreg_forward_1): Likewise.
3493 2017-10-13 Jakub Jelinek <jakub@redhat.com>
3496 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
3497 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
3498 =Q constraints to +Q and into insn condition add check
3499 that operands[0] and operands[1] are equal.
3500 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
3501 =Q constraints to +Q and into insn condition add check
3502 that operands[0] is equal to either operands[1] or operands[2].
3505 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
3506 instead of handling MINUS_EXPR twice (once for each argument),
3507 canonicalize operand order and handle just once, use rtype where
3508 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
3511 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
3512 any values of __C while still being pattern recognizable as a simple
3515 2017-10-13 Richard Biener <rguenther@suse.de>
3517 PR tree-optimization/82451
3519 2017-10-02 Richard Biener <rguenther@suse.de>
3521 PR tree-optimization/82355
3522 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
3523 a mapping for the enclosing loop but avoid generating one for
3525 (copy_bb_and_scalar_dependences): Remove premature codegen
3526 error on PHIs in blocks duplicated into multiple places.
3527 * graphite-scop-detection.c
3528 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
3529 in the region use it as loop and nest to analyze the DR in.
3530 (try_generate_gimple_bb): Likewise.
3531 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
3532 (add_loop_constraints): For blocks in a loop not in the region
3533 create a dimension with a single iteration.
3534 * sese.h (gbb_loop_at_index): Remove assert.
3536 * cfgloop.c (loop_preheader_edge): For the loop tree root
3537 return the single successor of the entry block.
3538 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3539 Reset the SCEV hashtable and niters.
3540 * graphite-scop-detection.c
3541 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
3542 assert that we only have POLYNOMIAL_CHREC that vary in loops
3543 contained in the region.
3544 (scop_detection::graphite_can_represent_expr): Adjust.
3545 (scop_detection::stmt_has_simple_data_refs_p): For loops
3546 not in the region set loop to NULL. The nest is now the
3547 entry edge to the region.
3548 (try_generate_gimple_bb): Likewise.
3549 * sese.c (scalar_evolution_in_region): Adjust for
3550 instantiate_scev change.
3551 * tree-data-ref.h (graphite_find_data_references_in_stmt):
3552 Make nest parameter the edge into the region.
3553 (create_data_ref): Likewise.
3554 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
3555 entry edge into a region and adjust instantiate_scev calls.
3556 (create_data_ref): Likewise.
3557 (graphite_find_data_references_in_stmt): Likewise.
3558 (find_data_references_in_stmt): Pass the loop preheader edge
3559 from the nest argument.
3560 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
3561 parameter the edge into the region.
3562 (instantiate_parameters): Use the loop preheader edge as entry.
3563 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
3565 (get_instantiated_value_entry): Make instantiate_below parameter
3566 the edge into the region.
3567 (instantiate_scev_name): Likewise. Adjust dominance checks,
3568 when we cannot use loop-based instantiation instantiate by
3569 walking use-def chains.
3570 (instantiate_scev_poly): Adjust.
3571 (instantiate_scev_binary): Likewise.
3572 (instantiate_scev_convert): Likewise.
3573 (instantiate_scev_not): Likewise.
3574 (instantiate_array_ref): Remove.
3575 (instantiate_scev_3): Likewise.
3576 (instantiate_scev_2): Likewise.
3577 (instantiate_scev_1): Likewise.
3578 (instantiate_scev_r): Do not blindly handle N-operand trees.
3579 Do not instantiate array-refs. Handle all constants and invariants.
3580 (instantiate_scev): Make instantiate_below parameter
3581 the edge into the region.
3582 (resolve_mixers): Use the loop preheader edge for the region
3583 parameter to instantiate_scev_r.
3584 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
3586 2017-10-13 Richard Biener <rguenther@suse.de>
3588 PR tree-optimization/82525
3589 * graphite-isl-ast-to-gimple.c
3590 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
3592 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
3593 Fail code generation when we cannot represent the isl integer.
3594 (binary_op_to_tree): Elide modulo operations that are no-ops
3595 in the type we code generate. Remove now superfluous code
3598 2017-10-13 Richard Biener <rguenther@suse.de>
3600 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
3601 (scop_detection::harmful_loop_in_region): Remove premature
3602 IV type restriction.
3603 (scop_detection::graphite_can_represent_scev): We can handle
3604 pointer IVs just fine.
3606 2017-10-13 Alan Modra <amodra@gmail.com>
3608 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
3609 "Clobbers and Scratch Registers". Add paragraph on
3610 alternative to clobbers for scratch registers and OpenBLAS
3613 2017-10-13 Alan Modra <amodra@gmail.com>
3615 * doc/extend.texi (Clobbers): Correct vax example. Delete old
3616 example of a memory input for a string of known length. Move
3617 commentary out of table. Add a number of new examples
3618 covering array memory inputs.
3620 2017-10-12 Martin Liska <mliska@suse.cz>
3622 PR tree-optimization/82493
3623 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
3624 (test_range_functions): New function.
3625 (sbitmap_c_tests): Likewise.
3626 * selftest-run-tests.c (selftest::run_tests): Run new tests.
3627 * selftest.h (sbitmap_c_tests): New function.
3629 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
3631 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
3633 * config/rs6000/amo.h: Fix spacing issue.
3635 2017-10-12 Jakub Jelinek <jakub@redhat.com>
3638 * config/i386/i386.md (*ashl<mode>3_mask_1,
3639 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
3640 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
3643 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
3645 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
3646 (profile_probability): Set max_probability
3647 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
3649 * profile-count.c (profile_probability::differs_from_p): Do not
3650 rely on max_probaiblity == 10000
3652 2017-10-12 Jeff Law <law@redhat.com>
3654 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
3657 2017-10-12 Martin Sebor <msebor@redhat.com>
3661 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
3662 (handle_alias_pairs): Call it.
3663 * common.opt (-Wattribute-alias): New option.
3664 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
3665 * doc/invoke.texi (-Wattribute-alias): Document.
3667 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
3670 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
3672 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3674 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3675 (make_hard_regno_born, make_hard_regno_dead): Update
3676 bb_killed_pseudos and bb_gen_pseudos.
3678 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
3680 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
3682 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
3684 * config/alpha/alpha.c (alpha_split_conditional_move):
3685 Use std::swap instead of manually swapping.
3686 (alpha_stdarg_optimize_hook): Ditto.
3687 (alpha_canonicalize_comparison): Ditto.
3689 2017-10-12 Bin Cheng <bin.cheng@arm.com>
3691 * tree-loop-distribution.c (struct builtin_info): New struct.
3692 (struct partition): Refactor fields into struct builtin_info.
3693 (partition_free): Free struct builtin_info.
3694 (build_size_arg_loc, build_addr_arg_loc): Delete.
3695 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
3696 information from struct builtin_info.
3697 (find_single_drs): New function refactored from classify_partition.
3698 Also moved builtin validity checks to this function.
3699 (compute_access_range, alloc_builtin): New functions.
3700 (classify_builtin_st, classify_builtin_ldst): New functions.
3701 (classify_partition): Refactor code into functions find_single_drs,
3702 classify_builtin_st and classify_builtin_ldst.
3703 (distribute_loop): Don't do runtime alias check when distributing
3705 (find_seed_stmts_for_distribution): New function.
3706 (pass_loop_distribution::execute): Refactor code finding seed
3707 stmts into above function. Support distribution for the innermost
3708 two-level loop nest. Adjust dump information.
3710 2017-10-12 Bin Cheng <bin.cheng@arm.com>
3712 * tree-loop-distribution.c: Adjust the general comment.
3713 (NUM_PARTITION_THRESHOLD): New macro.
3714 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
3715 (classify_partition): Skip builtin pattern of loop nest's inner loop.
3716 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
3717 in call to build_partition_graph.
3718 (finalize_partitions): New parameter. Make loop distribution more
3719 conservative by fusing more partitions.
3720 (distribute_loop): Don't do runtime alias check in case of loop nest
3722 (find_seed_stmts_for_distribution): New function.
3723 (prepare_perfect_loop_nest): New function.
3724 (pass_loop_distribution::execute): Refactor code finding seed stmts
3725 and loop nest into above functions. Support loop nest distribution.
3726 Adjust dump information accordingly.
3728 2017-10-12 Bin Cheng <bin.cheng@arm.com>
3730 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
3731 and set PTYPE_SEQUENTIAL for merged partition.
3733 2017-10-12 Richard Biener <rguenther@suse.de>
3735 PR tree-optimization/69728
3737 2017-09-19 Richard Biener <rguenther@suse.de>
3739 PR tree-optimization/69728
3740 * graphite-sese-to-poly.c (schedule_error): New global.
3741 (add_loop_schedule): Handle empty domain by failing the
3743 (build_original_schedule): Handle schedule_error.
3745 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
3746 domain by returning an unchanged schedule.
3748 2017-10-12 Jakub Jelinek <jakub@redhat.com>
3750 * genrecog.c (validate_pattern): For VEC_SELECT verify that
3751 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
3753 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
3755 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
3758 2017-10-12 Jakub Jelinek <jakub@redhat.com>
3761 * expr.c (store_field): Don't optimize away bitsize == 0 store
3762 from CALL_EXPR with addressable return type.
3764 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
3766 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
3767 * config/rs6000/rs6000.md (sel): Delete mode attribute.
3768 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
3769 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
3770 TARGET_ISEL instead of TARGET_ISEL<sel>.
3772 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
3774 * config/rs6000/rs6000.c
3775 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
3777 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
3779 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
3781 (reg_or_cint_operand): Fix comment.
3782 (reg_or_zero_operand): New predicate.
3783 * config/rs6000/rs6000-protos.h (output_isel): Delete.
3784 * config/rs6000/rs6000.c (output_isel): Delete.
3785 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
3786 instead of reg_or_cint_operand. Output instruction directly (not via
3788 (isel_unsigned_<mode>): Ditto.
3789 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
3790 gpc_reg_operand. Add an instruction alternative for this. Output
3791 instruction directly.
3792 (*isel_reversed_unsigned_<mode>): Ditto.
3794 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
3796 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
3797 (TARGET_CANONICALIZE_COMPARISON): Define.
3799 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
3802 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
3803 Check whether the dest is REG before adding REG_EQUIV note.
3805 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
3808 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3810 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3811 (make_hard_regno_born, make_hard_regno_dead): Update
3812 bb_killed_pseudos and bb_gen_pseudos.
3814 2017-10-11 Nathan Sidwell <nathan@acm.org>
3816 * incpath.h (enum incpath_kind): Name enum, prefix values.
3817 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
3818 * incpath.c (heads, tails): Use INC_MAX.
3819 (add_env_var_paths, add_standard_paths): Use incpath_kind.
3820 (merge_include_chains, split_quote_chain,
3821 register_include_chains): Update incpath_kind names.
3822 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
3823 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
3825 (add_framework_path, darwin_register_objc_includes): Likewise.
3826 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
3828 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
3830 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
3831 Do not use float_operator operator predicate.
3832 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
3833 * config/i386/predicates.md (float_operator): Remove predicate.
3835 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
3837 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
3838 (*jccxf_i387): Ditto.
3839 (*jcc<mode>_i387): Ditto.
3840 (*jccu<mode>_i387): Ditto.
3841 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
3842 (*jcc_*_i387 splitters): Remove.
3843 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
3844 * config/i386/i386.c (ix86_split_fp_branch): Remove.
3845 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
3848 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
3850 * profile-count.h (slow_safe_scale_64bit): New function.
3851 (safe_scale_64bit): New inline.
3852 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
3853 * profile-count.c: Include wide-int.h
3854 (slow_safe_scale_64bit): New.
3856 2017-10-11 Nathan Sidwell <nathan@acm.org>
3858 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
3859 HAS_DECL_ASSEMBLER_NAME_P.
3860 * gimple-expr.c (gimple_decl_printable_name: Check
3861 HAS_DECL_ASSEMBLER_NAME_P too.
3862 * ipa-utils.h (type_in_anonymous_namespace_p): Check
3863 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
3864 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
3865 * passes.c (rest_of_decl_compilation): Check
3866 HAS_DECL_ASSEMBLER_NAME_P too.
3867 * recog.c (verify_changes): Likewise.
3868 * tree-pretty-print.c (dump_decl_name): Likewise.
3869 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
3871 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
3872 (SET_DECL_ASSEMBLER_NAME): Use it.
3873 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
3874 (COPY_DECL_ASSEMBLER_NAME): Likewise.
3875 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
3877 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
3879 * config.gcc (i386, x86_64): Add extra objects.
3880 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
3881 (ix86_min_insn_size): Declare.
3882 (ix86_issue_rate): Declare.
3883 (ix86_adjust_cost): Declare.
3884 (ia32_multipass_dfa_lookahead): Declare.
3885 (ix86_macro_fusion_p): Declare.
3886 (ix86_macro_fusion_pair_p): Declare.
3887 (ix86_bd_has_dispatch): Declare.
3888 (ix86_bd_do_dispatch): Declare.
3889 (ix86_core2i7_init_hooks): Declare.
3890 (ix86_atom_sched_reorder): Declare.
3891 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
3892 (COSTS_N_BYTES): Move to x86-tune-costs.h.
3893 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
3894 (rip_relative_addr_p): Rename to ...
3895 (ix86_rip_relative_addr_p): ... this one; export.
3896 (memory_address_length): Update.
3897 (ix86_issue_rate): Move to x86-tune-sched.c.
3898 (ix86_flags_dependent): Move to x86-tune-sched.c.
3899 (ix86_agi_dependent): Move to x86-tune-sched.c.
3900 (exact_dependency_1): Move to x86-tune-sched.c.
3901 (exact_store_load_dependency): Move to x86-tune-sched.c.
3902 (ix86_adjust_cost): Move to x86-tune-sched.c.
3903 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
3904 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
3905 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
3906 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
3907 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
3908 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
3909 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
3910 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
3911 (min_insn_size): Rename to ...
3912 (ix86_min_insn_size): ... this one; export.
3913 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
3914 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
3915 (core2i7_first_cycle_multipass_backtrack): Move to
3916 x86-tune-sched-core.c.
3917 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
3918 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
3919 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
3920 (ix86_avoid_jump_mispredicts): Update.
3921 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
3922 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
3923 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
3924 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
3925 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
3926 (MAX_INSN): Move to ix86-tune-sched-bd.c.
3927 (MAX_IMM): Move to ix86-tune-sched-bd.c.
3928 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
3929 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
3930 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
3931 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
3932 (MAX_STORE): Move to ix86-tune-sched-bd.c.
3933 (BIG): Move to ix86-tune-sched-bd.c.
3934 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
3935 (enum insn_path): Move to ix86-tune-sched-bd.c.
3936 (get_mem_group): Move to ix86-tune-sched-bd.c.
3937 (is_cmp): Move to ix86-tune-sched-bd.c.
3938 (dispatch_violation): Move to ix86-tune-sched-bd.c.
3939 (is_branch): Move to ix86-tune-sched-bd.c.
3940 (is_prefetch): Move to ix86-tune-sched-bd.c.
3941 (init_window): Move to ix86-tune-sched-bd.c.
3942 (allocate_window): Move to ix86-tune-sched-bd.c.
3943 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
3944 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
3945 (process_end_window): Move to ix86-tune-sched-bd.c.
3946 (allocate_next_window): Move to ix86-tune-sched-bd.c.
3947 (find_constant): Move to ix86-tune-sched-bd.c.
3948 (get_num_immediates): Move to ix86-tune-sched-bd.c.
3949 (has_immediate): Move to ix86-tune-sched-bd.c.
3950 (get_insn_path): Move to ix86-tune-sched-bd.c.
3951 (get_insn_group): Move to ix86-tune-sched-bd.c.
3952 (count_num_restricted): Move to ix86-tune-sched-bd.c.
3953 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
3954 (add_insn_window): Move to ix86-tune-sched-bd.c.
3955 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
3956 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
3957 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
3958 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
3959 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
3960 (do_dispatch): Move to ix86-tune-sched-bd.c.
3961 (has_dispatch): Move to ix86-tune-sched-bd.c.
3962 * config/i386/t-i386: Add new object files.
3963 * config/i386/x86-tune-costs.h: New file.
3964 * config/i386/x86-tune-sched-atom.c: New file.
3965 * config/i386/x86-tune-sched-bd.c: New file.
3966 * config/i386/x86-tune-sched-core.c: New file.
3967 * config/i386/x86-tune-sched.c: New file.
3969 2017-10-11 Liu Hao <lh_mouse@126.com>
3971 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
3972 the generic version below instead.
3973 (should_colorize): Recognize Windows consoles as terminals
3975 * pretty-print.c [__MINGW32__] (write_all): New function.
3976 [__MINGW32__] (find_esc_head): Likewise.
3977 [__MINGW32__] (find_esc_terminator): Likewise.
3978 [__MINGW32__] (eat_esc_sequence): Likewise.
3979 [__MINGW32__] (mingw_ansi_fputs): New function that handles
3981 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
3984 2017-10-11 Richard Biener <rguenther@suse.de>
3986 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
3987 Properly call analyze_scalar_evolution with the loop of the stmt.
3989 2017-10-11 Richard Biener <rguenther@suse.de>
3991 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
3992 * tree-core.h (tree_base): Add chrec_var union member.
3993 * tree.h (CHREC_VAR): Remove.
3994 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
3995 * tree-chrec.h (build_polynomial_chrec): Adjust.
3996 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
3997 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
3999 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
4001 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
4002 * match.pd: ... here.
4003 ((T) X == (T) Y): Relax condition.
4005 2017-10-11 Bin Cheng <bin.cheng@arm.com>
4007 PR tree-optimization/82472
4008 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
4010 (break_alias_scc_partitions): Update postorder number.
4012 2017-10-11 Martin Liska <mliska@suse.cz>
4015 * opts.c (parse_no_sanitize_attribute): Do not use error_value
4017 * opts.h (parse_no_sanitize_attribute): Remove last argument.
4019 2017-10-11 Martin Liska <mliska@suse.cz>
4021 * print-rtl.c (print_insn): Move declaration of idbuf
4022 to same scope as name.
4024 2017-10-11 Martin Liska <mliska@suse.cz>
4029 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
4032 2017-10-11 Martin Liska <mliska@suse.cz>
4035 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
4038 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
4040 * config.gcc (default_gnu_indirect_function): Default to yes for
4041 arm*-*-linux* with glibc.
4043 2017-10-11 Richard Biener <rguenther@suse.de>
4045 * tree-scalar-evolution.c (get_scalar_evolution): Handle
4046 default-defs and types we do not want to analyze.
4047 (interpret_loop_phi): Replace unreachable code with an assert.
4048 (compute_scalar_evolution_in_loop): Remove and inline ...
4049 (analyze_scalar_evolution_1): ... here, replacing condition with
4050 what makes the intent clearer. Remove handling of cases
4051 get_scalar_evolution now handles.
4053 2017-10-10 Jim Wilson <wilson@tuliptree.org>
4055 PR rtl-optimization/81434
4056 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
4057 comment for main loop. In sched_group_found if, also add checks for
4058 pass and min_cost_group.
4060 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
4062 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
4063 (rs6000_insn_cost): New function.
4064 * config/rs6000/rs6000.md (cost): New attribute.
4066 2017-10-10 Jakub Jelinek <jakub@redhat.com>
4067 H.J. Lu <hongjiu.lu@intel.com>
4071 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
4072 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
4073 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
4074 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
4075 to OPTION_MASK_ISA_AVX512VL - builtins that have both
4076 OPTION_MASK_ISA_MMX and some other bit set require both
4077 mmx and the ISAs without the mmx bit.
4078 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
4079 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
4080 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
4081 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
4082 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
4083 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
4084 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
4085 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
4086 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
4087 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
4088 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
4089 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
4090 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
4091 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
4092 Add OPTION_MASK_ISA_MMX.
4094 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
4096 * config.gcc (armv7*-*-freebsd*): New target.
4097 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
4099 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
4101 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
4102 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
4105 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
4107 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
4108 a template parameter.
4109 (WIDE_INT_REF_FOR): Update accordingly.
4110 * tree.h (wi::int_traits <const_tree>): Delete.
4111 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
4112 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
4113 (wi::tree_to_wide_ref): New typedef.
4114 (wi::to_wide): New function.
4115 * calls.c (get_size_range): Use wi::to_wide when operating on
4117 * cgraph.c (cgraph_node::create_thunk): Likewise.
4118 * config/i386/i386.c (ix86_data_alignment): Likewise.
4119 (ix86_local_alignment): Likewise.
4120 * dbxout.c (stabstr_O): Likewise.
4121 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
4122 * expr.c (const_vector_from_tree): Likewise.
4123 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
4124 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
4125 (fold_negate_expr_1, int_const_binop_1, const_binop)
4126 (fold_convert_const_int_from_real, optimize_bit_field_compare)
4127 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
4128 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
4129 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
4130 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
4131 (fold_not_const, round_up_loc): Likewise.
4132 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
4133 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
4134 (alloca_call_type): Likewise.
4135 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
4136 * godump.c (go_output_typedef): Likewise.
4137 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
4138 * internal-fn.c (get_min_precision): Likewise.
4139 * ipa-cp.c (ipcp_store_vr_results): Likewise.
4140 * ipa-polymorphic-call.c
4141 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
4142 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
4143 (ipa_modify_call_arguments): Likewise.
4144 * match.pd: Likewise.
4145 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
4146 * print-tree.c (print_node_brief, print_node): Likewise.
4147 * stmt.c (expand_case): Likewise.
4148 * stor-layout.c (layout_type): Likewise.
4149 * tree-affine.c (tree_to_aff_combination): Likewise.
4150 * tree-cfg.c (group_case_labels_stmt): Likewise.
4151 * tree-data-ref.c (dr_analyze_indices): Likewise.
4152 (prune_runtime_alias_test_list): Likewise.
4153 * tree-dump.c (dequeue_and_dump): Likewise.
4154 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
4155 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
4156 * tree-pretty-print.c (dump_generic_node): Likewise.
4157 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
4158 (simple_iv_with_niters): Likewise.
4159 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
4160 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
4161 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
4162 * tree-ssa-loop-niter.c (split_to_var_and_offset)
4163 (refine_value_range_using_guard, number_of_iterations_ne_max)
4164 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
4165 (get_cst_init_from_scev, record_nonwrapping_iv)
4166 (scev_var_range_cant_overflow): Likewise.
4167 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
4168 * tree-ssa-pre.c (compute_avail): Likewise.
4169 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
4170 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
4171 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
4172 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
4173 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
4174 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
4175 (dump_case_nodes, try_switch_expansion): Likewise.
4176 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
4177 (vect_do_peeling): Likewise.
4178 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
4179 * tree-vect-stmts.c (vectorizable_load): Likewise.
4180 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
4181 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
4182 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
4183 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
4184 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
4185 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
4186 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
4187 (evrp_dom_walker::before_dom_children): Likewise.
4188 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
4189 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
4190 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
4191 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
4192 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
4193 (get_range_pos_neg): Likewise.
4194 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
4195 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
4196 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
4197 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
4198 * config/avr/avr.c (avr_fold_builtin): Likewise.
4199 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
4200 * config/msp430/msp430.c (msp430_attr): Likewise.
4201 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
4202 * config/powerpcspe/powerpcspe-c.c
4203 (altivec_resolve_overloaded_builtin): Likewise.
4204 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
4205 (rs6000_expand_ternop_builtin): Likewise.
4206 * config/rs6000/rs6000-c.c
4207 (altivec_resolve_overloaded_builtin): Likewise.
4208 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
4209 (rs6000_expand_ternop_builtin): Likewise.
4210 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
4212 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4214 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
4215 when copying loop nest with only one inner loop.
4217 2017-10-10 Richard Biener <rguenther@suse.de>
4219 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
4220 blocks if SCEV is active.
4221 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
4223 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
4224 (scev_initialize): Assert we are not yet initialized.
4226 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4228 * tree-loop-distribution.c (generate_loops_for_partition): Remove
4229 inner loop's exit stmt by making it always exit the loop, otherwise
4230 we would generate an infinite empty loop.
4232 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4234 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
4235 renaming variables in new preheader if it's deleted.
4237 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4239 * tree-loop-distribution.c (struct partition): Remove unused field
4240 loops of the structure.
4241 (partition_alloc, partition_free): Ditto.
4242 (build_rdg_partition_for_vertex): Ditto.
4244 2017-10-09 Jeff Law <law@redhat.com>
4246 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
4247 return type to match prototype and documentation.
4249 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4251 * config/rs6000/rs6000.c (processor_costs): Move to ...
4252 * config/rs6000/rs6000.h: ... here.
4253 (rs6000_cost): Declare.
4255 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
4257 * except.c (setjmp_fn): New global variable.
4258 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
4259 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
4260 if DONT_USE_BUILTIN_SETJMP is defined.
4262 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4264 * target.def (insn_cost): New hook.
4265 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
4266 * doc/tm.texi: Regenerate.
4267 * rtlanal.c (insn_cost): Use the new hook.
4269 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4271 * combine.c (combine_validate_cost): Compute the new insn_cost,
4272 not just pattern_cost.
4273 (try_combine): Adjust comment.
4275 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4277 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
4279 * combine.c (uid_insn_cost): Adjust comment.
4280 (combine_validate_cost): Adjust comment. Use pattern_cost instead
4282 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
4283 * dse.c (find_shift_sequence): Ditto.
4284 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
4285 (bb_valid_for_noce_process_p): Use pattern_cost.
4286 * rtl.h (insn_rtx_cost): Delete.
4287 (pattern_cost): New prototype.
4288 (insn_cost): New prototype.
4289 * rtlanal.c (insn_rtx_cost): Rename to...
4290 (pattern_cost): ... this.
4293 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
4295 * config/i386/i386.md (*jcc_2): Remove insn pattern.
4296 (*jcc<mode>_0_r_i387): Ditto.
4297 (*jccxf_r_i387): Ditto.
4298 (*jcc<mode>_r_i387): Ditto.
4299 (*jccu<mode>_r_i387): Ditto.
4300 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
4301 (*jcc): Rename from *jcc_1.
4303 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4305 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
4306 deferred rescans after the lvx/stvx recombination pre-pass.
4308 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
4310 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
4311 memory operation instruction support.
4312 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
4313 (rs6000-ibm-aix[789]*): Likewise.
4314 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
4315 Document new functions.
4317 2017-10-09 Richard Biener <rguenther@suse.de>
4319 PR tree-optimization/82397
4320 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
4321 equality only for semantically equal trees.
4323 2017-10-09 Richard Biener <rguenther@suse.de>
4325 PR tree-optimization/82449
4326 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
4327 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
4328 allow constant addresses.
4329 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
4332 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4334 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
4337 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4340 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
4343 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4346 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
4348 2017-10-09 Jakub Jelinek <jakub@redhat.com>
4351 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
4352 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
4354 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
4356 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
4357 (WI_BINARY_PREDICATE_RESULT): Likewise.
4358 (wi::binary_traits::operator_result): New type.
4359 (wi::binary_traits::predicate_result): Likewise.
4360 (generic_wide_int::operator~, unary generic_wide_int::operator-)
4361 (generic_wide_int::operator==, generic_wide_int::operator!=)
4362 (generic_wide_int::operator&, generic_wide_int::and_not)
4363 (generic_wide_int::operator|, generic_wide_int::or_not)
4364 (generic_wide_int::operator^, generic_wide_int::operator+
4365 (binary generic_wide_int::operator-, generic_wide_int::operator*):
4367 (operator~, unary operator-, operator==, operator!=, operator&)
4368 (operator|, operator^, operator+, binary operator-, operator*): New
4370 * expr.c (get_inner_reference): Use wi::bit_and_not.
4371 * fold-const.c (fold_binary_loc): Likewise.
4372 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
4373 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
4374 (bit_value_binop): Likewise.
4375 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
4376 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
4377 (extract_range_from_binary_expr_1): Likewise.
4378 (masked_increment): Likewise.
4379 (simplify_bit_ops_using_ranges): Likewise.
4381 2017-10-09 Martin Jambor <mjambor@suse.cz>
4384 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
4385 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
4386 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
4388 (hsa_fixup_mov_insn_type): New function.
4389 (hsa_op_with_type::get_in_type): Use it.
4390 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
4391 immediates in an assert.
4392 (hsa_op_with_type::extend_int_to_32bit): New method.
4393 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
4394 types. Convert to dest type if necessary.
4395 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
4396 (reg_for_gimple_ssa): Pass false as min32int to
4397 hsa_type_for_scalar_tree_type.
4398 (gen_hsa_addr): Fixup type when creating addresable temporary.
4399 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
4400 (gen_hsa_unary_operation): Extend operands and convert to dest type if
4401 necessary. Call hsa_fixup_mov_insn_type.
4402 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
4403 extend operands and convert to dest type if necessary.
4404 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
4405 to dest type if necessary.
4406 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
4407 if conversion nt necessary and size matches.
4408 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
4409 to dest type if necessary.
4410 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
4411 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
4413 (gen_hsa_clrsb): Likewise.
4414 (gen_hsa_ffs): Likewise.
4415 (gen_hsa_divmod): Extend operands and convert to dest type if
4417 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
4419 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
4421 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
4422 Remove empty default arguments. Use a brace block as output
4424 (conditional return): Ditto.
4426 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
4427 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
4429 (group_ending_nop): Ditto.
4430 (doloop_end): Ditto.
4431 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
4432 (splitters for those): Ditto.
4434 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
4436 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
4437 a conditional jump (and the compare for it) so that pc_rtx is the
4439 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
4440 for the deleted and renamed ctr<mode>_internal[234] patterns.
4441 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
4442 Delete second conditional return pattern.
4443 (ctr<mode>_internal2): Delete this second bdnz pattern.
4444 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
4445 (ctr<mode>_internal4): Delete this second bdz pattern.
4447 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
4449 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
4450 (always_initialized_rtx_for_ssa_name_p): New predicate.
4451 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
4452 (finish_out_of_ssa): Free new field of SA.
4453 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
4454 * tree-ssa-coalesce.c: Include tree-ssa.h.
4455 (get_parm_default_def_partitions): Remove extern keyword.
4456 (get_undefined_value_partitions): New function.
4457 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
4458 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
4460 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
4462 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
4464 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
4466 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
4468 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
4469 for targets that preffer 128bit.
4471 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
4473 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
4475 2017-10-08 Olivier Hainque <hainque@adacore.com>
4477 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
4478 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
4480 (thumb_set_return_address): Likewise.
4482 2017-10-08 Olivier Hainque <hainque@adacore.com>
4484 * common/config/arm/arm-common.c (arm_except_unwind_info):
4485 Handle DWARF2_UNWIND_INFO.
4487 2017-10-07 Michael Collison <michael.collison@arm.com>
4489 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
4492 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
4494 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
4495 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
4496 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
4497 defined, force the creation of a new block for a dispatch label.
4499 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
4501 * invoke.texi (Wsuggest-attribute=cold): Document.
4502 * common.opt (Wsuggest-attribute=cold): New
4503 * ipa-pure-const.c (warn_function_cold): New function.
4504 * predict.c (compute_function_frequency): Use it.
4505 * predict.h (warn_function_cold): Declare.
4507 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
4509 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
4512 2017-10-06 Martin Liska <mliska@suse.cz>
4514 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
4515 keyword for member functions.
4516 (struct sanopt_tree_couple): New struct.
4517 (struct sanopt_tree_couple_hash): New function.
4518 (struct sanopt_ctx): Add new hash_map.
4519 (has_dominating_ubsan_ptr_check): New function.
4520 (record_ubsan_ptr_check_stmt): Likewise.
4521 (maybe_optimize_ubsan_ptr_ifn): Likewise.
4522 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
4523 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
4525 2017-10-06 Sudakshina Das <sudi.das@arm.com>
4528 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
4529 aarch64_simd_valid_immediate on CONST_VECTORs.
4530 (aarch64_reg_or_bic_imm): Likewise.
4532 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
4534 PR rtl-optimization/82396
4535 * haifa-sched.c (ready_sort_real): Disable qsort checking.
4537 2017-10-06 Sebastian Pop <sebpop@gmail.com>
4539 * graphite-dependences.c (scop_get_reads): Move code to...
4540 (scop_get_must_writes): Move code to...
4541 (scop_get_may_writes): Move code to...
4542 (scop_get_reads_and_writes): ... here.
4543 (scop_get_dependences): Call scop_get_reads_and_writes.
4545 2017-10-06 Jakub Jelinek <jakub@redhat.com>
4547 PR tree-optimization/82434
4548 * fold-const.h (can_native_encode_type_p,
4549 can_native_encode_string_p): Remove.
4550 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
4551 don't encode anything, just return what would be otherwise returned.
4552 (native_encode_fixed, native_encode_complex, native_encode_vector):
4554 (native_encode_string): Likewise. Inline by hand
4555 can_native_encode_string_p.
4556 (can_native_encode_type_p): Remove.
4557 (can_native_encode_string_p): Remove.
4558 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
4559 STRING_CSTs using can_native_encode_string_p, test all
4560 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
4561 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
4562 argument from native_encode_expr.
4563 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
4564 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
4567 2017-10-06 Richard Biener <rguenther@suse.de>
4569 PR tree-optimization/82397
4570 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
4571 operand_equal_p but rely on data_ref_compare_tree for detecting
4573 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
4574 to match up with dr_group_sort_cmp.
4576 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4579 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
4581 * config/s390/s390-builtin-types.def: Regenerate.
4583 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4586 * config/s390/s390-builtin-types.def: Regenerate.
4587 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
4588 Change flag from B_VXE to B_VX.
4589 (s390_vec_min_dbl): Remove B_VXE flag.
4591 2017-10-06 Richard Biener <rguenther@suse.de>
4593 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
4594 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
4595 translate_isl_ast_to_gimple::is_valid_rename,
4596 translate_isl_ast_to_gimple::get_rename,
4597 translate_isl_ast_to_gimple::get_def_bb_for_const,
4598 translate_isl_ast_to_gimple::get_new_name,
4599 translate_isl_ast_to_gimple::collect_all_ssa_names,
4600 translate_isl_ast_to_gimple::copy_loop_phi_args,
4601 translate_isl_ast_to_gimple::collect_all_ssa_names,
4602 translate_isl_ast_to_gimple::copy_loop_phi_args,
4603 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
4604 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
4605 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
4606 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
4607 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
4608 translate_isl_ast_to_gimple::copy_cond_phi_args,
4609 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
4610 translate_isl_ast_to_gimple::edge_for_new_close_phis,
4611 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
4612 translate_isl_ast_to_gimple::rename_uses,
4613 translate_isl_ast_to_gimple::rename_all_uses): Remove.
4614 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
4615 (set_rename_for_each_def): Likewise.
4616 (graphite_copy_stmts_from_block): Handle debug stmt resetting
4617 here. Handle rewriting SCEV analyzable uses here.
4618 (copy_bb_and_scalar_dependences): Generate code for PHI
4620 (graphite_regenerate_ast_isl): Adjust.
4621 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
4622 (add_write, add_read): New functions.
4623 (build_cross_bb_scalars_def): Use it and simplify.
4624 (build_cross_bb_scalars_use): Likewise.
4625 (graphite_find_cross_bb_scalar_vars): Inline into...
4626 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
4627 simulating out-of-SSA. Compute liveout and add dependencies.
4628 (build_scops): Force an empty entry block.
4629 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
4631 (sese_build_liveouts): Declare.
4632 (sese_trivially_empty_bb_p): Likewise.
4633 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
4634 compute liveout and debug_liveout.
4635 (sese_bad_liveouts_use): Remove.
4636 (sese_reset_debug_liveouts_bb): Likewise.
4637 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
4638 (sese_build_liveouts): Build liveout and debug_liveout and store
4640 (new_sese_info): Adjust.
4641 (free_sese_info): Likewise.
4642 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
4643 do not build liveout here.
4644 (move_sese_in_condition): Adjust region entry.
4645 (scev_analyzable_p): Match up with chrec_apply requirements.
4646 (sese_trivially_empty_bb_p): New.
4647 * tree-into-ssa.c (get_reaching_def): Properly support generating
4648 default-defs for incremental rewrite of anonymous names.
4650 2017-10-06 Richard Biener <rguenther@suse.de>
4652 * graphite-sese-to-poly.c (extract_affine): For casts increasing
4653 precision do not perform modulo reduction.
4655 2017-10-06 Richard Biener <rguenther@suse.de>
4657 PR tree-optimization/82436
4658 * tree-vect-slp.c (vect_supported_load_permutation_p): More
4659 conservatively choose the vectorization factor when checking
4660 whether we can perform the required load permutation.
4661 (vect_transform_slp_perm_load): Assert when we may not fail.
4663 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
4665 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
4666 message for incompatible -msdata=* and -mcall-* options.
4668 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
4670 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
4671 rate for post-reload scheduling.
4673 2017-10-05 Tamar Christina <tamar.christina@arm.com>
4675 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
4677 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
4679 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
4680 to improve monte carlo in scimark.
4682 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
4684 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
4685 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
4686 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
4687 pentium4_cost, nocona_cost): Set reassociation width to 1.
4688 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
4689 width to 2 for fp operations and 1 otherwise.
4690 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
4692 (atom_cost): Set reassociation width to 2.
4693 (slm_cost, generic_cost): Set fp reassociation width
4694 to 2 and 1 otherwise.
4695 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
4696 (core_cost): Set fp reassociation width to 4 and vector to 2.
4697 (ix86_reassociation_width): Rewrite using cost table; special case
4698 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
4699 and TARGET_AVX128_OPTIMAL.
4700 * config/i386/i386.h (processor_costs): Add
4701 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
4702 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
4703 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
4704 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
4705 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
4706 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
4708 2017-10-05 Nathan Sidwell <nathan@acm.org>
4710 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
4712 2017-10-05 Tamar Christina <tamar.christina@arm.com>
4714 * config/arm/arm.c (arm_test_fpu_data): New.
4715 (arm_run_selftests): Call arm_test_fpu_data.
4717 2017-10-04 Nathan Sidwell <nathan@acm.org>
4719 * toplev.c (toplev::main): Remove excess parens on pretty_printer
4721 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
4723 2017-10-04 Sudakshina Das <sudi.das@arm.com>
4725 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
4726 check type for aarch64_simd_valid_immediate.
4727 (aarch64_output_simd_mov_immediate): Update prototype.
4728 (aarch64_simd_valid_immediate): Update prototype.
4729 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
4730 support for ORR-immediate.
4731 (and<mode>3): modified pattern to add support for BIC-immediate.
4732 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
4733 now checks for valid immediate for BIC and ORR based on new enum
4735 (aarch64_output_simd_mov_immediate): Function now used to output
4736 BIC/ORR imm as well based on new enum argument.
4737 * config/aarch64/constraints.md (Do): New vector immediate constraint.
4739 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
4740 (aarch64_reg_or_bic_imm): Likewise.
4742 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4744 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
4745 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
4747 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
4751 PR rtl-optimization/82396
4752 * haifa-sched.c (autopref_multipass_init): Simplify
4754 (autopref_rank_data): Simplify sort order.
4755 * sched-int.h (autopref_multipass_data_): Remove
4756 multi_mem_insn_p, min_offset and max_offset.
4758 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4760 * doc/sourcebuild.texi: Document vect_peeling_profitable.
4762 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4764 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
4767 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4769 * doc/sourcebuild.texi: Document vect_long_mult.
4771 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
4773 PR tree-optimization/82413
4774 * fold-const.c (build_range_check): Use widest_int when comparing
4775 the maximum ETYPE value with HIGH.
4777 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
4779 PR rtl-optimization/82396
4780 * haifa-sched.c (autopref_multipass_init): Simplify
4782 (autopref_rank_data): Simplify sort order.
4783 * sched-int.h (autopref_multipass_data_): Remove
4784 multi_mem_insn_p, min_offset and max_offset.
4786 2017-10-04 Jakub Jelinek <jakub@redhat.com>
4788 PR tree-optimization/82381
4789 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
4790 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
4793 PR tree-optimization/82374
4794 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
4795 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
4796 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
4797 current_function_decl to the new decl.
4799 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
4801 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
4802 helper macro for IEEE float128 hardware built-in functions.
4803 (SQRTF128_ODD): Add built-in functions with the round-to-odd
4805 (TRUNCF128_ODD): Likewise.
4806 (ADDF128_ODD): Likewise.
4807 (SUBF128_ODD): Likewise.
4808 (MULF128_ODD): Likewise.
4809 (DIVF128_ODD): Likewise.
4810 (FMAF128_ODD): Likewise.
4811 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
4812 UNSPEC_TRUNC_ROUND_TO_ODD.
4813 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
4814 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
4815 floating point round to odd instructions.
4816 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
4817 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
4818 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
4819 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
4820 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
4821 (trunc<mode>sf2_hw): Change the truncate with round to odd
4822 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
4823 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
4824 to odd hardware instructions.
4825 (sub<mode>3_odd): Likewise.
4826 (mul<mode>3_odd): Likewise.
4827 (div<mode>3_odd): Likewise.
4828 (sqrt<mode>2_odd): Likewise.
4829 (fma<mode>4_odd): Likewise.
4830 (fms<mode>4_odd): Likewise.
4831 (nfma<mode>4_odd): Likewise.
4832 (nfms<mode>4_odd): Likewise.
4833 (trunc<mode>df2_odd): Change the truncate with round to odd
4834 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
4836 * doc/extend.texi (PowerPC built-in functions): Update documentation
4837 for existing IEEE float128-bit built-in functions. Add built-in
4838 functions that generate the IEEE 128-bit floating point round to
4841 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
4843 PR rtl-optimization/77729
4844 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
4845 to (X&(C1&~C2))|C2 transformations.
4847 2017-10-03 Martin Jambor <mjambor@suse.cz>
4849 PR tree-optimization/82363
4850 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
4851 mismatch, mark lacc written regardless of racc.
4853 2017-10-03 Jakub Jelinek <jakub@redhat.com>
4855 PR tree-optimization/82381
4856 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
4857 stmt_to_insert nor wheather SSA_NAMEs are default defs.
4858 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
4859 fallthrough into reassoc_stmt_dominates_stmt_p.
4862 * combine.c (combine_instructions): Don't combine in unreachable
4865 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
4868 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
4869 function to not use the have_cpu variable. Do not set cpu_index,
4870 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
4872 (rs6000_valid_attribute_p): Remove duplicate initializations of
4873 old_optimize and func_optimize.
4874 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
4875 (rs6000_activate_target_options): Make global.
4876 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
4879 2017-10-02 Jakub Jelinek <jakub@redhat.com>
4881 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
4882 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
4883 Set *poffset to 0 and *psize and *pmax_size to -1 if
4884 *poffset + *psize overflows in HOST_WIDE_INT.
4886 PR tree-optimization/82387
4887 PR tree-optimization/82388
4888 PR tree-optimization/82389
4889 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
4890 instead of live_bytes non-NULL.
4892 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
4895 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
4898 2017-10-02 Richard Biener <rguenther@suse.de>
4900 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
4901 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
4902 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
4904 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
4906 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
4907 requested precision matches the type's.
4908 * calls.c (alloc_max_size): Calculate the new candidate size as
4909 a widest_int and use wi::to_widest when comparing it with the
4910 current candidate size.
4911 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
4912 zero rather than integer_zero_node.
4913 * match.pd: Check for a no-op conversion before using wi::add
4914 rather than after. Use tree_to_uhwi when summing small shift
4915 counts into an unsigned int.
4917 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
4918 Alan Hayward <alan.hayward@arm.com>
4919 David Sherwood <david.sherwood@arm.com>
4922 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
4923 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
4924 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
4925 POINTER_AND_FP_REGS.
4927 2017-10-02 Richard Biener <rguenther@suse.de>
4929 PR tree-optimization/82355
4930 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
4931 a mapping for the enclosing loop but avoid generating one for
4933 (copy_bb_and_scalar_dependences): Remove premature codegen
4934 error on PHIs in blocks duplicated into multiple places.
4935 * graphite-scop-detection.c
4936 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
4937 in the region use it as loop and nest to analyze the DR in.
4938 (try_generate_gimple_bb): Likewise.
4939 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
4940 (add_loop_constraints): For blocks in a loop not in the region
4941 create a dimension with a single iteration.
4942 * sese.h (gbb_loop_at_index): Remove assert.
4944 2017-10-01 Kevin Buettner <kevinb@redhat.com>
4946 * omp-expand.c (adjust_context_scope): New function.
4947 (expand_parallel_call): Call adjust_context_scope.
4949 2017-10-01 Jeff Law <law@redhat.com>
4951 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
4952 dom_opt_dom_walker class with direct access to private members.
4953 Add comments. Call test_for_singularity.
4954 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
4955 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
4956 m_dummy_cond anymore.
4957 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
4959 (pass_dominator:execute): Build the dummy_cond here and pass it
4960 to the dom_opt_dom_walker ctor.
4961 (test_for_singularity): New function.
4963 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
4964 Maya Rashish <coypu@sdf.org>
4966 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
4967 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
4968 (alpha*-*-netbsd*) Use nbsd_tm_file.
4969 (arm*-*-netbsdelf*) Likewise.
4970 (i[34567]86-*-netbsdelf*) Likewise.
4971 (x86_64-*-netbsd*) Likewise.
4972 (mips*-*-netbsd*) Likewise.
4973 (powerpc-*-netbsd*) Likewise.
4974 (sh*-*-netbsd*) Likewise.
4975 (sparc-*-netbsdelf*) Likewise.
4976 (sparc64-*-netbsd*) Likewise.
4977 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
4979 (vax-*-netbsdelf*) Likewise.
4980 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
4981 (UINT_FAST8_TYPE) Likewise.
4982 (INT_FAST16_TYPE) Check CHAR_FAST16.
4983 (UINT_FAST16_TYPE) Likewise.
4985 2017-09-30 Jakub Jelinek <jakub@redhat.com>
4988 * config/i386/i386.md
4989 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
4990 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
4991 *divmodsi4_zext_2): New define_insn_and_split.
4992 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
4993 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
4994 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
4995 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
4996 New define_insn_and_split.
4997 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
4998 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
4999 operands[1] having DImode when mode is SImode.
5001 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
5002 always SImode for DIV and MOD in REG_EQUAL notes.
5004 2017-09-29 Yury Gribov <tetra2005@gmail.com>
5007 * match.pd: Fix handling of NaNs in pattern.
5009 2017-09-29 Jeff Law <law@redhat.com>
5011 * sbitmap.c (bitmap_bit_in_range_p): New function.
5012 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
5013 * tree-ssa-dse.c (live_bytes_read): New function.
5014 (dse_classify_store): Ignore reads of dead bytes.
5016 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
5017 typos and whitespace errors.
5018 * config/i386/predicates.md (address_no_seg_operand): Likewise.
5019 * config/s390/s390.c (s390_emit_prologue): Likewise.
5021 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
5024 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
5025 with a symbol for LRA.
5027 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
5029 PR rtl-optimization/82338
5030 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
5032 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
5034 * genmodes.c (calc_wider_mode): Suppress qsort macro.
5035 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
5036 (qsort_chk): Declare.
5037 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
5038 (qsort_chk): New function.
5040 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5042 PR tree-optimization/82337
5043 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
5044 phi definition if the PHI result appears in an abnormal PHI.
5045 (find_basis_for_base_expr): Don't record a basis if the LHS of the
5046 basis appears in an abnormal PHI.
5048 2017-09-29 Richard Biener <rguenther@suse.de>
5050 * graphite-isl-ast-to-gimple.c
5051 (translate_isl_ast_to_gimple::set_codegen_error): New function.
5052 (binary_op_to_tree): Use it.
5053 (get_rename_from_scev): Likewise.
5054 (copy_loop_phi_nodes): Likewise.
5055 (copy_bb_and_scalar_dependences): Likewise.
5056 (translate_pending_phi_nodes): Likewise.
5058 2017-09-29 Jakub Jelinek <jakub@redhat.com>
5061 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
5062 for movabsq $(i32 << shift), r64.
5064 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
5066 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
5067 index when encoding %esp as %rsp to avoid 0x67 prefix.
5069 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
5071 * config/i386/i386.md (*movsf_internal, *movdf_internal):
5072 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
5074 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
5076 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
5077 Extensions with more than 16 double VFP registers.
5078 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
5079 to_clear_mask and all code related to it. Replace the remaining
5080 entry by a sbitmap and adapt code accordingly.
5082 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
5084 * brig-builtins.def: Change pure attributes to const.
5086 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5088 * config.gcc (default_gnu_indirect_function): Default to yes for
5089 sparc*-*-linux* with glibc.
5091 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5093 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
5094 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
5095 when creating .init_array and .fini_array sections with priority
5098 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
5101 * config/aarch64/aarch64.c
5102 (aarch64_builtin_support_vector_misalignment): Always return false
5103 when misalignment is unknown.
5105 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
5107 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
5108 this function to return false if the definition used by the swap
5109 instruction is artificial, or if the memory address from which the
5110 constant value is loaded is not represented by a base address held
5111 in a register or if the base address register is a frame or stack
5112 pointer. Additionally, return false if the base address of the
5113 loaded constant is a SYMBOL_REF but is not considered to be a
5115 (replace_swapped_load_constant): New function.
5116 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
5117 loaded constant vector with a load of a swapped constant vector.
5119 2017-09-27 Carl Love <cel@us.ibm.com>
5121 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
5122 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
5123 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
5126 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
5128 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
5129 first, always call autopref_rank_data otherwise.
5131 2017-09-27 Richard Biener <rguenther@suse.de>
5133 * graphite-scop-detection.c (find_scop_parameters): Move
5134 loop bound handling ...
5135 (gather_bbs::before_dom_children): ... here, avoiding the need
5136 to build scop_info->loop_nest.
5137 (record_loop_in_sese): Remove.
5138 * sese.h (sese_info_t::loop_nest): Remove.
5139 * sese.c (new_sese_info): Do not allocate loop_nest.
5140 (free_sese_info): Do not free loop_nest.
5142 2017-09-27 Jakub Jelinek <jakub@redhat.com>
5145 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
5146 lhs from calls if the lhs has addressable type.
5148 2017-09-27 Richard Biener <rguenther@suse.de>
5150 * graphite.h (scop::max_alias_set): New member.
5151 * graphite-scop-detection.c: Remove references to non-existing
5152 --param in comments.
5153 (build_alias_sets): Record the maximum alias set used for drs.
5154 (build_scops): Support zero as unlimited for
5155 --param graphite-max-arrays-per-scop.
5156 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
5158 (build_poly_sr_1): ... here. Compute alias set based on the
5159 maximum alias set used for drs rather than
5160 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
5162 2017-09-27 Richard Biener <rguenther@suse.de>
5164 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
5165 --param loop-block-tile-size=0 to disable tiling.
5167 2017-09-27 Richard Biener <rguenther@suse.de>
5169 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
5170 (graphite-max-nb-scop-params): Document special value zero.
5171 * domwalk.h (dom_walker::STOP): New symbolical constant.
5172 (dom_walker::dom_walker): Add optional parameter for bb to
5174 (dom_walker::~dom_walker): Declare.
5175 (dom_walker::before_dom_children): Document STOP return value.
5176 (dom_walker::m_user_bb_to_rpo): New member.
5177 (dom_walker::m_bb_to_rpo): Likewise.
5178 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
5179 mapping here if not provided by the user.
5180 (dom_walker::~dom_walker): Free bb to RPO mapping if not
5181 provided by the user.
5182 (dom_walker::STOP): Define.
5183 (dom_walker::walk): Do not compute bb to RPO mapping here.
5184 Support STOP return value from before_dom_children to stop
5186 * graphite-optimize-isl.c (optimize_isl): If the schedule
5187 is the same still generate code if -fgraphite-identity
5188 or -floop-parallelize-all are given.
5189 * graphite-scop-detection.c: Include cfganal.h.
5190 (gather_bbs::gather_bbs): Get and pass through bb to RPO
5192 (gather_bbs::before_dom_children): Return STOP for BBs
5194 (build_scops): Compute bb to RPO mapping and pass it to
5195 the domwalk. Treat --param graphite-max-nb-scop-params=0
5196 as not limiting the number of params.
5197 * graphite.c (graphite_initialize): Remove limit on the
5198 number of basic-blocks in a function.
5199 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
5200 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
5201 default value of 10.
5203 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5205 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
5206 Adjust code to eliminate needing to do the shift right 32-bits
5207 operation after XSCVDPSPN.
5209 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5211 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
5212 ((X / Y) != 0 -> X >= Y): Likewise.
5214 2017-09-26 Carl Love <cel@us.ibm.com>
5216 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
5217 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
5218 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
5219 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
5220 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
5221 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
5222 definitions and overloading.
5223 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
5224 statement for P9V_BUILTIN_XST_LEN_R.
5225 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
5226 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
5227 define_expand and define_insn for the instructions and builtins.
5228 * doc/extend.texi: Update the built-in documentation file for the new
5230 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
5231 define_insn for the instructions
5233 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
5236 * gcc/config/netbsd-protos.h: New file.
5237 * gcc/config/netbsd.c: New file.
5238 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
5239 * gcc/config/t-netbsd: New file.
5240 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
5241 (tmake_file) Add t-netbsd.
5242 (extra_objs) Add netbsd.o.
5244 2017-09-26 Janus Weil <janus@gcc.gnu.org>
5248 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
5250 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5252 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
5253 sign extension from a vector register to a GPR by doing a 32-bit
5254 direct move and then an EXTSW.
5255 (extendsi<mode>2 splitter): Likewise.
5256 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
5257 right or vector extract after doing XSCVDPSPN. Use
5258 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
5260 (movdi_from_sf_zero_ext): Likewise.
5261 (reload_gpr_from_vsxsf): Likewise.
5262 (p8_mfvsrd_4_disf): Delete, no longer used.
5263 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
5264 then needing to move the SFmode to a GPR to use the XSCVDPSP
5265 instruction instead of FRSP and XSCVDPSPN.
5266 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
5267 it is adjacent to the other XSCVSPDP insns.
5268 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
5269 SFmode to be in traditional Altivec registers.
5270 (vsx_xscvdpspn): Eliminate useless alternative constraint.
5271 (vsx_xscvspdpn): Likewise.
5272 (vsx_xscvspdpn_scalar): Likewise.
5274 2017-09-26 Martin Jambor <mjambor@suse.cz>
5276 * tree-sra.c (compare_access_positions): Put integral types first,
5277 stabilize sorting of integral types, remove conditions putting
5278 non-full-precision integers last.
5279 (sort_and_splice_var_accesses): Disable scalarization if a
5280 non-integert would be represented by a non-full-precision integer.
5282 2017-09-26 Joseph Myers <joseph@codesourcery.com>
5284 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
5285 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
5286 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
5287 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
5288 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
5289 conditionals inside the function instead of around it. Call
5290 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
5291 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
5293 2017-09-26 Richard Biener <rguenther@suse.de>
5295 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
5297 (scop_detection::build_scop_breadth): ... this. Removed.
5298 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
5299 (scop_detection::harmful_stmt_in_bb): Likewise.
5300 (scop_detection::graphite_can_represent_stmt): Likewise.
5301 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
5302 (scop_detection::can_represent_loop): Remove recursion, fold in ...
5303 (scop_detection::can_represent_loop_1): ... this. Removed.
5304 (scop_detection::harmful_loop_in_region): Simplify after inlining
5305 the above and remove more quadraticness.
5306 (build_scops): Adjust.
5307 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
5310 2017-09-26 Jakub Jelinek <jakub@redhat.com>
5313 * config/i386/i386.c (ix86_print_operand_address_as): Only test
5314 REGNO (base) == SP_REG if base is a REG.
5317 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
5318 if it is different SSA_NAME.
5319 (optimize_range_tests_cmp_bitwise): New function.
5320 (optimize_range_tests): Call it.
5322 2017-09-26 Richard Biener <rguenther@suse.de>
5324 PR tree-optimization/82321
5325 * graphite.c (canonicalize_loop_closed_ssa): Properly check
5326 for the def being inside the loop.
5328 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5330 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
5332 * config/s390/s390-builtins.def: Fix constraint on op4.
5334 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5336 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
5337 independent expanders.
5338 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
5339 ("vec_ordered", "vec_unordered"): New expanders.
5341 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5343 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
5346 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5348 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
5349 vec_unpacks_lo_v16qi.
5350 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
5352 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5354 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
5355 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
5356 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
5358 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5360 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
5362 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
5364 ("*vec_slb<mode>"): New insn pattern.
5365 ("vec_shr_<mode>"): New expander.
5366 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
5367 and force the shift count operand to V16QImode.
5368 ("vec_srb<mode>"): Set shift count mode to V16QI.
5370 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5372 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
5373 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
5374 ("vec_widen_smult_hi_<mode>"): New expander definitions.
5376 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
5379 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
5381 2017-09-26 Richard Biener <rguenther@suse.de>
5383 PR tree-optimization/82320
5384 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
5387 2017-09-25 Jeff Law <law@redhat.com>
5389 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
5390 prototype for new argument.
5391 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
5392 mostly extracted from rs6000_emit_allocate_stack.
5393 (rs6000_emit_probe_stack_range_stack_clash): New function.
5394 (rs6000_emit_allocate_stack): Call
5395 rs6000_emit_probe_stack_range_stack_clash as needed.
5396 (rs6000_emit_probe_stack_range): Add additional argument
5397 to call to gen_probe_stack_range{si,di}.
5398 (output_probe_stack_range): New.
5399 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
5400 (output_probe_stack_range_stack_clash): New.
5401 (rs6000_emit_prologue): Emit notes into dump file as requested.
5402 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
5403 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
5404 Add additional operand and pass it to output_probe_stack_range.
5406 2017-09-25 Bin Cheng <bin.cheng@arm.com>
5408 PR tree-optimization/82163
5409 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
5410 (checking_verify_loop_closed_ssa): New parameter.
5411 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
5412 (check_loop_closed_ssa_stmt): Delete.
5413 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
5414 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
5415 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
5418 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
5420 * brig-builtins.def: Treat HSAIL barrier builtins as
5421 setjmp/longjump style functions.
5423 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
5425 * target.def (constant_alignment): New hook.
5426 * defaults.h (CONSTANT_ALIGNMENT): Delete.
5427 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
5428 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
5429 * doc/tm.texi: Regenerate.
5430 * targhooks.h (default_constant_alignment): Declare.
5431 (constant_alignment_word_strings): Likewise.
5432 * targhooks.c (default_constant_alignment): New function.
5433 (constant_alignment_word_strings): Likewise.
5434 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
5435 instead of CONSTANT_ALIGNMENT.
5436 * varasm.c (align_variable, get_variable_align, build_constant_desc)
5437 (force_const_mem): Likewise.
5438 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
5439 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
5440 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
5441 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5442 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
5444 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
5445 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5446 constant_alignment_word_strings.
5447 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
5448 (CONSTANT_ALIGNMENT): Likewise.
5449 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5450 (arm_constant_alignment): New function.
5451 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
5452 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5453 constant_alignment_word_strings.
5454 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
5455 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5456 constant_alignment_word_strings.
5457 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
5458 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5459 (cris_constant_alignment): New function.
5460 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
5461 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5462 (epiphany_constant_alignment): New function.
5463 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
5464 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5465 constant_alignment_word_strings.
5466 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
5467 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5468 constant_alignment_word_strings.
5469 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
5470 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5471 constant_alignment_word_strings.
5472 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
5473 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
5474 * config/i386/i386.c (ix86_constant_alignment): Make static.
5475 Use the same interface as the target hook.
5476 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5477 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
5478 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5479 constant_alignment_word_strings.
5480 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
5481 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
5482 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5483 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
5484 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5485 constant_alignment_word_strings.
5486 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
5487 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5488 constant_alignment_word_strings.
5489 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
5490 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5491 constant_alignment_word_strings.
5492 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
5493 * config/microblaze/microblaze.c (microblaze_constant_alignment):
5495 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5496 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
5497 * config/mips/mips.c (mips_constant_alignment): New function.
5498 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5499 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
5500 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
5501 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5502 (mmix_constant_alignment): Make static. Use the same interface
5504 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
5505 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5506 constant_alignment_word_strings.
5507 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
5508 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5509 constant_alignment_word_strings.
5510 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
5511 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5512 constant_alignment_word_strings.
5513 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
5514 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5515 (rs6000_constant_alignment): New function.
5516 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
5517 * config/riscv/riscv.c (riscv_constant_alignment): New function.
5518 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5519 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
5520 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5521 (rs6000_constant_alignment): New function.
5522 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
5523 * config/s390/s390.c (s390_constant_alignment): New function.
5524 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5525 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
5526 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5527 constant_alignment_word_strings.
5528 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
5529 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5530 (sparc_constant_alignment): New function.
5531 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
5532 * config/spu/spu.c (spu_constant_alignment): New function.
5533 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5534 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
5535 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5536 constant_alignment_word_strings.
5537 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
5538 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5539 constant_alignment_word_strings.
5540 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
5541 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5542 constant_alignment_word_strings.
5543 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
5544 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5545 (visium_constant_alignment): New function.
5546 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
5547 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5548 (xtensa_constant_alignment): New function.
5549 * system.h (CONSTANT_ALIGNMENT): Poison.
5551 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
5553 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
5554 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
5555 (rs6000_builtin_valid_without_lhs): New helper function.
5556 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5557 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
5559 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
5561 * target.h (vec_perm_indices): Use unsigned short rather than
5563 (auto_vec_perm_indices): Likewise.
5564 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
5565 Use unsigned int rather than unsigned char.
5566 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
5568 2017-09-25 Richard Biener <rguenther@suse.de>
5570 * cfgloop.h (sort_sibling_loops): Declare.
5571 * cfgloop.c (sort_sibling_loops_cmp): New helper.
5572 (sort_sibling_loops): New function sorting the sibling loop list
5574 * graphite.c (graphite_transform_loops): Sort sibling loops.
5576 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
5578 * target.def (vec_perm_const_ok): Change sel parameter to
5580 * optabs-query.c (can_vec_perm_p): Update accordingly.
5581 * doc/tm.texi: Regenerate.
5582 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
5583 auto_vec_perm_indices and remove separate nelt field.
5584 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
5585 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
5586 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
5587 (aarch64_expand_vec_perm_const): Update accordingly.
5588 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
5589 to vec_perm_indices.
5590 * config/arm/arm.c (expand_vec_perm_d): Change perm to
5591 auto_vec_perm_indices and remove separate nelt field.
5592 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
5593 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
5594 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
5596 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
5597 to vec_perm_indices.
5598 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
5599 sel to vec_perm_indices.
5600 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
5601 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
5602 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
5604 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
5607 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
5610 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
5611 on the FUNCTION_DECL function context if it has a DIE that is a
5614 2017-09-25 Richard Biener <rguenther@suse.de>
5616 PR tree-optimization/82285
5617 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
5620 2017-09-25 Tom de Vries <tom@codesourcery.com>
5624 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
5627 2017-09-25 Richard Biener <rguenther@suse.de>
5629 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
5630 ISL errors other than isl_error_quota happen. Dump if the
5631 schedule is the same.
5632 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
5633 errors instead of aborting inside ISL.
5635 2017-09-25 Iain Sandoe <iain@codesourcery.com>
5638 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
5639 of libgcc_eh for m64.
5640 * config/i386/darwin64.h: Likewise.
5642 2017-09-25 Richard Biener <rguenther@suse.de>
5645 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
5646 attribute for incomplete types nor twice for complete ones.
5648 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
5651 * config/i386/i386.c (ix86_print_operand_address_as): Encode
5652 %esp as %rsp to avoid 0x67 prefix if there is no index or base
5655 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
5658 * config/i386/i386.opt (mprefer-avx256): Use
5659 ix86_target_flags variable.
5660 * config/i386/i386.c (ix86_target_string): Move
5661 -mprefer-avx256 to flag2_opts.
5663 2017-09-22 Jakub Jelinek <jakub@redhat.com>
5666 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
5667 and x != -1 | y != -1 into (x & y) != -1.
5669 2017-09-22 Steve Ellcey <sellcey@cavium.com>
5671 * config.gcc: Add new case statement to set
5672 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
5673 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
5674 s390x-*-linux* case statements. Added aarch64 to the list of
5675 supported architectures.
5677 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
5679 PR tree-optimization/82289
5680 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
5681 STMT_VINFO_RELEVANT_P.
5683 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
5684 Alan Hayward <alan.hayward@arm.com>
5685 David Sherwood <david.sherwood@arm.com>
5687 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
5688 for VR_RANGE only; don't allow VR_ANTI_RANGE.
5689 (extract_range_from_binary_expr_1): Don't call
5690 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
5692 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
5693 Alan Hayward <alan.hayward@arm.com>
5694 David Sherwood <david.sherwood@arm.com>
5696 * target.def (preferred_vector_alignment): New hook.
5697 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
5699 * doc/tm.texi: Regenerate.
5700 * targhooks.h (default_preferred_vector_alignment): Declare.
5701 * targhooks.c (default_preferred_vector_alignment): New function.
5702 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
5704 (DR_TARGET_ALIGNMENT): New macro.
5705 (aligned_access_p): Update commentary.
5706 (vect_known_alignment_in_bytes): New function.
5707 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
5709 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
5710 Calculate the misalignment based on the target alignment rather than
5712 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
5713 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
5714 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
5715 the target alignment, rather than masking the element misalignment
5716 with the number of elements in a vector. Also use the target
5717 alignment when calculating the maximum number of peels.
5718 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
5719 instead of TYPE_ALIGN_UNIT.
5720 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
5721 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
5722 (vect_create_addr_base_for_vector_ref): Update call accordingly.
5723 (vect_create_data_ref_ptr): Likewise.
5724 (vect_setup_realignment): Realign by ANDing with
5725 -DR_TARGET_MISALIGNMENT.
5726 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
5727 the number of peels based on DR_TARGET_ALIGNMENT.
5728 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
5729 with the guaranteed alignment boundary when deciding whether
5731 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
5732 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
5733 (ensure_base_align): Remove stmt_info parameter. Get the
5734 target base alignment from DR_TARGET_ALIGNMENT.
5735 (vectorizable_store): Update call accordingly. Interpret
5736 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
5738 (vectorizable_load): Likewise.
5740 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
5741 Alan Hayward <alan.hayward@arm.com>
5742 David Sherwood <david.sherwood@arm.com>
5744 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
5745 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
5746 (vect_enhance_data_refs_alignment): Likewise.
5748 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
5750 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
5751 error. Only quit immediately if parsing is complete.
5752 (BEGIN): Initialize fatal_err and parse_done.
5753 (begin fpu, end fpu): Check number of arguments.
5754 (begin arch, end arch): Likewise.
5755 (begin cpu, end cpu): Likewise.
5756 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
5757 (optalias): Likewise.
5759 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
5761 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
5762 * config/arm/arm-isa.h: Delete. Move definitions to ...
5763 * arm-cpus.in: ... here. Use new feature and fgroup values.
5764 * config/arm/arm.c (arm_option_override): Use lower case for feature
5766 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
5767 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
5768 * config/arm/parsecpu.awk (END): Add new command 'isa'.
5770 (print_isa_bits_for): New function.
5771 (gen_isa): New function.
5772 (gen_comm_data): Use print_isa_bits_for.
5773 (define feature): New keyword.
5774 (define fgroup): New keyword.
5775 * config/arm/t-arm (TM_H): Remove.
5776 (GTM_H): Add arm-isa.h.
5777 (arm-isa.h): Add rule to generate file.
5778 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
5779 case for feature bit names.
5781 2017-09-22 Richard Biener <rguenther@suse.de>
5783 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
5785 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
5786 print of no dependency loops ...
5787 * graphite.c (graphite_transform_loops): ... here.
5788 (canonicalize_loop_closed_ssa_form): Work from inner to outer
5790 (same_close_phi_node, remove_duplicate_close_phi,
5791 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
5792 (canonicalize_loop_closed_ssa): ... here and simplify.
5793 * graphite-optimize-isl.c: Include tree-vectorizer.h.
5794 (optimize_isl): Use dump_printf_loc to tell when we stopped
5795 optimizing because of an ISL timeout.
5797 2017-09-22 Richard Biener <rguenther@suse.de>
5799 PR tree-optimization/82291
5800 * tree-if-conv.c (predicate_mem_writes): Make sure to
5801 remove writes in blocks predicated with false.
5803 2017-09-22 Richard Biener <rguenther@suse.de>
5805 * sese.c: Include cfganal.h.
5806 (if_region_set_false_region): Remove.
5807 (create_if_region_on_edge): Likewise.
5808 (move_sese_in_condition): Re-implement without destroying
5811 2017-09-22 Richard Biener <rguenther@suse.de>
5813 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
5814 Verify both BBs contain loop PHI nodes before dispatching to
5816 (graphite_regenerate_ast_isl): Do not recompute dominators,
5817 do not verify three times. Restructure for clarity.
5818 * graphite-scop-detection.c (same_close_phi_node,
5819 remove_duplicate_close_phi, make_close_phi_nodes_unique,
5820 defined_in_loop_p, canonicalize_loop_closed_ssa,
5821 canonicalize_loop_closed_ssa_form): Simplify, remove excess
5822 checking and SSA rewrite, move to ...
5823 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
5824 (graphite_initialize): Do not pass in ctx, do not reset the
5825 SCEV cache, compute only dominators.
5826 (graphite_transform_loops): Allocate ISL ctx after
5827 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
5828 Maintain post-dominators only around build_scops.
5829 * sese.c (if_region_set_false_region): Make static. Free
5830 and recompute dominators.
5831 (move_sese_in_condition): Assert we don't get called with
5832 post-dominators computed.
5833 * sese.h (if_region_set_false_region): Remove.
5835 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
5837 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
5838 mode attribute for TARGET_AVX512VL.
5840 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
5842 * config/i386/i386.opt (mprefer-avx256): New option.
5843 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
5845 (ix86_preferred_simd_mode): Return 256-bit AVX modes
5846 for TARGET_PREFER_AVX256.
5847 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
5849 2017-09-21 Jeff Law <law@redhat.com>
5851 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
5852 Fix dump output if the only stack space is for pushed registers.
5854 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5856 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
5859 2017-09-21 Martin Sebor <msebor@redhat.com>
5862 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
5863 code (in C++) or code that triggers warnings.
5865 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
5867 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
5869 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
5871 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
5872 * sched-rgn.c: Ditto.
5873 * sel-sched-ir.c: Ditto.
5875 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
5877 * toplev.h (set_random_seed): Adjust return type.
5878 * toplev.c (init_local_tick): Move eager initialization of random_seed
5879 to get_random_seed. Adjust comment.
5880 (init_random_seed): Inline to get_random_seed, delete.
5881 (get_random_seed): Initialize random_seed lazily.
5882 (set_random_seed): Do not return previous value.
5883 (print_switch_value): Do not call get_random_seed.
5885 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
5887 * cgraph.c (delete_function_version): New, broken out from...
5888 (cgraph_node::delete_function_version): ...here. Rename to
5889 cgraph_node::delete_function_version_by_decl. Update all uses.
5890 (cgraph_node::remove): Call delete_function_version.
5892 2017-09-21 Jakub Jelinek <jakub@redhat.com>
5895 * tree-inline.c (expand_call_inline): Emit clobber stmts for
5896 VAR_DECLs to which addressable non-volatile parameters are mapped
5897 and for id->retvar after the return value assignment. Clear
5898 id->retval and id->retbnd after inlining.
5900 2017-09-21 Richard Biener <rguenther@suse.de>
5902 PR tree-optimization/82276
5903 PR tree-optimization/82244
5904 * tree-vrp.c (build_assert_expr_for): Set
5905 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
5907 (remove_range_assertions): Revert earlier change.
5909 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
5912 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
5914 2017-09-21 Richard Biener <rguenther@suse.de>
5916 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5917 Restore valid IL after code generation errors.
5918 * graphite.c (graphite_transform_loops): Diagnose code
5919 generation issues as MSG_MISSED_OPTIMIZATION and continue
5920 with processing SCOPs.
5922 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5923 Alan Hayward <alan.hayward@arm.com>
5924 David Sherwood <david.sherwood@arm.com>
5926 * calls.c (compute_argument_addresses): Use simplify_gen_binary
5927 rather than choosing between plus_constant and gen_rtx_<CODE>.
5928 * expr.c (emit_push_insn): Likewise.
5929 (expand_expr_real_2): Likewise.
5931 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5932 Alan Hayward <alan.hayward@arm.com>
5933 David Sherwood <david.sherwood@arm.com>
5935 * loop-unroll.c (split_iv): Call copy_rtx on the step.
5937 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5938 Alan Hayward <alan.hayward@arm.com>
5939 David Sherwood <david.sherwood@arm.com>
5941 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
5942 calling tree_to_uhwi.
5944 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5945 Alan Hayward <alan.hayward@arm.com>
5946 David Sherwood <david.sherwood@arm.com>
5948 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
5949 INTEGER_CST rather than a negative test for ADDR_EXPR.
5951 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5952 Alan Hayward <alan.hayward@arm.com>
5953 David Sherwood <david.sherwood@arm.com>
5955 * tree-vrp.c (extract_range_from_binary_expr_1): Check
5956 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
5958 2017-09-21 Richard Biener <rguenther@suse.de>
5960 PR tree-optimization/71351
5961 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
5962 graphite_create_new_loop_guard): Remove, fold remaining parts
5964 (translate_isl_ast_node_for): ... here and simplify.
5966 2017-09-21 Jakub Jelinek <jakub@redhat.com>
5969 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
5970 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
5971 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
5972 alternative always use QI mode, for -Os imov (=R,R) alternative
5973 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
5976 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5977 Jeff Law <law@redhat.com>
5979 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
5980 (allocate_stack_space): New function, partially extracted from
5982 (s390_emit_prologue): Track offset to most recent stack probe.
5983 Code to allocate space moved into allocate_stack_space.
5984 Dump actions when no stack is allocated.
5985 (s390_prologue_plus_offset): New function.
5986 (s390_emit_stack_probe): Likewise.
5988 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
5990 * common.opt (Wa, Wl, Wp, g, gz=): Add
5992 (gno-column-info): Remove.
5993 (gcolumn-info): Drop RejectNegative.
5995 (gno-record-gcc-switches): Remove.
5996 (grecord-gcc-switches): Drop RejectNegative.
5997 (gno-split-dwarf): Remove.
5998 (gsplit-dwarf): Drop RejectNegative.
5999 (gno-strict-dwarf): Remove.
6000 (gstrict-dwarf): Drop RejectNegative.
6001 * config/darwin.opt (gfull, gused): Add RejectNegative.
6002 * dwarf2out.c (gen_producer_string): Drop
6003 gno-record-gcc-switches handler.
6004 * optc-gen.awk: Add g to prefixes with negative forms.
6005 * opts-common.c (remapping_prefix_p): New.
6006 (find_opt): Check it.
6007 (generate_canonical_option): Test g prefix.
6008 (option_map): Add -gno- mapping.
6009 (add_misspelling_candidates): Check remapping_prefix_p.
6011 2017-09-20 Jeff Law <law@redhat.com>
6013 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
6014 thinko in stack clash protection support.
6016 * explow.c (compute_stack_clash_protection_loop_data): Use
6017 CONST_INT_P instead of explicit test. Verify object is a
6018 CONST_INT_P before looking at INTVAL.
6019 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
6020 instead of explicit test.
6022 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
6025 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
6026 address instead of to r1 and r11.
6028 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
6030 * config.gcc: Support "knm".
6031 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
6032 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
6034 * config/i386/i386.c (m_KNM): Define.
6035 (processor_target_table): Add "knm".
6037 (ix86_option_override_internal): Add "knm".
6038 (ix86_issue_rate): Add PROCESSOR_KNM.
6039 (ix86_adjust_cost): Ditto.
6040 (ia32_multipass_dfa_lookahead): Ditto.
6041 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
6042 (fold_builtin_cpu): Add M_INTEL_KNM.
6043 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
6044 (processor_type): Add PROCESSOR_KNM.
6045 * config/i386/x86-tune.def: Add m_KNM.
6046 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
6048 2017-09-20 Richard Biener <rguenther@suse.de>
6050 PR tree-optimization/80213
6051 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
6052 are allowed in empty BBs as well.
6053 (canonicalize_loop_closed_ssa): Also look for other complex
6055 (scop_detection::get_sese): Include the loop-closed PHI block
6057 (scop_detection::merge_sese): Remove code adding extra blocks.
6058 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
6059 (build_scops): Assert the final returned scop is invalid.
6061 2017-09-20 Richard Biener <rguenther@suse.de>
6063 PR tree-optimization/82264
6064 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
6066 (vn_phi_lookup): Likewise.
6067 (vn_phi_insert): Likewise.
6069 2017-09-20 Jakub Jelinek <jakub@redhat.com>
6071 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
6072 that fits into uhwi or shwi, add DW_AT_const_value regardless
6073 of early_dwarf without going through RTL, using add_AT_unsigned
6076 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
6077 (DEBUG_LTO_ABBREV_SECTION): Likewise.
6078 (DEBUG_LTO_MACINFO_SECTION): Likewise.
6079 (DEBUG_MACRO_SECTION): Likewise.
6080 (DEBUG_LTO_MACRO_SECTION): Likewise.
6081 (DEBUG_STR_DWO_SECTION): Likewise.
6082 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
6083 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
6084 (DEBUG_LTO_DWO_LINE_SECTION): Define.
6085 (DEBUG_LTO_LINE_STR_SECTION): Define.
6086 (init_sections_and_labels): Initialize debug_line_str_section
6087 variable. Initialize debug_loc_section for -gdwarf-5 to
6088 DEBUG_LOCLISTS_SECTION. Formatting fixes.
6090 2017-09-20 Richard Biener <rguenther@suse.de>
6092 * graphite-sese-to-poly.c (extract_affine): Properly handle
6093 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
6095 2017-09-20 Richard Biener <rguenther@suse.de>
6097 PR tree-optimization/81373
6098 * graphite-scop-detection.c (build_cross_bb_scalars_def):
6099 Force SESE live-out defs to be handled even if they are
6102 2017-09-19 Jeff Law <law@redhat.com>
6104 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
6105 nothing for stack adjustments with REG_STACK_CHECK.
6106 * sched-deps.c (parse_add_or_inc): Reject insns with
6107 REG_STACK_CHECK from dependency breaking.
6108 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
6109 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
6110 * reg-notes.def (STACK_CHECK): New note.
6112 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
6113 (ix86_expand_prologue): Dump stack clash info as needed.
6114 Call ix86_adjust_stack_and_probe_stack_clash as needed.
6116 * function.c (dump_stack_clash_frame_info): New function.
6117 * function.h (dump_stack_clash_frame_info): Prototype.
6118 (enum stack_clash_probes): New enum.
6120 * config/alpha/alpha.c (alpha_expand_prologue): Also check
6121 flag_stack_clash_protection.
6122 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
6123 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
6124 (arm_frame_pointer_required): Likewise.
6125 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
6126 (ia64_expand_prologue): Likewise.
6127 * config/mips/mips.c (mips_expand_prologue): Likewise.
6128 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
6129 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6130 (sparc_flat_expand_prologue): Likewise.
6131 * config/spu/spu.c (spu_expand_prologue): Likewise.
6133 * explow.c: Include "params.h".
6134 (anti_adjust_stack_and_probe_stack_clash): New function.
6135 (get_stack_check_protect): Likewise.
6136 (compute_stack_clash_protection_loop_data): Likewise.
6137 (emit_stack_clash_protection_loop_start): Likewise.
6138 (emit_stack_clash_protection_loop_end): Likewise.
6139 (allocate_dynamic_stack_space): Use get_stack_check_protect.
6140 Use anti_adjust_stack_and_probe_stack_clash.
6141 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
6142 (emit_stack_clash_protection_loop_start): Likewise.
6143 (emit_stack_clash_protection_loop_end): Likewise.
6144 * rtl.h (get_stack_check_protect): Prototype.
6145 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
6146 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
6147 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
6149 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
6151 * doc/tm.texi: Rebuilt.
6152 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
6153 get_stack_check_protect.
6154 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
6155 * config/arm/arm.c (arm_expand_prologue): Likewise.
6156 (arm_frame_pointer_required): Likewise.
6157 * config/i386/i386.c (ix86_expand_prologue): Likewise.
6158 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
6159 * config/mips/mips.c (mips_expand_prologue): Likewise.
6160 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
6161 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
6162 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6163 (sparc_flat_expand_prologue): Likewise.
6165 * common.opt (-fstack-clash-protection): New option.
6166 * flag-types.h (enum stack_check_type): Note difference between
6167 -fstack-check= and -fstack-clash-protection.
6168 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
6169 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
6170 * toplev.c (process_options): Issue warnings/errors for cases
6171 not handled with -fstack-clash-protection.
6172 * doc/invoke.texi (-fstack-clash-protection): Document new option.
6173 (-fstack-check): Note additional problem with -fstack-check=generic.
6174 Note that -fstack-check is primarily for Ada and refer users
6175 to -fstack-clash-protection for stack-clash-protection.
6176 Document new params for stack clash protection.
6178 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
6180 * config/i386/i386.c (ix86_split_long_move): Do not handle
6181 address used for LEA in a special way.
6183 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
6185 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
6187 2017-09-19 Martin Sebor <msebor@redhat.com>
6190 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
6191 of incompatible types.
6193 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
6195 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6196 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
6197 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6198 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
6200 2017-09-19 Richard Biener <rguenther@suse.de>
6202 PR tree-optimization/82244
6203 * tree-vrp.c (remove_range_assertions): Do not propagate
6204 a constant to abnormals but replace the assert with a copy.
6206 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
6208 PR rtl-optimization/57878
6209 PR rtl-optimization/68988
6210 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
6211 avoidance test involving non_reload_pseudos. Move frequency test
6212 below the general fragmentation avoidance test.
6214 2017-09-19 Richard Biener <rguenther@suse.de>
6216 PR tree-optimization/69728
6217 * graphite-sese-to-poly.c (schedule_error): New global.
6218 (add_loop_schedule): Handle empty domain by failing the
6220 (build_original_schedule): Handle schedule_error.
6222 2017-09-19 Richard Biener <rguenther@suse.de>
6224 * graphite-scop-detection.c (scop_detection::can_represent_loop):
6225 Do not iterate to sibling loops but only to siblings of inner
6228 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
6231 * config/m68k/m68k.md (moveq feeding equality comparison): Check
6232 that the registers are different.
6234 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
6236 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
6237 to processor_model and "amdfam17h" to arch_names_table.
6238 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
6240 2017-09-18 Jakub Jelinek <jakub@redhat.com>
6243 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
6245 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6246 Alan Hayward <alan.hayward@arm.com>
6247 David Sherwood <david.sherwood@arm.com>
6249 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
6251 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
6253 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
6254 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
6255 vect_schedule_slp_instance.
6256 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
6257 Update call to vect_slp_analyze_node_operations. Simplify return
6259 (vect_slp_analyze_bb_1): Update call accordingly.
6260 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
6261 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
6262 (vect_schedule_slp): Update call accordingly.
6264 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6265 Alan Hayward <alan.hayward@arm.com>
6266 David Sherwood <david.sherwood@arm.com>
6268 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
6269 with types that aren't in fact scalar.
6271 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6273 * tree-vect-slp.c (vect_record_max_nunits): New function,
6275 (vect_build_slp_tree_1): ...here.
6276 (vect_build_slp_tree_2): Call it for phis too.
6278 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6280 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
6281 to vect_get_vec_def_for_operand when getting the mask operand.
6283 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6284 Alan Hayward <alan.hayward@arm.com>
6285 David Sherwood <david.sherwood@arm.com>
6287 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
6290 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6291 Alan Hayward <alan.hayward@arm.com>
6292 David Sherwood <david.sherwood@arm.com>
6294 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
6295 calculation for vector booleans.
6297 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6298 Alan Hayward <alan.hayward@arm.com>
6299 David Sherwood <david.sherwood@arm.com>
6301 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
6303 (vect_transform_stmt): ...here.
6304 (vect_analyze_stmt): Use it instead of calling
6305 vectorizable_live_operation directly.
6307 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
6309 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
6310 non-SIMT targets in acc vector loops.
6312 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
6314 * configure.ac: Add arc and check if assembler supports gdwarf2.
6315 * configure: Regenerate.
6317 2017-09-18 Richard Biener <rguenther@suse.de>
6319 PR tree-optimization/82220
6320 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
6321 epilogue niters from the min_profitable_iters compute.
6323 2017-09-18 Jakub Jelinek <jakub@redhat.com>
6326 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
6327 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
6328 (ix86_init_pic_reg): Revert 2017-09-01 changes.
6330 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
6333 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
6334 switching to a new text section.
6336 2017-09-18 Richard Biener <rguenther@suse.de>
6338 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
6340 (build_alias_set): Reject aliases with no access function.
6342 2017-09-18 Richard Biener <rguenther@suse.de>
6344 PR tree-optimization/79622
6345 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
6347 (build_cross_bb_scalars_use): Likewise.
6349 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
6351 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
6353 2017-09-18 Alan Modra <amodra@gmail.com>
6356 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
6357 stack_pointer_rtx for count 0. Update comments. Break up
6358 large rtl expression.
6360 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
6362 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
6363 Increase to 20 bytes.
6364 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
6365 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
6366 or avx version of the stub.
6368 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
6371 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
6372 compute the minimum stack alignment. Also update preferred stack
6373 boundary for leaf functions.
6375 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
6377 PR tree-optimization/82228
6378 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
6381 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
6383 * common/config/nds32/nds32-common.c
6384 (nds32_option_optimization_table): Refine formatting.
6385 (nds32_option_optimization_table): Use -fsched-pressure and
6386 -fomit-frame-pointer for specific optimization level.
6388 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
6390 * config/nds32/nds32.c: Refine formatting and comments.
6391 * config/nds32/nds32.h: Likewise.
6392 * config/nds32/nds32.md: Likewise.
6393 * config/nds32/nds32-cost.c: Likewise.
6394 * config/nds32/nds32-isr.c: Likewise.
6395 * config/nds32/nds32-md-auxiliary.c: Likewise.
6396 * config/nds32/nds32-multiple.md: Likewise.
6397 * config/nds32/nds32-predicates.c: Likewise.
6399 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
6400 Jakub Jelinek <jakub@redhat.com>
6402 Add support for -std=c++2a.
6403 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
6405 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
6407 2017-09-15 Steve Ellcey <sellcey@cavium.com>
6410 * doc/extend.texi (Common Function Attributes): Add
6411 references to ARM, AArch64, and S/390 specific attributes.
6412 (Function Specific Option Pragmas): Add AArch64 and S/390
6413 to list of back ends that support the target pragma.
6415 2017-09-15 Nathan Sidwell <nathan@acm.org>
6417 * doc/standards.texi: Fix C++17 description. Update URLs for
6420 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
6422 * common.opt (Wcast-align=strict): New warning option.
6423 * doc/invoke.texi: Document -Wcast-align=strict.
6425 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
6427 * cgraph.h (cgraph_thunk_info): Add comments.
6428 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
6429 assert for VIRTUAL_* arguments stricter.
6431 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
6433 PR tree-optimization/71026
6434 * match.pd: Move RDIV patterns from fold-const.c
6435 * fold-const.c (distribute_real_division): Removed.
6436 (fold_binary_loc): Remove calls to distribute_real_divison.
6438 2017-09-15 Jakub Jelinek <jakub@redhat.com>
6440 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
6441 c++1z and gnu++1z as deprecated. Change other references to
6442 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
6443 Change -Wc++1z-compat to -Wc++17-compat.
6444 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
6445 * dwarf2out.c (highest_c_language): Handle C++17.
6446 (gen_compile_unit_die): Likewise.
6448 2017-09-15 Jakub Jelinek <jakub@redhat.com>
6450 PR rtl-optimization/82192
6451 * combine.c (make_extraction): Don't look through non-paradoxical
6452 SUBREGs or TRUNCATE if pos + len is or might be bigger than
6455 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6456 Alan Hayward <alan.hayward@arm.com>
6457 David Sherwood <david.sherwood@arm.com>
6459 * target.def (function_arg_offset): New hook.
6460 * targhooks.h (default_function_arg_offset): Declare.
6461 * targhooks.c (default_function_arg_offset): New function.
6462 * function.c (locate_and_pad_parm): Use
6463 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
6464 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
6465 (TARGET_FUNCTION_ARG_OFFSET): ...this.
6466 * doc/tm.texi: Regenerate.
6467 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
6468 * config/spu/spu.c (spu_function_arg_offset): New function.
6469 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
6470 * system.h (FUNCTION_ARG_OFFSET): Poison.
6472 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6473 Alan Hayard <alan.hayward@arm.com>
6474 David Sherwood <david.sherwood@arm.com>
6476 * target.def (truly_noop_truncation): New hook.
6477 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
6478 than TRULY_NOOP_TRUNCATION.
6479 * hooks.h (hook_bool_uint_uint_true): Declare.
6480 * hooks.c (hook_bool_uint_uint_true): New function.
6481 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
6482 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
6483 * doc/tm.texi: Regenerate.
6484 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
6485 rather than TRULY_NOOP_TRUNCATION in comments.
6486 (simplify_comparison): Likewise.
6487 (record_truncated_value): Likewise.
6488 * expmed.c (extract_bit_field_1): Likewise.
6489 (extract_split_bit_field): Likewise.
6490 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
6491 instead of TRULY_NOOP_TRUNCATION.
6492 * function.c (assign_parm_setup_block): Likewise.
6493 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
6494 * rtlhooks.c: Include target.h.
6495 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
6496 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
6497 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
6498 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
6499 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
6500 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
6501 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
6502 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
6503 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
6504 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
6505 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
6506 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
6507 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
6508 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
6509 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
6510 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
6511 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
6512 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
6513 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
6514 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
6515 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
6516 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
6517 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
6518 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
6519 * config/mips/mips.c (mips_truly_noop_truncation): New function.
6520 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
6521 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
6522 rather than TRULY_NOOP_TRUNCATION in comments.
6523 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
6524 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
6525 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
6526 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
6527 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
6528 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
6529 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
6530 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
6531 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
6532 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
6533 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
6534 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
6535 rather than TRULY_NOOP_TRUNCATION in comments.
6536 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
6537 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
6538 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
6539 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
6540 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
6541 TRULY_NOOP_TRUNCATION condition.
6542 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
6543 (TRULY_NOOP_TRUNCATION): Delete.
6544 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
6545 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
6546 * config/spu/spu.c (spu_truly_noop_truncation): New function.
6547 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
6548 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
6549 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
6550 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
6551 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
6552 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
6553 rather than TRULY_NOOP_TRUNCATION in comments.
6554 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
6555 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
6556 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
6557 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
6558 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
6559 * system.h (TRULY_NOOP_TRUNCATION): Poison.
6561 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
6564 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
6565 (*cmp_ior): Likewise.
6566 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
6567 (*ior_scc_scc_cmp): Likewise.
6568 (*and_scc_scc): Likewise.
6569 (*and_scc_scc_cmp): Likewise.
6571 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6572 Alan Hayard <alan.hayward@arm.com>
6573 David Sherwood <david.sherwood@arm.com>
6575 * target.def (can_change_mode_class): New hook.
6576 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
6577 (hard_regno_nregs): Likewise.
6578 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
6579 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
6580 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
6581 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
6582 (LOAD_EXTEND_OP): Update accordingly.
6583 * doc/tm.texi: Regenerate.
6584 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
6585 CANNOT_CHANGE_MODE_CLASS.
6586 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
6587 (REG_CAN_CHANGE_MODE_P): ...this new macro.
6588 * combine.c (simplify_set): Update accordingly.
6589 * emit-rtl.c (validate_subreg): Likewise.
6590 * recog.c (general_operand): Likewise.
6591 * regcprop.c (mode_change_ok): Likewise.
6592 * reload1.c (choose_reload_regs): Likewise.
6593 (inherit_piecemeal_p): Likewise.
6594 * rtlanal.c (simplify_subreg_regno): Likewise.
6595 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
6596 instead of CANNOT_CHANGE_MODE_CLASS.
6597 (reload_cse_simplify_operands): Likewise.
6598 * reload.c (push_reload): Use targetm.can_change_mode_class
6599 instead of CANNOT_CHANGE_MODE_CLASS.
6600 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
6601 REG_CANNOT_CHANGE_MODE_P.
6602 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6603 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
6604 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6605 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6606 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6607 (arm_can_change_mode_class): New function.
6608 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
6609 than CANNOT_CHANGE_MODE_CLASS in comments.
6610 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6611 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
6612 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
6613 (ix86_can_change_mode_class): ...this new function, inverting the
6614 sense of the return value.
6615 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6616 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6617 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6618 (ia64_can_change_mode_class): New function.
6619 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6620 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
6621 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
6622 (m32c_can_change_mode_class): ...this new function, inverting the
6623 sense of the return value.
6624 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6625 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6626 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
6627 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
6628 (mips_can_change_mode_class): ...this new function, inverting the
6629 sense of the return value.
6630 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6631 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6632 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6633 (msp430_can_change_mode_class): New function.
6634 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6635 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
6636 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6637 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6638 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6639 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
6640 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6641 (pa_cannot_change_mode_class): Replace with...
6642 (pa_can_change_mode_class): ...this new function, inverting the
6643 sense of the return value.
6644 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
6645 than CANNOT_CHANGE_MODE_CLASS in comments.
6646 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6647 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
6648 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6649 (pdp11_cannot_change_mode_class): Replace with...
6650 (pdp11_can_change_mode_class): ...this new function, inverting the
6651 sense of the return value.
6652 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6653 * config/powerpcspe/powerpcspe-protos.h
6654 (rs6000_cannot_change_mode_class_ptr): Delete.
6655 * config/powerpcspe/powerpcspe.c
6656 (rs6000_cannot_change_mode_class_ptr): Delete.
6657 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6658 (rs6000_option_override_internal): Assign to
6659 targetm.can_change_mode_class instead of
6660 rs6000_cannot_change_mode_class_ptr.
6661 (rs6000_cannot_change_mode_class): Replace with...
6662 (rs6000_can_change_mode_class): ...this new function, inverting the
6663 sense of the return value.
6664 (rs6000_debug_cannot_change_mode_class): Replace with...
6665 (rs6000_debug_can_change_mode_class): ...this new function.
6666 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6667 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
6668 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6669 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6670 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
6672 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
6673 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6674 (rs6000_option_override_internal): Assign to
6675 targetm.can_change_mode_class instead of
6676 rs6000_cannot_change_mode_class_ptr.
6677 (rs6000_cannot_change_mode_class): Replace with...
6678 (rs6000_can_change_mode_class): ...this new function, inverting the
6679 sense of the return value.
6680 (rs6000_debug_cannot_change_mode_class): Replace with...
6681 (rs6000_debug_can_change_mode_class): ...this new function.
6682 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6683 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
6684 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
6685 (s390_can_change_mode_class): ...this new function, inverting the
6686 sense of the return value.
6687 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6688 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6689 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
6690 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6691 (sh_cannot_change_mode_class): Replace with...
6692 (sh_can_change_mode_class): ...this new function, inverting the
6693 sense of the return value.
6694 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6695 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6696 (sparc_can_change_mode_class): New function.
6697 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6698 * config/spu/spu.c (spu_can_change_mode_class): New function.
6699 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6700 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6701 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6702 (visium_can_change_mode_class): New function.
6703 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
6705 2017-09-15 Richard Biener <rguenther@suse.de>
6707 PR tree-optimization/82217
6708 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
6709 but not undefined case.
6711 2017-09-15 Jakub Jelinek <jakub@redhat.com>
6714 * postreload.c (reload_cse_simplify_operands): Skip
6715 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
6717 2017-09-15 Richard Biener <rguenther@suse.de>
6719 PR tree-optimization/68823
6720 * graphite-scop-detection.c (build_alias_set): If we have a
6721 possible dependence check whether we can handle them by just
6722 looking at the DRs DR_ACCESS_FNs.
6723 (build_scops): If build_alias_set fails, fail the SCOP.
6725 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
6727 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
6728 to support float128 built-in functions that require the ISA 3.0
6730 (BU_FLOAT128_3_HW): Likewise.
6731 (SQRTF128): Add support for the IEEE 128-bit square root and fma
6733 (FMAF128): Likewise.
6735 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
6736 support for built-in functions that need the ISA 3.0 IEEE 128-bit
6737 floating point instructions.
6738 (rs6000_invalid_builtin): Likewise.
6739 (rs6000_builtin_mask_names): Likewise.
6740 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
6741 (RS6000_BTM_FLOAT128_HW): Likewise.
6742 (RS6000_BTM_COMMON): Likewise.
6743 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
6745 * doc/extend.texi (RS/6000 built-in functions): Document the
6746 IEEE 128-bit floating point square root and fused multiply-add
6749 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
6751 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
6752 reg (r2) isn't in the set of registers defined in the prologue.
6754 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6755 Alan Hayward <alan.hayward@arm.com>
6756 David Sherwood <david.sherwood@arm.com>
6758 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
6759 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
6760 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
6761 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
6762 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
6764 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
6765 max_vectorization_factor.
6766 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
6768 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6769 Alan Hayward <alan.hayward@arm.com>
6770 David Sherwood <david.sherwood@arm.com>
6772 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
6773 (vect_worthwhile_without_simd_p): Declare.
6774 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
6775 (vectorizable_reduction): Use it.
6776 * tree-vect-stmts.c (vectorizable_shift): Likewise.
6777 (vectorizable_operation): Likewise.
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-vectorizer.h (vect_get_num_copies): New function.
6784 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
6785 * tree-vect-loop.c (vectorizable_reduction): Likewise.
6786 (vectorizable_induction): Likewise.
6787 (vectorizable_live_operation): Likewise.
6788 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
6789 (vectorizable_bswap): Likewise.
6790 (vectorizable_call): Likewise.
6791 (vectorizable_conversion): Likewise.
6792 (vectorizable_assignment): Likewise.
6793 (vectorizable_shift): Likewise.
6794 (vectorizable_operation): Likewise.
6795 (vectorizable_store): Likewise.
6796 (vectorizable_load): Likewise.
6797 (vectorizable_condition): Likewise.
6798 (vectorizable_comparison): Likewise.
6799 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
6801 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6802 Alan Hayward <alan.hayward@arm.com>
6803 David Sherwood <david.sherwood@arm.com>
6805 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
6806 of vect_init_vector.
6808 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6809 Alan Hayward <alan.hayward@arm.com>
6810 David Sherwood <david.sherwood@arm.com>
6812 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
6813 an inline wrapper that provides a location.
6814 (gimple_build_vector): Likewise.
6815 * gimple-fold.c (gimple_build_vector_from_val): New function.
6816 (gimple_build_vector): Likewise.
6817 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
6818 functions to build the initial value. Always return a gimple value.
6819 (get_initial_defs_for_reduction): Likewise. Only compute
6821 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
6822 vect_init_vector on the results from get_initial_def(s)_for_reduction.
6823 (vectorizable_induction): Use gimple_build_vector rather than
6826 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6827 Alan Hayward <alan.hayward@arm.com>
6828 David Sherwood <david.sherwood@arm.com>
6830 * target.h (vec_perm_indices): New typedef.
6831 (auto_vec_perm_indices): Likewise.
6832 * optabs-query.h: Include target.h
6833 (can_vec_perm_p): Take a vec_perm_indices *.
6834 * optabs-query.c (can_vec_perm_p): Likewise.
6835 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
6836 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
6837 * tree-vect-generic.c (lower_vec_perm): Likewise.
6838 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
6839 (vect_grouped_load_supported): Likewise.
6840 (vect_shift_permute_load_chain): Likewise.
6841 (vect_permute_store_chain): Use auto_vec_perm_indices.
6842 (vect_permute_load_chain): Likewise.
6843 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
6844 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
6845 Update uses of can_vec_perm_p.
6846 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
6847 mode with a number of elements. Take a vec_perm_indices *.
6848 (vect_create_epilog_for_reduction): Update accordingly.
6849 Use auto_vec_perm_indices.
6850 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
6851 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
6852 (vect_transform_slp_perm_load): Likewise.
6853 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
6854 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
6855 (vect_gen_perm_mask_checked): Likewise.
6856 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
6857 (vect_gen_perm_mask_checked): Likewise.
6858 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
6859 (vectorizable_store): Likewise.
6860 (vectorizable_load): Likewise.
6861 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
6862 (vectorizable_bswap): Likewise.
6864 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6865 Alan Hayward <alan.hayward@arm.com>
6866 David Sherwood <david.sherwood@arm.com>
6868 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
6869 * tree.c (build_vector): Likewise.
6870 (build_vector_from_ctor): Update accordingly.
6871 (build_vector_from_val): Likewise.
6872 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
6873 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
6874 * tree-vect-generic.c (add_rshift): Likewise.
6875 (expand_vector_divmod): Likewise.
6876 (optimize_vector_constructor): Likewise.
6877 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
6878 (vect_transform_slp_perm_load): Likewise.
6879 (vect_schedule_slp_instance): Likewise.
6880 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
6881 (vectorizable_call): Likewise.
6882 (vect_gen_perm_mask_any): Likewise. Add elements in order.
6883 * expmed.c (make_tree): Likewise.
6884 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
6885 a vector passed to build_vector.
6886 (fold_convert_const): Likewise.
6887 (exact_inverse): Likewise.
6888 (fold_ternary_loc): Likewise.
6889 (fold_relational_const): Likewise.
6890 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
6891 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
6892 (const_unop): Likewise. Store the reduction accumulator in a
6893 variable rather than an array.
6894 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
6895 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
6896 the new vector, rather than constructing it after the input arrays.
6897 (native_interpret_vector): Use auto_vec<tree> when building
6898 a vector passed to build_vector. Add elements in order.
6899 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
6900 auto_vec<tree> when building a vector passed to build_vector.
6901 (vect_create_epilog_for_reduction): Likewise.
6902 (vectorizable_induction): Likewise.
6903 (get_initial_def_for_reduction): Likewise. Fix indentation of
6905 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
6907 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
6908 passed to build_vector.
6910 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6911 Alan Hayward <alan.hayward@arm.com>
6912 David Sherwood <david.sherwood@arm.com>
6914 * tree-core.h (tree_base::u): Add an "nelts" field.
6915 (tree_vector): Use VECTOR_CST_NELTS as the length.
6916 * tree.c (tree_size): Likewise.
6917 (make_vector): Initialize VECTOR_CST_NELTS.
6918 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
6919 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
6920 TYPE_VECTOR_SUBPARTS.
6921 * expr.c (const_vector_mask_from_tree): Consistently use "units"
6922 as the number of units, setting it from VECTOR_CST_NELTS.
6923 (const_vector_from_tree): Likewise.
6924 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
6925 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
6926 (fold_negate_expr_1): Likewise.
6927 (fold_convert_const): Likewise.
6928 (const_binop): Likewise. Differentiate the number of output and
6930 (const_unop): Likewise.
6931 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
6932 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
6933 in cases that did the opposite.
6935 2017-09-14 Richard Biener <rguenther@suse.de>
6937 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
6940 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
6942 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
6944 2017-09-14 Jakub Jelinek <jakub@redhat.com>
6947 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
6948 if and where to split a bb, except for splitting before debug insn
6949 sequences followed by non-label real insn. Delete debug insns
6950 in between basic blocks.
6952 * combine.c (make_compound_operation_int): Formatting fixes.
6954 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
6955 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6956 * config/netbsd.h (LINK_EH_SPEC): Likewise.
6957 * config/sol2.h (LINK_EH_SPEC): Likewise.
6958 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6959 * config/s390/linux.h (LINK_SPEC): Likewise.
6960 * config/freebsd.h (LINK_EH_SPEC): Likewise.
6961 * config/openbsd.h (LINK_EH_SPEC): Likewise.
6962 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6963 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
6964 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
6965 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6966 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
6967 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
6969 2017-09-13 Jakub Jelinek <jakub@redhat.com>
6971 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
6973 (ENDFILE_LINUX_SPEC): Likewise.
6974 (LINK_EH_SPEC): Likewise.
6975 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
6976 (LINK_OS_LINUX_SPEC32): Likewise.
6977 (LINK_OS_LINUX_SPEC64): Likewise.
6978 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
6979 (LINK_OS_LINUX_SPEC): Likewise.
6981 2017-09-13 Martin Liska <mliska@suse.cz>
6984 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
6985 CASE_HIGH is NULL_TREE.
6987 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
6988 Alan Hayward <alan.hayward@arm.com>
6989 David Sherwood <david.sherwood@arm.com>
6991 * target.def (secondary_memory_needed): New hook.
6992 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
6993 instead of SECONDARY_MEMORY_NEEDED.
6994 (secondary_memory_needed_mode): Likewise.
6995 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
6996 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
6997 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
6998 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
6999 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
7000 * doc/tm.texi: Regenerate.
7001 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
7002 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
7003 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7004 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
7005 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
7006 * config/i386/i386.c (inline_secondary_memory_needed): Put the
7007 mode argument first and change the reg_class arguments to reg_class_t.
7008 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
7009 Make static. Update the call to inline_secondary_memory_needed.
7010 (ix86_register_move_cost): Update the call to
7011 inline_secondary_memory_needed.
7012 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7013 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
7015 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
7016 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7018 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
7019 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
7020 * config/mips/mips.c (mips_secondary_memory_needed): Make static
7021 and match hook interface. Add comment from mips.h.
7022 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7023 * config/mmix/mmix.md (truncdfsf2): Refer to
7024 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7026 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
7027 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
7028 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7029 (pa_secondary_memory_needed): New function.
7030 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
7031 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
7032 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7033 (pdp11_secondary_memory_needed): Make static and match hook interface.
7034 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
7035 * config/powerpcspe/powerpcspe-protos.h
7036 (rs6000_secondary_memory_needed_ptr): Delete.
7037 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
7039 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7040 (rs6000_option_override_internal): Assign to
7041 targetm.secondary_memory_needed rather than
7042 rs6000_secondary_memory_needed_ptr.
7043 (rs6000_secondary_memory_needed): Match hook interface.
7044 (rs6000_debug_secondary_memory_needed): Likewise.
7045 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
7046 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
7047 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
7048 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7049 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
7050 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
7052 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
7053 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7054 (rs6000_option_override_internal): Assign to
7055 targetm.secondary_memory_needed rather than
7056 rs6000_secondary_memory_needed_ptr.
7057 (rs6000_secondary_memory_needed): Match hook interface.
7058 (rs6000_debug_secondary_memory_needed): Likewise.
7059 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
7060 * config/s390/s390.c (s390_secondary_memory_needed): New function.
7061 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7062 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
7063 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7064 (sparc_secondary_memory_needed): New function.
7065 * lra-constraints.c (check_and_process_move): Refer to
7066 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7068 (curr_insn_transform): Likewise.
7069 (process_alt_operands): Use targetm.secondary_memory_needed
7070 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7071 (check_secondary_memory_needed_p): Likewise.
7072 (choose_split_class): Likewise.
7073 * reload.c: Unconditionally include code that was previously
7074 conditional on SECONDARY_MEMORY_NEEDED.
7075 (push_secondary_reload): Use targetm.secondary_memory_needed
7076 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7077 (push_reload): Likewise.
7078 * reload1.c: Unconditionally include code that was previously
7079 conditional on SECONDARY_MEMORY_NEEDED.
7080 (choose_reload_regs): Use targetm.secondary_memory_needed
7081 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7082 (gen_reload): Likewise.
7083 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
7085 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7086 Alan Hayward <alan.hayward@arm.com>
7087 David Sherwood <david.sherwood@arm.com>
7089 * target.def (secondary_memory_needed_mode): New hook:
7090 * targhooks.c (default_secondary_memory_needed_mode): Declare.
7091 * targhooks.h (default_secondary_memory_needed_mode): New function.
7092 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
7093 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
7094 * doc/tm.texi: Regenerate.
7095 * lra-constraints.c (check_and_process_move): Use
7096 targetm.secondary_memory_needed_mode instead of
7097 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
7098 (curr_insn_transform): Likewise.
7099 * reload.c (get_secondary_mem): Likewise.
7100 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7101 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
7103 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7104 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7105 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
7106 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7107 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
7109 * config/powerpcspe/powerpcspe-protos.h
7110 (rs6000_secondary_memory_needed_mode): Delete.
7111 * config/powerpcspe/powerpcspe.c
7112 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7113 (rs6000_secondary_memory_needed_mode): Make static.
7114 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7115 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
7117 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7119 (rs6000_secondary_memory_needed_mode): Make static.
7120 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7121 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
7122 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7123 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7124 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7126 (sparc_secondary_memory_needed_mode): New function.
7127 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
7129 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
7131 * config/aarch64/constraints.md (Umq): New constraint.
7132 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
7134 (mov<mode>): Update condition.
7136 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7138 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
7139 when bitposition is the same.
7141 2017-09-13 Richard Biener <rguenther@suse.de>
7143 * dwarf2out.c (output_die_symbol): Remove.
7144 (output_die): Do not output a DIEs symbol.
7146 2017-09-13 Richard Biener <rguenther@suse.de>
7149 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
7150 default-def to avoid breaking iterator update with the weird
7151 interaction with cgraph_update_edges_for_call_stmt_node.
7153 2017-09-13 Richard Biener <rguenther@suse.de>
7155 * tree-cfg.c (verify_gimple_assign_binary): Add verification
7156 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
7157 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
7158 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
7160 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
7162 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
7163 Disable pc relative literal load irrespective of
7164 TARGET_FIX_ERR_A53_84341 for default.
7166 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
7168 * config/sparc/sparc.c (output_return): Output the source location of
7169 the insn in the delay slot, if any.
7170 (output_sibcall): Likewise.
7172 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
7175 * common.opt (-static-pie): New alias.
7176 (shared): Negate static-pie.
7177 (-no-pie): Update help text.
7179 (static-pie): New option.
7180 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
7181 -static-pie support.
7182 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
7183 (LINK_EH_SPEC): Likewise.
7184 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7185 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7186 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7187 * gcc.c (LINK_COMMAND_SPEC): Likewise.
7188 (init_gcc_specs): Likewise.
7189 (init_spec): Likewise.
7190 (display_help): Update help message for -pie.
7191 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
7194 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
7196 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
7197 (movdi_aarch64): Likewise.
7198 (movti_aarch64): Likewise.
7200 2017-09-12 Simon Wright <simon@pushface.org>
7203 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
7204 calculation of the minor version, always output as 0.
7206 2017-09-12 Jakub Jelinek <jakub@redhat.com>
7209 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
7210 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
7211 on it early, rather than manual conversion late. For
7212 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
7213 instead of performing manual conversion.
7215 2017-09-12 Carl Love <cel@us.ibm.com>
7217 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
7218 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
7220 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7221 VMULOSW): Add definitions.
7222 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7223 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7224 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7225 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7226 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7228 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7230 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
7232 (movti_aarch64): Likewise.
7233 (movdf_aarch64): Likewise.
7234 (movtf_aarch64): Likewise.
7235 (load_pairdi): Likewise.
7236 (store_pairdi): Likewise.
7237 (load_pairdf): Likewise.
7238 (store_pairdf): Likewise.
7239 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
7240 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
7241 (ldr_got_small_<mode>): Likewise.
7242 (ldr_got_small_28k_<mode>): Likewise.
7243 (ldr_got_tiny): Likewise.
7244 * config/aarch64/iterators.md (ldst_sz): New.
7245 (ldpstp_sz): Likewise.
7246 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
7248 (thunderx_load): Split load_8 to load_16.
7249 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
7251 (thunderx2t99_storepair_basic): Split store_8 to store_16.
7252 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
7253 (xgene1_store_pair): Split store_8 to store_16.
7254 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
7255 (falkor_st_0_st_sd): Split store_8 to store_16.
7257 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7259 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
7260 and store1/2/3/4 to store_4/8/12/16.
7261 * config/aarch64/aarch64.md: Update for rename.
7262 * config/arm/arm.md: Likewise.: Likewise.
7263 * config/arm/arm.c: Likewise.
7264 * config/arm/thumb1.md: Likewise.
7265 * config/arm/thumb2.md: Likewise.
7266 * config/arm/vfp.md: Likewise.
7267 * config/arm/arm-generic.md: Likewise.
7268 * config/arm/arm1020e.md: Likewise.
7269 * config/arm/arm1026ejs.md: Likewise.
7270 * config/arm/arm1136jfs.md: Likewise.
7271 * config/arm/arm926ejs.md: Likewise.
7272 * config/arm/cortex-a15.md: Likewise.
7273 * config/arm/cortex-a17.md: Likewise.
7274 * config/arm/cortex-a5.md: Likewise.
7275 * config/arm/cortex-a53.md: Likewise.
7276 * config/arm/cortex-a57.md: Likewise.
7277 * config/arm/cortex-a7.md: Likewise.
7278 * config/arm/cortex-a8.md: Likewise.
7279 * config/arm/cortex-a9.md: Likewise.
7280 * config/arm/cortex-m4.md: Likewise.
7281 * config/arm/cortex-m7.md: Likewise.
7282 * config/arm/cortex-r4.md: Likewise.
7283 * config/arm/exynos-m1.md: Likewise.
7284 * config/arm/fa526.md: Likewise.
7285 * config/arm/fa606te.md: Likewise.
7286 * config/arm/fa626te.md: Likewise.
7287 * config/arm/fa726te.md: Likewise.
7288 * config/arm/fmp626.md: Likewise.
7289 * config/arm/iwmmxt.md: Likewise.
7290 * config/arm/ldmstm.md: Likewise.
7291 * config/arm/marvell-pj4.md: Likewise.
7292 * config/arm/xgene1.md: Likewise.
7293 * config/aarch64/thunderx.md: Likewise.
7294 * config/aarch64/thunderx2t99.md: Likewise.
7295 * config/aarch64/falkor.md: Likewise.
7297 2017-09-12 Martin Liska <mliska@suse.cz>
7299 * attribs.c (private_lookup_attribute): New function.
7300 * attribs.h (private_lookup_attribute): Declared here.
7301 (lookup_attribute): Called from this place.
7303 2017-09-12 Richard Biener <rguenther@suse.de>
7305 PR tree-optimization/82157
7306 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
7307 stmts with side-effects.
7309 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7310 Alan Hayward <alan.hayward@arm.com>
7311 David Sherwood <david.sherwood@arm.com>
7313 * target.def (hard_regno_nregs): New hook.
7314 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
7315 * targhooks.h (default_hard_regno_nregs): Declare.
7316 * targhooks.c (default_hard_regno_nregs): New function.
7317 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
7318 (TARGET_HARD_REGNO_NREGS): ...this hook.
7319 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
7320 (CLASS_MAX_NREGS): Likewise.
7321 * doc/tm.texi: Regenerate.
7322 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
7323 instead of HARD_REGNO_NREGS.
7324 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
7325 HARD_REGNO_NREGS in the comment.
7326 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
7327 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
7328 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
7329 Return an unsigned int.
7330 (TARGET_HARD_REGNO_NREGS): Redefine.
7331 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
7332 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
7333 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
7334 (arc_hard_regno_nregs): New function.
7335 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
7336 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
7337 (arm_hard_regno_nregs): New function.
7338 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
7339 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
7340 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
7341 (TARGET_HARD_REGNO_NREGS): Redefine.
7342 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
7343 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
7344 (HARD_REGNO_NREGS): Delete.
7345 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
7346 (cr16_hard_regno_nregs): New function.
7347 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
7348 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
7349 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
7350 (cris_hard_regno_nregs): New function.
7351 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
7352 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
7353 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
7354 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
7355 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
7356 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
7357 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
7358 (frv_hard_regno_nregs): Make static. Take and return an
7360 (frv_class_max_nregs): Remove outdated copy of documentation.
7361 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
7362 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
7363 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
7364 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
7365 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
7366 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
7367 (TARGET_HARD_REGNO_NREGS): Redefine.
7368 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
7369 (CLASS_MAX_NREGS): Update comment.
7370 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
7371 (ia64_hard_regno_nregs): New function.
7372 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
7373 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
7374 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
7375 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
7376 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
7378 (m32c_hard_regno_nregs): Likewise. Make static.
7379 (TARGET_HARD_REGNO_NREGS): Redefine.
7380 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
7381 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
7382 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
7383 (m68k_hard_regno_nregs): New function.
7384 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
7385 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
7386 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
7387 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
7388 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
7389 Take and return an unsigned int.
7390 (TARGET_HARD_REGNO_NREGS): Redefine.
7391 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
7392 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
7393 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
7394 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
7395 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
7396 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
7397 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
7398 (msp430_hard_regno_nregs): Make static. Take and return an
7400 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
7401 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
7402 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
7403 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
7404 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
7405 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
7406 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
7407 (TARGET_HARD_REGNO_NREGS): Redefine.
7408 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
7409 (PA_HARD_REGNO_NREGS): ...this.
7410 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
7411 (PA_HARD_REGNO_NREGS): ...this.
7412 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
7413 (pa_hard_regno_nregs): New function.
7414 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
7415 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
7416 (pdp11_hard_regno_nregs): New function.
7417 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
7418 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
7419 (rs6000_hard_regno_nregs_hook): New function.
7420 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
7421 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
7422 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
7423 Take and return an unsigned int. Move earlier in file.
7424 (TARGET_HARD_REGNO_NREGS): Redefine.
7425 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
7426 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
7427 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
7428 (rl78_hard_regno_nregs): Make static. Take and return an
7430 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
7431 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
7432 (rs6000_hard_regno_nregs_hook): New function.
7433 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
7434 * config/rx/rx.c (rx_hard_regno_nregs): New function.
7435 (TARGET_HARD_REGNO_NREGS): Redefine.
7436 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
7437 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
7438 instead of HARD_REGNO_NREGS.
7439 (s390_hard_regno_nregs): New function.
7440 (s390_hard_regno_mode_ok): Add comment from s390.h.
7441 (TARGET_HARD_REGNO_NREGS): Redefine.
7442 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
7443 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
7444 (sh_hard_regno_nregs): New function.
7445 (sh_pass_in_reg_p): Use it.
7446 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
7447 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
7448 (sparc_hard_regno_nregs): New function.
7449 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
7450 * config/spu/spu.c (spu_hard_regno_nregs): New function.
7451 (spu_function_arg_advance): Use it, supplying a valid register number.
7452 (TARGET_HARD_REGNO_NREGS): Redefine.
7453 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
7454 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
7455 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
7456 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
7457 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
7458 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
7459 (CLASS_MAX_NREGS): Remove copy of old documentation.
7460 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
7461 (visium_hard_regno_nregs): New function.
7462 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
7463 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
7464 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
7465 xtensa_hard_regno_nregs): New function.
7466 * system.h (HARD_REGNO_NREGS): Poison.
7468 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7470 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
7471 hard_regno_nregs instead of HARD_REGNO_NREGS.
7472 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
7473 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
7474 (c6x_expand_epilogue): Likewise.
7475 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
7476 (frv_read_iacc_argument): Likewise.
7477 * config/sh/sh.c: Include regs.h.
7478 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
7479 (regs_used): Likewise.
7480 (output_stack_adjust): Likewise.
7481 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
7482 * expmed.c: Include regs.h.
7483 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
7484 * ree.c: Include regs.h.
7485 (combine_reaching_defs): Use hard_regno_nregs instead of
7487 (add_removable_extension): Likewise.
7489 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7491 * regs.h (hard_regno_nregs): Turn into a function.
7492 (end_hard_regno): Update accordingly.
7493 * caller-save.c (setup_save_areas): Likewise.
7494 (save_call_clobbered_regs): Likewise.
7495 (replace_reg_with_saved_mem): Likewise.
7496 (insert_restore): Likewise.
7497 (insert_save): Likewise.
7498 * combine.c (can_change_dest_mode): Likewise.
7499 (move_deaths): Likewise.
7500 (distribute_notes): Likewise.
7501 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
7502 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
7503 (rs6000_split_multireg_move): Likewise.
7504 (rs6000_register_move_cost): Likewise.
7505 (rs6000_memory_move_cost): Likewise.
7506 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
7507 (rs6000_split_multireg_move): Likewise.
7508 (rs6000_register_move_cost): Likewise.
7509 (rs6000_memory_move_cost): Likewise.
7510 * cselib.c (cselib_reset_table): Likewise.
7511 (cselib_lookup_1): Likewise.
7512 * emit-rtl.c (set_mode_and_regno): Likewise.
7513 * function.c (aggregate_value_p): Likewise.
7514 * ira-color.c (setup_profitable_hard_regs): Likewise.
7515 (check_hard_reg_p): Likewise.
7516 (calculate_saved_nregs): Likewise.
7517 (assign_hard_reg): Likewise.
7518 (improve_allocation): Likewise.
7519 (calculate_spill_cost): Likewise.
7520 * ira-emit.c (modify_move_list): Likewise.
7521 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
7522 (ira_hard_reg_in_set_p): Likewise.
7523 * ira.c (setup_reg_mode_hard_regset): Likewise.
7524 (clarify_prohibited_class_mode_regs): Likewise.
7525 (check_allocation): Likewise.
7526 * lra-assigns.c (find_hard_regno_for_1): Likewise.
7527 (lra_setup_reg_renumber): Likewise.
7528 (setup_try_hard_regno_pseudos): Likewise.
7529 (spill_for): Likewise.
7530 (assign_hard_regno): Likewise.
7531 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
7532 * lra-constraints.c (in_class_p): Likewise.
7533 (lra_constraint_offset): Likewise.
7534 (simplify_operand_subreg): Likewise.
7535 (lra_constraints): Likewise.
7536 (split_reg): Likewise.
7537 (split_if_necessary): Likewise.
7538 (invariant_p): Likewise.
7539 (inherit_in_ebb): Likewise.
7540 * lra-lives.c (process_bb_lives): Likewise.
7541 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
7542 (get_hard_regs): Likewise.
7543 (do_remat): Likewise.
7544 * lra-spills.c (assign_spill_hard_regs): Likewise.
7545 * mode-switching.c (create_pre_exit): Likewise.
7546 * postreload.c (reload_combine_recognize_pattern): Likewise.
7547 * recog.c (peep2_find_free_register): Likewise.
7548 * regcprop.c (kill_value_regno): Likewise.
7549 (set_value_regno): Likewise.
7550 (copy_value): Likewise.
7551 (maybe_mode_change): Likewise.
7552 (find_oldest_value_reg): Likewise.
7553 (copyprop_hardreg_forward_1): Likewise.
7554 * regrename.c (check_new_reg_p): Likewise.
7555 (regrename_do_replace): Likewise.
7556 * reload.c (push_reload): Likewise.
7557 (combine_reloads): Likewise.
7558 (find_dummy_reload): Likewise.
7559 (operands_match_p): Likewise.
7560 (find_reloads): Likewise.
7561 (find_equiv_reg): Likewise.
7562 (reload_adjust_reg_for_mode): Likewise.
7563 * reload1.c (count_pseudo): Likewise.
7564 (count_spilled_pseudo): Likewise.
7565 (find_reg): Likewise.
7566 (clear_reload_reg_in_use): Likewise.
7567 (free_for_value_p): Likewise.
7568 (allocate_reload_reg): Likewise.
7569 (choose_reload_regs): Likewise.
7570 (reload_adjust_reg_for_temp): Likewise.
7571 (emit_reload_insns): Likewise.
7572 (delete_output_reload): Likewise.
7573 * rtlanal.c (subreg_get_info): Likewise.
7574 * sched-deps.c (sched_analyze_reg): Likewise.
7575 * sel-sched.c (init_regs_for_mode): Likewise.
7576 (mark_unavailable_hard_regs): Likewise.
7577 (choose_best_reg_1): Likewise.
7578 (verify_target_availability): Likewise.
7579 * valtrack.c (dead_debug_insert_temp): Likewise.
7580 * var-tracking.c (track_loc_p): Likewise.
7581 (emit_note_insn_var_location): Likewise.
7582 * varasm.c (make_decl_rtl): Likewise.
7583 * reginfo.c (choose_hard_reg_mode): Likewise.
7584 (init_reg_modes_target): Refer directly to
7585 this_target_regs->x_hard_regno_nregs.
7587 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7589 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
7590 instead of hard_regno_nregs.
7592 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7594 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
7595 end_hard_regno instead of hard_regno_nregs.
7596 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
7597 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
7598 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
7599 * ira-color.c (improve_allocation): Likewise.
7600 * lra-assigns.c (find_hard_regno_for_1): Likewise.
7601 * lra-lives.c (mark_regno_live): Likewise.
7602 (mark_regno_dead): Likewise.
7603 * lra-remat.c (operand_to_remat): Likewise.
7604 * lra.c (collect_non_operand_hard_regs): Likewise.
7605 * postreload.c (reload_combine_note_store): Likewise.
7606 (move2add_valid_value_p): Likewise.
7607 * reload.c (regno_clobbered_p): Likewise.
7609 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7611 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
7613 * config/v850/v850.c (v850_reorg): Likewise.
7614 * reload.c (refers_to_regno_for_reload_p): Likewise.
7615 (find_equiv_reg): Likewise.
7616 * reload1.c (reload_reg_reaches_end_p): Likewise.
7618 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7620 * caller-save.c (add_used_regs): Use REG_NREGS instead of
7622 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
7623 * config/arm/arm.c (output_move_neon): Likewise.
7624 (arm_attr_length_move_neon): Likewise.
7625 (neon_split_vcombine): Likewise.
7626 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
7627 (c6x_mark_reg_written): Likewise.
7628 (c6x_dwarf_register_span): Likewise.
7629 * config/i386/i386.c (ix86_save_reg): Likewise.
7630 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
7631 (rws_access_reg): Likewise.
7632 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
7633 * mode-switching.c (create_pre_exit): Likewise.
7634 * ree.c (combine_reaching_defs): Likewise.
7635 (add_removable_extension): Likewise.
7636 * regcprop.c (find_oldest_value_reg): Likewise.
7637 (copyprop_hardreg_forward_1): Likewise.
7638 * reload.c (reload_inner_reg_of_subreg): Likewise.
7639 (push_reload): Likewise.
7640 (combine_reloads): Likewise.
7641 (find_dummy_reload): Likewise.
7642 (reload_adjust_reg_for_mode): Likewise.
7643 * reload1.c (find_reload_regs): Likewise.
7644 (forget_old_reloads_1): Likewise.
7645 (reload_reg_free_for_value_p): Likewise.
7646 (reload_adjust_reg_for_temp): Likewise.
7647 (emit_reload_insns): Likewise.
7648 (delete_output_reload): Likewise.
7649 * sel-sched.c (choose_best_reg_1): Likewise.
7650 (choose_best_pseudo_reg): Likewise.
7652 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7653 Alan Hayward <alan.hayward@arm.com>
7654 David Sherwood <david.sherwood@arm.com>
7656 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
7657 * target.def (slow_unaligned_access): New hook.
7658 * targhooks.h (default_slow_unaligned_access): Declare.
7659 * targhooks.c (default_slow_unaligned_access): New function.
7660 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
7661 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
7662 * doc/tm.texi: Regenerate.
7663 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
7664 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
7665 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
7667 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
7668 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
7670 (rs6000_slow_unaligned_access): New function.
7671 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
7672 (expand_block_compare): Likewise.
7673 (expand_strn_compare): Likewise.
7674 (rs6000_rtx_costs): Likewise.
7675 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
7676 (riscv_slow_unaligned_access): Likewise.
7677 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
7678 (riscv_slow_unaligned_access_p): ...this and make static.
7679 (riscv_option_override): Update accordingly.
7680 (riscv_slow_unaligned_access): New function.
7681 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
7682 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
7683 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
7684 (rs6000_slow_unaligned_access): New function.
7685 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
7686 (rs6000_rtx_costs): Likewise.
7687 * config/rs6000/rs6000-string.c (expand_block_compare)
7688 (expand_strn_compare): Use targetm.slow_unaligned_access instead
7689 of SLOW_UNALIGNED_ACCESS.
7690 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
7691 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
7692 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
7693 of SLOW_UNALIGNED_ACCESS.
7694 * expmed.c (simple_mem_bitfield_p): Likewise.
7695 * expr.c (alignment_for_piecewise_move): Likewise.
7696 (emit_group_load_1): Likewise.
7697 (emit_group_store): Likewise.
7698 (copy_blkmode_from_reg): Likewise.
7699 (emit_push_insn): Likewise.
7700 (expand_assignment): Likewise.
7701 (store_field): Likewise.
7702 (expand_expr_real_1): Likewise.
7703 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
7704 * lra-constraints.c (simplify_operand_subreg): Likewise.
7705 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
7706 * gimple-ssa-store-merging.c: Likewise in block comment at start
7708 * tree-ssa-strlen.c: Include target.h.
7709 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
7710 of SLOW_UNALIGNED_ACCESS.
7711 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
7713 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7715 PR rtl-optimization/82185
7716 * expmed.c (emit_store_flag_int): Only test tem if it has been
7719 2017-09-12 Richard Biener <rguenther@suse.de>
7722 * match.pd ((FTYPE) N CMP CST): Fix typo.
7724 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
7726 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
7728 (mips_near_type_p): Add 'short_call' attribute as a synonym
7730 * doc/extend.texi (short_call): Document new function attribute.
7732 2017-09-12 Jakub Jelinek <jakub@redhat.com>
7735 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
7736 assertion check that in the condition.
7737 (get_atomic_generic_size): Likewise. Before testing if parameter
7738 has pointer type, if it has array type, call for C++
7739 default_conversion to perform array-to-pointer conversion.
7741 2017-09-12 Richard Biener <rguenther@suse.de>
7743 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
7744 for operations we cannot scalarize.
7746 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
7748 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
7749 vectors heap vectors. Clean up comments.
7750 Make visited_bbs a reference.
7751 (profitable_jump_thread_path): Make GC
7752 vectors heap vectors. Clean up comments.
7754 (convert_and_register_jump_thread_path): Make GC vectors heap
7756 (check_subpath_and_update_thread_path): Same. Clean up comments.
7757 Make visited_bbs a reference.
7758 (handle_phi): Abstract common code to to
7759 register_jump_thread_path_if_profitable.
7760 Rename VAR_BB to DEF_BB.
7762 Make GC vectors heap vectors.
7763 Make visited_bbs a reference.
7764 (handle_assignment): Same.
7765 (register_jump_thread_path_if_profitable): New.
7766 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
7768 Make GC vectors heap vectors. Clean up comments.
7769 Make visited_bbs a reference.
7770 (find_jump_threads_backwards): Make visited_bbs live in the stack.
7771 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
7774 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
7777 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
7778 words of E_DImode object are reachable by xtensa_uimm8x4 access.
7780 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
7782 Revert r251800 and r251799.
7784 2017-09-11 Martin Jambor <mjambor@suse.cz>
7787 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
7788 arguments in advance.
7789 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
7790 use it to find predecessor edges.
7791 (naive_outof_ssa): Collect vector of predecessors.
7793 2017-09-08 Jason Merrill <jason@redhat.com>
7795 PR c++/70029 - ICE with ref-qualifier and -flto
7796 * langhooks.h (struct lang_hooks_for_types): Add
7797 copy_lang_qualifiers.
7798 * attribs.c (build_type_attribute_qual_variant): Use it.
7799 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
7801 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
7802 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
7804 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
7807 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
7808 (*mulsi3_sp64): New instruction.
7809 (mulsi3): New expander.
7811 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
7813 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
7815 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7817 * sancov.c: Include memmodel.h.
7819 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
7822 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
7825 2017-09-07 Carl Love <cel@us.ibm.com>
7827 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
7828 the sldi instruction.
7830 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
7832 * sancov.c: Include tm_p.h.
7834 2017-09-07 Jakub Jelinek <jakub@redhat.com>
7837 * output.h (switch_to_other_text_partition): New declaration.
7838 * varasm.c (switch_to_other_text_partition): New function.
7839 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
7840 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
7841 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
7842 to the other text partition before emitting LCL label and switch back
7843 after emitting the word after it.
7845 2017-09-07 Richard Biener <rguenther@suse.de>
7847 * passes.def (pass_split_crit_edges): Remove instance before PRE.
7848 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
7849 critical edges here, after loop init.
7850 (pass_data_pre): Remove PROP_no_crit_edges flags.
7851 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
7852 for valueization of call args to avoid leaking VN_TOP.
7853 (visit_use): Assert we do not visit default defs.
7854 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
7855 Use error_mark_node to more easily detect leaking VN_TOP.
7856 All default-defs are varying, not VN_TOP. Mark them visited.
7857 (run_scc_vn): Make code match comment.
7859 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
7861 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
7862 OPTION_MASK_FLOAT128_KEYWORD.
7863 (POWERPC_MASKS): Likewise.
7864 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
7865 support for the -mfloat128-type option, and make -mfloat128
7866 default on PowerPC Linux systems. Define or undefine
7867 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
7868 Define __float128 to be __ieee128 if IEEE 128-bit support is
7869 enabled, or undefine it.
7870 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
7871 Delete defining __FLOAT128_TYPE__.
7872 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
7873 -mfloat128-type option and make -mfloat128 default on PowerPC
7875 (TARGET_FLOAT128_TYPE): Likewise.
7876 (-mfloat128-type): Likewise.
7877 * config/rs6000/rs6000.c (rs6000_option_override_internal):
7878 Delete the -mfloat128-type option and make -mfloat128 default on
7879 PowerPC Linux systems. Always use __ieee128 to be the keyword for
7880 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
7881 128-bit floating point is enabled. Change tests from using
7882 -mfloat128-type to -mfloat128.
7883 (rs6000_mangle_type): Use the correct mangling for the __float128
7884 type even if normal long double is restricted to 64-bits.
7885 (floatn_mode): Enable the _Float128 type by default on VSX Linux
7887 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
7888 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
7889 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
7891 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
7892 documentation for -mfloat128.
7894 2017-09-06 Olivier Hainque <hainque@adacore.com>
7896 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
7898 2017-09-06 Wish Wu <wishwu007@gmail.com>
7899 Jakub Jelinek <jakub@redhat.com>
7901 * asan.c (initialize_sanitizer_builtins): Add
7902 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
7903 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
7904 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
7905 BT_FN_VOID_UINT64_PTR variables.
7906 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
7907 (BT_FN_VOID_UINT16_UINT16): Likewise.
7908 (BT_FN_VOID_UINT32_UINT32): Likewise.
7909 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
7910 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
7911 (BT_FN_VOID_UINT64_PTR): Likewise.
7912 * common.opt (flag_sanitize_coverage): New variable.
7913 (fsanitize-coverage=trace-pc): Remove.
7914 (fsanitize-coverage=): Add.
7915 * flag-types.h (enum sanitize_coverage_code): New enum.
7916 * fold-const.c (fold_range_test): Disable non-short-circuit
7917 optimization if flag_sanitize_coverage.
7918 (fold_truth_andor): Likewise.
7919 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
7920 * opts.c (COVERAGE_SANITIZER_OPT): Define.
7921 (coverage_sanitizer_opts): New array.
7922 (get_closest_sanitizer_option): Add OPTS argument, handle also
7923 OPT_fsanitize_coverage_.
7924 (parse_sanitizer_options): Adjusted to also handle
7925 OPT_fsanitize_coverage_.
7926 (common_handle_option): Add OPT_fsanitize_coverage_.
7927 * sancov.c (instrument_comparison, instrument_switch): New function.
7928 (sancov_pass): Add trace-cmp support.
7929 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
7930 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
7931 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
7932 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
7933 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
7934 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
7935 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
7936 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
7937 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
7938 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
7940 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
7942 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
7943 error. Only quit immediately if parsing is complete.
7944 (BEGIN): Initialize fatal_err and parse_done.
7945 (begin fpu, end fpu): Check number of arguments.
7946 (begin arch, end arch): Likewise.
7947 (begin cpu, end cpu): Likewise.
7948 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
7949 (optalias): Likewise.
7951 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
7953 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
7954 * config/arm/arm-isa.h: Delete. Move definitions to ...
7955 * arm-cpus.in: ... here. Use new feature and fgroup values.
7956 * config/arm/arm.c (arm_option_override): Use lower case for feature
7958 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
7959 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
7960 * config/arm/parsecpu.awk (END): Add new command 'isa'.
7962 (print_isa_bits_for): New function.
7963 (gen_isa): New function.
7964 (gen_comm_data): Use print_isa_bits_for.
7965 (define feature): New keyword.
7966 (define fgroup): New keyword.
7967 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
7968 (arm-isa.h): Add rule to generate file.
7969 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
7970 case for feature bit names.
7972 2017-09-06 Richard Biener <rguenther@suse.de>
7974 * tree-ssa-pre.c (NECESSARY): Remove.
7975 (create_expression_by_pieces): Do not touch pass-local flags.
7976 (insert_into_preds_of_block): Likewise.
7977 (do_pre_regular_insertion): Likewise.
7978 (eliminate_insert): Likewise.
7979 (eliminate_dom_walker::before_dom_children): Likewise.
7980 (fini_eliminate): Do not look at inserted_exprs.
7981 (mark_operand_necessary): Remove.
7982 (remove_dead_inserted_code): Replace with simple work-list
7983 algorithm based on inserted_exprs and SSA uses.
7984 (pass_pre::execute): Re-order fini_eliminate and
7985 remove_dead_inserted_code.
7987 2017-09-06 Olivier Hainque <hainque@adacore.com>
7989 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
7990 for VxWorks 7. Adjust surrounding comments.
7992 2017-09-06 Richard Biener <rguenther@suse.de>
7994 * gimple-ssa-strength-reduction.c
7995 (find_candidates_dom_walker::before_dom_children): Also allow
7998 2017-09-06 Richard Biener <rguenther@suse.de>
8000 PR tree-optimization/82108
8001 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
8002 for gap in the non-permutation SLP case.
8004 2017-09-06 Martin Jambor <mjambor@suse.cz>
8006 PR tree-optimization/82078
8007 * tree-sra.c (sort_and_splice_var_accesses): Move call to
8008 add_access_to_work_queue...
8009 (build_accesses_from_assign): ...here.
8010 (propagate_all_subaccesses): Make sure racc is the group
8011 representative, if there is one.
8013 2017-09-06 Jakub Jelinek <jakub@redhat.com>
8016 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
8019 2017-09-06 Richard Biener <rguenther@suse.de>
8021 * gimple-ssa-strength-reduction.c
8022 (find_candidates_dom_walker::before_doom_children): Use a
8023 type and not a mode check.
8025 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
8028 * config/arm/predicates.md (arm_general_adddi_operand): Create new
8030 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
8032 2017-09-05 Jeff Law <law@redhat.com>
8034 PR tree-optimization/64910
8035 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
8036 cases where we have 3 or more operands.
8038 2017-09-05 Jakub Jelinek <jakub@redhat.com>
8041 * omp-low.c (lower_omp_for): Recompute tree invariant if
8042 gimple_omp_for_initial/final is ADDR_EXPR.
8045 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
8046 into gimple val before gimplification fo the COND_EXPR.
8048 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
8050 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
8051 REGION_COPY argument.
8052 (thread_through_all_blocks): Remove unused argument to
8053 duplicate_thread_path.
8055 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8056 Alan Hayward <alan.hayward@arm.com>
8057 David Sherwood <david.sherwood@arm.com>
8059 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
8060 Take a scalar_mode rather than a machine_mode.
8061 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8062 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
8063 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8064 (aarch64_gen_adjusted_ldpstp): Likewise.
8065 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
8067 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8068 Alan Hayward <alan.hayward@arm.com>
8069 David Sherwood <david.sherwood@arm.com>
8071 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
8072 Take a scalar_int_mode instead of a machine_mode.
8073 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8074 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8075 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8076 (aarch64_simd_attr_length_rglist): Delete.
8077 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
8078 a scalar_int_mode instead of a machine_mode.
8079 (aarch64_add_offset): Likewise.
8080 (aarch64_internal_mov_immediate): Likewise
8081 (aarch64_add_constant_internal): Likewise.
8082 (aarch64_add_constant): Likewise.
8083 (aarch64_movw_imm): Likewise.
8084 (aarch64_rtx_arith_op_extract_p): Likewise.
8085 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8086 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8087 Remove assert that the mode isn't a vector.
8088 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8089 (aarch64_expand_mov_immediate): Update calls after above changes.
8090 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
8091 (aarch64_and_bitmask_imm): Check for scalar integer modes.
8092 (aarch64_move_imm): Likewise.
8093 (aarch64_can_const_movi_rtx_p): Likewise.
8094 (aarch64_strip_extend): Likewise.
8095 (aarch64_extr_rtx_p): Likewise.
8096 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
8097 a CONST_INT when the mode parameter is VOIDmode.
8098 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
8100 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8102 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
8103 * stor-layout.c (bitwise_mode_for_mode): Likewise.
8104 (bitwise_type_for_mode): Update accordingly.
8106 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8108 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
8109 * stor-layout.c (mode_for_size_tree): Likewise.
8110 (mode_for_array): Update accordingly.
8111 (layout_decl): Likewise.
8112 (compute_record_mode): Likewise. Only set the mode once.
8114 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8116 * target.def (get_mask_mode): Change return type to opt_mode.
8118 * doc/tm.texi: Regenerate.
8119 * targhooks.h (default_get_mask_mode): Return an opt_mode.
8120 * targhooks.c (default_get_mask_mode): Likewise.
8121 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
8122 * optabs-query.c (can_vec_mask_load_store_p): Update use of
8123 targetm.get_mask_mode.
8124 * tree.c (build_truth_vector_type): Likewise.
8126 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8128 * machmode.h (mode_for_vector): Return an opt_mode.
8129 * stor-layout.c (mode_for_vector): Likewise.
8130 (mode_for_int_vector): Update accordingly.
8131 (layout_type): Likewise.
8132 * config/i386/i386.c (emit_memmov): Likewise.
8133 (ix86_expand_set_or_movmem): Likewise.
8134 (ix86_expand_vector_init): Likewise.
8135 (ix86_get_mask_mode): Likewise.
8136 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
8138 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
8139 * expmed.c (extract_bit_field_1): Likewise.
8140 * expr.c (expand_expr_real_2): Likewise.
8141 * optabs-query.c (can_vec_perm_p): Likewise.
8142 (can_vec_mask_load_store_p): Likewise.
8143 * optabs.c (expand_vec_perm): Likewise.
8144 * targhooks.c (default_get_mask_mode): Likewise.
8145 * tree-vect-stmts.c (vectorizable_store): Likewise.
8146 (vectorizable_load): Likewise.
8147 (get_vectype_for_scalar_type_and_size): Likewise.
8149 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8151 * machmode.h (mode_for_int_vector): New function.
8152 * stor-layout.c (mode_for_int_vector): Likewise.
8153 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
8154 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
8155 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
8156 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
8157 (s390_expand_vcond): Likewise.
8159 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8161 * machmode.h (opt_machine_mode): New type.
8162 (opt_mode<T>): Allow construction from anything that can be
8164 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
8165 (mode_for_size): Return an opt_machine_mode.
8166 * stor-layout.c (mode_for_size): Likewise.
8167 (mode_for_size_tree): Update call accordingly.
8168 (bitwise_mode_for_mode): Likewise.
8169 (make_fract_type): Likewise.
8170 (make_accum_type): Likewise.
8171 * caller-save.c (replace_reg_with_saved_mem): Update call
8173 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8174 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8175 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8176 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8177 * expmed.c (extract_bit_field_1): Likewise.
8178 * reload.c (get_secondary_mem): Likewise.
8179 * varasm.c (assemble_integer): Likewise.
8180 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
8183 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8185 * machmode.h (decimal_float_mode_for_size): New function.
8186 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
8187 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
8188 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
8189 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
8190 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
8191 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
8193 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8195 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
8196 (get_builtin_sync_mode): Likewise.
8197 (expand_ifn_atomic_compare_exchange): Likewise.
8198 (expand_builtin_atomic_clear): Likewise.
8199 (expand_builtin_atomic_test_and_set): Likewise.
8200 (fold_builtin_atomic_always_lock_free): Likewise.
8201 * calls.c (compute_argument_addresses): Likewise.
8202 (emit_library_call_value_1): Likewise.
8203 (store_one_arg): Likewise.
8204 * combine.c (combine_instructions): Likewise.
8205 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
8206 * config/arm/arm.c (arm_function_value): Likewise.
8207 (aapcs_allocate_return_reg): Likewise.
8208 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
8209 * config/i386/i386.c (construct_container): Likewise.
8210 (ix86_gimplify_va_arg): Likewise.
8211 (ix86_expand_sse_cmp): Likewise.
8212 (emit_memmov): Likewise.
8213 (emit_memset): Likewise.
8214 (expand_small_movmem_or_setmem): Likewise.
8215 (ix86_expand_pextr): Likewise.
8216 (ix86_expand_pinsr): Likewise.
8217 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
8218 * config/microblaze/microblaze.c (microblaze_block_move_straight):
8220 * config/mips/mips.c (mips_function_value_1) Likewise.
8221 (mips_block_move_straight): Likewise.
8222 (mips_expand_ins_as_unaligned_store): Likewise.
8223 * config/powerpcspe/powerpcspe.c
8224 (rs6000_darwin64_record_arg_advance_flush): Likewise.
8225 (rs6000_darwin64_record_arg_flush): Likewise.
8226 * config/rs6000/rs6000.c
8227 (rs6000_darwin64_record_arg_advance_flush): Likewise.
8228 (rs6000_darwin64_record_arg_flush): Likewise.
8229 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
8230 (sparc_function_value_1): Likewise.
8231 * config/spu/spu.c (adjust_operand): Likewise.
8232 (spu_emit_branch_or_set): Likewise.
8233 (arith_immediate_p): Likewise.
8234 * emit-rtl.c (gen_lowpart_common): Likewise.
8235 * expr.c (expand_expr_real_1): Likewise.
8236 * function.c (assign_parm_setup_block): Likewise.
8237 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8238 * reload1.c (alter_reg): Likewise.
8239 * stor-layout.c (mode_for_vector): Likewise.
8240 (layout_type): Likewise.
8242 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8244 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
8245 (spu_convert_move): Likewise.
8246 * lower-subreg.c (resolve_simple_move): Likewise.
8248 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8251 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
8252 define_insn to a define_expand.
8253 (altivec_vsum2sws_direct): New define_insn.
8254 (altivec_vsumsws): Convert from a define_insn to a define_expand.
8256 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8258 * config/arm/arm.c (arm_option_params_internal): Improve setting of
8261 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
8266 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
8267 realign stack if stack alignment needed is less than incoming
8270 2017-09-05 Marek Polacek <polacek@redhat.com>
8273 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
8276 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8278 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
8280 2017-09-05 Richard Biener <rguenther@suse.de>
8282 PR tree-optimization/82084
8283 * fold-const.c (can_native_encode_string_p): Handle wide characters.
8285 2017-09-05 Richard Biener <rguenther@suse.de>
8287 PR tree-optimization/82102
8288 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
8290 2017-09-05 Martin Liska <mliska@suse.cz>
8292 PR tree-optimization/82032
8293 * tree-cfg.c (generate_range_test): New function.
8294 * tree-cfg.h (generate_range_test): Declared here.
8295 * tree-cfgcleanup.c (convert_single_case_switch): New function.
8296 (cleanup_control_expr_graph): Use it.
8297 * tree-switch-conversion.c (try_switch_expansion): Remove
8299 (emit_case_nodes): Use generate_range_test.
8301 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
8304 * config/i386/i386.md (*<btsc><mode>_mask): Add
8305 TARGET_USE_BT to insn constraint.
8306 (*btr<mode>_mask): Ditto.
8308 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
8310 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
8311 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
8313 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
8316 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
8317 TARGET_NEON and TARGET_IWMMXT.
8318 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
8319 TARGET_NEON and TARGET_IWMMXT.
8320 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
8322 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
8324 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
8325 (ix86_rewrite_tls_address): Ditto.
8326 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
8327 (ix86_rewrite_tls_address_1): Ditto.
8328 (ix86_rewrite_tls_address): Ditto.
8329 * config/i386/predicates.md (tls_address_pattern): New predicate.
8330 * config/i386/i386.md (TLS address splitter): New splitter.
8332 2017-09-04 Richard Biener <rguenther@suse.de>
8334 PR tree-optimization/82084
8335 * fold-const.h (can_native_encode_string_p): Declare.
8336 * fold-const.c (can_native_encode_string_p): Factor out from ...
8337 (native_encode_string): ... here.
8338 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
8339 vectorizing stores from constants we later cannot handle.
8341 2017-09-04 Marek Polacek <polacek@redhat.com>
8344 * doc/invoke.texi: Update -Wtautological-compare documentation.
8346 2017-09-04 Jeff Law <law@redhat.com>
8348 PR tree-optimization/64910
8349 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
8350 swap the first and last operand if the last is a constant.
8352 2017-09-04 Marek Polacek <polacek@redhat.com>
8355 * convert.c (do_narrow): When sanitizing signed integer overflows,
8356 bail out for signed types.
8357 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
8359 2017-09-04 Richard Biener <rguenther@suse.de>
8361 PR tree-optimization/82060
8362 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
8363 Move devirtualization after stmt folding and before EH/AB/noreturn
8364 cleanup to get the stmt refs canonicalized. Use a bool instead
8365 of gimple_modified_p since that doesn't work for NOPs. Schedule
8366 NOPs generated by folding for removal.
8368 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8369 Alan Hayward <alan.hayward@arm.com>
8370 David Sherwood <david.sherwood@arm.com>
8372 * coretypes.h (pad_direction): New enum.
8373 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
8374 (FUNCTION_ARG_PADDING): Likewise.
8375 * target.def (function_arg_padding): New hook.
8376 * targhooks.h (default_function_arg_padding): Declare.
8377 * targhooks.c (default_function_arg_padding): New function.
8378 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
8379 (TARGET_FUNCTION_ARG_PADDING): ...this.
8380 * doc/tm.texi: Regenerate.
8381 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
8382 instead of direction.
8383 (compute_argument_addresses): Likewise.
8384 (load_register_parameters): Likewise.
8385 (emit_library_call_value_1): Likewise.
8386 (store_one_arg): Use targetm.calls.function_arg_padding instead
8387 of FUNCTION_ARG_PADDING.
8388 (must_pass_in_stack_var_size_or_pad): Likewise.
8389 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
8390 (emit_group_store): Likewise.
8391 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
8392 instead of FUNCTION_ARG_PADDING.
8393 (emit_push_insn): Likewise, and propagate enum change throughout
8395 * function.h (direction): Delete.
8396 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
8398 * function.c (assign_parm_find_stack_rtl): Likewise.
8399 (assign_parm_setup_block_p): Likewise.
8400 (assign_parm_setup_block): Likewise.
8401 (gimplify_parameters): Likewise.
8402 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
8403 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
8405 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
8406 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8407 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
8408 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
8409 (aarch64_function_arg_padding): ...this new function.
8410 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
8411 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8412 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
8413 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8414 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
8415 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8416 (arm_pad_arg_upward): Replace with...
8417 (arm_function_arg_padding): ...this new function.
8418 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
8420 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
8421 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
8422 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8423 (ia64_hpux_function_arg_padding): Replace with...
8424 (ia64_function_arg_padding): ...this new function. Use pad_direction
8425 instead of direction. Check for TARGET_HPUX.
8426 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
8427 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8428 (iq2000_function_arg_padding): New function.
8429 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
8430 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
8431 (mips_function_arg_padding): ...this new function.
8432 (mips_pad_reg_upward): Update accordingly.
8433 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8434 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
8435 targetm.calls.function_arg_padding.
8436 (FUNCTION_ARG_PADDING): Delete.
8437 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8438 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
8439 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8440 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
8441 (nios2_block_reg_padding): Return pad_direction instead of direction.
8442 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
8443 instead of direction.
8444 (nios2_function_arg_padding): Likewise. Make static.
8445 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8446 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
8447 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
8448 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
8449 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8450 (pa_function_arg_padding): Make static. Return pad_direction instead
8452 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
8453 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8454 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
8455 instead of direction. Use targetm.calls.function_arg_padding.
8456 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
8457 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
8458 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
8459 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
8460 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
8462 (function_arg_padding): Rename to...
8463 (rs6000_function_arg_padding): ...this. Make static. Return
8464 pad_direction instead of direction.
8465 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
8466 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
8467 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8468 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
8469 instead of direction. Use targetm.calls.function_arg_padding.
8470 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
8471 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
8472 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
8473 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
8474 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8475 (function_arg_padding): Rename to...
8476 (rs6000_function_arg_padding): ...this. Make static. Return
8477 pad_direction instead of direction.
8478 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
8479 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
8480 * config/s390/s390.c (s390_function_arg_padding): New function.
8481 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8482 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
8483 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
8484 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8485 (function_arg_padding): Rename to...
8486 (sparc_function_arg_padding): ...this. Make static. Return
8487 pad_direction instead of direction.
8488 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
8489 * config/spu/spu.c (spu_function_arg_padding): New function.
8490 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8491 * system.h (FUNCTION_ARG_PADDING): Poison.
8493 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8494 Alan Hayward <alan.hayward@arm.com>
8495 David Sherwood <david.sherwood@arm.com>
8497 * target.def (modes_tieable_p): New hook.
8498 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
8499 (TARGET_MODES_TIEABLE_P): ...this.
8500 * doc/tm.texi.in: Regenerate.
8501 * hooks.h (hook_bool_mode_mode_true): Declare.
8502 * hooks.c (hook_bool_mode_mode_true): New function.
8503 * combine.c (subst): Use targetm.modes_tieable_p instead of
8505 * dse.c (find_shift_sequence): Likewise.
8506 * expmed.c (extract_low_bits): Likewise.
8507 * lower-subreg.c: Include target.h.
8508 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
8510 * rtlanal.c (rtx_cost): Likewise.
8511 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
8512 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
8513 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
8514 (TARGET_MODES_TIEABLE_P): Redefine.
8515 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
8516 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
8517 (TARGET_MODES_TIEABLE_P): Redefine.
8518 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
8519 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
8520 (arc_modes_tieable_p): New function.
8521 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
8522 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
8523 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
8524 (arm_modes_tieable_p): Make static.
8525 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
8526 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
8527 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
8528 (TARGET_MODES_TIEABLE_P): Redefine.
8529 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
8530 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
8531 (TARGET_MODES_TIEABLE_P): Redefine.
8532 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
8533 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
8534 (cr16_modes_tieable_p): New function.
8535 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
8536 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
8537 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
8538 (TRULY_NOOP_TRUNCATION): Update comment.
8539 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
8540 (TRULY_NOOP_TRUNCATION): Update comment.
8541 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
8542 (frv_modes_tieable_p): New function.
8543 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
8544 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
8545 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
8546 (TARGET_MODES_TIEABLE_P): Redefine.
8547 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
8548 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
8549 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
8550 (TARGET_MODES_TIEABLE_P): Redefine.
8551 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
8552 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
8553 (ia64_modes_tieable_p): New function.
8554 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
8555 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
8556 (iq2000_modes_tieable_p): New function.
8557 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
8558 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
8559 (lm32_modes_tieable_p): New function.
8560 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
8561 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
8562 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
8563 (TARGET_MODES_TIEABLE_P): Redefine.
8564 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
8565 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
8566 (m32r_modes_tieable_p): New function.
8567 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
8568 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
8569 (m68k_modes_tieable_p): New function.
8570 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
8571 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
8572 (mcore_modes_tieable_p): New function.
8573 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
8574 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
8576 (TARGET_MODES_TIEABLE_P): Redefine.
8577 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
8578 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
8579 * config/mips/mips.c (mips_modes_tieable_p): Make static.
8580 (TARGET_MODES_TIEABLE_P): Redefine.
8581 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
8582 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
8583 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
8584 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
8585 (mn10300_modes_tieable_p): ...this and make static.
8586 (TARGET_MODES_TIEABLE_P): Redefine.
8587 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
8588 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
8589 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
8590 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
8591 (msp430_modes_tieable_p): Make static.
8592 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
8593 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
8594 (TARGET_MODES_TIEABLE_P): Redefine.
8595 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
8596 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
8597 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
8598 (TARGET_MODES_TIEABLE_P): Redefine.
8599 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
8600 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
8601 * config/pa/pa.c (pa_modes_tieable_p): Make static.
8602 (TARGET_MODES_TIEABLE_P): Redefine.
8603 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
8604 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
8605 (pdp11_modes_tieable_p): New function.
8606 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
8607 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
8608 (rs6000_modes_tieable_p): New function.
8609 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
8610 * config/powerpcspe/powerpcspe.md: Update comment.
8611 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
8612 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
8613 (TARGET_MODES_TIEABLE_P): Redefine.
8614 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
8615 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
8616 (rl78_modes_tieable_p): New function.
8617 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
8618 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
8619 (rs6000_modes_tieable_p): New function.
8620 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
8621 * config/rs6000/rs6000.md: Update comment.
8622 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
8623 * config/rx/rx.c (rx_modes_tieable_p): New function.
8624 (TARGET_MODES_TIEABLE_P): Redefine.
8625 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
8626 * config/s390/s390.c (s390_modes_tieable_p): New function.
8627 (TARGET_MODES_TIEABLE_P): Redefine.
8628 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
8629 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
8630 (sh_modes_tieable_p): New function.
8631 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
8632 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
8633 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
8634 (sparc_modes_tieable_p): Make static.
8635 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
8636 * config/spu/spu.c (spu_modes_tieable_p): New function.
8637 (TARGET_MODES_TIEABLE_P): Redefine.
8638 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
8639 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
8640 (TARGET_MODES_TIEABLE_P): Redefine.
8641 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
8642 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
8643 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
8644 * config/v850/v850.c (v850_modes_tieable_p): New function.
8645 (TARGET_MODES_TIEABLE_P): Redefine.
8646 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
8647 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
8648 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
8649 (visium_modes_tieable_p): New function.
8650 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
8651 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
8652 (xtensa_modes_tieable_p): New function.
8653 * system.h (MODES_TIEABLE_P): Poison.
8655 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8656 Alan Hayward <alan.hayward@arm.com>
8657 David Sherwood <david.sherwood@arm.com>
8659 * target.def (hard_regno_mode_ok): New hook.
8660 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
8661 (TARGET_HARD_REGNO_MODE_OK): ...this.
8662 * doc/tm.texi.in: Regenerate.
8663 * hooks.h (hook_bool_uint_mode_true): Declare.
8664 * hooks.c (hook_bool_uint_mode_true): New function.
8665 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
8667 * genpreds.c (write_insn_preds_c): Add an include of target.h.
8668 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
8669 instead of HARD_REGNO_MODE_OK.
8670 * caller-save.c: Include target.h.
8671 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
8673 * combine.c (can_combine_p): Likewise.
8674 (combinable_i3pat): Likewise.
8675 (can_change_dest_mode): Likewise.
8676 * expr.c (init_expr_target): Likewise.
8677 (convert_move): Likewise.
8678 (convert_modes): Likewise.
8679 * ira.c (setup_prohibited_class_mode_regs): Likewise.
8680 (setup_prohibited_mode_move_regs): Likewise.
8681 * ira.h (target_ira): Likewise.
8682 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8683 * lra-constraints.c (process_alt_operands): Likewise.
8684 (split_reg): Likewise.
8685 * recog.c (peep2_find_free_register): Likewise.
8686 * ree.c (combine_reaching_defs): Likewise.
8687 * regcprop.c (maybe_mode_change): Likewise.
8688 * reginfo.c (init_reg_sets_1): Likewise.
8689 (choose_hard_reg_mode): Likewise.
8690 (simplifiable_subregs): Likewise.
8691 * regrename.c (check_new_reg_p): Likewise.
8692 * reload.c (find_valid_class): Likewise.
8693 (find_valid_class_1): Likewise.
8694 (reload_inner_reg_of_subreg): Likewise.
8695 (push_reload): Likewise.
8696 (combine_reloads): Likewise.
8697 (find_dummy_reload): Likewise.
8698 (find_reloads): Likewise.
8699 * reload1.c (find_reg): Likewise.
8700 (set_reload_reg): Likewise.
8701 (allocate_reload_reg): Likewise.
8702 (choose_reload_regs): Likewise.
8703 (reload_adjust_reg_for_temp): Likewise.
8704 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
8705 (simplify_subreg_regno): Likewise.
8706 * sel-sched.c (init_regs_for_mode): Likewise.
8707 * varasm.c (make_decl_rtl): Likewise.
8708 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
8709 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
8711 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
8712 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
8713 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8714 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
8715 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
8716 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8717 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
8718 (arc_mode_class): Delete.
8719 (HARD_REGNO_MODE_OK): Delete.
8720 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8721 (arc_hard_regno_mode_ok): Rename old array to...
8722 (arc_hard_regno_mode_ok_modes): ...this.
8723 (arc_conditional_register_usage): Update accordingly.
8724 (arc_mode_class): Make static.
8725 (arc_hard_regno_mode_ok): New function.
8726 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
8727 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
8728 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8729 (arm_hard_regno_mode_ok): Make static.
8730 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
8732 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
8733 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
8734 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
8736 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8737 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
8738 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
8739 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
8740 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
8741 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8742 * config/bfin/predicates.md (valid_reg_operand): Use
8743 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
8744 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
8745 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
8746 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8747 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
8748 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
8749 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8750 (cr16_hard_regno_mode_ok): Make static and return a bool.
8751 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
8752 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8753 (cris_hard_regno_mode_ok): New function.
8754 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
8755 (epiphany_mode_class): Delete.
8756 (HARD_REGNO_MODE_OK): Delete.
8757 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
8758 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8759 (hard_regno_mode_ok): Rename to...
8760 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
8761 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
8762 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
8764 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
8765 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
8766 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8767 (frv_hard_regno_mode_ok): Make static and return a bool.
8768 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
8770 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
8771 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
8772 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
8773 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
8775 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8776 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
8777 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
8778 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
8780 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8781 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
8782 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8783 (ia64_hard_regno_mode_ok): New function.
8784 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
8785 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8786 (iq2000_hard_regno_mode_ok): New function.
8787 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
8788 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8789 (lm32_hard_regno_mode_ok): New function.
8790 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
8791 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
8792 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
8793 instead of HARD_REGNO_MODE_OK.
8794 (m32c_hard_regno_ok): Rename to...
8795 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
8796 (m32c_cannot_change_mode_class): Update accordingly.
8797 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8798 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
8799 (m32r_mode_class): Delete.
8800 (HARD_REGNO_MODE_OK): Delete.
8801 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8802 (m32r_hard_regno_mode_ok): Rename to...
8803 (m32r_hard_regno_modes): ...this.
8804 (m32r_mode_class): Make static.
8805 (m32r_hard_regno_mode_ok): New function.
8806 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
8807 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
8808 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8809 (m68k_hard_regno_mode_ok): Make static.
8810 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
8811 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8812 (mcore_hard_regno_mode_ok): New function.
8813 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
8814 (HARD_REGNO_MODE_OK): Delete.
8815 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
8817 (microblaze_hard_regno_mode_ok_p): ...this and make static.
8818 (microblaze_hard_regno_mode_ok): New function.
8819 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8820 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
8821 (mips_hard_regno_mode_ok): Delete.
8822 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
8823 (mips_hard_regno_mode_ok_p): ...this and make static.
8824 (mips_hard_regno_mode_ok_p): Rename to...
8825 (mips_hard_regno_mode_ok_uncached): ...this.
8826 (mips_hard_regno_mode_ok): New function.
8827 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
8828 of HARD_REGNO_MODE_OK.
8829 (mips_option_override): Update after above name changes.
8830 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8831 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
8832 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
8833 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
8834 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
8835 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8836 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
8837 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
8838 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
8839 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8840 (msp430_hard_regno_mode_ok): Make static and return a bool.
8841 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
8842 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
8843 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
8845 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8846 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
8847 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
8848 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
8849 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
8850 (PA_HARD_REGNO_MODE_OK): ...this
8851 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
8852 (PA_HARD_REGNO_MODE_OK): ...this.
8853 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8854 (pa_hard_regno_mode_ok): New function.
8855 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
8856 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8857 (pdp11_hard_regno_mode_ok): New function.
8858 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
8859 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
8861 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
8863 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8864 (rs6000_hard_regno_mode_ok): Rename to...
8865 (rs6000_hard_regno_mode_ok_uncached): ...this.
8866 (rs6000_init_hard_regno_mode_ok): Update accordingly.
8867 (rs6000_hard_regno_mode_ok): New function.
8868 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
8869 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
8870 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
8871 (riscv_hard_regno_mode_ok): ...this and make static.
8872 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8873 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
8874 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
8875 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8876 (rl78_hard_regno_mode_ok): Make static and return bool.
8877 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
8878 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
8880 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
8881 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8882 (rs6000_hard_regno_mode_ok): Rename to...
8883 (rs6000_hard_regno_mode_ok_uncached): ...this.
8884 (rs6000_init_hard_regno_mode_ok): Update accordingly.
8885 (rs6000_hard_regno_mode_ok): New function.
8886 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
8887 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
8888 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8889 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
8890 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
8891 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
8892 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8893 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
8894 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
8895 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8896 (sh_hard_regno_mode_ok): Make static.
8897 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
8898 instead of HARD_REGNO_MODE_OK.
8899 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
8900 (sparc_mode_class): Delete.
8901 (HARD_REGNO_MODE_OK): Delete.
8902 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8903 (hard_regno_mode_classes): Make static.
8904 (sparc_mode_class): Likewise.
8905 (sparc_hard_regno_mode_ok): New function.
8906 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
8907 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
8908 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
8910 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8911 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
8912 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
8913 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
8914 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
8915 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8916 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
8917 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
8918 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8919 (visium_hard_regno_mode_ok): New function.
8920 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
8921 instead of HARD_REGNO_MODE_OK.
8922 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
8923 (HARD_REGNO_MODE_OK): Delete.
8924 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
8925 (xtensa_hard_regno_mode_ok_p): ...this and make static.
8926 (xtensa_option_override): Update accordingly.
8927 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8928 (xtensa_hard_regno_mode_ok): New function.
8929 * system.h (HARD_REGNO_MODE_OK): Poison.
8931 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8932 Alan Hayward <alan.hayward@arm.com>
8933 David Sherwood <david.sherwood@arm.com>
8935 * target.def (hard_regno_call_part_clobbered): New hook.
8936 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
8937 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
8938 * doc/tm.texi: Regenerate.
8939 * hooks.h (hook_bool_uint_mode_false): Declare.
8940 * hooks.c (hook_bool_uint_mode_false): New function.
8941 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8942 * cselib.c (cselib_process_insn): Use
8943 targetm.hard_regno_call_part_clobbered instead of
8944 HARD_REGNO_CALL_PART_CLOBBERED.
8945 * ira-conflicts.c (ira_build_conflicts): Likewise.
8946 * ira-costs.c (ira_tune_allocno_costs): Likewise.
8947 * lra-constraints.c (need_for_call_save_p): Likewise.
8948 * lra-lives.c: Include target.h.
8949 (check_pseudos_live_through_calls): Use
8950 targetm.hard_regno_call_part_clobbered instead of
8951 HARD_REGNO_CALL_PART_CLOBBERED.
8952 * regcprop.c: Include target.h.
8953 (copyprop_hardreg_forward_1): Use
8954 targetm.hard_regno_call_part_clobbered instead of
8955 HARD_REGNO_CALL_PART_CLOBBERED.
8956 * reginfo.c (choose_hard_reg_mode): Likewise.
8957 * regrename.c (check_new_reg_p): Likewise.
8958 * reload.c (find_equiv_reg): Likewise.
8959 * reload1.c (emit_reload_insns): Likewise.
8960 * sched-deps.c (deps_analyze_insn): Likewise.
8961 * sel-sched.c (init_regs_for_mode): Likewise.
8962 (mark_unavailable_hard_regs): Likewise.
8963 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
8964 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8965 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
8967 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8968 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8969 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
8971 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
8973 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8974 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8975 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
8977 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8978 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8979 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
8981 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8982 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
8984 * config/powerpcspe/powerpcspe.c
8985 (rs6000_hard_regno_call_part_clobbered): New function.
8986 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8987 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8988 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
8990 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8991 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8992 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
8994 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8995 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8996 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
8998 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8999 Alan Hayward <alan.hayward@arm.com>
9000 David Sherwood <david.sherwood@arm.com>
9002 * rtl.h (subreg_memory_offset): Declare.
9003 * emit-rtl.c (subreg_memory_offset): New function.
9004 * expmed.c (store_bit_field_1): Use it.
9005 * expr.c (undefined_operand_subword_p): Likewise.
9006 * simplify-rtx.c (simplify_subreg): Likewise.
9008 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
9010 PR rtl-optimization/57448
9013 * optabs.c (expand_atomic_load): Place compiler memory barriers if
9014 using atomic_load pattern.
9015 (expand_atomic_store): Likewise.
9017 2017-09-04 Jakub Jelinek <jakub@redhat.com>
9020 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
9021 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
9022 handling. Use replace_call_with_value with NULL instead of
9023 gsi_replace, unlink_stmt_vdef and release_defs.
9025 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
9028 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
9030 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9033 * rtl.h (emit_library_call_value_1): Declare.
9034 (emit_library_call): Replace declaration with a series of overloads.
9035 Remove the parameter count argument.
9036 (emit_library_call_value): Likewise.
9037 * calls.c (emit_library_call_value_1): Make global. Replace varargs
9038 with an "rtx_mode_t *".
9039 (emit_library_call_value): Delete.
9040 (emit_library_call): Likewise.
9041 * asan.c (asan_emit_stack_protection): Update calls accordingly.
9042 (asan_emit_allocas_unpoison): Likewise.
9043 * builtins.c (expand_builtin_powi): Likewise.
9044 (expand_asan_emit_allocas_unpoison): Likewise.
9045 * cfgexpand.c (expand_main_function): Likewise.
9046 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
9047 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
9048 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
9049 * config/arm/arm.c (arm_trampoline_init): Likewise.
9050 (arm_call_tls_get_addr): Likewise.
9051 (arm_expand_divmod_libfunc): Likewise.
9052 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
9053 (smulsi3_highpart): Likewise.
9054 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
9055 (c6x_expand_compare): Likewise.
9056 (c6x_expand_movmem): Likewise.
9057 * config/frv/frv.c (frv_trampoline_init): Likewise.
9058 * config/i386/i386.c (ix86_trampoline_init): Likewise.
9059 (ix86_expand_divmod_libfunc): Likewise.
9060 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
9061 (ia64_expand_compare): Likewise.
9062 (ia64_profile_hook): Likewise.
9063 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
9064 (nonlocal_goto): Likewise.
9065 (restore_stack_nonlocal): Likewise.
9066 * config/m32r/m32r.c (block_move_call): Likewise.
9067 (m32r_trampoline_init): Likewise.
9068 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
9069 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
9070 (m68k_call_m68k_read_tp): Likewise.
9071 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
9072 (microblaze_expand_divide): Likewise.
9073 * config/mips/mips.h (mips_args): Likewise.
9074 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
9075 (MIPS_ICACHE_SYNC): Likewise.
9076 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
9077 (nios2_trampoline_init): Likewise.
9078 * config/pa/pa.c (hppa_tls_call): Likewise.
9079 (pa_trampoline_init): Likewise.
9080 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
9081 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
9082 (expand_strn_compare): Likewise.
9083 (rs6000_generate_compare): Likewise.
9084 (rs6000_expand_float128_convert): Likewise.
9085 (output_profile_hook): Likewise.
9086 (rs6000_trampoline_init): Likewise.
9087 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
9088 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
9089 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
9090 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
9091 (rs6000_generate_compare): Likewise.
9092 (rs6000_expand_float128_convert): Likewise.
9093 (output_profile_hook): Likewise.
9094 (rs6000_trampoline_init): Likewise.
9095 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
9096 * config/sh/sh.c (sh_trampoline_init): Likewise.
9097 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
9098 (sparc_emit_float_lib_cmp): Likewise.
9099 (sparc32_initialize_trampoline): Likewise.
9100 (sparc64_initialize_trampoline): Likewise.
9101 (sparc_profile_hook): Likewise.
9102 * config/spu/spu.c (ea_load_store): Likewise.
9103 * config/spu/spu.md (floatunssidf2): Likewise.
9104 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
9105 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
9106 * config/visium/visium.c (expand_block_move_4): Likewise.
9107 (expand_block_move_2): Likewise.
9108 (expand_block_move_1): Likewise.
9109 (expand_block_set_4): Likewise.
9110 (expand_block_set_2): Likewise.
9111 (expand_block_set_1): Likewise.
9112 (visium_trampoline_init): Likewise.
9113 (visium_profile_hook): Likewise.
9114 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
9115 (xtensa_setup_frame_addresses): Likewise.
9116 (xtensa_trampoline_init): Likewise.
9117 * except.c (sjlj_emit_function_enter): Likewise.
9118 (sjlj_emit_function_exit): Likewise.
9119 * explow.c (allocate_dynamic_stack_space): Likewise.
9120 (probe_stack_range): Likewise.
9121 * expr.c (convert_mode_scalar): Likewise.
9122 * optabs.c (expand_binop): Likewise.
9123 (expand_twoval_binop_libfunc): Likewise.
9124 (expand_unop): Likewise.
9125 (prepare_cmp_insn): Likewise.
9126 (prepare_float_lib_cmp): Likewise.
9127 (expand_float): Likewise.
9128 (expand_fix): Likewise.
9129 (expand_fixed_convert): Likewise.
9130 (maybe_emit_sync_lock_test_and_set): Likewise.
9131 (expand_atomic_compare_and_swap): Likewise.
9132 (expand_mem_thread_fence): Likewise.
9133 (expand_atomic_fetch_op): Likewise.
9135 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
9137 * doc/generic.texi (OpenACC): Adjust URL.
9138 * doc/invoke.texi (C Dialect Options): Ditto.
9140 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
9142 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
9143 predicate for operand 1. Add (m,<S>) constraint.
9144 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
9145 Prevent memory operand 1 with register operand 2.
9147 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
9149 PR rtl-optimization/82024
9150 * combine.c (try_combine): If the combination result is a PARALLEL,
9151 and we only need to retain the SET in there that would be placed
9152 at I2, check that we can place that at I3 instead, before doing so.
9154 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9157 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
9159 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
9160 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
9163 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
9164 Jeff Law <law@redhat.com>
9166 * varasm.c (bss_initializer_p): Do not put constants into .bss
9167 (categorize_decl_for_section): Handle bss_initializer_p returning
9168 false when DECL_INITIAL is NULL.
9170 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9173 * config/s390/s390.c (s390_can_inline_p): New function.
9175 2017-09-01 Jeff Law <law@redhat.com>
9177 PR tree-optimization/82052
9178 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
9179 Always initialize the returned slot after a hash table miss
9180 when INSERT is true.
9182 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
9184 * config/s390/s390.md (mem_signal_fence): Remove.
9185 * doc/md.texi (mem_signal_fence): Remove.
9186 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
9188 * target-insns.def (mem_signal_fence): Remove.
9190 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9193 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
9196 * asan.c (create_odr_indicator): Strip name encoding from assembler
9197 name before appending it after __odr_asan_.
9199 2017-09-01 Martin Liska <mliska@suse.cz>
9201 PR tree-optimization/82059
9202 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
9203 frequency only when an edge is redirected.
9205 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9207 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
9208 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
9209 (arc_conditional_register_usage): Remove ARC600 lp_count
9211 (arc_file_start): Emit Tag_ARC_CPU_variation.
9212 (arc_can_use_doloop_p): New conditions to use ZOLs.
9213 (hwloop_fail): New function.
9214 (hwloop_optimize): Likewise.
9215 (hwloop_pattern_reg): Likewise.
9216 (arc_doloop_hooks): New struct, to be used with reorg_loops.
9217 (arc_reorg_loops): New function, calls reorg_loops.
9218 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
9219 (arc600_corereg_hazard): Remove ZOL checking, case handled by
9221 (arc_loop_hazard): Remove function, functionality moved into
9223 (arc_hazard): Remove arc_loop_hazard call.
9224 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
9225 into hwloop_optimize.
9226 (arc_label_align): Remove ZOL handling.
9227 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
9228 * config/arc/arc.md (doloop_begin): Remove pattern.
9229 (doloop_begin_i): Likewise.
9230 (doloop_end_i): Likewise.
9231 (doloop_fallback): Likewise.
9232 (doloop_fallback_m): Likewise.
9233 (doloop_end): Reimplement expand.
9234 (arc_lp): New pattern for LP instruction.
9235 (loop_end): New pattern.
9236 (loop_fail): Likewise.
9237 (decrement_and_branch_until_zero): Likewise.
9238 * config/arc/arc.opt (mlpc-width): New option.
9239 * doc/invoke.texi (mlpc-width): Document option.
9241 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9243 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
9244 (arc_ccfsm_advance): Fix checking for delay slots.
9245 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
9247 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9249 * config/arc/arc.md (movqi_insn): Add stores to save constant long
9251 (movhi_insn): Update store instruction constraint which are saving
9252 6-bit short immediates.
9253 (movsi_insn): Consider also short scaled load operations.
9254 (zero_extendhisi2_i): Use Usd constraint instead of T.
9255 (extendhisi2_i): Add q constraint.
9256 (arc_clzsi2): Add type and length attributes.
9257 (arc_ctzsi2): Likewise.
9258 * config/arc/constraints.md (Usc): Update constraint, the
9259 assembler can parse two relocations for a single instruction.
9261 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9263 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
9264 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
9266 2017-08-31 Olivier Hainque <hainque@adacore.com>
9268 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
9269 match as powerpc-wrs-vxworks*.
9271 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
9273 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
9274 register constraint for by-element operand.
9275 (aarch64_mls_elt_merge<mode>): Likewise.
9277 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9279 * config/arc/arc.c (arc_can_follow_jump): Check for short
9282 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9284 * config.gcc: Use g.opt for arc.
9285 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
9286 functionality moved to ...
9287 (legitimate_scaled_address_p): New function, ...here.
9288 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
9289 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
9290 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
9292 (arc_override_options): Handle G option.
9293 (arc_output_pic_addr_const): Correct function definition.
9294 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
9295 (arc_decl_anon_ns_mem_p): Delete.
9296 (arc_in_small_data_p): Overhaul this function to take into
9297 consideration the value given via G option.
9298 (arc_rewrite_small_data_1): Renamed and corrected old
9299 arc_rewrite_small_data function.
9300 (arc_rewrite_small_data): New function.
9301 (small_data_pattern): Don't use pic_offset_table_rtx.
9302 * config/arc/arc.h (CC1_SPEC): Recognize G option.
9303 * config/arc/simdext.md (movmisalignv2hi): Use
9304 prepare_move_operands function.
9306 (movmisalign*): Likewise.
9307 * doc/invoke.texi (ARC options): Document -G option.
9309 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9311 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
9313 * config/arc/arc.c (arc_print_operand): Output scalled address for
9314 sdata whenever is possible.
9315 (arc_in_small_data_p): Allow sdata for 64bit datum when double
9316 load/stores are available.
9317 (compact_sda_memory_operand): Check for the alignment required by
9318 code density instructions.
9319 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
9321 * config/arc/constraints.md (Usd): Update constraint.
9322 (Us0): New constraint.
9323 (Usc): Update constraint.
9325 2017-08-31 Richard Biener <rguenther@suse.de>
9328 * dwarf2out.c (dwarf2out_early_global_decl): Process each
9331 2017-08-31 Tamar Christina <tamar.christina@arm.com>
9333 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
9334 Resize type_signature.
9336 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9337 Alan Hayward <alan.hayward@arm.com>
9338 David Sherwood <david.sherwood@arm.com>
9340 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
9341 subregs whose inner modes can be stored in GPRs.
9342 (aarch64_classify_index): Likewise.
9344 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9345 Alan Hayward <alan.hayward@arm.com>
9346 David Sherwood <david.sherwood@arm.com>
9348 * config/aarch64/iterators.md (V_cmp_result): Rename to...
9349 (V_INT_EQUIV): ...this.
9350 (v_cmp_result): Rename to...
9351 (v_int_equiv): ...this.
9352 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
9353 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
9354 (copysign<mode>3): Likewise.
9355 (aarch64_simd_bsl<mode>_internal): Likewise.
9356 (aarch64_simd_bsl<mode>): Likewise.
9357 (vec_cmp<mode><mode>): Likewise.
9358 (vcond<mode><mode>): Likewise.
9359 (vcond<v_cmp_mixed><mode>): Likewise.
9360 (vcondu<mode><v_cmp_mixed>): Likewise.
9361 (aarch64_cm<optab><mode>): Likewise.
9362 (aarch64_cmtst<mode>): Likewise.
9363 (aarch64_fac<optab><mode>): Likewise.
9364 (vec_perm_const<mode>): Likewise.
9365 (vcond_mask_<mode><v_cmp_result>): Rename to...
9366 (vcond_mask_<mode><v_int_equiv>): ...this.
9367 (vec_cmp<mode><v_cmp_result>): Rename to...
9368 (vec_cmp<mode><v_int_equiv>): ...this.
9370 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9371 Alan Hayward <alan.hayward@arm.com>
9372 David Sherwood <david.sherwood@arm.com>
9374 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
9376 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
9377 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
9378 (UNSPEC_LD4_DREG): New unspecs.
9379 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
9380 (aarch64_ld2<mode>_dreg_be): Replace with...
9381 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
9383 (aarch64_ld3<mode>_dreg_le)
9384 (aarch64_ld3<mode>_dreg_be): Replace with...
9385 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
9387 (aarch64_ld4<mode>_dreg_le)
9388 (aarch64_ld4<mode>_dreg_be): Replace with...
9389 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
9392 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9394 PR tree-optimization/81987
9395 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
9396 insert an initializer in a location not dominated by the stride
9399 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
9401 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
9402 on the entire header of the finally block in the fallthru case.
9404 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
9406 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
9408 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
9410 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
9411 rs6000_emit_move_from_cr and call renamed function.
9412 (rs6000_emit_prologue): Call renamed functions.
9413 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
9414 movesi_from_cr, remove volatile CRs.
9416 2017-08-30 Jon Beniston <jon@beniston.com>
9417 Richard Biener <rguenther@suse.de>
9419 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
9420 of VECTOR_MODE_P check.
9421 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
9422 element vector types.
9424 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9426 * df.h (df_read_modify_subreg_p): Remove in favor of...
9427 * rtl.h (read_modify_subreg_p): ...this new function. Take a
9428 const_rtx instead of an rtx.
9429 * cprop.c (local_cprop_find_used_regs): Update accordingly.
9430 * df-problems.c (df_word_lr_mark_ref): Likewise.
9431 * ira-lives.c (mark_pseudo_reg_live): Likewise.
9432 (mark_pseudo_reg_dead): Likewise.
9433 (mark_ref_dead): Likewise.
9434 * reginfo.c (init_subregs_of_mode): Likewise.
9435 * sched-deps.c (sched_analyze_1): Likewise.
9436 * df-scan.c (df_def_record_1): Likewise.
9437 (df_uses_record): Likewise.
9438 (df_read_modify_subreg_p): Remove in favor of...
9439 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
9440 const_rtx instead of an rtx.
9442 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9443 Alan Hayward <alan.hayward@arm.com>
9444 David Sherwood <david.sherwood@arm.com>
9446 * rtl.h (partial_subreg_p): New function.
9447 * caller-save.c (save_call_clobbered_regs): Use it.
9448 * calls.c (expand_call): Likewise.
9449 * combine.c (combinable_i3pat): Likewise.
9450 (simplify_set): Likewise.
9451 (make_extraction): Likewise.
9452 (make_compound_operation_int): Likewise.
9453 (gen_lowpart_or_truncate): Likewise.
9454 (force_to_mode): Likewise.
9455 (make_field_assignment): Likewise.
9456 (reg_truncated_to_mode): Likewise.
9457 (record_truncated_value): Likewise.
9458 (move_deaths): Likewise.
9459 * cse.c (record_jump_cond): Likewise.
9460 (cse_insn): Likewise.
9461 * cselib.c (cselib_lookup_1): Likewise.
9462 * expmed.c (extract_bit_field_using_extv): Likewise.
9463 * function.c (assign_parm_setup_reg): Likewise.
9464 * ifcvt.c (noce_convert_multiple_sets): Likewise.
9465 * ira-build.c (create_insn_allocnos): Likewise.
9466 * lra-coalesce.c (merge_pseudos): Likewise.
9467 * lra-constraints.c (match_reload): Likewise.
9468 (simplify_operand_subreg): Likewise.
9469 (curr_insn_transform): Likewise.
9470 * lra-lives.c (process_bb_lives): Likewise.
9471 * lra.c (new_insn_reg): Likewise.
9472 (lra_substitute_pseudo): Likewise.
9473 * regcprop.c (mode_change_ok): Likewise.
9474 (maybe_mode_change): Likewise.
9475 (copyprop_hardreg_forward_1): Likewise.
9476 * reload.c (push_reload): Likewise.
9477 (find_reloads): Likewise.
9478 (find_reloads_subreg_address): Likewise.
9479 * reload1.c (alter_reg): Likewise.
9480 (eliminate_regs_1): Likewise.
9481 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9483 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
9485 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
9486 back to if statements, including unpack.
9488 2017-08-30 Martin Liska <mliska@suse.cz>
9491 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
9493 (func_checker::compare_asm_inputs_outputs): ... this function.
9494 (func_checker::compare_gimple_asm): Use the function to compare
9495 also ASM constrains.
9496 * ipa-icf-gimple.h: Rename the function.
9498 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9499 Alan Hayward <alan.hayward@arm.com>
9500 David Sherwood <david.sherwood@arm.com>
9502 * coretypes.h (complex_mode): New type.
9503 * gdbhooks.py (build_pretty_printer): Handle it.
9504 * machmode.h (complex_mode): New class.
9505 (complex_mode::includes_p): New function.
9506 (is_complex_int_mode): Likewise.
9507 (is_complex_float_mode): Likewise.
9508 * genmodes.c (get_mode_class): Handle complex mode classes.
9509 * function.c (expand_function_end): Use is_complex_int_mode.
9511 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9512 Alan Hayward <alan.hayward@arm.com>
9513 David Sherwood <david.sherwood@arm.com>
9515 * coretypes.h (scalar_mode_pod): New typedef.
9516 * gdbhooks.py (build_pretty_printer): Handle it.
9517 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
9518 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
9519 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
9520 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
9523 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9524 Alan Hayward <alan.hayward@arm.com>
9525 David Sherwood <david.sherwood@arm.com>
9527 * machmode.h (mode_for_vector): Take a scalar_mode instead
9529 * stor-layout.c (mode_for_vector): Likewise.
9530 * explow.c (promote_mode): Use as_a <scalar_mode>.
9531 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
9533 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9534 Alan Hayward <alan.hayward@arm.com>
9535 David Sherwood <david.sherwood@arm.com>
9537 * target.def (preferred_simd_mode): Take a scalar_mode
9538 instead of a machine_mode.
9539 * targhooks.h (default_preferred_simd_mode): Likewise.
9540 * targhooks.c (default_preferred_simd_mode): Likewise.
9541 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
9542 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
9543 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
9544 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
9545 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
9546 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
9547 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
9548 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
9550 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
9551 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
9552 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
9553 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
9554 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
9555 * doc/tm.texi: Regenerate.
9556 * optabs-query.c (can_vec_mask_load_store_p): Return false for
9559 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9560 Alan Hayward <alan.hayward@arm.com>
9561 David Sherwood <david.sherwood@arm.com>
9563 * target.def (scalar_mode_supported_p): Take a scalar_mode
9564 instead of a machine_mode.
9565 * targhooks.h (default_scalar_mode_supported_p): Likewise.
9566 * targhooks.c (default_scalar_mode_supported_p): Likewise.
9567 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
9568 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
9569 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
9570 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
9571 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
9572 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
9573 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
9574 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
9575 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
9576 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
9577 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
9578 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
9580 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
9581 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
9582 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
9583 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
9584 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
9586 * doc/tm.texi: Regenerate.
9588 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9589 Alan Hayward <alan.hayward@arm.com>
9590 David Sherwood <david.sherwood@arm.com>
9592 * coretypes.h (opt_scalar_mode): New typedef.
9593 * gdbhooks.py (build_pretty_printers): Handle it.
9594 * machmode.h (mode_iterator::get_2xwider): Add overload for
9596 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
9598 * expr.c (convert_mode_scalar): Likewise.
9599 * omp-low.c (omp_clause_aligned_alignment): Likewise.
9600 * optabs.c (expand_float): Likewise.
9601 (expand_fix): Likewise.
9602 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
9604 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9605 Alan Hayward <alan.hayward@arm.com>
9606 David Sherwood <david.sherwood@arm.com>
9608 * optabs.c (expand_float): Explicitly check for scalars before
9609 using a branching expansion.
9610 (expand_fix): Likewise.
9612 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9613 Alan Hayward <alan.hayward@arm.com>
9614 David Sherwood <david.sherwood@arm.com>
9616 * expr.c (convert_mode): Split scalar handling out into...
9617 (convert_mode_scalar): ...this new function. Treat the modes
9620 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9621 Alan Hayward <alan.hayward@arm.com>
9622 David Sherwood <david.sherwood@arm.com>
9624 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
9626 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
9628 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9629 Alan Hayward <alan.hayward@arm.com>
9630 David Sherwood <david.sherwood@arm.com>
9632 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
9633 rather than a machine_mode.
9634 (fixed_from_string): Likewise.
9635 (fixed_convert): Likewise.
9636 (fixed_convert_from_int): Likewise.
9637 (fixed_convert_from_real): Likewise.
9638 (real_convert_from_fixed): Likewise.
9639 * fixed-value.c (fixed_from_double_int): Likewise.
9640 (fixed_from_string): Likewise.
9641 (fixed_convert): Likewise.
9642 (fixed_convert_from_int): Likewise.
9643 (fixed_convert_from_real): Likewise.
9644 (real_convert_from_fixed): Likewise.
9645 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
9647 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9648 Alan Hayward <alan.hayward@arm.com>
9649 David Sherwood <david.sherwood@arm.com>
9651 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
9652 of separate mode class checks. Do not allow vector modes here.
9653 (immed_wide_int_const): Use as_a <scalar_mode>.
9654 * explow.c (trunc_int_for_mode): Likewise.
9655 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
9656 (wi::shwi): Likewise.
9657 (wi::min_value): Likewise.
9658 (wi::max_value): Likewise.
9659 * dwarf2out.c (loc_descriptor): Likewise.
9660 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
9663 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9664 Alan Hayward <alan.hayward@arm.com>
9665 David Sherwood <david.sherwood@arm.com>
9667 * tree.h (SCALAR_TYPE_MODE): New macro.
9668 * expr.c (expand_expr_addr_expr_1): Use it.
9669 (expand_expr_real_2): Likewise.
9670 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
9671 (fold_convert_const_fixed_from_int): Likewise.
9672 (fold_convert_const_fixed_from_real): Likewise.
9673 (native_encode_fixed): Likewise
9674 (native_encode_complex): Likewise
9675 (native_encode_vector): Likewise.
9676 (native_interpret_fixed): Likewise.
9677 (native_interpret_real): Likewise.
9678 (native_interpret_complex): Likewise.
9679 (native_interpret_vector): Likewise.
9680 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
9681 (simd_clone_adjust_argument_types): Likewise.
9682 (simd_clone_init_simd_arrays): Likewise.
9683 (simd_clone_adjust): Likewise.
9684 * stor-layout.c (layout_type): Likewise.
9685 * tree.c (build_minus_one_cst): Likewise.
9686 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
9687 * tree-inline.c (estimate_move_cost): Likewise.
9688 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
9689 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
9690 (vectorizable_reduction): Likewise.
9691 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
9692 (vect_recog_mixed_size_cond_pattern): Likewise.
9693 (check_bool_pattern): Likewise.
9694 (adjust_bool_pattern): Likewise.
9695 (search_type_for_mask_1): Likewise.
9696 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
9697 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
9698 (vectorizable_load): Likewise.
9699 (vectorizable_store): Likewise.
9700 * ubsan.c (ubsan_encode_value): Likewise.
9701 * varasm.c (output_constant): Likewise.
9703 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9704 Alan Hayward <alan.hayward@arm.com>
9705 David Sherwood <david.sherwood@arm.com>
9707 * coretypes.h (scalar_mode): New class.
9708 * machmode.h (scalar_mode): Likewise.
9709 (scalar_mode::includes_p): New function.
9710 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
9711 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
9712 * genmodes.c (get_mode_class): Handle remaining scalar modes.
9713 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
9714 * expmed.c (store_bit_field_1): Likewise.
9715 (extract_bit_field_1): Likewise.
9716 * expr.c (write_complex_part): Likewise.
9717 (read_complex_part): Likewise.
9718 (emit_move_complex_push): Likewise.
9719 (expand_expr_real_2): Likewise.
9720 * function.c (assign_parm_setup_reg): Likewise.
9721 (assign_parms_unsplit_complex): Likewise.
9722 * optabs.c (expand_binop): Likewise.
9723 * rtlanal.c (subreg_get_info): Likewise.
9724 * simplify-rtx.c (simplify_immed_subreg): Likewise.
9725 * varasm.c (output_constant_pool_2): Likewise.
9727 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9728 Alan Hayward <alan.hayward@arm.com>
9729 David Sherwood <david.sherwood@arm.com>
9731 * expmed.c (extract_high_half): Use scalar_int_mode and remove
9733 (expmed_mult_highpart_optab): Likewise.
9734 (expmed_mult_highpart): Likewise.
9736 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9737 Alan Hayward <alan.hayward@arm.com>
9738 David Sherwood <david.sherwood@arm.com>
9740 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
9741 instead of a machine_mode.
9742 (builtin_memset_read_str): Likewise.
9743 * builtins.c (c_readstr): Likewise.
9744 (builtin_memcpy_read_str): Likewise.
9745 (builtin_strncpy_read_str): Likewise.
9746 (builtin_memset_read_str): Likewise.
9747 (builtin_memset_gen_str): Likewise.
9748 (expand_builtin_signbit): Use scalar_int_mode for local variables.
9749 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
9750 instead of a machine_mode.
9751 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
9753 (make_extraction): Likewise.
9754 (try_widen_shift_mode): Take and return scalar_int_modes instead
9756 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
9757 a scalar_int_mode instead of a machine_mode.
9758 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
9759 (avr_addr_space_pointer_mode): Likewise.
9760 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
9761 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
9762 (msp430_unwind_word_mode): Likewise.
9763 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
9764 (spu_addr_space_pointer_mode): Likewise.
9765 (spu_addr_space_address_mode): Likewise.
9766 (spu_libgcc_cmp_return_mode): Likewise.
9767 (spu_libgcc_shift_count_mode): Likewise.
9768 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
9769 (rl78_addr_space_pointer_mode): Likewise.
9770 (fl78_unwind_word_mode): Likewise.
9771 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
9773 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
9774 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
9775 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
9776 (mips_valid_pointer_mode): Likewise.
9777 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
9778 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
9779 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
9781 (ft32_addr_space_address_mode): Likewise.
9782 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
9783 scalar_int_mode instead of a machine_mode.
9784 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
9786 (m32c_addr_space_address_mode): Likewise.
9787 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
9788 (rs6000_eh_return_filter_mode): Likewise.
9789 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
9790 (rs6000_eh_return_filter_mode): Likewise.
9791 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
9792 (s390_libgcc_shift_count_mode): Likewise.
9793 (s390_unwind_word_mode): Likewise.
9794 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
9796 * target.def (mode_rep_extended): Likewise.
9797 (valid_pointer_mode): Likewise.
9798 (addr_space.valid_pointer_mode): Likewise.
9799 (eh_return_filter_mode): Return a scalar_int_mode rather than
9801 (libgcc_cmp_return_mode): Likewise.
9802 (libgcc_shift_count_mode): Likewise.
9803 (unwind_word_mode): Likewise.
9804 (addr_space.pointer_mode): Likewise.
9805 (addr_space.address_mode): Likewise.
9806 * doc/tm.texi: Regenerate.
9807 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
9809 (do_jump): Use scalar_int_mode for local variables.
9810 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
9811 rather than a machine_mode.
9812 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
9813 (scompare_loc_descriptor_wide): Likewise.
9814 (scompare_loc_descriptor_narrow): Likewise.
9815 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
9817 * except.c (sjlj_emit_dispatch_table): Likewise.
9818 (expand_builtin_eh_copy_values): Likewise.
9819 * explow.c (convert_memory_address_addr_space_1): Likewise.
9820 Take a scalar_int_mode rather than a machine_mode.
9821 (convert_memory_address_addr_space): Take a scalar_int_mode rather
9822 than a machine_mode.
9823 (memory_address_addr_space): Use scalar_int_mode for local variables.
9824 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
9825 rather than a machine_mode.
9826 * expmed.c (mask_rtx): Likewise.
9827 (init_expmed_one_conv): Likewise.
9828 (expand_mult_highpart_adjust): Likewise.
9829 (extract_high_half): Likewise.
9830 (expmed_mult_highpart_optab): Likewise.
9831 (expmed_mult_highpart): Likewise.
9832 (expand_smod_pow2): Likewise.
9833 (expand_sdiv_pow2): Likewise.
9834 (emit_store_flag_int): Likewise.
9835 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
9837 (extract_low_bits): Likewise.
9838 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
9840 * expr.c (pieces_addr::adjust): Likewise.
9841 (can_store_by_pieces): Likewise.
9842 (store_by_pieces): Likewise.
9843 (clear_by_pieces_1): Likewise.
9844 (expand_expr_addr_expr_1): Likewise.
9845 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
9846 (expand_expr_real_1): Likewise.
9847 (try_casesi): Likewise.
9848 * final.c (shorten_branches): Likewise.
9849 * fold-const.c (fold_convert_const_int_from_fixed): Change the
9850 type of "mode" to machine_mode.
9851 * internal-fn.c (expand_arith_overflow_result_store): Take a
9852 scalar_int_mode rather than a machine_mode.
9853 (expand_mul_overflow): Use scalar_int_mode for local variables.
9854 * loop-doloop.c (doloop_modify): Likewise.
9855 (doloop_optimize): Likewise.
9856 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
9857 than a machine_mode.
9858 (expand_doubleword_shift_condmove): Likewise.
9859 (expand_doubleword_shift): Likewise.
9860 (expand_doubleword_clz): Likewise.
9861 (expand_doubleword_popcount): Likewise.
9862 (expand_doubleword_parity): Likewise.
9863 (expand_absneg_bit): Use scalar_int_mode for local variables.
9864 (prepare_float_lib_cmp): Likewise.
9865 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
9866 rather than a machine_mode.
9867 (convert_memory_address_addr_space): Likewise.
9868 (get_mode_bounds): Likewise.
9869 (get_address_mode): Return a scalar_int_mode rather than a
9871 * rtlanal.c (get_address_mode): Likewise.
9872 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
9873 than a machine_mode.
9874 * targhooks.c (default_mode_rep_extended): Likewise.
9875 (default_valid_pointer_mode): Likewise.
9876 (default_addr_space_valid_pointer_mode): Likewise.
9877 (default_eh_return_filter_mode): Return a scalar_int_mode rather
9878 than a machine_mode.
9879 (default_libgcc_cmp_return_mode): Likewise.
9880 (default_libgcc_shift_count_mode): Likewise.
9881 (default_unwind_word_mode): Likewise.
9882 (default_addr_space_pointer_mode): Likewise.
9883 (default_addr_space_address_mode): Likewise.
9884 * targhooks.h (default_eh_return_filter_mode): Likewise.
9885 (default_libgcc_cmp_return_mode): Likewise.
9886 (default_libgcc_shift_count_mode): Likewise.
9887 (default_unwind_word_mode): Likewise.
9888 (default_addr_space_pointer_mode): Likewise.
9889 (default_addr_space_address_mode): Likewise.
9890 (default_mode_rep_extended): Take a scalar_int_mode rather than
9892 (default_valid_pointer_mode): Likewise.
9893 (default_addr_space_valid_pointer_mode): Likewise.
9894 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
9896 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
9897 rather than a machine_mode.
9898 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
9899 for local variables.
9900 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
9901 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
9902 than a machine_mode.
9904 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9905 Alan Hayward <alan.hayward@arm.com>
9906 David Sherwood <david.sherwood@arm.com>
9908 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
9909 the mode argument to scalar_int_mode.
9910 (do_jump_by_parts_zero_rtx): Likewise.
9911 (do_jump_by_parts_equality_rtx): Likewise.
9912 (do_jump_by_parts_greater): Take a mode argument.
9913 (do_jump_by_parts_equality): Likewise.
9914 (do_jump_1): Update calls accordingly.
9916 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9917 Alan Hayward <alan.hayward@arm.com>
9918 David Sherwood <david.sherwood@arm.com>
9920 * is-a.h (safe_dyn_cast): New function.
9921 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
9922 (jump_table_for_label): Likewise.
9923 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
9924 instead of an rtx_insn *.
9925 (shorten_branches): Use dyn_cast instead of LABEL_P and
9926 JUMP_TABLE_DATA_P. Use jump_table_for_label and
9927 rtx_jump_table_data::get_data_mode.
9928 (final_scan_insn): Likewise.
9930 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9931 Alan Hayward <alan.hayward@arm.com>
9932 David Sherwood <david.sherwood@arm.com>
9934 * combine.c (try_combine): Use is_a <scalar_int_mode> when
9935 trying to combine a full-register integer set with a subreg
9938 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9939 Alan Hayward <alan.hayward@arm.com>
9940 David Sherwood <david.sherwood@arm.com>
9942 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
9943 that is always either address_mode or pointer_mode.
9945 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9946 Alan Hayward <alan.hayward@arm.com>
9947 David Sherwood <david.sherwood@arm.com>
9949 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
9950 when the two are known to be equal.
9952 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9953 Alan Hayward <alan.hayward@arm.com>
9954 David Sherwood <david.sherwood@arm.com>
9956 * simplify-rtx.c (simplify_const_unary_operation): Use
9957 is_a <scalar_int_mode> instead of checking for a nonzero
9958 precision. Forcibly convert op_mode to a scalar_int_mode
9959 in that case. More clearly differentiate the operand and
9960 result modes and use the former when deciding what the value
9961 of a count-bits operation should be. Use is_int_mode instead
9962 of checking for a MODE_INT. Remove redundant check for whether
9963 this mode has a zero precision.
9965 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9966 Alan Hayward <alan.hayward@arm.com>
9967 David Sherwood <david.sherwood@arm.com>
9969 * optabs.c (widen_leading): Change the type of the mode argument
9970 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
9971 (widen_bswap): Likewise.
9972 (expand_parity): Likewise.
9973 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
9974 (expand_ffs): Likewise.
9975 (epand_unop): Check for scalar integer modes before calling the
9978 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9979 Alan Hayward <alan.hayward@arm.com>
9980 David Sherwood <david.sherwood@arm.com>
9982 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
9984 (expand_expr_real_1): Update call accordingly.
9986 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9987 Alan Hayward <alan.hayward@arm.com>
9988 David Sherwood <david.sherwood@arm.com>
9990 * expmed.c (store_bit_field_using_insv): Add op0_mode and
9991 value_mode arguments. Use scalar_int_mode internally.
9992 (store_bit_field_1): Rename the new integer mode from imode
9993 to op0_mode and use it instead of GET_MODE (op0). Update calls
9994 to store_split_bit_field, store_bit_field_using_insv and
9995 store_fixed_bit_field.
9996 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
9997 Use scalar_int_mode internally. Use a bit count rather than a mode
9998 when calculating the largest bit size for get_best_mode.
9999 Update calls to store_split_bit_field and store_fixed_bit_field_1.
10000 (store_fixed_bit_field_1): Add mode and value_mode arguments.
10001 Remove assertion that OP0 has a scalar integer mode.
10002 (store_split_bit_field): Add op0_mode and value_mode arguments.
10003 Update calls to extract_fixed_bit_field.
10004 (extract_bit_field_using_extv): Add an op0_mode argument.
10005 Use scalar_int_mode internally.
10006 (extract_bit_field_1): Rename the new integer mode from imode to
10007 op0_mode and use it instead of GET_MODE (op0). Update calls to
10008 extract_split_bit_field, extract_bit_field_using_extv and
10009 extract_fixed_bit_field.
10010 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
10011 to extract_split_bit_field and extract_fixed_bit_field_1.
10012 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
10013 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
10014 on the target mode.
10015 (extract_split_bit_field): Add an op0_mode argument. Update call
10016 to extract_fixed_bit_field.
10018 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10019 Alan Hayward <alan.hayward@arm.com>
10020 David Sherwood <david.sherwood@arm.com>
10022 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
10023 * explow.c (hard_function_value): Likewise.
10024 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
10025 convert_to_mode call outside the loop.
10026 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
10027 for the mode iterator. Require the mode specified by max_pieces
10029 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
10031 (copy_blkmode_to_reg): Likewise.
10032 (set_storage_via_setmem): Likewise.
10033 * optabs.c (prepare_cmp_insn): Likewise.
10034 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
10035 * stor-layout.c (finish_bitfield_representative): Likewise.
10037 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10038 Alan Hayward <alan.hayward@arm.com>
10039 David Sherwood <david.sherwood@arm.com>
10041 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
10042 * expr.c (convert_move): Use them.
10043 (convert_modes): Likewise.
10044 (store_expr_with_bounds): Likewise.
10046 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10047 Alan Hayward <alan.hayward@arm.com>
10048 David Sherwood <david.sherwood@arm.com>
10050 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
10051 parameter for the mode of "x". Remove the "known_x", "known_mode"
10052 and "known_ret" arguments. Change the type of the mode argument
10053 to scalar_int_mode.
10054 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
10055 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
10056 (reg_num_sign_bit_copies_for_combine): Likewise.
10057 * rtlanal.c (nonzero_bits1): Likewise.
10058 (num_sign_bit_copies1): Likewise.
10059 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
10060 (reg_num_sign_bit_copies_general): Likewise.
10061 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
10062 (reg_nonzero_bits_general): Likewise.
10064 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10065 Alan Hayward <alan.hayward@arm.com>
10066 David Sherwood <david.sherwood@arm.com>
10068 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
10069 than in subroutines. Return 1 for non-integer modes.
10070 (cached_num_sign_bit_copies): Change the type of the mode parameter
10071 to scalar_int_mode.
10072 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
10073 classes. Handle CONST_INT_P first and then check whether X also
10074 has a scalar integer mode. Check the same thing for inner registers
10075 of a SUBREG and for values that are being extended or truncated.
10077 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10078 Alan Hayward <alan.hayward@arm.com>
10079 David Sherwood <david.sherwood@arm.com>
10081 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
10082 in subroutines. Return the mode mask for non-integer modes.
10083 (cached_nonzero_bits): Change the type of the mode parameter
10084 to scalar_int_mode.
10085 (nonzero_bits1): Likewise. Remove early exit for other mode
10086 classes. Handle CONST_INT_P first and then check whether X
10087 also has a scalar integer mode.
10089 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10090 Alan Hayward <alan.hayward@arm.com>
10091 David Sherwood <david.sherwood@arm.com>
10093 * expr.c (widest_int_mode_for_size): Make the comment match the code.
10094 Return a scalar_int_mode and assert that the size is greater than
10096 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
10097 (op_by_pieces_d::op_by_pieces_d): Likewise.
10098 (op_by_pieces_d::run): Likewise.
10099 (can_store_by_pieces): Likewise.
10101 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10102 Alan Hayward <alan.hayward@arm.com>
10103 David Sherwood <david.sherwood@arm.com>
10105 * combine.c (extract_left_shift): Add a mode argument and update
10107 (make_compound_operation_int): Change the type of the mode parameter
10108 to scalar_int_mode and update the call to extract_left_shift.
10110 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10111 Alan Hayward <alan.hayward@arm.com>
10112 David Sherwood <david.sherwood@arm.com>
10114 * combine.c (simplify_and_const_int): Change the type of the mode
10115 parameter to scalar_int_mode.
10116 (simplify_and_const_int_1): Likewise. Update recursive call.
10118 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10119 Alan Hayward <alan.hayward@arm.com>
10120 David Sherwood <david.sherwood@arm.com>
10122 * combine.c (simplify_compare_const): Check that the mode is a
10123 scalar_int_mode (rather than VOIDmode) before testing its
10125 (simplify_comparison): Move COMPARISON_P handling out of the
10126 loop and restrict the latter part of the loop to scalar_int_modes.
10127 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
10128 and when considering SUBREG_REGs. Use is_int_mode instead of
10129 checking GET_MODE_CLASS against MODE_INT.
10131 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10132 Alan Hayward <alan.hayward@arm.com>
10133 David Sherwood <david.sherwood@arm.com>
10135 * combine.c (try_widen_shift_mode): Move check for equal modes to...
10136 (simplify_shift_const_1): ...here. Use scalar_int_mode for
10137 shift_unit_mode and for modes involved in scalar shifts.
10139 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10140 Alan Hayward <alan.hayward@arm.com>
10141 David Sherwood <david.sherwood@arm.com>
10143 * combine.c (force_int_to_mode): New function, split out from...
10144 (force_to_mode): ...here. Keep xmode up-to-date and use it
10145 instead of GET_MODE (x).
10147 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10148 Alan Hayward <alan.hayward@arm.com>
10149 David Sherwood <david.sherwood@arm.com>
10151 * optabs-query.h (extraction_insn::struct_mode): Change type to
10152 opt_scalar_int_mode and update comment.
10153 (extraction_insn::field_mode): Change type to scalar_int_mode.
10154 (extraction_insn::pos_mode): Likewise.
10155 * combine.c (make_extraction): Update accordingly.
10156 * optabs-query.c (get_traditional_extraction_insn): Likewise.
10157 (get_optab_extraction_insn): Likewise.
10158 * recog.c (simplify_while_replacing): Likewise.
10159 * expmed.c (narrow_bit_field_mem): Change the type of the mode
10160 parameter to opt_scalar_int_mode.
10162 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10163 Alan Hayward <alan.hayward@arm.com>
10164 David Sherwood <david.sherwood@arm.com>
10166 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
10167 to a scalar_int_mode instead of a machine_mode.
10168 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
10169 (get_best_mode): Return a boolean and use a pointer argument to store
10170 the selected mode. Replace the limit mode parameter with a bit limit.
10171 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
10172 for the values returned by bit_field_mode_iterator::next_mode.
10173 (store_bit_field): Update call to get_best_mode.
10174 (store_fixed_bit_field): Likewise.
10175 (extract_fixed_bit_field): Likewise.
10176 * expr.c (optimize_bitfield_assignment_op): Likewise.
10177 * fold-const.c (optimize_bit_field_compare): Likewise.
10178 (fold_truth_andor_1): Likewise.
10179 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
10180 Update for new type of m_mode.
10181 (get_best_mode): As above.
10183 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10184 Alan Hayward <alan.hayward@arm.com>
10185 David Sherwood <david.sherwood@arm.com>
10187 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
10188 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
10189 (store_bit_field): Check is_a <scalar_int_mode> before calling
10190 strict_volatile_bitfield_p.
10191 (extract_bit_field): Likewise.
10193 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10194 Alan Hayward <alan.hayward@arm.com>
10195 David Sherwood <david.sherwood@arm.com>
10197 * target.def (cstore_mode): Return a scalar_int_mode.
10198 * doc/tm.texi: Regenerate.
10199 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
10200 * targhooks.h (default_cstore_mode): Likewise.
10201 * targhooks.c (default_cstore_mode): Likewise, using a forced
10203 * expmed.c (emit_cstore): Expect the target of the cstore to be
10206 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10207 Alan Hayward <alan.hayward@arm.com>
10208 David Sherwood <david.sherwood@arm.com>
10210 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
10212 (niter_desc): Likewise mode.
10213 (iv_analyze): Add a mode parameter.
10215 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
10216 and change its type to scalar_int_mode.
10217 * loop-iv.c: Update commentary at head of file.
10218 (iv_constant): Pass the mode paraeter before the rtx it describes
10219 and change its type to scalar_int_mode. Remove VOIDmode handling.
10220 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
10221 (iv_extend): Likewise.
10222 (shorten_into_mode): Likewise.
10223 (iv_add): Use scalar_int_mode.
10224 (iv_mult): Likewise.
10225 (iv_shift): Likewise.
10226 (canonicalize_iv_subregs): Likewise.
10227 (get_biv_step_1): Pass the outer_mode parameter before the rtx
10228 it describes and change its mode to scalar_int_mode. Also change
10229 the type of the returned inner_mode to scalar_int_mode.
10230 (get_biv_step): Likewise, turning outer_mode from a pointer
10231 into a direct parameter. Update call to get_biv_step_1.
10232 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
10233 iv_constant and get_biv_step.
10234 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
10235 and change its type to scalar_int_mode. Don't initialise iv->mode
10236 to VOIDmode and remove later checks for its still being VOIDmode.
10237 Update calls to iv_analyze_op and iv_analyze_expr. Check
10238 is_a <scalar_int_mode> when changing the mode under consideration.
10239 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
10240 Update call to iv_analyze_expr.
10241 (iv_analyze_op): Add a mode parameter. Reject subregs whose
10242 inner register is not also a scalar_int_mode. Update call to
10244 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
10245 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
10246 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
10247 separate mode class checks. Update calls to iv_analyze. Remove
10248 fix-up of VOIDmodes after iv_analyze_biv.
10249 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
10250 don't have a scalar_int_mode. Update call to biv_p.
10252 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10253 Alan Hayward <alan.hayward@arm.com>
10254 David Sherwood <david.sherwood@arm.com>
10256 * cfgexpand.c (convert_debug_memory_address): Use
10257 as_a <scalar_int_mode>.
10258 * combine.c (expand_compound_operation): Likewise.
10259 (make_extraction): Likewise.
10260 (change_zero_ext): Likewise.
10261 (simplify_comparison): Likewise.
10262 * cse.c (cse_insn): Likewise.
10263 * dwarf2out.c (minmax_loc_descriptor): Likewise.
10264 (mem_loc_descriptor): Likewise.
10265 (loc_descriptor): Likewise.
10266 * expmed.c (init_expmed_one_mode): Likewise.
10267 (synth_mult): Likewise.
10268 (emit_store_flag_1): Likewise.
10269 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
10270 of a comparison with size.
10271 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
10272 (reduce_to_bit_field_precision): Likewise.
10273 * function.c (expand_function_end): Likewise.
10274 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
10275 * loop-doloop.c (doloop_modify): Likewise.
10276 * optabs.c (expand_binop): Likewise.
10277 (expand_unop): Likewise.
10278 (expand_copysign_absneg): Likewise.
10279 (prepare_cmp_insn): Likewise.
10280 (maybe_legitimize_operand): Likewise.
10281 * recog.c (const_scalar_int_operand): Likewise.
10282 * rtlanal.c (get_address_mode): Likewise.
10283 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10284 (simplify_cond_clz_ctz): Likewise.
10285 * tree-nested.c (get_nl_goto_field): Likewise.
10286 * tree.c (build_vector_type_for_mode): Likewise.
10287 * var-tracking.c (use_narrower_mode): Likewise.
10289 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10290 Alan Hayward <alan.hayward@arm.com>
10291 David Sherwood <david.sherwood@arm.com>
10293 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
10294 * builtins.c (expand_builtin_signbit): Use it.
10295 * cfgexpand.c (expand_debug_expr): Likewise.
10296 * dojump.c (do_jump): Likewise.
10297 (do_compare_and_jump): Likewise.
10298 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
10299 * expmed.c (make_tree): Likewise.
10300 * expr.c (expand_expr_real_2): Likewise.
10301 (expand_expr_real_1): Likewise.
10302 (try_casesi): Likewise.
10303 * fold-const-call.c (fold_const_call_ss): Likewise.
10304 * fold-const.c (unextend): Likewise.
10305 (extract_muldiv_1): Likewise.
10306 (fold_single_bit_test): Likewise.
10307 (native_encode_int): Likewise.
10308 (native_encode_string): Likewise.
10309 (native_interpret_int): Likewise.
10310 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
10311 * internal-fn.c (expand_addsub_overflow): Likewise.
10312 (expand_neg_overflow): Likewise.
10313 (expand_mul_overflow): Likewise.
10314 (expand_arith_overflow): Likewise.
10315 * match.pd: Likewise.
10316 * stor-layout.c (layout_type): Likewise.
10317 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10318 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
10319 * tree-ssanames.c (get_range_info): Likewise.
10320 * tree-switch-conversion.c (array_value_type) Likewise.
10321 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
10322 (vect_recog_divmod_pattern): Likewise.
10323 (vect_recog_mixed_size_cond_pattern): Likewise.
10324 * tree-vrp.c (extract_range_basic): Likewise.
10325 (simplify_float_conversion_using_ranges): Likewise.
10326 * tree.c (int_fits_type_p): Likewise.
10327 * ubsan.c (instrument_bool_enum_load): Likewise.
10328 * varasm.c (mergeable_string_section): Likewise.
10329 (narrowing_initializer_constant_valid_p): Likewise.
10330 (output_constant): Likewise.
10332 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10333 Alan Hayward <alan.hayward@arm.com>
10334 David Sherwood <david.sherwood@arm.com>
10336 * machmode.h (NARROWEST_INT_MODE): New macro.
10337 * expr.c (alignment_for_piecewise_move): Use it instead of
10338 GET_CLASS_NARROWEST_MODE (MODE_INT).
10339 (push_block): Likewise.
10340 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
10342 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10344 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10345 Alan Hayward <alan.hayward@arm.com>
10346 David Sherwood <david.sherwood@arm.com>
10348 * postreload.c (move2add_valid_value_p): Change the type of the
10349 mode parameter to scalar_int_mode.
10350 (move2add_use_add2_insn): Add a mode parameter and use it instead
10352 (move2add_use_add3_insn): Likewise.
10353 (reload_cse_move2add): Update accordingly.
10355 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10356 Alan Hayward <alan.hayward@arm.com>
10357 David Sherwood <david.sherwood@arm.com>
10359 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
10361 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
10362 * optabs.c (expand_unop): Likewise.
10364 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10365 Alan Hayward <alan.hayward@arm.com>
10366 David Sherwood <david.sherwood@arm.com>
10368 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
10369 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
10370 (popcount_loc_descriptor): Likewise.
10371 (bswap_loc_descriptor): Likewise.
10372 (rotate_loc_descriptor): Likewise.
10373 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
10374 calling the functions above.
10376 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10377 Alan Hayward <alan.hayward@arm.com>
10378 David Sherwood <david.sherwood@arm.com>
10380 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
10382 (try_combine): Likewise.
10383 (simplify_if_then_else): Likewise.
10384 * cse.c (cse_insn): Likewise.
10385 * dwarf2out.c (mem_loc_descriptor): Likewise.
10386 * emit-rtl.c (gen_lowpart_common): Likewise.
10387 * simplify-rtx.c (simplify_truncation): Likewise.
10388 (simplify_binary_operation_1): Likewise.
10389 (simplify_const_relational_operation): Likewise.
10390 (simplify_ternary_operation): Likewise.
10391 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
10393 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10394 Alan Hayward <alan.hayward@arm.com>
10395 David Sherwood <david.sherwood@arm.com>
10397 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
10398 * reload.c (push_reload): Likewise.
10399 (find_reloads): Likewise.
10401 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10402 Alan Hayward <alan.hayward@arm.com>
10403 David Sherwood <david.sherwood@arm.com>
10405 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
10406 (make_compound_operation_int): Likewise.
10407 (change_zero_ext): Likewise.
10408 * expr.c (convert_move): Likewise.
10409 (convert_modes): Likewise.
10410 * fwprop.c (forward_propagate_subreg): Likewise.
10411 * loop-iv.c (get_biv_step_1): Likewise.
10412 * optabs.c (widen_operand): Likewise.
10413 * postreload.c (move2add_valid_value_p): Likewise.
10414 * recog.c (simplify_while_replacing): Likewise.
10415 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10416 (simplify_binary_operation_1): Likewise. Remove redundant
10417 mode equality check.
10419 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10420 Alan Hayward <alan.hayward@arm.com>
10421 David Sherwood <david.sherwood@arm.com>
10423 * combine.c (combine_simplify_rtx): Add checks for
10424 is_a <scalar_int_mode>.
10425 (simplify_if_then_else): Likewise.
10426 (make_field_assignment): Likewise.
10427 (simplify_comparison): Likewise.
10428 * ifcvt.c (noce_try_bitop): Likewise.
10429 * loop-invariant.c (canonicalize_address_mult): Likewise.
10430 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10432 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10433 Alan Hayward <alan.hayward@arm.com>
10434 David Sherwood <david.sherwood@arm.com>
10436 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
10437 is_a <scalar_int_mode> instead of != BLKmode.
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 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
10444 instead of != VOIDmode.
10445 * combine.c (if_then_else_cond): Likewise.
10446 (change_zero_ext): Likewise.
10447 * dwarf2out.c (mem_loc_descriptor): Likewise.
10448 (loc_descriptor): Likewise.
10449 * rtlanal.c (canonicalize_condition): Likewise.
10450 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
10452 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10453 Alan Hayward <alan.hayward@arm.com>
10454 David Sherwood <david.sherwood@arm.com>
10456 * simplify-rtx.c (simplify_binary_operation_1): Use
10457 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
10459 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10460 Alan Hayward <alan.hayward@arm.com>
10461 David Sherwood <david.sherwood@arm.com>
10463 * wide-int.h (int_traits<unsigned char>) New class.
10464 (int_traits<unsigned short>) Likewise.
10465 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
10466 Use GET_MODE_UNIT_PRECISION and remove redundant test for
10468 * combine.c (set_nonzero_bits_and_sign_copies): Use
10469 is_a <scalar_int_mode>.
10470 (find_split_point): Likewise.
10471 (combine_simplify_rtx): Likewise.
10472 (simplify_logical): Likewise.
10473 (expand_compound_operation): Likewise.
10474 (expand_field_assignment): Likewise.
10475 (make_compound_operation): Likewise.
10476 (extended_count): Likewise.
10477 (change_zero_ext): Likewise.
10478 (simplify_comparison): Likewise.
10479 * dwarf2out.c (scompare_loc_descriptor): Likewise.
10480 (ucompare_loc_descriptor): Likewise.
10481 (minmax_loc_descriptor): Likewise.
10482 (mem_loc_descriptor): Likewise.
10483 (loc_descriptor): Likewise.
10484 * expmed.c (init_expmed_one_mode): Likewise.
10485 * lra-constraints.c (lra_constraint_offset): Likewise.
10486 * optabs.c (prepare_libcall_arg): Likewise.
10487 * postreload.c (move2add_note_store): Likewise.
10488 * reload.c (operands_match_p): Likewise.
10489 * rtl.h (load_extend_op): Likewise.
10490 * rtlhooks.c (gen_lowpart_general): Likewise.
10491 * simplify-rtx.c (simplify_truncation): Likewise.
10492 (simplify_unary_operation_1): Likewise.
10493 (simplify_binary_operation_1): Likewise.
10494 (simplify_const_binary_operation): Likewise.
10495 (simplify_const_relational_operation): Likewise.
10496 (simplify_subreg): Likewise.
10497 * stor-layout.c (bitwise_mode_for_mode): Likewise.
10498 * var-tracking.c (adjust_mems): Likewise.
10499 (prepare_call_arguments): Likewise.
10501 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10502 Alan Hayward <alan.hayward@arm.com>
10503 David Sherwood <david.sherwood@arm.com>
10505 * machmode.h (is_int_mode): New fuction.
10506 * combine.c (find_split_point): Use it.
10507 (combine_simplify_rtx): Likewise.
10508 (simplify_if_then_else): Likewise.
10509 (simplify_set): Likewise.
10510 (simplify_shift_const_1): Likewise.
10511 (simplify_comparison): Likewise.
10512 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
10513 * cse.c (notreg_cost): Likewise.
10514 (cse_insn): Likewise.
10515 * cselib.c (cselib_lookup_1): Likewise.
10516 * dojump.c (do_jump_1): Likewise.
10517 (do_compare_rtx_and_jump): Likewise.
10518 * dse.c (get_call_args): Likewise.
10519 * dwarf2out.c (rtl_for_decl_init): Likewise.
10520 (native_encode_initializer): Likewise.
10521 * expmed.c (emit_store_flag_1): Likewise.
10522 (emit_store_flag): Likewise.
10523 * expr.c (convert_modes): Likewise.
10524 (store_field): Likewise.
10525 (expand_expr_real_1): Likewise.
10526 * fold-const.c (fold_read_from_constant_string): Likewise.
10527 * gimple-ssa-sprintf.c (get_format_string): Likewise.
10528 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
10529 * optabs.c (expand_binop): Likewise.
10530 (expand_unop): Likewise.
10531 (expand_abs_nojump): Likewise.
10532 (expand_one_cmpl_abs_nojump): Likewise.
10533 * simplify-rtx.c (mode_signbit_p): Likewise.
10534 (val_signbit_p): Likewise.
10535 (val_signbit_known_set_p): Likewise.
10536 (val_signbit_known_clear_p): Likewise.
10537 (simplify_relational_operation_1): Likewise.
10538 * tree.c (vector_type_mode): Likewise.
10540 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10541 Alan Hayward <alan.hayward@arm.com>
10542 David Sherwood <david.sherwood@arm.com>
10544 * machmode.h (smallest_mode_for_size): Fix formatting.
10545 (smallest_int_mode_for_size): New function.
10546 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
10547 instead of smallest_mode_for_size.
10548 * combine.c (make_extraction): Likewise.
10549 * config/arc/arc.c (arc_expand_movmem): Likewise.
10550 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
10551 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
10552 * config/s390/s390.c (s390_expand_insv): Likewise.
10553 * config/sparc/sparc.c (assign_int_registers): Likewise.
10554 * config/spu/spu.c (spu_function_value): Likewise.
10555 (spu_function_arg): Likewise.
10556 * coverage.c (get_gcov_type): Likewise.
10557 (get_gcov_unsigned_t): Likewise.
10558 * dse.c (find_shift_sequence): Likewise.
10559 * expmed.c (store_bit_field_1): Likewise.
10560 * expr.c (convert_move): Likewise.
10561 (store_field): Likewise.
10562 * internal-fn.c (expand_arith_overflow): Likewise.
10563 * optabs-query.c (get_best_extraction_insn): Likewise.
10564 * optabs.c (expand_twoval_binop_libfunc): Likewise.
10565 * stor-layout.c (layout_type): Likewise.
10566 (initialize_sizetypes): Likewise.
10567 * targhooks.c (default_get_mask_mode): Likewise.
10568 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
10570 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10571 Alan Hayward <alan.hayward@arm.com>
10572 David Sherwood <david.sherwood@arm.com>
10574 * machmode.h (opt_mode::else_blk): New function.
10575 (int_mode_for_mode): Declare.
10576 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
10577 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
10579 * cfgexpand.c (expand_debug_expr): Likewise.
10580 * combine.c (gen_lowpart_or_truncate): Likewise.
10581 (gen_lowpart_for_combine): Likewise.
10582 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
10583 * config/avr/avr.c (avr_to_int_mode): Likewise.
10584 (avr_out_plus_1): Likewise.
10585 (avr_out_plus): Likewise.
10586 (avr_out_round): Likewise.
10587 * config/i386/i386.c (ix86_split_to_parts): Likewise.
10588 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
10589 (s390_expand_vcond): Likewise.
10590 * config/spu/spu.c (spu_split_immediate): Likewise.
10591 (spu_expand_mov): Likewise.
10592 * dse.c (get_stored_val): Likewise.
10593 * expmed.c (store_bit_field_1): Likewise.
10594 (convert_extracted_bit_field): Use int_mode_for_mode instead of
10596 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
10597 (extract_low_bits): Likewise.
10598 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
10599 handling rather than repeating the check.
10600 (emit_group_store): Likewise.
10601 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
10602 * optabs.c (expand_absneg_bit): Likewise.
10603 (expand_copysign_absneg): Likewise.
10604 (expand_copysign_bit): Likewise.
10605 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
10606 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
10607 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
10608 * var-tracking.c (prepare_call_arguments): Likewise.
10609 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
10610 int_mode_for_mode instead of mode_for_size.
10611 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
10613 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10614 Alan Hayward <alan.hayward@arm.com>
10615 David Sherwood <david.sherwood@arm.com>
10617 * machmode.h (int_mode_for_size): New function.
10618 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
10619 instead of mode_for_size.
10620 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
10622 * combine.c (expand_field_assignment): Use int_mode_for_size
10623 instead of mode_for_size.
10624 (make_extraction): Likewise.
10625 (simplify_shift_const_1): Likewise.
10626 (simplify_comparison): Likewise.
10627 * dojump.c (do_jump): Likewise.
10628 * dwarf2out.c (mem_loc_descriptor): Likewise.
10629 * emit-rtl.c (init_derived_machine_modes): Likewise.
10630 * expmed.c (flip_storage_order): Likewise.
10631 (convert_extracted_bit_field): Likewise.
10632 * expr.c (copy_blkmode_from_reg): Likewise.
10633 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
10634 * internal-fn.c (expand_mul_overflow): Likewise.
10635 * lower-subreg.c (simple_move): Likewise.
10636 * optabs-libfuncs.c (init_optabs): Likewise.
10637 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10638 * tree.c (vector_type_mode): Likewise.
10639 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
10640 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
10641 * tree-vect-generic.c (expand_vector_parallel): Likewise.
10642 * tree-vect-stmts.c (vectorizable_load): Likewise.
10643 (vectorizable_store): Likewise.
10645 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10646 Alan Hayward <alan.hayward@arm.com>
10647 David Sherwood <david.sherwood@arm.com>
10649 * coretypes.h (pod_mode): New type.
10650 (scalar_int_mode_pod): New typedef.
10651 * machmode.h (pod_mode): New class.
10652 (int_n_data_t::m): Change type to scalar_int_mode_pod.
10653 * genmodes.c (emit_mode_int_n): Update accordingly.
10654 * lower-subreg.h (target_lower_subreg): Change type to
10655 scalar_int_mode_pod.
10656 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
10657 scalar_int_mode_pod.
10659 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10660 Alan Hayward <alan.hayward@arm.com>
10661 David Sherwood <david.sherwood@arm.com>
10663 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
10664 machine_mode to scalar_int_mode.
10665 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
10666 (rs6000_option_override_internal): Remove cast to int.
10667 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
10668 machine_mode to scalar_int_mode.
10669 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
10670 (rs6000_option_override_internal): Remove cast to int.
10671 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
10672 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
10674 * config/s390/s390.c (s390_expand_builtin): Likewise.
10675 * coretypes.h (scalar_int_mode): New type.
10676 (opt_scalar_int_mode): New typedef.
10677 * machmode.h (scalar_int_mode): New class.
10678 (scalar_int_mode::includes_p): New function.
10679 (byte_mode): Change type to scalar_int_mode.
10680 (word_mode): Likewise.
10681 (ptr_mode): Likewise.
10682 * emit-rtl.c (byte_mode): Likewise.
10683 (word_mode): Likewise.
10684 (ptr_mode): Likewise.
10685 (init_derived_machine_modes): Update accordingly.
10686 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
10687 and MODE_PARTIAL_INT.
10688 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
10689 opt_scalar_int_mode.
10691 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10692 Alan Hayward <alan.hayward@arm.com>
10693 David Sherwood <david.sherwood@arm.com>
10695 * target.def (libgcc_floating_mode_supported_p): Take a
10697 * doc/tm.texi: Regenerate.
10698 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
10700 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
10701 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
10704 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10705 Alan Hayward <alan.hayward@arm.com>
10706 David Sherwood <david.sherwood@arm.com>
10708 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
10709 * doc/tm.texi: Regenerate.
10710 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
10711 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
10712 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
10713 * targhooks.h (default_floatn_mode): Likewise.
10714 * targhooks.c (default_floatn_mode): Likewise.
10715 * tree.c (build_common_tree_nodes): Update accordingly.
10717 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10718 Alan Hayward <alan.hayward@arm.com>
10719 David Sherwood <david.sherwood@arm.com>
10721 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
10722 (mode_iterator::iterate_p): Likewise.
10723 (mode_iterator::get_wider): Likewise.
10724 * expr.c (init_expr_target): Use opt_scalar_float_mode.
10726 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10727 Alan Hayward <alan.hayward@arm.com>
10728 David Sherwood <david.sherwood@arm.com>
10730 * coretypes.h (opt_scalar_float_mode): New typedef.
10731 * machmode.h (float_mode_for_size): New function.
10732 * emit-rtl.c (double_mode): Delete.
10733 (init_emit_once): Use float_mode_for_size.
10734 * stor-layout.c (layout_type): Likewise.
10735 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
10737 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10738 Alan Hayward <alan.hayward@arm.com>
10739 David Sherwood <david.sherwood@arm.com>
10741 * output.h (assemble_real): Take a scalar_float_mode.
10742 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
10743 * config/arm/arm.md (consttable_4): Likewise.
10744 (consttable_8): Likewise.
10745 (consttable_16): Likewise.
10746 * config/mips/mips.md (consttable_float): Likewise.
10747 * config/s390/s390.c (s390_output_pool_entry): Likewise.
10748 * varasm.c (assemble_real): Take a scalar_float_mode.
10749 (output_constant_pool_2): Update accordingly.
10750 (output_constant): Likewise.
10752 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10753 Alan Hayward <alan.hayward@arm.com>
10754 David Sherwood <david.sherwood@arm.com>
10756 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
10757 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
10758 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
10759 (native_encode_real): Likewise.
10760 (native_interpret_real): Likewise.
10761 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
10762 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10764 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10765 Alan Hayward <alan.hayward@arm.com>
10766 David Sherwood <david.sherwood@arm.com>
10768 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
10769 <scalar_float_mode>. Simplify.
10770 (gen_extend_conv_libfunc): Likewise.
10772 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10773 Alan Hayward <alan.hayward@arm.com>
10774 David Sherwood <david.sherwood@arm.com>
10776 * coretypes.h (scalar_float_mode): New type.
10777 * machmode.h (mode_traits::from_int): Use machine_mode if
10778 USE_ENUM_MODES is defined.
10779 (is_a): New function.
10781 (dyn_cast): Likewise.
10782 (scalar_float_mode): New class.
10783 (scalar_float_mode::includes_p): New function.
10784 (is_float_mode): Likewise.
10785 * gdbhooks.py (MachineModePrinter): New class.
10786 (build_pretty_printer): Use it for scalar_float_mode.
10787 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
10788 (format_helper::format_helper): Turn into a template.
10789 * genmodes.c (get_mode_class): New function.
10790 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
10791 or machine_mode if none.
10792 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
10793 as_a <scalar_float_mode>.
10794 * dwarf2out.c (mem_loc_descriptor): Likewise.
10795 (insert_float): Likewise.
10796 (add_const_value_attribute): Likewise.
10797 * simplify-rtx.c (simplify_immed_subreg): Likewise.
10798 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
10799 (expand_unop): Update accordingly.
10800 (expand_abs_nojump): Likewise.
10801 (expand_copysign_absneg): Take a scalar_float_mode.
10802 (expand_copysign_bit): Likewise.
10803 (expand_copysign): Update accordingly.
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 * coretypes.h (opt_mode): New class.
10810 * machmode.h (opt_mode): Likewise.
10811 (opt_mode::else_void): New function.
10812 (opt_mode::require): Likewise.
10813 (opt_mode::exists): Likewise.
10814 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
10815 (GET_MODE_2XWIDER_MODE): Likewise.
10816 (mode_iterator::get_wider): Update accordingly.
10817 (mode_iterator::get_2xwider): Likewise.
10818 (mode_iterator::get_known_wider): Likewise, turning into a template.
10819 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
10820 forcing a wider mode to exist.
10821 * config/cr16/cr16.h (LONG_REG_P): Likewise.
10822 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
10823 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
10824 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
10825 * lower-subreg.c (init_lower_subreg): Likewise.
10826 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
10827 on the final iteration.
10828 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
10829 a wider mode exists before asking for a move pattern.
10830 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
10831 forcing a wider mode to exist.
10832 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
10833 returning false if no such mode exists.
10834 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
10835 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
10836 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
10837 Avoid checking for a MODE_INT if we already know the mode is not a
10839 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
10840 forcing a wider mode to exist.
10841 (expmed_mult_highpart_optab): Likewise.
10842 (expmed_mult_highpart): Likewise.
10843 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
10845 * lto-streamer-in.c (lto_input_mode_table): Likewise.
10846 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
10847 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
10848 * internal-fn.c (expand_mul_overflow): Update use of
10849 GET_MODE_2XWIDER_MODE.
10850 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10851 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
10852 GET_MODE_WIDER_MODE.
10853 (convert_plusminus_to_widen): Likewise.
10854 * tree-switch-conversion.c (array_value_type): Likewise.
10855 * var-tracking.c (emit_note_insn_var_location): Likewise.
10856 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10857 Return false inside rather than outside the loop if no wider mode
10859 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
10860 and GET_MODE_2XWIDER_MODE
10861 (can_compare_p): Use else_void.
10862 * gdbhooks.py (OptMachineModePrinter): New class.
10863 (build_pretty_printer): Use it for opt_mode.
10865 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10866 Alan Hayward <alan.hayward@arm.com>
10867 David Sherwood <david.sherwood@arm.com>
10869 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
10870 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
10872 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10873 Alan Hayward <alan.hayward@arm.com>
10874 David Sherwood <david.sherwood@arm.com>
10876 * machmode.h (mode_traits): New structure.
10877 (get_narrowest_mode): New function.
10878 (mode_iterator::start): Likewise.
10879 (mode_iterator::iterate_p): Likewise.
10880 (mode_iterator::get_wider): Likewise.
10881 (mode_iterator::get_known_wider): Likewise.
10882 (mode_iterator::get_2xwider): Likewise.
10883 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
10884 (FOR_EACH_MODE): Likewise.
10885 (FOR_EACH_MODE_FROM): Likewise.
10886 (FOR_EACH_MODE_UNTIL): Likewise.
10887 (FOR_EACH_WIDER_MODE): Likewise.
10888 (FOR_EACH_2XWIDER_MODE): Likewise.
10889 * builtins.c (expand_builtin_strlen): Use new mode iterators.
10890 * combine.c (simplify_comparison): Likewise
10891 * config/i386/i386.c (type_natural_mode): Likewise.
10892 * cse.c (cse_insn): Likewise.
10893 * dse.c (find_shift_sequence): Likewise.
10894 * emit-rtl.c (init_derived_machine_modes): Likewise.
10895 (init_emit_once): Likewise.
10896 * explow.c (hard_function_value): Likewise.
10897 * expmed.c (extract_fixed_bit_field_1): Likewise.
10898 (extract_bit_field_1): Likewise.
10899 (expand_divmod): Likewise.
10900 (emit_store_flag_1): Likewise.
10901 * expr.c (init_expr_target): Likewise.
10902 (convert_move): Likewise.
10903 (alignment_for_piecewise_move): Likewise.
10904 (widest_int_mode_for_size): Likewise.
10905 (emit_block_move_via_movmem): Likewise.
10906 (copy_blkmode_to_reg): Likewise.
10907 (set_storage_via_setmem): Likewise.
10908 (compress_float_constant): Likewise.
10909 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10910 * optabs-query.c (get_best_extraction_insn): Likewise.
10911 * optabs.c (expand_binop): Likewise.
10912 (expand_twoval_unop): Likewise.
10913 (expand_twoval_binop): Likewise.
10914 (widen_leading): Likewise.
10915 (widen_bswap): Likewise.
10916 (expand_parity): Likewise.
10917 (expand_unop): Likewise.
10918 (prepare_cmp_insn): Likewise.
10919 (prepare_float_lib_cmp): Likewise.
10920 (expand_float): Likewise.
10921 (expand_fix): Likewise.
10922 (expand_sfix_optab): Likewise.
10923 * postreload.c (move2add_use_add2_insn): Likewise.
10924 * reg-stack.c (reg_to_stack): Likewise.
10925 * reginfo.c (choose_hard_reg_mode): Likewise.
10926 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
10927 * stor-layout.c (mode_for_size): Likewise.
10928 (smallest_mode_for_size): Likewise.
10929 (mode_for_vector): Likewise.
10930 (finish_bitfield_representative): Likewise.
10931 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
10932 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
10933 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10934 * var-tracking.c (prepare_call_arguments): Likewise.
10936 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10937 Alan Hayward <alan.hayward@arm.com>
10938 David Sherwood <david.sherwood@arm.com>
10940 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
10941 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
10942 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
10943 * machmode.h (mode_size): Move earlier in file.
10944 (mode_precision): Likewise.
10945 (mode_inner): Likewise.
10946 (mode_nunits): Likewise.
10947 (mode_unit_size): Likewise.
10948 (unit_unit_precision): Likewise.
10949 (mode_wider): Likewise.
10950 (mode_2xwider): Likewise.
10951 (machine_mode): New class.
10952 (mode_to_bytes): New function.
10953 (mode_to_bits): Likewise.
10954 (mode_to_precision): Likewise.
10955 (mode_to_inner): Likewise.
10956 (mode_to_unit_size): Likewise.
10957 (mode_to_unit_precision): Likewise.
10958 (mode_to_nunits): Likewise.
10959 (GET_MODE_SIZE): Use mode_to_bytes.
10960 (GET_MODE_BITSIZE): Use mode_to_bits.
10961 (GET_MODE_PRECISION): Use mode_to_precision.
10962 (GET_MODE_INNER): Use mode_to_inner.
10963 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
10964 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
10965 (GET_MODE_NUNITS): Use mode_to_nunits.
10966 * system.h (ALWAYS_INLINE): New macro.
10967 * config/powerpcspe/powerpcspe-c.c
10968 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
10969 int for arg1_mode and arg2_mode.
10971 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10972 Alan Hayward <alan.hayward@arm.com>
10973 David Sherwood <david.sherwood@arm.com>
10975 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
10976 Prefix mode names with E_ in case statements.
10977 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10978 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
10979 (aarch64_split_simd_move): Likewise.
10980 (aarch64_gen_storewb_pair): Likewise.
10981 (aarch64_gen_loadwb_pair): Likewise.
10982 (aarch64_gen_store_pair): Likewise.
10983 (aarch64_gen_load_pair): Likewise.
10984 (aarch64_get_condition_code_1): Likewise.
10985 (aarch64_constant_pool_reload_icode): Likewise.
10986 (get_rsqrte_type): Likewise.
10987 (get_rsqrts_type): Likewise.
10988 (get_recpe_type): Likewise.
10989 (get_recps_type): Likewise.
10990 (aarch64_gimplify_va_arg_expr): Likewise.
10991 (aarch64_simd_container_mode): Likewise.
10992 (aarch64_emit_load_exclusive): Likewise.
10993 (aarch64_emit_store_exclusive): Likewise.
10994 (aarch64_expand_compare_and_swap): Likewise.
10995 (aarch64_gen_atomic_cas): Likewise.
10996 (aarch64_emit_bic): Likewise.
10997 (aarch64_emit_atomic_swap): Likewise.
10998 (aarch64_emit_atomic_load_op): Likewise.
10999 (aarch64_evpc_trn): Likewise.
11000 (aarch64_evpc_uzp): Likewise.
11001 (aarch64_evpc_zip): Likewise.
11002 (aarch64_evpc_ext): Likewise.
11003 (aarch64_evpc_rev): Likewise.
11004 (aarch64_evpc_dup): Likewise.
11005 (aarch64_gen_ccmp_first): Likewise.
11006 (aarch64_gen_ccmp_next): Likewise.
11007 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11008 (alpha_emit_xfloating_libcall): Likewise.
11009 (emit_insxl): Likewise.
11010 (alpha_arg_type): Likewise.
11011 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
11012 (arc_preferred_simd_mode): Likewise.
11013 (arc_secondary_reload): Likewise.
11014 (get_arc_condition_code): Likewise.
11015 (arc_print_operand): Likewise.
11016 (arc_legitimate_constant_p): Likewise.
11017 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11018 * config/arc/arc.md (casesi_load): Likewise.
11019 (casesi_compact_jump): Likewise.
11020 * config/arc/predicates.md (proper_comparison_operator): Likewise.
11021 (cc_use_register): Likewise.
11022 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11023 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
11024 (arm_init_iwmmxt_builtins): Likewise.
11025 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
11026 (neon_expand_vector_init): Likewise.
11027 (arm_attr_length_move_neon): Likewise.
11028 (maybe_get_arm_condition_code): Likewise.
11029 (arm_emit_vector_const): Likewise.
11030 (arm_preferred_simd_mode): Likewise.
11031 (arm_output_iwmmxt_tinsr): Likewise.
11032 (thumb1_output_casesi): Likewise.
11033 (thumb2_output_casesi): Likewise.
11034 (arm_emit_load_exclusive): Likewise.
11035 (arm_emit_store_exclusive): Likewise.
11036 (arm_expand_compare_and_swap): Likewise.
11037 (arm_evpc_neon_vuzp): Likewise.
11038 (arm_evpc_neon_vzip): Likewise.
11039 (arm_evpc_neon_vrev): Likewise.
11040 (arm_evpc_neon_vtrn): Likewise.
11041 (arm_evpc_neon_vext): Likewise.
11042 (arm_validize_comparison): Likewise.
11043 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
11044 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
11045 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
11046 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
11047 (c6x_preferred_simd_mode): Likewise.
11048 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
11049 (epiphany_rtx_costs): Likewise.
11050 * config/epiphany/predicates.md (proper_comparison_operator):
11052 * config/frv/frv.c (condexec_memory_operand): Likewise.
11053 (frv_emit_move): Likewise.
11054 (output_move_single): Likewise.
11055 (output_condmove_single): Likewise.
11056 (frv_hard_regno_mode_ok): Likewise.
11057 (frv_matching_accg_mode): Likewise.
11058 * config/h8300/h8300.c (split_adds_subs): Likewise.
11059 (h8300_rtx_costs): Likewise.
11060 (h8300_print_operand): Likewise.
11061 (compute_mov_length): Likewise.
11062 (output_logical_op): Likewise.
11063 (compute_logical_op_length): Likewise.
11064 (compute_logical_op_cc): Likewise.
11065 (h8300_shift_needs_scratch_p): Likewise.
11066 (output_a_shift): Likewise.
11067 (compute_a_shift_length): Likewise.
11068 (compute_a_shift_cc): Likewise.
11069 (expand_a_rotate): Likewise.
11070 (output_a_rotate): Likewise.
11071 * config/i386/i386.c (classify_argument): Likewise.
11072 (function_arg_advance_32): Likewise.
11073 (function_arg_32): Likewise.
11074 (function_arg_64): Likewise.
11075 (function_value_64): Likewise.
11076 (ix86_gimplify_va_arg): Likewise.
11077 (ix86_legitimate_constant_p): Likewise.
11078 (put_condition_code): Likewise.
11079 (split_double_mode): Likewise.
11080 (ix86_avx256_split_vector_move_misalign): Likewise.
11081 (ix86_expand_vector_logical_operator): Likewise.
11082 (ix86_split_idivmod): Likewise.
11083 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
11084 (ix86_build_const_vector): Likewise.
11085 (ix86_build_signbit_mask): Likewise.
11086 (ix86_match_ccmode): Likewise.
11087 (ix86_cc_modes_compatible): Likewise.
11088 (ix86_expand_branch): Likewise.
11089 (ix86_expand_sse_cmp): Likewise.
11090 (ix86_expand_sse_movcc): Likewise.
11091 (ix86_expand_int_sse_cmp): Likewise.
11092 (ix86_expand_vec_perm_vpermi2): Likewise.
11093 (ix86_expand_vec_perm): Likewise.
11094 (ix86_expand_sse_unpack): Likewise.
11095 (ix86_expand_int_addcc): Likewise.
11096 (ix86_split_to_parts): Likewise.
11097 (ix86_vectorize_builtin_gather): Likewise.
11098 (ix86_vectorize_builtin_scatter): Likewise.
11099 (avx_vpermilp_parallel): Likewise.
11100 (inline_memory_move_cost): Likewise.
11101 (ix86_tieable_integer_mode_p): Likewise.
11102 (x86_maybe_negate_const_int): Likewise.
11103 (ix86_expand_vector_init_duplicate): Likewise.
11104 (ix86_expand_vector_init_one_nonzero): Likewise.
11105 (ix86_expand_vector_init_one_var): Likewise.
11106 (ix86_expand_vector_init_concat): Likewise.
11107 (ix86_expand_vector_init_interleave): Likewise.
11108 (ix86_expand_vector_init_general): Likewise.
11109 (ix86_expand_vector_set): Likewise.
11110 (ix86_expand_vector_extract): Likewise.
11111 (emit_reduc_half): Likewise.
11112 (ix86_emit_i387_round): Likewise.
11113 (ix86_mangle_type): Likewise.
11114 (ix86_expand_round_sse4): Likewise.
11115 (expand_vec_perm_blend): Likewise.
11116 (canonicalize_vector_int_perm): Likewise.
11117 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
11118 (expand_vec_perm_1): Likewise.
11119 (expand_vec_perm_interleave3): Likewise.
11120 (expand_vec_perm_even_odd_pack): Likewise.
11121 (expand_vec_perm_even_odd_1): Likewise.
11122 (expand_vec_perm_broadcast_1): Likewise.
11123 (ix86_vectorize_vec_perm_const_ok): Likewise.
11124 (ix86_expand_vecop_qihi): Likewise.
11125 (ix86_expand_mul_widen_hilo): Likewise.
11126 (ix86_expand_sse2_abs): Likewise.
11127 (ix86_expand_pextr): Likewise.
11128 (ix86_expand_pinsr): Likewise.
11129 (ix86_preferred_simd_mode): Likewise.
11130 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
11131 * config/i386/sse.md (*andnot<mode>3): Likewise.
11132 (<mask_codefor><code><mode>3<mask_name>): Likewise.
11133 (*<code><mode>3): Likewise.
11134 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
11135 (ia64_expand_atomic_op): Likewise.
11136 (ia64_arg_type): Likewise.
11137 (ia64_mode_to_int): Likewise.
11138 (ia64_scalar_mode_supported_p): Likewise.
11139 (ia64_vector_mode_supported_p): Likewise.
11140 (expand_vec_perm_broadcast): Likewise.
11141 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
11142 (iq2000_function_arg_advance): Likewise.
11143 (iq2000_function_arg): Likewise.
11144 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
11145 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
11146 (m68k_libcall_value): Likewise.
11147 (m68k_function_value): Likewise.
11148 (sched_attr_op_type): Likewise.
11149 * config/mcore/mcore.c (mcore_output_move): Likewise.
11150 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
11152 (microblaze_function_arg): Likewise.
11153 * config/mips/mips.c (mips16_build_call_stub): Likewise.
11154 (mips_print_operand): Likewise.
11155 (mips_mode_ok_for_mov_fmt_p): Likewise.
11156 (mips_vector_mode_supported_p): Likewise.
11157 (mips_preferred_simd_mode): Likewise.
11158 (mips_expand_vpc_loongson_even_odd): Likewise.
11159 (mips_expand_vec_unpack): Likewise.
11160 (mips_expand_vi_broadcast): Likewise.
11161 (mips_expand_vector_init): Likewise.
11162 (mips_expand_vec_reduc): Likewise.
11163 (mips_expand_msa_cmp): Likewise.
11164 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
11165 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
11166 (cc_flags_for_mode): Likewise.
11167 * config/msp430/msp430.c (msp430_print_operand): Likewise.
11168 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
11169 (nds32_output_casesi_pc_relative): Likewise.
11170 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11171 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
11172 (nvptx_gen_unpack): Likewise.
11173 (nvptx_gen_pack): Likewise.
11174 (nvptx_gen_shuffle): Likewise.
11175 (nvptx_gen_wcast): Likewise.
11176 (nvptx_preferred_simd_mode): Likewise.
11177 * config/pa/pa.c (pa_secondary_reload): Likewise.
11178 * config/pa/predicates.md (base14_operand): Likewise.
11179 * config/powerpcspe/powerpcspe-c.c
11180 (altivec_resolve_overloaded_builtin): Likewise.
11181 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
11183 (rs6000_preferred_simd_mode): Likewise.
11184 (output_vec_const_move): Likewise.
11185 (rs6000_expand_vector_extract): Likewise.
11186 (rs6000_split_vec_extract_var): Likewise.
11187 (reg_offset_addressing_ok_p): Likewise.
11188 (rs6000_legitimate_offset_address_p): Likewise.
11189 (rs6000_legitimize_address): Likewise.
11190 (rs6000_emit_set_const): Likewise.
11191 (rs6000_const_vec): Likewise.
11192 (rs6000_emit_move): Likewise.
11193 (spe_build_register_parallel): Likewise.
11194 (rs6000_darwin64_record_arg_recurse): Likewise.
11195 (swap_selector_for_mode): Likewise.
11196 (spe_init_builtins): Likewise.
11197 (paired_init_builtins): Likewise.
11198 (altivec_init_builtins): Likewise.
11199 (do_load_for_compare): Likewise.
11200 (rs6000_generate_compare): Likewise.
11201 (rs6000_expand_float128_convert): Likewise.
11202 (emit_load_locked): Likewise.
11203 (emit_store_conditional): Likewise.
11204 (rs6000_output_function_epilogue): Likewise.
11205 (rs6000_handle_altivec_attribute): Likewise.
11206 (rs6000_function_value): Likewise.
11207 (emit_fusion_gpr_load): Likewise.
11208 (emit_fusion_p9_load): Likewise.
11209 (emit_fusion_p9_store): Likewise.
11210 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
11211 (fusion_gpr_mem_load): Likewise.
11212 (fusion_addis_mem_combo_load): Likewise.
11213 (fusion_addis_mem_combo_store): Likewise.
11214 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
11215 (fusion_gpr_mem_load): Likewise.
11216 (fusion_addis_mem_combo_load): Likewise.
11217 (fusion_addis_mem_combo_store): Likewise.
11218 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11220 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
11221 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
11222 (rs6000_preferred_simd_mode): Likewise.
11223 (output_vec_const_move): Likewise.
11224 (rs6000_expand_vector_extract): Likewise.
11225 (rs6000_split_vec_extract_var): Likewise.
11226 (reg_offset_addressing_ok_p): Likewise.
11227 (rs6000_legitimate_offset_address_p): Likewise.
11228 (rs6000_legitimize_address): Likewise.
11229 (rs6000_emit_set_const): Likewise.
11230 (rs6000_const_vec): Likewise.
11231 (rs6000_emit_move): Likewise.
11232 (rs6000_darwin64_record_arg_recurse): Likewise.
11233 (swap_selector_for_mode): Likewise.
11234 (paired_init_builtins): Likewise.
11235 (altivec_init_builtins): Likewise.
11236 (rs6000_expand_float128_convert): Likewise.
11237 (emit_load_locked): Likewise.
11238 (emit_store_conditional): Likewise.
11239 (rs6000_output_function_epilogue): Likewise.
11240 (rs6000_handle_altivec_attribute): Likewise.
11241 (rs6000_function_value): Likewise.
11242 (emit_fusion_gpr_load): Likewise.
11243 (emit_fusion_p9_load): Likewise.
11244 (emit_fusion_p9_store): Likewise.
11245 * config/rx/rx.c (rx_gen_move_template): Likewise.
11246 (flags_from_mode): Likewise.
11247 * config/s390/predicates.md (s390_alc_comparison): Likewise.
11248 (s390_slb_comparison): Likewise.
11249 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
11250 (s390_vector_mode_supported_p): Likewise.
11251 (s390_cc_modes_compatible): Likewise.
11252 (s390_match_ccmode_set): Likewise.
11253 (s390_canonicalize_comparison): Likewise.
11254 (s390_emit_compare_and_swap): Likewise.
11255 (s390_branch_condition_mask): Likewise.
11256 (s390_rtx_costs): Likewise.
11257 (s390_secondary_reload): Likewise.
11258 (__SECONDARY_RELOAD_CASE): Likewise.
11259 (s390_expand_cs): Likewise.
11260 (s390_preferred_simd_mode): Likewise.
11261 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
11262 * config/sh/sh.c (sh_print_operand): Likewise.
11263 (dump_table): Likewise.
11264 (sh_secondary_reload): Likewise.
11265 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11266 * config/sh/sh.md (casesi_worker_1): Likewise.
11267 (casesi_worker_2): Likewise.
11268 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
11269 (fcc_comparison_operator): Likewise.
11270 * config/sparc/sparc.c (sparc_expand_move): Likewise.
11271 (emit_soft_tfmode_cvt): Likewise.
11272 (sparc_preferred_simd_mode): Likewise.
11273 (output_cbranch): Likewise.
11274 (sparc_print_operand): Likewise.
11275 (sparc_expand_vec_perm_bmask): Likewise.
11276 (vector_init_bshuffle): Likewise.
11277 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11278 (spu_vector_mode_supported_p): Likewise.
11279 (spu_expand_insv): Likewise.
11280 (spu_emit_branch_or_set): Likewise.
11281 (spu_handle_vector_attribute): Likewise.
11282 (spu_builtin_splats): Likewise.
11283 (spu_builtin_extract): Likewise.
11284 (spu_builtin_promote): Likewise.
11285 (spu_expand_sign_extend): Likewise.
11286 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11287 (tilegx_simd_int): Likewise.
11288 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
11289 (tilepro_simd_int): Likewise.
11290 * config/v850/v850.c (const_double_split): Likewise.
11291 (v850_print_operand): Likewise.
11292 (ep_memory_offset): Likewise.
11293 * config/vax/vax.c (vax_rtx_costs): Likewise.
11294 (vax_output_int_move): Likewise.
11295 (vax_output_int_add): Likewise.
11296 (vax_output_int_subtract): Likewise.
11297 * config/visium/predicates.md (visium_branch_operator): Likewise.
11298 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
11299 (visium_print_operand_address): Likewise.
11300 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11301 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
11302 (xtensa_expand_conditional_branch): Likewise.
11303 (xtensa_copy_incoming_a7): Likewise.
11304 (xtensa_output_literal): Likewise.
11305 * dfp.c (decimal_real_maxval): Likewise.
11306 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11308 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11309 Alan Hayward <alan.hayward@arm.com>
11310 David Sherwood <david.sherwood@arm.com>
11312 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
11313 (mode_nunits_inline): Likewise.
11314 (mode_inner_inline): Likewise.
11315 (mode_unit_size_inline): Likewise.
11316 (mode_unit_precision_inline): Likewise.
11317 (emit_insn_modes_h): Likewise. Also emit a #define of the
11319 (emit_mode_wider): Add an E_ prefix to mode names.
11320 (emit_mode_complex): Likewise.
11321 (emit_mode_inner): Likewise.
11322 (emit_mode_adjustments): Likewise.
11323 (emit_mode_int_n): Likewise.
11324 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
11325 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
11326 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
11327 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
11328 (CRC32_BUILTIN, ENTRY): Likewise.
11329 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
11330 (aarch64_pop_regs): Likewise.
11331 (aarch64_process_components): Likewise.
11332 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
11333 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
11334 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
11335 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
11336 * config/arm/arm.c (arm_init_libfuncs): Likewise.
11337 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
11339 * config/i386/i386-builtin.def (pcmpestr): Likewise.
11340 (pcmpistr): Likewise.
11341 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
11342 * config/mmix/mmix.c (mmix_output_condition): Likewise.
11343 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
11345 * config/rl78/rl78.c (mduc_regs): Likewise.
11346 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
11347 (htm_expand_builtin): Likewise.
11348 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
11349 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
11350 * config/xtensa/xtensa.c (print_operand): Likewise.
11351 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
11352 (NUM_MODE_VECTOR_INT): Likewise.
11353 * genoutput.c (null_operand): Likewise.
11354 (output_operand_data): Likewise.
11355 * genrecog.c (print_parameter_value): Likewise.
11356 * lra.c (debug_operand_data): Likewise.
11358 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11360 * dwarf2out.c (scompare_loc_descriptor_wide)
11361 (scompare_loc_descriptor_narrow): New functions, split out from...
11362 (scompare_loc_descriptor): ...here.
11363 * expmed.c (emit_store_flag_int): New function, split out from...
11364 (emit_store_flag): ...here.
11366 2017-08-30 Richard Biener <rguenther@suse.de>
11368 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
11369 (dwarf2out_early_finish): Move setting of AT_pubnames from
11370 early debug output to early finish.
11372 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11374 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
11375 and -mdata-region to the assembler.
11377 2017-08-30 Richard Biener <rguenther@suse.de>
11379 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
11381 (gen_subprogram_die): Add DW_AT_object_pointer only early.
11382 (dwarf2out_early_global_decl): Only generate a DIE for the
11383 abstract origin if it doesn't already exist or is a declaration DIE.
11384 (resolve_addr): Do not add the linkage name twice when
11385 generating a stub DIE for the DW_TAG_GNU_call_site target.
11387 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11389 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11390 Use machine_mode rather than int for arg1_mode.
11392 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
11395 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
11396 that the second argument of the built-in functions to unpack
11397 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
11398 switch statement instead a lot of if statements.
11399 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
11400 Allow 64-bit values to be in Altivec registers as well as
11401 traditional floating point registers.
11402 (pack<mode>, FMOVE128_VSX iterator): Likewise.
11404 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
11406 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
11407 MAX_REGS_PER_ADDRESS == 1.
11409 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
11411 * config/i386/i386.opt (flag_fentry): Do not init to -1.
11412 * config/i386/i386.c (ix86_option_override_internal): Simplify
11413 setting of opts->x_flag_entry.
11415 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11416 Jakub Jelinek <jakub@redhat.com>
11417 Richard Biener <rguenther@suse.de>
11419 PR tree-optimization/81503
11420 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
11421 folded constant fits in the target type; reorder tests for clarity.
11423 2017-08-29 Martin Liska <mliska@suse.cz>
11425 * passes.def: Include pass_lower_switch.
11426 * stmt.c (dump_case_nodes): Remove and move to
11427 tree-switch-conversion.
11428 (case_values_threshold): Likewise.
11429 (expand_switch_as_decision_tree_p): Likewise.
11430 (emit_case_decision_tree): Likewise.
11431 (expand_case): Likewise.
11432 (balance_case_nodes): Likewise.
11433 (node_has_low_bound): Likewise.
11434 (node_has_high_bound): Likewise.
11435 (node_is_bounded): Likewise.
11436 (emit_case_nodes): Likewise.
11437 (struct simple_case_node): New struct.
11438 (add_case_node): Remove.
11439 (emit_case_dispatch_table): Use vector instead of case_list.
11440 (reset_out_edges_aux): Remove.
11441 (compute_cases_per_edge): Likewise.
11442 (expand_case): Build list of simple_case_node.
11443 (expand_sjlj_dispatch_table): Use it.
11444 * tree-switch-conversion.c (struct case_node): Moved from
11445 stmt.c and adjusted.
11446 (emit_case_nodes): Likewise.
11447 (node_has_low_bound): Likewise.
11448 (node_has_high_bound): Likewise.
11449 (node_is_bounded): Likewise.
11450 (case_values_threshold): Likewise.
11451 (reset_out_edges_aux): Likewise.
11452 (compute_cases_per_edge): Likewise.
11453 (add_case_node): Likewise.
11454 (dump_case_nodes): Likewise.
11455 (balance_case_nodes): Likewise.
11456 (expand_switch_as_decision_tree_p): Likewise.
11457 (emit_jump): Likewise.
11458 (emit_case_decision_tree): Likewise.
11459 (try_switch_expansion): Likewise.
11460 (do_jump_if_equal): Likewise.
11461 (emit_cmp_and_jump_insns): Likewise.
11462 (fix_phi_operands_for_edge): New function.
11463 (record_phi_operand_mapping): Likewise.
11464 (class pass_lower_switch): New pass.
11465 (pass_lower_switch::execute): New function.
11466 (make_pass_lower_switch): Likewise.
11467 (conditional_probability):
11468 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
11469 * tree-pass.h: Add make_pass_lower_switch.
11471 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11474 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
11477 2017-08-29 Richard Biener <rguenther@suse.de>
11479 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
11480 we do not add a DW_AT_inline attribute twice.
11481 (gen_subprogram_die): Remove code setting DW_AT_inline on
11482 DECL_ABSTRACT_P nodes.
11484 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
11486 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
11487 calls to internal functions.
11488 (gimplify_modify_expr): Likewise.
11489 * tree-call-cdce.c (use_internal_fn): Likewise.
11490 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
11491 (convert_to_divmod): Set the nothrow flag.
11492 * tree-if-conv.c (predicate_mem_writes): Likewise.
11493 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
11494 (vectorizable_call): Likewise.
11495 (vectorizable_store): Likewise.
11496 (vectorizable_load): Likewise.
11497 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
11498 (vect_recog_mask_conversion_pattern): Likewise.
11500 2017-08-29 Martin Liska <mliska@suse.cz>
11503 * gcc.c (driver_handle_option): Add new argument.
11504 * opts-common.c (handle_option): Pass
11505 target_option_override_hook.
11506 * opts-global.c (lang_handle_option): Add new option.
11507 (set_default_handlers): Add new argument.
11508 (decode_options): Likewise.
11509 * opts.c (target_handle_option): Likewise.
11510 (common_handle_option): Call target_option_override_hook.
11511 * opts.h (struct cl_option_handler_func): Add hook for
11512 target option override.
11513 (struct cl_option_handlers): Likewise.
11514 (set_default_handlers): Add new argument.
11515 (decode_options): Likewise.
11516 (common_handle_option): Likewise.
11517 (target_handle_option): Likewise.
11518 * toplev.c (toplev::main): Pass targetm.target_option.override
11521 2017-08-29 Richard Biener <rguenther@suse.de>
11522 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
11524 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
11525 life to the active subtree.
11527 2017-08-28 Jeff Law <law@redhat.com>
11529 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
11530 derive_equivalences.
11531 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
11532 Code moved into....
11533 (edge_info::derive_equivalences): New private member function
11535 * tree-ssa-dom.c (class edge_info): Changed from a struct
11536 to a class. Add ctor/dtor, methods and data members.
11537 (edge_info::edge_info): Renamed from allocate_edge_info.
11538 Initialize additional members.
11539 (edge_info::~edge_info): New.
11540 (free_dom_edge_info): Delete the edge info.
11541 (record_edge_info): Use new class & associated member functions.
11542 Tighten forms for testing for edge equivalences.
11543 (record_temporary_equivalences): Iterate over the simple
11544 equivalences rather than assuming there's only one per edge.
11545 (cprop_into_successor_phis): Iterate over the simple
11546 equivalences rather than assuming there's only one per edge.
11547 (optimize_stmt): Use operand_equal_p rather than pointer
11548 equality for mini-DSE code.
11550 2017-08-28 Nathan Sidwell <nathan@acm.org>
11552 * gcc.c (execute): Fold SIGPIPE handling into switch
11553 statement. Adjust internal error message.
11555 2017-08-28 Richard Biener <rguenther@suse.de>
11558 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
11559 Do nothing for removed DIEs.
11561 2017-08-28 Richard Biener <rguenther@suse.de>
11563 PR tree-optimization/81977
11564 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
11567 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
11570 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
11571 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
11572 using targetm.gen_mem_thread_fence.
11574 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
11577 * config/i386/i386.md (*<btsc><mode>): Change operand 2
11578 predicate to register_operand. Reorder operands.
11579 (*btr<mode>): Ditto.
11580 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
11581 (*btr<mode>_mask): Ditto.
11583 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
11585 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
11586 * config/rs6000/xmmintrin.h: New file.
11587 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
11589 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11592 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
11593 parameter and_insn and return it.
11594 (recombine_lvx_pattern): Insert a copy to ensure availability of
11595 the base register of the copied masking operation at the point of
11596 the instruction replacement.
11597 (recombine_stvx_pattern): Likewise.
11599 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
11601 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
11602 undocumented switches.
11603 (-mpower9-dform-vector): Likewise.
11604 (-mpower9-dform): Likewise.
11605 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
11606 comments to delete references to -mpower9-dform* switches.
11607 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
11608 Delete reference to -mpower9-dform* switches, test for
11609 -mpower9-vector instead.
11610 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
11611 (OTHER_P9_VECTOR_MASKS): Likewise.
11612 (POWERPC_MASKS): Likewise.
11613 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
11614 tests against -mpower9-dform* to -mpower9-vector. Delete code
11615 that checked for -mpower9-dform* consistancy with other options.
11616 Add test for -mpower9-misc to enable other power9 switches.
11617 (rs6000_init_hard_regno_mode_ok): Likewise.
11618 (rs6000_option_override_internal): Likewise.
11619 (rs6000_emit_prologue): Likewise.
11620 (rs6000_emit_epilogue): Likewise.
11621 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
11622 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
11623 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
11625 (emit_fusion_p9_store): Likewise.
11626 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
11627 resetting these macros if the assembler does not support ISA 3.0
11629 (TARGET_P9_DFORM_VECTOR): Likewise.
11630 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
11631 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
11633 2017-08-25 Alan Modra <amodra@gmail.com>
11635 PR rtl-optimization/81747
11636 * cse.c (cse_extended_basic_block): Don't attempt to record
11637 equivalences for degenerate conditional branches that branch
11638 to their fall-through.
11640 2017-08-24 Martin Sebor <msebor@redhat.com>
11642 PR middle-end/81908
11643 * gimple-fold.c (size_must_be_zero_p): New function.
11644 (gimple_fold_builtin_memory_op): Call it.
11646 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
11648 * config/rs6000/mm_malloc.h: New file.
11650 2017-08-24 Bin Cheng <bin.cheng@arm.com>
11652 PR tree-optimization/81913
11653 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
11654 analysis when either IVs in condition can wrap.
11656 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
11658 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
11659 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
11661 2017-08-24 Richard Biener <rguenther@suse.de>
11664 * targhooks.c (default_target_can_inline_p): Properly
11665 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
11666 is present and always compare.
11667 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
11668 infer -mfpmath=sse from TARGET_SSE_P.
11669 (ix86_can_inline_p): Properly use target_option_default_node when
11670 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
11672 2017-08-24 Richard Biener <rguenther@suse.de>
11675 * dwarf2out.c (output_die): Handle flag_generate_offload like
11677 (output_comp_unit): Likewise.
11678 (gen_array_type_die): Likewise.
11679 (dwarf2out_early_finish): Likewise.
11680 (note_variable_value_in_expr): Likewise.
11681 (dwarf2out_finish): Likewise. Adjust assert.
11682 * cgraphunit.c (symbol_table::compile): Move setting of
11683 flag_generate_offload earlier ...
11684 (symbol_table::finalize_compilation_unit): ... here, before
11685 early debug finalization.
11687 2017-08-24 Richard Biener <rguenther@suse.de>
11689 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
11690 and ipa-fnsummary.h.
11691 (ix86_can_inline_p): When ix86_fpmath flags do not match
11692 check whether the callee uses FP math at all.
11694 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
11696 PR middle-end/81931
11697 * tree-ssanames.c (get_nonzero_bits): Use element_precision
11698 instead of TYPE_PRECISION.
11700 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
11701 Alan Hayward <alan.hayward@arm.com>
11702 David Sherwood <david.sherwood@arm.com>
11704 * combine.c (make_extraction): Use subreg_offset_from_lsb.
11706 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
11708 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
11710 (ix86_frame::stack_realign_allocate): New field.
11711 (struct machine_frame_state): Modify comments.
11712 (machine_frame_state::sp_realigned_fp_end): New field.
11713 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
11714 layout calculation.
11715 (sp_valid_at): Add assertion to assure no attempt to access invalid
11716 offset of a realigned stack.
11717 (fp_valid_at): Likewise.
11718 (choose_baseaddr): Modify comments.
11719 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
11720 ix86_expand_prologue.
11721 (ix86_expand_prologue): Modify stack realignment and allocation.
11722 (ix86_expand_epilogue): Modify comments.
11723 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
11724 avx2_runtime, avx512f, and avx512f_runtime.
11726 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
11728 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
11729 (mstackrealign): Do not init to -1.
11730 * config/i386/i386.c (ix86_option_override_internal):
11731 Check opts_set, not opts when setting default value of
11732 opts->x_ix86_force_align_arg_pointer.
11734 2017-08-23 Richard Biener <rguenther@suse.de>
11736 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
11737 lang_hooks.decl_printable_name.
11738 * print-rtl-function.c (print_rtx_function): Likewise.
11739 * tree-pretty-print.c (dump_function_header): Likewise.
11741 2017-08-23 Richard Biener <rguenther@suse.de>
11744 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
11745 -g0 at compile-time.
11747 2017-08-23 Tamar Christina <tamar.christina@arm.com>
11749 PR middle-end/19706
11750 * doc/sourcebuild.texi (Other hardware attributes):
11753 2017-08-23 Tamar Christina <tamar.christina@arm.com>
11755 PR middle-end/19706
11756 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
11757 Fix single-use check.
11759 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
11761 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
11763 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
11765 * doc/install.texi: Modify to add more details on running selected
11768 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
11770 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
11771 is combined with -mabi=ms.
11772 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
11775 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11777 PR tree-optimization/81488
11778 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
11779 and cached_basis fields.
11780 (MAX_SPREAD): New constant.
11781 (alloc_cand_and_find_basis): Initialize new fields.
11782 (clear_visited): New function.
11783 (create_phi_basis_1): Rename from create_phi_basis, set visited
11784 and cached_basis fields.
11785 (create_phi_basis): New wrapper function.
11786 (phi_add_costs_1): Rename from phi_add_costs, add spread
11787 parameter, set visited field, short-circuit when limits reached.
11788 (phi_add_costs): New wrapper function.
11789 (record_phi_increments_1): Rename from record_phi_increments, set
11791 (record_phi_increments): New wrapper function.
11792 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
11793 (phi_incr_cost): New wrapper function.
11794 (all_phi_incrs_profitable_1): Rename from
11795 all_phi_incrs_profitable, set visited field.
11796 (all_phi_incrs_profitable): New wrapper function.
11798 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
11799 Alan Hayward <alan.hayward@arm.com>
11800 David Sherwood <david.sherwood@arm.com>
11802 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
11803 that takes the outer and inner modes.
11804 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
11805 comparison as the canonical test for a paradoxical subreg.
11806 * combine.c (simplify_set): Use paradoxical_subreg_p.
11807 (make_extraction): Likewise.
11808 (force_to_mode): Likewise.
11809 (rtx_equal_for_field_assignment_p): Likewise.
11810 (gen_lowpart_for_combine): Likewise.
11811 (simplify_comparison): Likewise.
11812 * cse.c (equiv_constant): Likewise.
11813 * expmed.c (store_bit_field_1): Likewise.
11814 * final.c (alter_subreg): Likewise.
11815 * fwprop.c (propagate_rtx): Likewise.
11816 (forward_propagate_subreg): Likewise.
11817 * ira-conflicts.c (ira_build_conflicts): Likewise.
11818 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
11819 * lra-constraints.c (curr_insn_transform): Likewise.
11820 (split_reg): Likewise.
11821 * lra-eliminations.c (move_plus_up): Likewise.
11822 (lra_eliminate_regs_1): Likewise.
11823 * recog.c (general_operand): Likewise.
11824 * ree.c (combine_reaching_defs): Likewise.
11825 * reload.c (push_reload): Likewise.
11826 (find_reloads): Likewise.
11827 * reload1.c (elimination_effects): Likewise.
11828 (compute_reload_subreg_offset): Likewise.
11829 (choose_reload_regs): Likewise.
11830 * rtlanal.c (subreg_lsb_1): Likewise.
11831 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11832 (simplify_subreg): Likewise.
11833 * var-tracking.c (track_loc_p): Likewise.
11834 * emit-rtl.c (byte_lowpart_offset): Likewise.
11835 (paradoxical_subreg_p): Delete out-of-line definition.
11837 2017-08-22 Jeff Law <law@redhat.com>
11839 PR tree-optimization/81741
11840 PR tree-optimization/71947
11841 * tree-ssa-dom.c: Include tree-inline.h.
11842 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
11843 equivalences if one is more expensive to compute than the other.
11844 * tree-ssa-scopedtables.h (class const_or_copies): Make
11845 record_const_or_copy_raw method private.
11846 (class avail_exprs_stack): New method simplify_binary_operation.
11847 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
11848 avail_exprs_stack::simplify_binary_operation as needed.
11849 (avail_exprs_stack::simplify_binary_operation): New function.
11851 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
11853 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
11854 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
11855 (DOT_SYMBOLS): Likewise.
11856 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
11857 (RELOCATABLE_NEEDS_FIXUP): Likewise.
11858 (RS6000_ABI_NAME): Likewise.
11859 (TARGET_CMODEL): Likewise.
11860 (TOC_SECTION_ASM_OP): Likewise.
11861 (SET_CMODEL): New macro.
11862 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
11864 2017-08-22 Richard Biener <rguenther@suse.de>
11866 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
11867 to allow for free-lang-data replacements similar to verify_type_variant.
11869 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
11870 Alan Hayward <alan.hayward@arm.com>
11871 David Sherwood <david.sherwood@arm.com>
11873 * config/aarch64/aarch64.md (casesi): Use DImode rather than
11874 VOIDmode for the LABEL_REF.
11876 2017-08-22 Richard Biener <rguenther@suse.de>
11878 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
11880 2017-08-22 Richard Biener <rguenther@suse.de>
11882 * common.opt (feliminate-dwarf2-dups): Ignore.
11883 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
11884 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
11885 same_die_p_wrap, compute_section_prefix,
11886 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
11887 (comdat_symbol_id, comdat_symbol_number): Likewise.
11888 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
11890 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
11891 (output_die): Mark unreachable path unreachable.
11892 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
11893 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
11894 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
11895 (dwarf2out_early_finish): Likewise.
11897 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
11899 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
11901 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
11904 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
11905 not VAR_P. Filter attribute warnings with OPT_Wattributes.
11906 (avr_attribute_table) <io, io_low, address>: Initialize
11907 .decl_required with true.
11909 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
11911 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
11912 undocumented debugging options.
11913 (-mvsx-scalar-double): Likewise.
11914 (-mallow-df-permute): Likewise.
11915 (-mvectorize-builtins): Likewise.
11916 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
11917 (rs6000_builtin_vectorized_function): Likewise.
11918 (rs6000_builtin_md_vectorized_function): Likewise.
11919 (rs6000_opt_vars): Likewise.
11921 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
11924 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
11925 (*btrq_imm): Rename from *btrq.
11926 (*btcq_imm): Rename from *btcq.
11927 (btsc): New code attribute.
11928 (*<btsc><mode>): New insn pattern.
11929 (*btr<mode>): Ditto.
11930 (*<btsc><mode>_mask): New insn_and_split pattern.
11931 (*btr<mode>_mask): Ditto.
11933 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11934 Alan Hayward <alan.hayward@arm.com>
11935 David Sherwood <david.sherwood@arm.com>
11937 * function.c (pad_below): Simplify padding calculation.
11939 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11940 Alan Hayward <alan.hayward@arm.com>
11941 David Sherwood <david.sherwood@arm.com>
11943 * target.def (function_prologue): Remove frame size argument.
11944 (function_epilogue): Likewise.
11945 * doc/tm.texi: Regenerate.
11946 * final.c (final_start_function): Update call to function_prologue.
11947 (final_end_function): Update call to function_epilogue.
11948 (default_function_pro_epilogue): Remove frame size argument.
11949 * output.h (default_function_pro_epilogue): Likewise.
11950 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
11951 (arm_output_function_prologue): Likewise.
11952 * config/frv/frv.c (frv_function_prologue): Likewise.
11953 (frv_function_epilogue): Likewise.
11954 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
11955 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
11956 (ia64_output_function_epilogue): Likewise.
11957 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
11958 (m32r_output_function_epilogue): Likewise.
11959 * config/microblaze/microblaze.c (microblaze_function_prologue)
11960 (microblaze_function_epilogue): Likewise.
11961 * config/mips/mips.c (mips_output_function_prologue): Likewise.
11962 (mips_output_function_epilogue): Likewise.
11963 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
11964 (mmix_target_asm_function_epilogue): Likewise.
11965 * config/msp430/msp430.c (msp430_start_function): Likewise.
11966 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
11967 (nds32_asm_function_epilogue): Likewise.
11968 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
11969 * config/pa/pa.c (pa_output_function_prologue): Likewise.
11970 (pa_output_function_epilogue): Likewise.
11971 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
11972 (rs6000_output_function_epilogue): Likewise.
11973 * config/rl78/rl78.c (rl78_start_function): Likewise.
11974 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
11975 (rs6000_output_function_epilogue): Likewise.
11976 * config/rx/rx.c (rx_output_function_prologue): Likewise.
11977 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
11978 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
11979 (sparc_asm_function_epilogue): Likewise.
11981 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11983 * tree.h (type_has_mode_precision_p): New function.
11984 * convert.c (convert_to_integer_1): Use it.
11985 * expr.c (expand_expr_real_2): Likewise.
11986 (expand_expr_real_1): Likewise.
11987 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
11988 * match.pd: Likewise.
11989 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
11990 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
11991 * tree-tailcall.c (process_assignment): Likewise.
11992 * tree-vect-loop.c (vectorizable_reduction): Likewise.
11993 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
11994 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
11995 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11996 (vectorizable_assignment): Likewise.
11997 (vectorizable_shift): Likewise.
11998 (vectorizable_operation): Likewise.
11999 * tree-vrp.c (register_edge_assert_for_2): Likewise.
12001 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
12003 * match.pd: Add pow (C, x) simplification.
12005 2017-08-21 Richard Biener <rguenther@suse.de>
12007 PR tree-optimization/81900
12008 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
12009 for blocks with abnormal predecessors.
12010 (compute_antic): Do not set visited flag prematurely.
12012 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
12015 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
12017 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12019 * stor-layout.h (vector_type_mode): Move to...
12020 * tree.h (vector_type_mode): ...here.
12021 * stor-layout.c (vector_type_mode): Move to...
12022 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
12024 2017-08-21 Richard Biener <rguenther@suse.de>
12026 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
12027 register_external_die hooks.
12028 (debug_false_tree_charstarstar_uhwistar): Declare.
12029 (debug_nothing_tree_charstar_uhwi): Likewise.
12030 * debug.c (do_nothing_debug_hooks): Adjust.
12031 (debug_false_tree_charstarstar_uhwistar): New do nothing.
12032 (debug_nothing_tree_charstar_uhwi): Likewise.
12033 * dbxout.c (dbx_debug_hooks): Adjust.
12034 (xcoff_debug_hooks): Likewise.
12035 * sdbout.c (sdb_debug_hooks): Likewise.
12036 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
12037 * dwarf2out.c (macinfo_label_base): New global.
12038 (dwarf2out_register_external_die): New function for the
12039 register_external_die hook.
12040 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
12041 (dwarf2_debug_hooks): Use them.
12042 (dwarf2_lineno_debug_hooks): Adjust.
12043 (struct die_struct): Add with_offset flag.
12044 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
12045 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
12046 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
12047 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
12048 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
12049 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
12050 defining section names for the early LTO debug variants.
12051 (reset_indirect_string): New helper.
12052 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
12053 (print_dw_val): Add support for offsetted symbol references.
12054 (get_ultimate_context): Split out from is_cxx.
12055 (is_cxx): Use get_ultimate_context.
12056 (is_fortran): Add decl overload.
12057 (compute_comp_unit_symbol): Split out worker from
12058 compute_section_prefix.
12059 (compute_section_prefix): Call compute_comp_unit_symbol and
12060 set comdat_type_p here.
12061 (output_die): Skip DIE symbol output for the LTO added one.
12062 Handle DIE symbol references with offset.
12063 (output_comp_unit): Guard section name mangling properly.
12064 For LTO debug sections emit a symbol at the section beginning
12065 which we use to refer to its DIEs.
12066 (add_abstract_origin_attribute): For DIEs registered via
12067 dwarf2out_register_external_die directly refer to the early
12068 DIE rather than indirectly through the shadow one we created.
12069 Remove obsolete call to dwarf2out_abstract_function for
12070 non-function/block origins.
12071 (gen_array_type_die): When generating early LTO debug do
12072 not emit DW_AT_string_length.
12073 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
12074 late when in LTO. As suggested place a gcc_unreachable for
12075 the DECL_ABSTRACT_P case.
12076 (gen_subprogram_die): Avoid another specification DIE
12077 for early built declarations/definitions for the late LTO case.
12078 (gen_variable_die): Add type references for late duplicated VLA dies
12080 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
12081 we have the abstract instance already.
12082 (process_scope_var): Adjust decl DIE contexts in LTO which
12083 first puts them in limbo.
12084 (gen_decl_die): Do not generate type DIEs late apart from
12085 types for VLAs or for decls we do not yet have a DIE. Do not
12086 call dwarf2out_abstract_function late.
12087 (dwarf2out_early_global_decl): Make sure to create DIEs
12088 for abstract instances of a decl first.
12089 (dwarf2out_late_global_decl): Adjust comment.
12090 (output_macinfo_op): With multiple macro sections use
12091 macinfo_label_base to distinguish labels.
12092 (output_macinfo): Likewise. Update macinfo_label_base.
12093 Pass in the line info label.
12094 (note_variable_value_in_expr): When generating LTO resolve
12095 all variable values here by generating DIEs as needed.
12096 (init_sections_and_labels): Add early LTO debug flag parameter
12097 and generate different sections and names if set. Add generation
12098 counter for the labels so we can have multiple of them.
12099 (reset_dies): Helper to allow DIEs to be output multiple times.
12100 (dwarf2out_finish): When outputting DIEs to the fat part of an
12101 LTO object first reset DIEs.
12102 (dwarf2out_early_finish): Output early DIEs when generating LTO.
12103 (modified_type_die): Check for decl_ultimate_origin being self
12105 (gen_type_die_with_usage): Likewise.
12106 (gen_typedef_die): Allow decl_ultimate_origin being self.
12107 (set_decl_abstract_flags): Remove.
12108 (set_block_abstract_flags): Likewise.
12109 (dwarf2out_abstract_function): Treat the early generated DIEs
12110 as the abstract copy and only add DW_AT_inline and
12111 DW_AT_artificial here and call set_decl_origin_self.
12112 If the DIE has an abstract origin don't do anything.
12113 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
12114 if we have none yet (Go fails to build one, PR78628).
12115 (variably_modified_type_p): Prevent endless recursion for Ada
12116 cyclic pointer types.
12117 * lto-streamer-in.c: Include debug.h.
12118 (dref_queue): New global.
12119 (lto_read_tree_1): Stream in DIE references.
12120 (lto_input_tree): Register DIE references.
12121 (input_function): Stream DECL_DEBUG_ARGS.
12122 * lto-streamer-out.c: Include debug.h.
12123 (lto_write_tree_1): Output DIE references.
12124 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
12125 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
12126 (output_function): Stream DECL_DEBUG_ARGS.
12127 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
12128 Stream DECL_ABSTRACT_ORIGIN.
12129 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
12130 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
12131 DECL_CONTEXT for file-scope decls.
12132 * lto-streamer.h (struct dref_entry): Declare.
12133 (dref_queue): Likewise.
12134 * cfgexpand.c (pass_expand::execute): Do not call the
12135 outlining_inline_function hook here.
12136 * lto-wrapper.c (debug_obj): New global.
12137 (tool_cleanup): Unlink it if required.
12138 (debug_objcopy): New function.
12139 (run_gcc): Handle early debug sections in the IL files by
12140 extracting them to separate files, partially linkin them and
12141 feeding the result back as result to the linker.
12142 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
12143 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
12144 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
12145 sections into a separate segment.
12146 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
12148 (darwin_asm_dwarf_section): Likewise.
12149 (darwin_asm_output_dwarf_offset): Likewise.
12150 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
12152 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12153 Alan Hayward <alan.hayward@arm.com>
12154 David Sherwood <david.sherwood@arm.com>
12156 * read-md.h (md_reader::record_potential_iterator_use): Replace
12157 pointer argument with an rtx and an index.
12158 * read-rtl.c (iterator_group::apply_iterator): Likewise.
12159 (apply_mode_iterator): Likewise.
12160 (apply_code_iterator): Likewise.
12161 (apply_int_iterator): Likewise.
12162 (apply_subst_iterator): Likewise.
12163 (record_iterator_use): Likewise.
12164 (record_attribute_use): Likewise.
12165 (md_reader::record_potential_iterator_use): Likewise. Update calls
12166 to record_iterator_use and apply_iterator.
12167 (iterator_use): Replace ptr with x and index.
12168 (attribute_use): Likewise.
12169 (apply_attribute_uses): Update calls to apply_iterator.
12170 (apply_iterators): Likewise. Update initialization of iterator_use.
12171 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
12172 and record_potential_iterator_use.
12173 (rtx_reader::read_rtx_operand): Likewise.
12175 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12176 Alan Hayward <alan.hayward@arm.com>
12177 David Sherwood <david.sherwood@arm.com>
12179 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
12182 2017-08-21 Richard Biener <rguenther@suse.de>
12184 PR middle-end/81884
12185 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
12186 at struct end conservatively when comparing common bases.
12188 2017-08-21 Richard Biener <rguenther@suse.de>
12190 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
12191 (mem_ref_in_stmt): Remove.
12192 (determine_max_movement): Use ref index to get at the reference.
12193 (invariantness_dom_walker::before_dom_children): Deal with
12194 lim data already initialized.
12195 (gather_mem_refs_stmt): Initialize lim data and record ref index.
12197 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12199 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
12200 (TARGET_ISA_ROUND): Ditto.
12201 (TARGET_ROUND): Ditto.
12202 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
12203 * config/i386/i386.md: Ditto.
12204 * config/i386/sse.md: Ditto.
12205 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
12206 with OPTION_MASK_ISA_SSE4_1.
12208 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12211 * doc/extend.texi (x86 Built-in Functions): Correct the name of
12212 __builtin_ia32_lzcnt_u16.
12214 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
12217 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
12218 (rs6000_set_current_function): Rewrite function to use it.
12220 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
12223 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
12224 and TYPE_WARN_IF_NOT_ALIGN.
12225 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
12226 (handle_warn_if_not_align): New.
12227 (place_union_field): Call handle_warn_if_not_align.
12228 (place_field): Call handle_warn_if_not_align.
12229 Copy TYPE_WARN_IF_NOT_ALIGN.
12230 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
12231 (layout_type): Likewise.
12232 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
12234 (tree_decl_common): Add warn_if_not_align.
12235 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
12236 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
12237 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
12238 (DECL_WARN_IF_NOT_ALIGN): Likewise.
12239 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
12240 * doc/extend.texi: Document warn_if_not_aligned attribute.
12241 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
12243 2017-08-17 Martin Liska <mliska@suse.cz>
12246 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
12247 (get_data_dependence): Use it as pointer type.
12248 (distribute_loop): Likewise.
12250 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12252 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
12253 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
12254 (p8_vmrgow_<mode>_direct): New define_insn.
12255 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
12256 handle endianness for vmrgew and vmrgow permute patterns.
12258 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
12260 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
12261 * config/rs6000/rs6000-cpus.def: Remove comment.
12262 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
12263 (POWERPC_MASKS): Likewise.
12264 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
12265 use of TARGET_VSX_TIMODE.
12266 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
12267 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
12268 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
12269 (rs6000_option_override_internal): Remove dead code.
12270 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
12271 (rs6000_legitimize_reload_address): Likewise.
12272 (rs6000_legitimate_address_p): Likewise.
12273 (rs6000_opt_masks): Delete "vsx-timode".
12274 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
12275 from function comment.
12276 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
12277 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
12278 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
12280 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
12281 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
12282 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
12283 (VSX_TI): Likewise.
12285 (define_peephole2): Likewise.
12287 2017-08-17 Martin Sebor <msebor@redhat.com>
12290 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
12291 past the end of an array.
12292 (test_pp_format): Add test cases.
12294 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
12296 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
12297 missing ECF_NOTHROW flags.
12299 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
12302 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
12303 operands residing in integer registers.
12304 (*vsx_le_perm_load_<mode>): Likewise.
12305 (*vsx_le_perm_store_<mode>): Likewise.
12306 (define_peephole2): Add peepholes to optimize the above.
12308 2017-08-17 Marek Polacek <polacek@redhat.com>
12310 PR middle-end/81814
12311 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
12312 to mimic what shorten_compare did. Change the return type to bool.
12313 (fold_cond_expr_with_comparison): Update call to
12314 operand_equal_for_comparison_p.
12315 (fold_ternary_loc): Likewise.
12317 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
12319 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
12321 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
12323 2017-08-17 Richard Biener <rguenther@suse.de>
12325 * tree-ssa-structalias.c (solve_graph): When propagating
12326 to successors update the graphs succ edges and avoid duplicate work.
12328 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
12331 * config/i386/i386.c (ix86_option_override_internal): Save target
12332 specific options after ix86_stack_protector_guard_reg was changed.
12334 2017-08-17 Richard Biener <rguenther@suse.de>
12336 PR tree-optimization/81827
12337 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
12338 (new_var_info): Initialize it conservatively.
12339 (get_call_vi): Mark register vars.
12340 (new_scalar_tmp_constraint_exp): Likewise.
12341 (handle_rhs_call): Likewise.
12342 (handle_const_call): Likewise.
12343 (create_function_info_for): Likewise.
12344 (solve_constraints): Sort varinfos to separate register from
12345 non-register vars to pack points-to solution bitmaps during
12348 2017-08-17 Marek Polacek <polacek@redhat.com>
12350 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
12352 2017-08-17 Richard Biener <rguenther@suse.de>
12354 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
12355 to true when overflow is undefined and we saturated the result.
12357 2017-08-17 Alan Modra <amodra@gmail.com>
12360 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
12361 Don't use store multiple if only one reg needs saving.
12362 (interesting_frame_related_regno): New function.
12363 (rs6000_frame_related): Don't emit frame info for regs that
12365 (rs6000_emit_epilogue): Likewise.
12367 2017-08-16 Nathan Sidwell <nathan@acm.org>
12369 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
12370 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
12371 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
12372 * tree.c (free_lang_data_in_type): Use else-if chain. Always
12373 clear TYPE_LANG_1. Remove obsolete member-function stripping.
12374 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
12375 (verify_type): Adjust for TYPE_BINFO move.
12376 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
12377 process TYPE_BINFO directly.
12378 (hash_tree): Likewise.
12379 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
12381 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
12384 2017-08-16 David Malcolm <dmalcolm@redhat.com>
12386 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
12388 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
12391 * config/i386/i386.md (*anddi_1_btr): Change predicates of
12392 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
12393 Add ix86_binary_operator_ok to insn constraint.
12394 (*iordi_1_bts): Ditto.
12395 (*xordi_1_btc): Ditto.
12396 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
12397 Update corresponding peephole2 pattern.
12401 2017-08-16 Bin Cheng <bin.cheng@arm.com>
12403 PR tree-optimization/81832
12404 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
12405 copy loop header which has IFN_LOOP_DIST_ALIAS call.
12407 2017-08-16 Marek Polacek <polacek@redhat.com>
12410 * fold-const.c (fold_indirect_ref_1): Restore original behavior
12411 regarding size_zero_node.
12413 2017-08-16 Martin Liska <mliska@suse.cz>
12416 * config.gcc: Respect previously set extra_objs in case
12419 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
12421 PR tree-optimization/81835
12422 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
12423 the phi SSA_NAME. Check that the condition in a COND_EXPR does
12424 not depend on the phi.
12426 2017-08-16 Alan Modra <amodra@gmail.com>
12428 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
12431 2017-08-16 Alan Modra <amodra@gmail.com>
12433 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
12434 (save_reg_p): ..into this. Update all callers.
12435 (first_reg_to_save): Simplify.
12437 2017-08-16 Alan Modra <amodra@gmail.com>
12439 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
12442 2017-08-15 Joseph Myers <joseph@codesourcery.com>
12446 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
12447 constant count if that count is less than 32.
12449 2017-08-15 Nathan Sidwell <nathan@acm.org>
12451 * gcc.c (execute): Emit friendlier message if inferior is killed
12452 by an external cause.
12454 2017-08-15 Richard Biener <rguenther@suse.de>
12456 PR tree-optimization/81790
12457 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
12458 CONSTRUCTORs from simplifying and VN.
12460 2017-08-14 Martin Sebor <msebor@redhat.com>
12462 * builtin-attrs.def: Add comments.
12464 2017-08-14 Martin Sebor <msebor@redhat.com>
12467 * doc/extend.texi (attribute nonstring): Document new attribute.
12469 2017-08-14 Martin Sebor <msebor@redhat.com>
12472 * tree-diagnostic.c (default_tree_printer): Handle %G.
12473 * gimple-pretty-print.h (percent_G_format): Declare new function.
12474 * gimple-pretty-print.c (percent_G_format): Define.
12475 * tree-pretty-print.c (percent_K_format): Add argument.
12477 2017-08-14 Martin Sebor <msebor@redhat.com>
12479 PR translation/79998
12480 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
12481 Remove a stray space.
12483 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
12486 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
12487 (*iordi_1_bts): Ditto.
12488 (*xordi_1_btc): Ditto.
12490 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12493 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
12494 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12496 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
12497 quoted strings, and make more translator-friendly.
12498 (darwin_rs6000_override_options): Likewise.
12499 (rs6000_option_override_internal): Likewise.
12500 (rs6000_return_in_memory): Fix overlong line.
12501 (init_cmulative_args): Use quoted strings, and make more
12502 translator-friendly.
12503 (rs6000_pass_by_reference): Fix overlong line.
12504 (def_builtin): Use quoted strings.
12505 (altivec_expand_predicate_builtin): Use quoted strings, and make
12506 more translator-friendly.
12507 (htm_expand_builtin): Use quoted strings.
12508 (cpu_expand_builtin): Use quoted strings, and make more
12509 translator-friendly.
12510 (altivec_expand_builtin): Likewise.
12511 (paired_expand_predicate_builtin): Likewise.
12512 (rs6000_invalid_builtin): Likewise.
12513 (builtin_function_type): Use quoted strings.
12514 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
12515 more translator-friendly.
12516 (rs6000_trampoline_init): Likewise.
12517 (rs6000_handle_altivec_attribute): Likewise.
12518 (rs6000_inner_target_options): Use quoted strings.
12519 (rs6000_disable_incompatible_switches): Likewise.
12520 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
12521 strings, and make more translator-friendly.
12522 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
12524 2017-08-14 Bin Cheng <bin.cheng@arm.com>
12526 PR tree-optimization/81799
12527 * tree-loop-distribution.c (version_loop_by_alias_check): Force
12528 cond_expr to simple gimple operand.
12530 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
12532 PR middle-end/46932
12533 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
12535 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
12539 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
12540 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
12541 TARGET_GASISR_PROLOGUES.
12542 * config/avr/avr.c (avr_option_override): Same.
12543 (avr_pass_pre_proep::execute): Same.
12545 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
12548 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
12549 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
12551 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
12553 * config/i386/i386.md (*load_tp_<mode>): Redefine as
12554 define_insn_and_split. Split to a memory load from 0 in
12555 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
12556 using PTR mode iterator.
12557 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
12558 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
12559 (*add_tp_<mode>): Redefine as define_insn_and_split.
12560 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
12561 address space. Merge with *add_tp_x32 using PTR mode iterator.
12562 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
12563 Split to an add with a memory load from 0 in
12564 DEFAULT_TLS_SEG_REG address space.
12566 2017-08-12 Andrew Pinski <apinski@cavium.com>
12568 * config/aarch64/aarch64-option-extensions.def (rdma):
12569 Fix feature string to what Linux prints out in /proc/cpuinfo.
12571 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
12574 * dwarf2out.c (modified_type_die): For C typedef types that have
12575 an ultimate origin, process the ultimate origin instead of the
12577 (gen_typedef_die): Assert that input DECLs have no ultimate
12579 (gen_type_die_with_usage): For typedef variants that have an
12580 ultimate origin, just call gen_decl_die on the original DECL.
12581 (process_scope_var): Avoid creating DIEs for local typedefs and
12582 concrete static variables.
12584 2017-08-12 Alan Modra <amodra@gmail.com>
12588 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
12589 match gnu-user.h startfile.
12590 (ENDFILE_LINUX_SPEC): Similarly.
12592 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
12595 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
12597 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
12599 2017-08-11 Tamar Christina <tamar.christina@arm.com>
12600 * config/aarch64/aarch64.md (mov<mode>): Change.
12601 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
12602 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
12603 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
12605 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
12607 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
12608 for storage order barriers.
12610 2017-08-11 Martin Liska <mliska@suse.cz>
12613 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
12614 variables of void type.
12616 2017-08-11 Martin Liska <mliska@suse.cz>
12618 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
12619 TARGET_SUPPORTS_ALIASES.
12620 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
12621 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
12622 (optimize_weakref): Likewise.
12623 * symtab.c (symtab_node::noninterposable_alias): Likewise.
12624 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
12625 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
12627 2017-08-11 Martin Liska <mliska@suse.cz>
12630 * config/i386/i386.c (make_resolver_func): Do complete
12631 refactoring of the function.
12633 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
12636 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
12637 * config/i386/i386.c (ix86_stack_protect_guard): Use
12638 ix86_stack_protect_guard_symbol_str to generate varible declaration.
12639 * doc/invoke.texi (x86 Options): Document
12640 -mstack-protector-guard-symbol= option.
12642 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
12644 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
12645 * config/i386/i386.c (ix86_split_stack_guard): New function.
12646 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
12647 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
12648 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
12649 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
12650 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
12651 (split_stack_space_check): Call ix86_split_stack_guard.
12653 2017-08-10 Martin Sebor <msebor@redhat.com>
12655 * print-tree.c (print_node): Print location using the established
12657 Replace spaces with colons.
12658 (debug_raw, debug): Ditto.
12660 2017-08-10 Martin Sebor <msebor@redhat.com>
12663 * pretty-print.c (pp_format): Correct the handling of %s precision.
12665 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
12668 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
12670 (ix86_finalize_stack_frame_flags): This. Also clear
12671 frame_pointer_needed if -fno-omit-frame-pointer is used without
12673 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
12674 with ix86_finalize_stack_frame_flags.
12675 (ix86_expand_epilogue): Likewise.
12676 (ix86_expand_split_stack_prologue): Likewise.
12677 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
12679 2017-08-10 Martin Liska <mliska@suse.cz>
12682 * c-attribs.c (handle_target_attribute):
12683 Report warning for an empty string argument of target attribute.
12685 2017-08-09 Jakub Jelinek <jakub@redhat.com>
12688 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
12690 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
12691 or DECL_NONLOCAL labels.
12692 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
12693 or DECL_NONLOCAL labels here.
12695 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
12697 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
12698 to indicate when early gimple folding has been disabled.
12699 (rs6000_gimple_fold_builtin): Add debug content.
12700 (rs6000_invalid_builtin): Fix whitespace.
12701 (rs6000_expand_builtin): Fix whitespace.
12702 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
12704 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
12707 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
12708 SAVE_MULTIPLE if not all the registers that saves, should be saved.
12710 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
12712 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
12713 (qdf24xx): Likewise.
12714 * config/aarch64/aarch64.md: Include falkor.md.
12715 * config/aarch64/falkor.md: New.
12717 2017-08-09 Marek Polacek <polacek@redhat.com>
12720 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
12721 * diagnostic.c (emit_diagnostic): Add a comment.
12722 (emit_diagnostic_valist): New function.
12724 2017-08-09 Marek Polacek <polacek@redhat.com>
12727 * input.c (make_location): New overload.
12728 * input.h (make_location): Declare.
12730 2017-08-08 Alan Modra <amodra@gmail.com>
12731 H.J. Lu <hongjiu.lu@intel.com>
12734 * gcc.c (NO_PIE_SPEC): Delete.
12735 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
12737 (LINK_PIE_SPEC): ..to here.
12738 (LINK_COMMAND_SPEC): Support -no-pie.
12739 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
12740 chain of crtbegin*.o selection, update for PIE_SPEC changes and
12742 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
12743 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
12744 (ENDFILE_CRTEND_SPEC): Similarly.
12746 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
12749 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
12750 (mstack-protector-guard-offset=): Ditto.
12751 * config/i386/i386.c (ix86_option_override): Handle
12752 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
12754 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
12755 ix86_stack_protect_guard_offset variables.
12756 (TARGET_STACK_PROTECT_GUARD): Always define.
12757 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
12758 and -mstack-protector-guard-offset= options.
12760 2017-08-08 Bin Cheng <bin.cheng@arm.com>
12762 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
12763 boundary case for the last candidate.
12765 2017-08-08 Bin Cheng <bin.cheng@arm.com>
12767 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
12768 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
12770 2017-08-08 Tamar Christina <tamar.christina@arm.com>
12772 PR middle-end/19706
12773 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
12774 * config/aarch64/aarch64-builtins.c
12775 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
12776 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
12777 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
12779 2017-08-08 Tamar Christina <tamar.christina@arm.com>
12780 Andrew Pinski <pinskia@gmail.com>
12782 PR middle-end/19706
12783 * internal-fn.def (XORSIGN): New.
12784 * optabs.def (xorsign_optab): New.
12785 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
12786 (convert_expand_mult_copysign): New.
12787 (pass_optimize_widening_mul::execute): Call
12788 convert_expand_mult_copysign.
12790 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12792 PR tree-optimization/81354
12793 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
12794 Insert on edges rather than explicitly creating landing pads.
12795 (analyze_candidates_and_replace): Commit edge inserts.
12797 2017-08-08 Richard Biener <rguenther@suse.de>
12799 PR middle-end/81719
12800 * tree-ssa-loop-niter.c: Include tree-dfa.h.
12801 (expand_simple_operations): Also look through ADDR_EXPRs with
12802 MEM_REF bases treating them as POINTER_PLUS_EXPR.
12804 2017-08-08 Richard Biener <rguenther@suse.de>
12806 PR tree-optimization/81723
12807 * tree-vect-slp.c (struct bst_traits): New hash traits.
12808 (bst_fail): New global.
12809 (vect_build_slp_tree_2): New worker, split out from ...
12810 (vect_build_slp_tree): ... this now wrapping it with using
12811 bst_fail set to cache SLP tree build fails. Properly handle
12813 (vect_analyze_slp_instance): Allocate and free bst_fail.
12815 2017-08-08 Martin Liska <mliska@suse.cz>
12818 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
12819 LABEL_DECLs that can be from a different function.
12821 2017-08-08 Bin Cheng <bin.cheng@arm.com>
12823 PR tree-optimization/81744
12824 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
12825 loop's number of iterations.
12827 2017-08-08 Martin Liska <mliska@suse.cz>
12829 * asan.c: Include header files.
12830 * attribs.c (build_decl_attribute_variant): New function moved
12832 (build_type_attribute_qual_variant): Likewise.
12833 (cmp_attrib_identifiers): Likewise.
12834 (simple_cst_list_equal): Likewise.
12835 (omp_declare_simd_clauses_equal): Likewise.
12836 (attribute_value_equal): Likewise.
12837 (comp_type_attributes): Likewise.
12838 (build_type_attribute_variant): Likewise.
12839 (lookup_ident_attribute): Likewise.
12840 (remove_attribute): Likewise.
12841 (merge_attributes): Likewise.
12842 (merge_type_attributes): Likewise.
12843 (merge_decl_attributes): Likewise.
12844 (merge_dllimport_decl_attributes): Likewise.
12845 (handle_dll_attribute): Likewise.
12846 (attribute_list_equal): Likewise.
12847 (attribute_list_contained): Likewise.
12848 * attribs.h (lookup_attribute): New function moved from tree.[ch].
12849 (lookup_attribute_by_prefix): Likewise.
12850 * bb-reorder.c: Include header files.
12851 * builtins.c: Likewise.
12852 * calls.c: Likewise.
12853 * cfgexpand.c: Likewise.
12854 * cgraph.c: Likewise.
12855 * cgraphunit.c: Likewise.
12856 * convert.c: Likewise.
12857 * dwarf2out.c: Likewise.
12858 * final.c: Likewise.
12859 * fold-const.c: Likewise.
12860 * function.c: Likewise.
12861 * gimple-expr.c: Likewise.
12862 * gimple-fold.c: Likewise.
12863 * gimple-pretty-print.c: Likewise.
12864 * gimple.c: Likewise.
12865 * gimplify.c: Likewise.
12866 * hsa-common.c: Likewise.
12867 * hsa-gen.c: Likewise.
12868 * internal-fn.c: Likewise.
12869 * ipa-chkp.c: Likewise.
12870 * ipa-cp.c: Likewise.
12871 * ipa-devirt.c: Likewise.
12872 * ipa-fnsummary.c: Likewise.
12873 * ipa-inline.c: Likewise.
12874 * ipa-visibility.c: Likewise.
12876 * lto-cgraph.c: Likewise.
12877 * omp-expand.c: Likewise.
12878 * omp-general.c: Likewise.
12879 * omp-low.c: Likewise.
12880 * omp-offload.c: Likewise.
12881 * omp-simd-clone.c: Likewise.
12882 * opts-global.c: Likewise.
12883 * passes.c: Likewise.
12884 * predict.c: Likewise.
12885 * sancov.c: Likewise.
12886 * sanopt.c: Likewise.
12887 * symtab.c: Likewise.
12888 * toplev.c: Likewise.
12889 * trans-mem.c: Likewise.
12890 * tree-chkp.c: Likewise.
12891 * tree-eh.c: Likewise.
12892 * tree-into-ssa.c: Likewise.
12893 * tree-object-size.c: Likewise.
12894 * tree-parloops.c: Likewise.
12895 * tree-profile.c: Likewise.
12896 * tree-ssa-ccp.c: Likewise.
12897 * tree-ssa-live.c: Likewise.
12898 * tree-ssa-loop.c: Likewise.
12899 * tree-ssa-sccvn.c: Likewise.
12900 * tree-ssa-structalias.c: Likewise.
12901 * tree-ssa.c: Likewise.
12902 * tree-streamer-in.c: Likewise.
12903 * tree-vectorizer.c: Likewise.
12904 * tree-vrp.c: Likewise.
12905 * tsan.c: Likewise.
12906 * ubsan.c: Likewise.
12907 * varasm.c: Likewise.
12908 * varpool.c: Likewise.
12909 * tree.c: Remove functions moved to attribs.[ch].
12910 * tree.h: Likewise.
12911 * config/aarch64/aarch64.c: Add attrs.h header file.
12912 * config/alpha/alpha.c: Likewise.
12913 * config/arc/arc.c: Likewise.
12914 * config/arm/arm.c: Likewise.
12915 * config/avr/avr.c: Likewise.
12916 * config/bfin/bfin.c: Likewise.
12917 * config/c6x/c6x.c: Likewise.
12918 * config/cr16/cr16.c: Likewise.
12919 * config/cris/cris.c: Likewise.
12920 * config/darwin.c: Likewise.
12921 * config/epiphany/epiphany.c: Likewise.
12922 * config/fr30/fr30.c: Likewise.
12923 * config/frv/frv.c: Likewise.
12924 * config/ft32/ft32.c: Likewise.
12925 * config/h8300/h8300.c: Likewise.
12926 * config/i386/winnt.c: Likewise.
12927 * config/ia64/ia64.c: Likewise.
12928 * config/iq2000/iq2000.c: Likewise.
12929 * config/lm32/lm32.c: Likewise.
12930 * config/m32c/m32c.c: Likewise.
12931 * config/m32r/m32r.c: Likewise.
12932 * config/m68k/m68k.c: Likewise.
12933 * config/mcore/mcore.c: Likewise.
12934 * config/microblaze/microblaze.c: Likewise.
12935 * config/mips/mips.c: Likewise.
12936 * config/mmix/mmix.c: Likewise.
12937 * config/mn10300/mn10300.c: Likewise.
12938 * config/moxie/moxie.c: Likewise.
12939 * config/msp430/msp430.c: Likewise.
12940 * config/nds32/nds32-isr.c: Likewise.
12941 * config/nds32/nds32.c: Likewise.
12942 * config/nios2/nios2.c: Likewise.
12943 * config/nvptx/nvptx.c: Likewise.
12944 * config/pa/pa.c: Likewise.
12945 * config/pdp11/pdp11.c: Likewise.
12946 * config/powerpcspe/powerpcspe.c: Likewise.
12947 * config/riscv/riscv.c: Likewise.
12948 * config/rl78/rl78.c: Likewise.
12949 * config/rx/rx.c: Likewise.
12950 * config/s390/s390.c: Likewise.
12951 * config/sh/sh.c: Likewise.
12952 * config/sol2.c: Likewise.
12953 * config/sparc/sparc.c: Likewise.
12954 * config/spu/spu.c: Likewise.
12955 * config/stormy16/stormy16.c: Likewise.
12956 * config/tilegx/tilegx.c: Likewise.
12957 * config/tilepro/tilepro.c: Likewise.
12958 * config/v850/v850.c: Likewise.
12959 * config/vax/vax.c: Likewise.
12960 * config/visium/visium.c: Likewise.
12961 * config/xtensa/xtensa.c: Likewise.
12963 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
12966 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
12967 constraints since the -mupper-regs-* switches have been
12969 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
12970 into a vector from a double word element that was extracted from
12971 another vector, and eliminate extra XXPERMDI instructions.
12972 (vsx_concat_<mode>_2): Likewise.
12973 (vsx_concat_<mode>_3): Likewise.
12974 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
12975 concat to allow optimizing inserts from previous extracts.
12977 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
12979 * config/i386/i386.c (ix86_stack_protect_guard): Generate
12980 memory reference to a SSP offset in TLS address space.
12981 (ix86_print_operand) <case '@'>: Remove.
12982 (ix86_print_operand_punct_valid_p): Remove '@' code.
12983 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
12984 UNSPEC_SP_TLS_TEST.
12985 (stack_tls_protect_set_<mode>): Remove.
12986 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
12987 (stack_tls_protect_test_<mode>): Remove.
12988 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
12990 2017-08-07 Olivier Hainque <hainque@adacore.com>
12993 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
12995 2017-08-07 Douglas Rupp <rupp@adacore.com>
12997 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
12998 variable was referenced as multidir in command.
13000 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13003 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
13006 2017-08-07 Martin Liska <mliska@suse.cz>
13008 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
13009 * config/rl78/rl78.c: Add include of attribs.h.
13010 * config/sh/sh.c: Likewise.
13011 * config/v850/v850.c: Likewise.
13013 2017-08-07 Tom de Vries <tom@codesourcery.com>
13015 PR middle-end/78266
13016 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
13018 2017-08-07 Martin Liska <mliska@suse.cz>
13020 * config/mips/mips.c: Include attribs.h.
13022 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
13025 * doc/invoke.texi: Document change in behvaior for -Ofast for
13028 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
13030 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
13032 (aarch64_pop_regs): Likewise.
13033 (aarch64_gen_load_pair): Likewise.
13034 (aarch64_save_callee_saves): Likewise.
13035 (aarch64_restore_callee_saves): Likewise.
13037 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13039 * config/i386/i386.c: Revert the last change.
13041 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13044 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13046 (ix86_finalize_stack_frame_flags): This. Also clear
13047 frame_pointer_needed if -fno-omit-frame-pointer is used without
13049 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13050 with ix86_finalize_stack_frame_flags.
13051 (ix86_expand_epilogue): Likewise.
13052 (ix86_expand_split_stack_prologue): Likewise.
13054 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13057 * config/i386/i386.c (get_builtin_code_for_version): Set priority
13058 to P_AES for Westmere.
13060 2017-08-07 Jonathan Yong <10walls@gmail.com>
13062 * config/i386/mingw.opt (fset-stack-executable): Removed.
13063 * config/i386/cygming.opt (fset-stack-executable): Moved
13065 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
13067 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
13069 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
13071 2017-08-07 Marek Polacek <polacek@redhat.com>
13073 PR middle-end/81737
13074 * fold-const.c (fold_indirect_ref_1): Check type_domain.
13076 2017-08-07 Martin Liska <mliska@suse.cz>
13078 * attribs.h (canonicalize_attr_name): New function.
13079 (cmp_attribs): Move from c-format.c and adjusted.
13080 (is_attribute_p): Moved from tree.h.
13081 * tree-inline.c: Add new includes.
13082 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
13083 (private_is_attribute_p): Remove.
13084 (private_lookup_attribute): Likewise.
13085 (private_lookup_attribute_by_prefix): Simplify.
13086 (remove_attribute): Use is_attribute_p.
13087 * tree.h: Remove removed declarations.
13089 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13091 PR middle-end/81698
13092 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
13093 instead of computing it in the function. Formatting fix.
13094 (expand_case): Don't rely on default_edge being the first edge,
13095 clear it if removing it, pass default_edge to
13096 emit_case_dispatch_table.
13097 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
13100 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
13102 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
13103 insn in the function, emit NOP after the insn.
13105 2017-08-06 Tom de Vries <tom@codesourcery.com>
13107 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
13110 2017-08-06 Tom de Vries <tom@codesourcery.com>
13112 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
13115 2017-08-04 Yury Gribov <tetra2005@gmail.com>
13117 PR tree-optimization/57371
13118 * match.pd: New pattern.
13120 2017-08-04 Marek Polacek <polacek@redhat.com>
13122 PR middle-end/81695
13123 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
13124 perform the computation in offset_int.
13126 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13128 PR tree-optimization/81136
13129 * tree-vectorizer.h: Include tree-hash-traits.h.
13130 (vec_base_alignments): New typedef.
13131 (vec_info): Add a base_alignments field.
13132 (vect_record_base_alignments): Declare.
13133 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
13135 (DR_IS_CONDITIONAL_IN_STMT): New macro.
13136 (create_data_ref): Add an is_conditional_in_stmt argument.
13137 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
13138 the is_conditional_in_stmt field.
13139 (data_ref_loc): Add an is_conditional_in_stmt field.
13140 (get_references_in_stmt): Set the is_conditional_in_stmt field.
13141 (find_data_references_in_stmt): Update call to create_data_ref.
13142 (graphite_find_data_references_in_stmt): Likewise.
13143 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
13144 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
13145 (vect_record_base_alignment): New function.
13146 (vect_record_base_alignments): Likewise.
13147 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
13148 for nested statements even if we fail to compute a misalignment.
13149 Use pooled base alignments for unconditional references.
13150 (vect_find_same_alignment_drs): Compare base addresses instead
13152 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
13153 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
13155 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13157 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
13158 Add an explicit name for the enum. Use auto_vec for slp_instances
13159 and grouped_stores.
13160 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
13162 (_bb_vec_info): Add a constructor and destructor.
13163 (vinfo_for_stmt): Return NULL for uids of -1 as well.
13164 (destroy_loop_vec_info): Delete.
13165 (vect_destroy_datarefs): Likewise.
13166 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
13167 (vec_info::vec_info): New function.
13168 (vec_info::~vec_info): Likewise.
13169 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
13170 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
13171 destroy_loop_vec_info.
13172 * tree-vect-loop.c (new_loop_vec_info): Replace with...
13173 (_loop_vec_info::_loop_vec_info): ...this.
13174 (destroy_loop_vec_info): Replace with...
13175 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
13176 the stmt_vec_infos. Leave handling of vec_info information to its
13177 destructor. Remove explicit vector releases.
13178 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
13179 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
13180 * tree-vect-slp.c (new_bb_vec_info): Replace with...
13181 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
13182 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
13183 (destroy_bb_vec_info): Replace with...
13184 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
13185 information to its destructor.
13186 (vect_slp_analyze_bb_1): Use new and delete instead of
13187 new_bb_vec_info and destroy_bb_vec_info.
13188 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
13191 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13193 * tree-data-ref.h (subscript): Add access_fn field.
13194 (data_dependence_relation): Add could_be_independent_p.
13195 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
13196 (same_access_functions): Move to tree-data-ref.c.
13197 * tree-data-ref.c (ref_contains_union_access_p): New function.
13198 (access_fn_component_p): Likewise.
13199 (access_fn_components_comparable_p): Likewise.
13200 (dr_analyze_indices): Add a reference to access_fn_component_p.
13201 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
13203 (constant_access_functions): Likewise.
13204 (add_other_self_distances): Likewise.
13205 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
13206 (initialize_data_dependence_relation): Use XCNEW and remove
13207 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
13208 of access functions that have the same type. Allow the
13209 subsequence to end with different bases in some circumstances.
13210 Record the chosen access functions in SUB_ACCESS_FN.
13211 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
13212 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
13213 (subscript_dependence_tester_1): Likewise dra and drb.
13214 (build_classic_dist_vector): Update calls accordingly.
13215 (subscript_dependence_tester): Likewise.
13216 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
13217 DDR_COULD_BE_INDEPENDENT_P.
13218 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
13219 comp_alias_ddrs instead of may_alias_ddrs.
13220 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
13222 (vect_analyze_data_ref_dependence): Use it if
13223 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
13224 distance vectors if that fails.
13225 (dependence_distance_ge_vf): New function.
13226 (vect_prune_runtime_alias_test_list): Use it. Don't clear
13227 LOOP_VINFO_MAY_ALIAS_DDRS.
13229 2017-08-04 Richard Biener <rguenther@suse.de>
13231 PR middle-end/81705
13232 * fold-const.c (fold_binary_loc): Properly restrict
13233 minus_var0 && minus_var1 case when associating undefined overflow
13236 2017-08-04 Tom de Vries <tom@codesourcery.com>
13238 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
13240 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13242 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13243 Don't start diagnostic messages with a capital letter.
13244 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13246 (rs6000_invalid_builtin): Likewise.
13247 (rs6000_trampoline_init): Likewise.
13249 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13252 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
13253 after setting changeable df flags.
13255 2017-08-03 Richard Biener <rguenther@suse.de>
13257 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
13258 up if the use is in USE - X.
13260 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
13262 * toplev.c (dumpfile.h): New include.
13263 (internal_error_reentered): New static function. Use it...
13264 (internal_error_function): ...here to handle reentered internal_error.
13266 2017-08-03 Richard Biener <rguenther@suse.de>
13268 PR middle-end/81148
13269 * fold-const.c (split_tree): Add minus_var and minus_con
13270 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
13271 here but always use minus_*.
13272 (associate_trees): Assert we never associate with MINUS_EXPR
13273 and NULL first operand. Do not recurse for PLUS_EXPR operands
13274 when associating as MINUS_EXPR either.
13275 (fold_binary_loc): Track minus_var and minus_con.
13277 2017-08-03 Tom de Vries <tom@codesourcery.com>
13280 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
13281 ACCEL_COMPILER, apply finish_options on
13282 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
13284 2017-08-03 Tom de Vries <tom@codesourcery.com>
13287 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
13288 function_entry_patch_area_size > 0.
13290 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13293 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
13294 instead of 10??LU, perform unit multiplication in wide_int,
13295 don't change alloc_object_size_limit if the limit is larger
13298 PR tree-optimization/81655
13299 PR tree-optimization/81588
13300 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
13301 the case when ranges[i].low and high are 1 for unsigned type with
13304 PR middle-end/81052
13305 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
13306 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
13308 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13310 * tree-vrp.h: Add include guard.
13312 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
13315 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
13316 (ud2): New insn pattern.
13317 * config/i386/i386.c (ix86_expand_epilogue):
13318 For naked functions, generate ud2 instead of trap insn.
13320 2017-08-02 Marek Polacek <polacek@redhat.com>
13323 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
13325 2017-08-02 Tom de Vries <tom@codesourcery.com>
13326 Cesar Philippidis <cesar@codesourcery.com>
13328 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
13329 Add missing edge probabilities.
13331 2017-08-02 Tamar Christina <tamar.christina@arm.com>
13333 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
13334 Correct endianness.
13336 2017-08-02 Jakub Jelinek <jakub@redhat.com>
13338 PR middle-end/79499
13339 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
13340 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
13341 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
13342 prologue_seq sequences - if any.
13344 2017-08-02 Richard Biener <rguenther@suse.de>
13346 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
13347 via vectors if supported, integer extracts via punning if supported
13348 or otherwise vector extracts.
13350 2017-08-02 Richard Biener <rguenther@suse.de>
13352 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
13354 (bitmap_insert_into_set): ... this.
13356 2017-08-02 Richard Biener <rguenther@suse.de>
13358 PR tree-optimization/81633
13360 2015-08-17 Alan Hayward <alan.hayward@arm.com>
13362 PR tree-optimization/71752
13363 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
13365 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
13367 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
13368 (machine_function::call_ms2sysv_pad_out): Remove field.
13369 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
13370 (ix86_compute_frame_layout): Likewise.
13372 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
13375 * config/i386/i386.c (ix86_set_func_type): Disallow naked
13376 attribute with interrupt attribute.
13378 2017-08-01 Andrew Pinski <apinski@cavium.com>
13380 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
13381 BIT_INSERT_EXPR's operand 1
13382 to see if the types precision matches.
13384 2017-08-01 Martin Liska <mliska@suse.cz>
13386 PR middle-end/70140
13387 * builtins.c (expand_builtin_memcpy_args): Remove.
13388 (expand_builtin_memcpy): Call newly added function
13389 expand_builtin_memory_copy_args.
13390 (expand_builtin_memcpy_with_bounds): Likewise.
13391 (expand_builtin_mempcpy): Remove last argument.
13392 (expand_builtin_mempcpy_with_bounds): Likewise.
13393 (expand_builtin_memory_copy_args): New function created from
13394 expand_builtin_mempcpy_args with small modifications.
13395 (expand_builtin_mempcpy_args): Remove.
13396 (expand_builtin_stpcpy): Remove unused argument.
13397 (expand_builtin): Likewise.
13398 (expand_builtin_with_bounds): Likewise.
13400 2017-08-01 Martin Liska <mliska@suse.cz>
13403 Make mempcpy more optimal (PR middle-end/70140).
13405 2017-08-01 Jakub Jelinek <jakub@redhat.com>
13408 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
13409 __builtin_vec_cmpne verify both arguments are compatible vectors
13410 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
13411 verify arg1_type is a pointer or array type. For __builtin_vec_st,
13412 move computation of aligned to after checking the argument types.
13416 * config/rs6000/vsx.md (vextract_fp_from_shorth,
13417 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
13420 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
13421 Doug Rupp <rupp@adacore.com>
13422 Olivier Hainque <hainque@adacore.com>
13424 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
13425 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
13427 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
13428 for TARGET_OS_CPP_BUILTIN.
13429 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
13430 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
13432 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
13433 passing required abi options to the assembler for EABI configurations.
13434 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
13435 of .text.hot and .text.unlikely sections for kernel modules when
13436 using ARM style exceptions.
13437 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
13438 options. Add EXTRA_CC1_SPEC.
13439 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
13441 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
13443 (ARM_TARGET2_DWARF_FORMAT): Define.
13444 * config/arm/t-vxworks: Adjust multilib control to removal of the
13445 Diab command line options.
13447 2017-08-01 Martin Liska <mliska@suse.cz>
13449 PR gcov-profile/81561
13450 * gcov.c (unblock): Make unblocking safe as we need to preserve
13451 index correspondence of blocks and block_lists.
13453 2017-08-01 Richard Biener <rguenther@suse.de>
13455 PR tree-optimization/81181
13456 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
13457 (compute_antic): ... end of iteration here.
13459 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
13461 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
13462 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
13463 (ftree-slp-vectorize): Likewise.
13464 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
13465 can no longer be set independent of flag_tree_loop_vectorize.
13466 * omp-general.c (emp_max_vf): Likewise.
13467 * opts.c (enable_fdo_optimizations): Remove references to
13468 flag_tree_vectorize, these are now implicit.
13469 (common_handle_option): Remove handling for OPT_ftree_vectorize,
13470 and leave it for the options machinery.
13472 2017-08-01 Martin Liska <mliska@suse.cz>
13474 PR middle-end/70140
13475 * builtins.c (expand_builtin_memcpy_args): Remove.
13476 (expand_builtin_memcpy): Call newly added function
13477 expand_builtin_memory_copy_args.
13478 (expand_builtin_memcpy_with_bounds): Likewise.
13479 (expand_builtin_mempcpy): Remove last argument.
13480 (expand_builtin_mempcpy_with_bounds): Likewise.
13481 (expand_builtin_memory_copy_args): New function created from
13482 expand_builtin_mempcpy_args with small modifications.
13483 (expand_builtin_mempcpy_args): Remove.
13484 (expand_builtin_stpcpy): Remove unused argument.
13485 (expand_builtin): Likewise.
13486 (expand_builtin_with_bounds): Likewise.
13488 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
13491 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
13492 print "ds:" only for immediates in generic address space.
13494 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
13497 * config/i386/i386.c (ix86_funciton_naked): New prototype.
13498 (ix86_function_ok_for_sibcall): Return false for naked functions.
13500 2017-08-01 Richard Biener <rguenther@suse.de>
13502 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
13503 (compute_antic): Seed worklist with exit block predecessors.
13504 * cfganal.c (dfs_find_deadend): For a cycle return the source
13505 of the edge closing it.
13507 2017-08-01 Tamar Christina <tamar.christina@arm.com>
13509 * config/aarch64/aarch64.c
13510 (aarch64_can_const_movi_rtx_p): Move 0 check.
13512 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13514 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
13515 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
13517 * match.pd: Ditto in address comparison pattern.
13519 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13521 PR tree-optimization/81627
13522 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
13523 closed ssa form for store-store chain.
13525 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13527 PR tree-optimization/81620
13528 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
13529 for store-store chain.
13531 2017-08-01 Jakub Jelinek <jakub@redhat.com>
13533 PR tree-optimization/81588
13534 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
13535 ranges[i].in_p, invert comparison code ccode. For >/>=,
13536 swap rhs1 and rhs2 and comparison code unconditionally,
13537 for </<= don't do that. Don't swap rhs1/rhs2 again if
13538 ranges[i].in_p, instead invert comparison code ccode if
13539 opcode or oe->rank is BIT_IOR_EXPR.
13542 * optabs.def (vec_extract_optab, vec_init_optab): Change from
13543 a direct optab to conversion optab.
13544 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
13545 with GET_MODE_INNER as last argument instead of optab_handler.
13546 * expmed.c (extract_bit_field_1): Likewise. Use vector from
13547 vector extraction if possible and optab is available.
13548 * expr.c (store_constructor): Use convert_optab_handler instead
13549 of optab_handler. Use vector initialization from smaller
13550 vectors if possible and optab is available.
13551 * tree-vect-stmts.c (vectorizable_load): Likewise.
13552 * doc/md.texi (vec_extract, vec_init): Document that the optabs
13553 now have two modes.
13554 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
13555 of vec_init from half-sized vectors with the same element mode.
13556 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
13557 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
13558 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
13559 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
13560 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
13561 after mode in gen_vec_extract* calls.
13562 (vec_extract<mode>): Renamed to ...
13563 (vec_extract<mode><ssescalarmodelower>): ... this.
13564 (vec_extract<mode><ssehalfvecmodelower>): New expander.
13565 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
13566 element mode after mode in gen_vec_init* calls.
13567 (VEC_INIT_HALF_MODE): New mode iterator.
13568 (vec_init<mode>): Renamed to ...
13569 (vec_init<mode><ssescalarmodelower>): ... this.
13570 (vec_init<mode><ssehalfvecmodelower>): New expander.
13571 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
13572 (vec_extractv2sfsf): ... this.
13573 (vec_initv2sf): Renamed to ...
13574 (vec_initv2sfsf): ... this.
13575 (vec_extractv2si): Renamed to ...
13576 (vec_extractv2sisi): ... this.
13577 (vec_initv2si): Renamed to ...
13578 (vec_initv2sisi): ... this.
13579 (vec_extractv4hi): Renamed to ...
13580 (vec_extractv4hihi): ... this.
13581 (vec_initv4hi): Renamed to ...
13582 (vec_initv4hihi): ... this.
13583 (vec_extractv8qi): Renamed to ...
13584 (vec_extractv8qiqi): ... this.
13585 (vec_initv8qi): Renamed to ...
13586 (vec_initv8qiqi): ... this.
13587 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
13588 (vec_init<mode>): Renamed to ...
13589 (vec_init<mode><VEC_base_l>): ... this.
13590 (vec_extract<mode>): Renamed to ...
13591 (vec_extract<mode><VEC_base_l>): ... this.
13592 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
13593 (vec_initv2sfsf): ... this.
13594 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
13595 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
13596 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
13597 element mode after mode in gen_vec_init* calls.
13598 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
13599 (vec_init<mode><Vel>): ... this.
13600 (vec_extract<mode>): Renamed to ...
13601 (vec_extract<mode><Vel>): ... this.
13602 * config/aarch64/iterators.md (Vel): New mode attribute.
13603 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
13604 Add element mode after mode in gen_vec_extract* calls.
13605 * config/s390/vector.md (non_vec_l): New mode attribute.
13606 (vec_extract<mode>): Renamed to ...
13607 (vec_extract<mode><non_vec_l>): ... this.
13608 (vec_init<mode>): Renamed to ...
13609 (vec_init<mode><non_vec_l>): ... this.
13610 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
13611 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
13613 * config/arm/iterators.md (V_elem_l): New mode attribute.
13614 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
13615 (vec_extract<mode><V_elem_l>): ... this.
13616 (vec_extractv2di): Renamed to ...
13617 (vec_extractv2didi): ... this.
13618 (vec_init<mode>): Renamed to ...
13619 (vec_init<mode><V_elem_l>): ... this.
13620 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
13621 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
13622 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
13623 Add element mode after gen_vec_extract* calls.
13624 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
13625 (vec_init<mode><unitmode>): ... this.
13626 (vec_extract<mode>): Renamed to ...
13627 (vec_extract<mode><unitmode>): ... this.
13628 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
13629 (vec_init<mode><unitmode>): ... this.
13630 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
13631 (vec_initv2sfsf): ... this.
13632 (vec_extractv2sf): Renamed to ...
13633 (vec_extractv2sfsf): ... this.
13634 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
13635 Add element mode after gen_vec_extract* calls.
13636 * config/mips/mips.md (unitmode): New mode iterator.
13637 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
13638 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
13639 * config/spu/spu.md (inner_l): New mode attribute.
13640 (vec_init<mode>): Renamed to ...
13641 (vec_init<mode><inner_l>): ... this.
13642 (vec_extract<mode>): Renamed to ...
13643 (vec_extract<mode><inner_l>): ... this.
13644 * config/sparc/sparc.md (veltmode): New mode iterator.
13645 (vec_init<VMALL:mode>): Renamed to ...
13646 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
13647 * config/ia64/vect.md (vec_initv2si): Renamed to ...
13648 (vec_initv2sisi): ... this.
13649 (vec_initv2sf): Renamed to ...
13650 (vec_initv2sfsf): ... this.
13651 (vec_extractv2sf): Renamed to ...
13652 (vec_extractv2sfsf): ... this.
13653 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
13654 (vec_init<mode>): Renamed to ...
13655 (vec_init<mode><VEC_base_l>): ... this.
13656 (vec_extract<mode>): Renamed to ...
13657 (vec_extract<mode><VEC_base_l>): ... this.
13658 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
13659 (vec_initv2sfsf): ... this.
13660 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
13661 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
13662 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
13663 gen_vec_init* calls.
13665 2017-08-01 Richard Biener <rguenther@suse.de>
13667 PR tree-optimization/81297
13668 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
13669 TREE_OVERFLOW from INTEGER_CSTs.
13671 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
13673 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
13675 2017-07-31 Carl Love <cel@us.ibm.com>
13677 * config/rs6000/rs6000-c: Add support for built-in functions
13678 vector signed char vec_xl_be (signed long long, signed char *);
13679 vector unsigned char vec_xl_be (signed long long, unsigned char *);
13680 vector signed int vec_xl_be (signed long long, signed int *);
13681 vector unsigned int vec_xl_be (signed long long, unsigned int *);
13682 vector signed long long vec_xl_be (signed long long, signed long long *);
13683 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
13684 vector signed short vec_xl_be (signed long long, signed short *);
13685 vector unsigned short vec_xl_be (signed long long, unsigned short *);
13686 vector double vec_xl_be (signed long long, double *);
13687 vector float vec_xl_be (signed long long, float *);
13688 * config/rs6000/altivec.h (vec_xl_be): Add #define.
13689 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
13690 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
13692 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
13693 (altivec_expand_builtin): Add switch statement to call
13694 altivec_expand_xl_be for each builtin.
13695 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
13696 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
13697 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
13698 __builtin_vsx_le_be_v16qi.
13699 * doc/extend.texi: Update the built-in documentation file for the
13700 new built-in functions.
13702 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
13705 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
13707 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
13709 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13711 * config.gcc: Add z14.
13712 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
13713 CPU model numbers for z13s and z14.
13714 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
13716 * config/s390/s390-opts.h (enum processor_type): Rename
13717 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
13718 * config/s390/s390.c (processor_table): Add field for CPU name to
13719 be passed to Binutils.
13720 (s390_asm_output_machine_for_arch): Use the new field in
13721 processor_table for Binutils.
13722 (s390_expand_builtin): Replace arch12 with z14.
13723 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
13724 (s390_get_sched_attrmask): Likewise.
13725 (s390_get_unit_mask): Likewise.
13726 * config/s390/s390.opt: Add z14 to processor_type enum.
13728 2017-07-31 Martin Jambor <mjambor@suse.cz>
13731 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
13732 regardless of optimization level.
13734 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
13735 Martin Liska <mliska@suse.cz>
13737 * predict.def: Remove old comment and adjust probability.
13738 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
13739 PREDICT statements.
13741 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
13744 * config/i386/i386.c (ix86_function_naked): New function.
13745 (ix86_can_use_return_insn_p): Return false for naked functions.
13746 (ix86_expand_prologue): Skip prologue for naked functions.
13747 (ix86_expand_epilogue): Skip epilogue for naked functions
13748 and emit trap instruction.
13749 (ix86_warn_func_return): New function.
13750 (ix86_attribute_table): Add "naked" attribute specification.
13751 (TARGET_WARN_FUNC_RETURN): Define.
13752 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
13754 2017-07-31 Martin Liska <mliska@suse.cz>
13756 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
13757 (dump_gimple_bb_header): Always dump BB info.
13758 (pp_cfg_jump): Do not append info about BB when dumping a jump.
13760 2017-07-31 Martin Liska <mliska@suse.cz>
13763 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
13764 also with current_function_decl non-null equality.
13766 2017-07-31 Jakub Jelinek <jakub@redhat.com>
13769 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
13770 change type to the element type, instead add eltype variable and
13771 use it where we are interested in the element type.
13773 PR tree-optimization/81603
13774 * ipa-polymorphic-call.c
13775 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
13776 offset arithmetic in offset_int, bail out if the resulting bit offset
13777 doesn't fit into shwi.
13779 2017-07-31 Martin Liska <mliska@suse.cz>
13781 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
13782 (gimplify_save_expr): Fix comment.
13784 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
13787 * config/i386/i386.c (ix86_function_arg): Update arguments for
13789 (ix86_compute_frame_layout): Set the initial stack offset to
13790 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
13791 INCOMING_FRAME_SP_OFFSET.
13792 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
13793 stack before exception handler returns.
13794 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
13795 the 'ERROR_CODE' for exception handler.
13797 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
13799 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
13800 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
13801 (ASM_OUTPUT_REG_POP): Ditto.
13802 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
13803 instead of asm_fprintf to output pure string.
13805 2017-07-29 Jakub Jelinek <jakub@redhat.com>
13807 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
13808 to imported_module_or_decl hook.
13809 (debug_nothing_tree_tree_tree_bool): Remove.
13810 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
13811 * debug.c (do_nothing_debug_hooks): Use
13812 debug_nothing_tree_tree_tree_bool_bool instead of
13813 debug_nothing_tree_tree_tree_bool.
13814 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
13815 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
13816 * sdbout.c (sdb_debug_hooks): Likewise.
13817 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
13818 (gen_namespace_die): Add DW_AT_export_symbols attribute if
13820 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
13821 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
13822 attribute, don't add anything.
13824 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13826 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
13827 (fold_build2_stat_loc): Likewise.
13828 (fold_build3_stat_loc): Likewise.
13829 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
13830 (fold_build1_loc): Remove macro.
13831 (fold_build2_loc): Likewise.
13832 (fold_build3_loc): Likewise.
13834 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13836 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
13837 (gimple_build_debug_bind_source_stat): Likewise.
13838 * gimple.h (gimple_build_debug_bind): Remove macro.
13839 (gimple_build_debug_bind_source): Likewise.
13841 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13843 * bitmap.c (bitmap_alloc): Adjust.
13844 (bitmap_gc_alloc): Likewise.
13845 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
13847 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13849 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
13850 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
13851 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
13852 (bitmap_gc_alloc_stat): Likewise.
13853 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
13855 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13857 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
13858 * rtl.h (shallow_copy_rtx): Remove macro.
13860 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13862 * emit-rtl.c (gen_raw_REG): Adjust.
13863 * gengenrtl.c (gendef): Likewise.
13864 * rtl.c (rtx_alloc_stat): Remove _stat from name.
13865 * rtl.h (rtx_alloc): Remove macro.
13867 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13869 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
13870 (build_tree_list_stat): Likewise.
13871 * tree.h (build_tree_list): Remove macro.
13872 (build_tree_list_vec): Likewise.
13874 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13876 * tree.c (make_vector_stat): Remove _stat from name.
13877 (build_vector_stat): Likewise.
13878 * tree.h (make_vector_stat): Remove macro.
13879 (build_vector_stat): Likewise.
13881 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13883 * tree.h (build_var_debug_value): Remove prototype.
13885 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13887 * tree.c (tree_cons_stat): Remove _stat from name.
13888 * tree.h (tree_cons): Remove macro.
13890 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13892 * tree.c (build_vl_exp_stat): Remove _stat from name.
13893 * tree.h (build_vl_exp): Remove macro.
13895 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13897 * tree.c (build_decl_stat): Remove _stat from name.
13898 * tree.h (build_decl): Remove macro.
13900 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13902 * gimple.c (gimple_build_with_ops_stat): Adjust.
13903 (gimple_alloc_stat): Remove _stat from name.
13904 * gimple.h (gimple_alloc): Remove macro.
13906 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13908 * tree.c (make_tree_vec_stat): Remove _stat from name.
13909 (grow_tree_vec_stat): Likewise.
13910 * tree.h (make_tree_vec_stat): Adjust prototype.
13911 (grow_tree_vec_stat): Likewise.
13912 (make_tree_vec): Remove macro.
13913 (grow_tree_vec): Likewise.
13915 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13917 * fold-const.c (fold_build1_stat_loc): Adjust.
13918 (fold_build2_stat_loc): Likewise.
13919 (fold_build3_stat_loc): Likewise.
13920 * tree.c (build0_stat): Remove _stat from name.
13921 (build1_stat): Likewise.
13922 (build2_stat): Likewise.
13923 (build3_stat): Likewise.
13924 (build4_stat): Likewise.
13925 (build5_stat): Likewise.
13926 * tree.h (build1_loc): Remove macro, and rename _stat function
13928 (build2_loc): Likewise.
13929 (build3_loc): Likewise.
13930 (build4_loc): Likewise.
13931 (build5_loc): Likewise.
13933 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13935 * tree.c (make_int_cst_stat): Remove _stat from name.
13936 * tree.h (make_int_cst_stat): Adjust prototype.
13937 (make_int_cst): Remove macro.
13939 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13941 * tree.c (make_tre_binfo_stat): Remove _stat from name.
13942 * tree.h (make_tree_binfo_stat): Adjust prototype.
13943 (make_tree_binfo): Remove.
13945 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13947 * tree.c (copy_node_stat): Rename to copy_node.
13948 (build_distinct_type_copy): Adjust.
13949 * tree.h (copy_node_stat): Adjust prototype.
13950 (copy_node): Remove macro.
13952 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13954 * tree.c (make_node_stat): rename to make_node.
13955 (build_tree_list_stat): Adjust.
13956 (build0_stat): Likewise.
13957 (build2_stat): Likewise.
13958 (build3_stat): Likewise.
13959 (build4_stat): Likewise.
13960 (build5_stat): Likewise.
13961 (build_decl_stat): Likewise.
13962 * tree.h (make_node_stat): Adjust prototype.
13963 (make_node): remove macro.
13965 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
13967 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
13968 (PPC_FEATURE2_SCV): Likewise.
13969 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
13971 2017-07-28 Tamar Christina <tamar.christina@arm.com>
13973 * config/aarch64/aarch64.c
13974 (aarch64_internal_mov_immediate): Add new special pattern.
13975 * config/aarch64/aarch64.md (*movdi_aarch64):
13976 Add reg/32bit const mov case.
13978 2017-07-28 Tamar Christina <tamar.christina@arm.com>
13979 Richard Sandiford <richard.sandiford@linaro.org>
13981 * config/aarch64/aarch64.md (mov<mode>): Generalize.
13982 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
13983 Add integer and movi cases.
13984 (movi-split-hf-df-sf split, fp16): New.
13985 (enabled): Added TARGET_FP_F16INST.
13986 * config/aarch64/iterators.md (GPF_HF): New.
13987 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
13989 2017-07-28 Tamar Christina <tamar.christina@arm.com>
13991 * config/aarch64/aarch64.c
13992 (aarch64_simd_container_mode): Add prototype.
13993 (aarch64_expand_mov_immediate): Add HI support.
13994 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
13995 (aarch64_can_const_movi_rtx_p): New.
13996 (aarch64_preferred_reload_class):
13997 Remove restrictions of using FP registers for certain SIMD operations.
13998 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
13999 (aarch64_valid_floating_const): Add integer move validation.
14000 (aarch64_simd_imm_scalar_p): Remove.
14001 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
14002 (aarch64_legitimate_constant_p): Expand list of supported cases.
14003 * config/aarch64/aarch64-protos.h
14004 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
14005 (aarch64_reinterpret_float_as_int): New.
14006 (aarch64_simd_imm_scalar_p): Remove.
14007 * config/aarch64/constraints.md (Uvi): New.
14008 (Dd): Split into Ds and new Dd.
14009 * config/aarch64/aarch64.md (*movsi_aarch64):
14011 (*movdi_aarch64): Add SIMD mov case.
14013 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14015 * tree-predcom.c: (struct chain): Handle store-store chain in which
14016 stores for elimination only store loop invariant values.
14017 (execute_pred_commoning_chain): Ditto.
14018 (prepare_initializers_chain_store_elim): Ditto.
14019 (prepare_finalizers): Ditto.
14020 (is_inv_store_elimination_chain): New function.
14021 (initialize_root_vars_store_elim_1): New function.
14023 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14025 * tree-predcom.c: Revise general description of the pass.
14026 (enum chain_type): New enum type for store elimination.
14027 (struct chain): New field supporting store elimination.
14028 (struct component): Ditto.
14029 (dump_chain): Dump store-stores chain.
14030 (release_chain): Release resources.
14031 (split_data_refs_to_components): Compute and create component
14032 contains only stores for elimination.
14033 (get_chain_last_ref_at): New function.
14034 (make_invariant_chain): Initialization.
14035 (make_rooted_chain): Specify chain type in parameter and record it.
14036 (add_looparound_copies): Skip for store-stores chain.
14037 (determine_roots_comp): Compute type of chain and pass it to
14039 (initialize_root_vars_store_elim_2): New function.
14040 (finalize_eliminated_stores): New function.
14041 (remove_stmt): Handle store for elimination.
14042 (execute_pred_commoning_chain): Execute predictive commoning on
14043 store-store chains.
14044 (determine_unroll_factor): Skip unroll for store-stores chain.
14045 (prepare_initializers_chain_store_elim): New function.
14046 (prepare_initializers_chain): Hanlde store-store chain.
14047 (prepare_finalizers_chain, prepare_finalizers): New function.
14048 (tree_predictive_commoning_loop): Return integer value indicating
14049 if loop is unrolled or lcssa form is corrupted.
14050 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
14052 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14054 * tree-predcom.c (initialize_root): Delete.
14055 (execute_pred_commoning_chain): Initialize root vars and replace
14056 reference of non-combined chain directly, rather than call above
14059 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14061 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
14062 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
14064 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14066 * tree-predcom.c (struct chain): New field init_seq.
14067 (release_chain): Release init_seq.
14068 (prepare_initializers_chain): Record intialization stmts in above
14070 (insert_init_seqs): New function.
14071 (tree_predictive_commoning_loop): Call insert_init_seqs.
14073 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14075 * tree-predcom.c (determine_roots_comp): Skip trivial components.
14077 2017-07-28 Richard Biener <rguenther@suse.de>
14079 * match.pd: Remove superfluous :c.
14080 * genmatch.c (simplify::id): Add member.
14081 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
14083 (current_id): New global.
14084 (dt_node::parent): Move from ...
14085 (dt_operand::parent): ... here. Add for_id member.
14086 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
14087 (decision_tree::find_node): Relax order requirement when
14088 merging DT_TRUE nodes to ones inbetween the current simplify
14089 and the one we try to merge with. Add diagnostic whenever
14090 we need to enforce pattern order by not merging.
14091 (decision_tree::insert): Set current_id.
14092 (decision_tree::print_node): Dump parent node and for_id.
14093 (parser::last_id): Add member.
14094 (parser::push_simplify): Assign unique id.
14095 (parser::parser): Initialize last_id.
14097 2017-07-28 Martin Liska <mliska@suse.cz>
14100 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
14101 gimple_build_debug_bind.
14103 2017-07-28 Richard Biener <rguenther@suse.de>
14105 PR tree-optimization/81502
14106 * match.pd: Add pattern combining BIT_INSERT_EXPR with
14108 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
14110 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
14111 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
14112 for BIT_FIELD_REF args.
14113 * fold-const.c (make_bit_field_ref): Likewise.
14114 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
14116 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14119 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
14120 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
14121 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
14122 Or it into SANITIZER_UNDEFINED.
14123 * ubsan.c: Include gimple-fold.h and varasm.h.
14124 (ubsan_expand_ptr_ifn): New function.
14125 (instrument_pointer_overflow): New function.
14126 (maybe_instrument_pointer_overflow): New function.
14127 (instrument_object_size): Formatting fix.
14128 (pass_ubsan::execute): Call instrument_pointer_overflow
14129 and maybe_instrument_pointer_overflow.
14130 * internal-fn.c (expand_UBSAN_PTR): New function.
14131 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
14132 * sanitizer.def (__ubsan_handle_pointer_overflow,
14133 __ubsan_handle_pointer_overflow_abort): New builtins.
14134 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
14135 * internal-fn.def (UBSAN_PTR): New internal function.
14136 * opts.c (sanitizer_opts): Add pointer-overflow.
14137 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
14138 * fold-const.c (build_range_check): Compute pointer range check in
14139 integral type if pointer arithmetics would be needed. Formatting
14142 2017-07-28 Martin Liska <mliska@suse.cz>
14145 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
14146 parameters that are of a variable-length.
14148 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
14150 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
14152 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
14153 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
14154 (CRT_CALL_STATIC_FUNCTION): Likewise.
14155 (ASM_DEFAULT_SPEC): New define.
14156 (ASM_SPEC32): Likewise.
14157 (ASM_SPEC64): Likewise.
14158 (ASM_SPEC_COMMON): Likewise.
14159 (ASM_SPEC): Likewise.
14160 (INVALID_64BIT): Likewise.
14161 (LINK_OS_DEFAULT_SPEC): Likewise.
14162 (LINK_OS_SPEC32): Likewise.
14163 (LINK_OS_SPEC64): Likewise.
14164 (POWERPC_LINUX): Likewise.
14165 (PTRDIFF_TYPE): Likewise.
14166 (RESTORE_FP_PREFIX): Likewise.
14167 (RESTORE_FP_SUFFIX): Likewise.
14168 (SAVE_FP_PREFIX): Likewise.
14169 (SAVE_FP_SUFFIX): Likewise.
14170 (SIZE_TYPE): Likewise.
14171 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
14172 (TARGET_64BIT): Likewise.
14173 (TARGET_64BIT): Likewise.
14174 (TARGET_AIX): Likewise.
14175 (WCHAR_TYPE_SIZE): Likewise.
14176 (WCHAR_TYPE): Undefine.
14177 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
14178 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
14179 (CPP_OS_RTEMS_SPEC): Delete.
14180 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
14181 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
14183 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
14185 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14187 PR tree-optimization/81578
14188 * tree-parloops.c (build_new_reduction): Bail out if
14189 reduction_code isn't one of the standard OpenMP reductions.
14190 Move the details printing after that decision.
14192 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14194 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
14195 related to reload_in_progress.
14196 (splat_input_operand): Likewise.
14197 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
14199 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
14201 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
14202 (TARGET_INSTANTIATE_DECLS): Likewise.
14203 (legitimate_indexed_address_p): Delete reload_in_progress code.
14204 (rs6000_debug_legitimate_address_p): Likewise.
14205 (rs6000_eliminate_indexed_memrefs): Likewise.
14206 (rs6000_emit_le_vsx_store): Likewise.
14207 (rs6000_emit_move_si_sf_subreg): Likewise.
14208 (rs6000_emit_move): Likewise.
14209 (register_to_reg_type): Likewise.
14210 (rs6000_pre_atomic_barrier): Likewise.
14211 (rs6000_machopic_legitimize_pic_address): Likewise.
14212 (rs6000_allocate_stack_temp): Likewise.
14213 (rs6000_address_for_fpconvert): Likewise.
14214 (rs6000_address_for_altivec): Likewise.
14215 (rs6000_secondary_memory_needed_rtx): Delete function.
14216 (rs6000_check_sdmode): Likewise.
14217 (rs6000_alloc_sdmode_stack_slot): Likewise.
14218 (rs6000_instantiate_decls): Likewise.
14219 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
14220 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
14221 Delete reload_in_progress.
14222 (*vec_reload_and_plus_<mptrsize>): Likewise.
14223 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
14224 (vsx_div_v2di): Likewise.
14225 (vsx_udiv_v2di): Likewise.
14227 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14229 * config/rs6000/rs6000.opt (mlra): Replace with stub.
14230 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
14231 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
14232 (rs6000_debug_reg_global): Delete print of LRA status.
14233 (rs6000_option_override_internal): Delete dead LRA related code.
14234 (rs6000_lra_p): Delete function.
14235 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
14237 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
14239 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
14240 * config/riscv/rtems.h: New file.
14242 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14243 Sudakshina Das <sudi.das@arm.com>
14245 * config/aarch64/aarch64.md
14246 (define_split for and<mode>3nr_compare): Move
14247 non aarch64_logical_operand to a register.
14248 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
14249 register immediate operand to a register.
14250 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
14252 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14254 PR middle-end/81564
14255 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
14257 2017-07-27 Richard Biener <rguenther@suse.de>
14259 PR tree-optimization/81573
14260 PR tree-optimization/81494
14261 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
14262 multi defuse cycle case.
14264 2017-07-27 Richard Biener <rguenther@suse.de>
14266 PR tree-optimization/81571
14267 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
14270 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
14272 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
14273 earlier and only if MASK_FPU is set. Adjust formatting.
14275 2017-07-27 Martin Liska <mliska@suse.cz>
14277 * opt-functions.awk: Add validation of value of Init.
14278 * optc-gen.awk: Pass new argument.
14280 2017-07-27 Martin Liska <mliska@suse.cz>
14282 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
14283 Fix wrong condition.
14285 2017-07-27 Martin Liska <mliska@suse.cz>
14287 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
14288 BBs and edges seen by autoFDO.
14290 2017-07-27 Richard Biener <rguenther@suse.de>
14292 PR tree-optimization/81502
14293 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
14294 with incompatible but same sized type.
14295 (execute_update_addresses_taken): Likewise.
14297 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
14299 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
14300 flag_tree_loop_vectorize rather than flag_tree_vectorize.
14302 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14305 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
14306 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
14307 Change s_operand to memory_operand.
14309 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
14311 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
14312 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
14313 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
14314 Emit instructions rather than returning an expression. Handle TFmode
14315 and KFmode by casting to TImode.
14316 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
14317 (rs6000_emit_le_vsx_store): Likewise.
14318 * config/rs6000/vsx.md (VSX_TI): New iterator.
14319 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
14320 (*vsx_le_undo_permute_<mode>): Likewise.
14321 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
14322 emit the split sequence.
14323 (*vsx_le_perm_store_<mode>): Likewise.
14325 2017-07-27 Jakub Jelinek <jakub@redhat.com>
14327 PR tree-optimization/81555
14328 PR tree-optimization/81556
14329 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
14330 if true, force CHANGED for the recursive invocation.
14331 (reassociate_bb): Remember original length of ops array, pass
14332 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
14334 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
14335 attributes for noipa attribute. For naked attribute use
14336 lookup_attribute first before lookup_attribute_spec.
14337 * final.c (rest_of_handle_final): Disable IPA RA for functions with
14339 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
14340 for functions with noipa attribute.
14341 (cgraph_externally_visible_p): Return true for functions with noipa
14343 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
14344 for functions with noipa attribute.
14345 * doc/extend.texi: Document noipa function attribute.
14346 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
14347 also for functions with noipa attribute.
14348 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
14350 2017-07-26 Andrew Pinski <apinski@cavium.com>
14352 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
14353 vec_unalign_load_cost and vec_unalign_store_cost.
14355 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14357 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
14358 -mvsx-small-integer option.
14359 (ISA_3_0_MASKS_IEEE): Likewise.
14360 (OTHER_VSX_VECTOR_MASKS): Likewise.
14361 (POWERPC_MASKS): Likewise.
14362 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
14363 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
14364 code, only testing for DImode being allowed in non-VSX floating
14366 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
14367 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
14369 (rs6000_option_override_internal): Delete -mvsx-small-integer.
14370 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
14371 TARGET_P8_VECTOR test.
14372 (rs6000_secondary_reload_simple_move): Likewise.
14373 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
14374 since TARGET_P9_VECTOR was already tested.
14375 (rs6000_opt_masks): Remove -mvsx-small-integer.
14376 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
14377 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14379 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
14380 test for TARGET_VEXTRACTUB was used, and that uses
14382 (p9 extract splitter): Likewise.
14383 (vsx_extract_<mode>_di_p9): Likewise.
14384 (vsx_extract_<mode>_store_p9): Likewise.
14385 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
14386 for TARGET_P9_VECTOR was used. Delete code that is now dead with
14387 the elimination of TARGET_VSX_SMALL_INTEGER.
14388 (vsx_extract_<mode>_p8): Likewise.
14389 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
14390 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
14391 (vsx_set_<mode>_p9): Likewise.
14392 (vsx_set_v4sf_p9): Likewise.
14393 (vsx_set_v4sf_p9_zero): Likewise.
14394 (vsx_insert_extract_v4sf_p9): Likewise.
14395 (vsx_insert_extract_v4sf_p9_2): Likewise.
14396 * config/rs6000/rs6000.md (sign extend splitter): Change
14397 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
14398 (floatsi<mode>2_lfiwax_mem): Likewise.
14399 (floatunssi<mode>2_lfiwzx_mem): Likewise.
14400 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
14401 since a test for TARGET_P9_VECTOR was used.
14402 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14403 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
14404 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14405 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
14406 TARGET_P8_VECTOR test.
14407 (fix_trunc<mode>si2_stfiwx): Likewise.
14408 (fix_trunc<mode>si2_internal): Likewise.
14409 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
14410 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14412 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
14413 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
14414 TARGET_P8_VECTOR test.
14415 (fixuns_trunc<mode>si2_stfiwx): Likewise.
14416 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
14417 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14419 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
14420 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
14421 since a test for TARGET_P9_VECTOR was used.
14422 (splitter for loading small constants): Likewise.
14424 2017-07-26 Andrew Pinski <apinski@cavium.com>
14426 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
14429 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
14432 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
14433 (fp_valid_at): Likewise.
14435 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
14437 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
14438 (qdf24xx_addrcost_table): Likewise.
14439 (cortexa57_tunings): Update to use generic_branch_cost.
14440 (cortexa72_tunings): Likewise.
14441 (cortexa73_tunings): Likewise.
14442 (qdf24xx_tunings): Likewise.
14444 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
14446 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
14447 (thunderx2t99_branch_cost): Likewise.
14448 (cortexa35_tunings): Update to use generic_branch_cost.
14449 (cortexa53_tunings): Likewise.
14450 (cortexa57_tunings): Likewise.
14451 (cortexa72_tunings): Likewise.
14452 (cortexa73_tunings): Likewise.
14453 (thunderx2t99_tunings): Likewise.
14455 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
14457 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
14458 (sparc_option_override): Honour MASK_FSMULD.
14459 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
14460 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
14461 * config/sparc/sparc.opt (mfsmuld): New option.
14462 * doc/invoke.texi (mfsmuld): Document option.
14464 2017-07-26 Marek Polacek <polacek@redhat.com>
14466 PR middle-end/70992
14467 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
14469 2017-07-26 Richard Biener <rguenther@suse.de>
14471 * gimple-match-head.c (do_valueize): Return OP if valueize
14473 (get_def): New helper to get at the def stmt of a SSA name
14474 if valueize allows.
14475 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
14476 do_valueize to get at the def stmt.
14477 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
14479 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
14481 PR middle-end/46932
14482 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
14484 2017-07-26 Martin Liska <mliska@suse.cz>
14487 * function.c (expand_function_start): Make expansion of
14488 nonlocal_goto_save_area after parm_birth_insn.
14490 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
14492 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
14493 from all CPU target flags enable members.
14495 2017-07-26 Richard Biener <rguenther@suse.de>
14497 * genmatch.c (dt_simplify::gen): Make iterator vars const.
14498 (decision_tree::gen): Make 'type' const.
14499 (write_predicate): Likewise.
14501 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
14503 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
14504 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
14505 (rs6000_option_override_internal): Likewise.
14506 (rs6000_expand_vector_set): Likewise.
14507 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
14508 (TARGET_UPPER_REGS_SF): Likewise.
14509 (TARGET_UPPER_REGS_DI): Likewise.
14510 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
14511 (TARGET_DIRECT_MOVE_64BIT): Likewise.
14512 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
14513 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14514 (Splitters for DI constants in Altivec registers): Likewise.
14515 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
14516 (vsx_set_v4sf_p9): Likewise.
14517 (vsx_set_v4sf_p9_zero): Likewise.
14518 (vsx_insert_extract_v4sf_p9): Likewise.
14519 (vsx_insert_extract_v4sf_p9_2): Likewise.
14521 2017-07-25 Carl Love <cel@us.ibm.com>
14523 * doc/extend.texi: Update the built-in documentation file for the
14524 existing built-in functions
14525 vector signed char vec_cnttz (vector signed char);
14526 vector unsigned char vec_cnttz (vector unsigned char);
14527 vector signed short vec_cnttz (vector signed short);
14528 vector unsigned short vec_cnttz (vector unsigned short);
14529 vector signed int vec_cnttz (vector signed int);
14530 vector unsigned int vec_cnttz (vector unsigned int);
14531 vector signed long long vec_cnttz (vector signed long long);
14532 vector unsigned long long vec_cnttz (vector unsigned long long);
14534 2017-07-25 Andrew Pinski <apinski@cavium.com>
14536 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
14537 accesses where the use is for the first operand of a BIT_INSERT.
14539 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
14542 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
14543 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
14545 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
14547 * config/i386/gstabs.h: Delete.
14548 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
14550 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
14552 * config/i386/i386.c (ix86_decompose_address): Do not check for
14553 register RTX when looking at index_reg or base_reg.
14554 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
14556 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
14558 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
14559 to update EH info here.
14561 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
14563 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
14565 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
14567 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
14569 2017-07-25 Torsten Duwe <duwe@suse.de>
14571 * common.opt: Introduce -fpatchable-function-entry
14572 command line option, and its variables function_entry_patch_area_size
14573 and function_entry_patch_area_start.
14574 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
14575 including a two-value parser.
14576 * target.def (print_patchable_function_entry): New target hook.
14577 * targhooks.h (default_print_patchable_function_entry): New function.
14578 * targhooks.c (default_print_patchable_function_entry): Likewise.
14579 * toplev.c (process_options): Switch off IPA-RA if
14580 patchable function entries are being generated.
14581 * varasm.c (assemble_start_function): Look at the
14582 patchable-function-entry command line switch and current
14583 function attributes and maybe generate NOP instructions by
14584 calling the print_patchable_function_entry hook.
14585 * doc/extend.texi: Document patchable_function_entry attribute.
14586 * doc/invoke.texi: Document -fpatchable_function_entry
14587 command line option.
14588 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
14590 * doc/tm.texi: Re-generate.
14592 2017-07-25 Jakub Jelinek <jakub@redhat.com>
14595 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
14596 TARGET_AVX512DQ rather than TARGET_AVX512BW.
14598 2017-07-25 Tamar Christina <tamar.christina@arm.com>
14600 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
14602 2017-07-25 Richard Biener <rguenther@suse.de>
14604 PR tree-optimization/81455
14605 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
14606 not walk in cycles when looking for guards.
14608 2017-07-25 Richard Biener <rguenther@suse.de>
14610 PR tree-optimization/81529
14611 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
14612 when optimizing backedge uses.
14614 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
14616 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
14618 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
14619 to dl_section_ref. On AIX, append an expression to subtract
14620 the size of the section length to dl_section_ref.
14622 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
14624 * configure.ac: If any of the config.* scripts fail, exit 1.
14625 * configure: Regenerate.
14627 2017-07-25 Richard Biener <rguenther@suse.de>
14629 PR middle-end/81546
14630 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
14631 of immediate uses, be more verbose on errors.
14633 2017-07-25 Richard Biener <rguenther@suse.de>
14635 PR tree-optimization/81510
14636 * tree-vect-loop.c (vect_is_simple_reduction): When the
14637 reduction stmt is not inside the loop bail out.
14639 2017-07-25 Richard Biener <rguenther@suse.de>
14641 PR tree-optimization/81303
14642 * tree-vect-loop-manip.c (vect_loop_versioning): Build
14643 profitability check against LOOP_VINFO_NITERSM1.
14645 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
14647 * domwalk.c (cmp_bb_postorder): Simplify.
14648 (sort_bbs_postorder): New function. Use it...
14649 (dom_walker::walk): ...here to optimize common cases.
14651 2017-07-25 Martin Liska <mliska@suse.cz>
14654 * ipa-visibility.c (function_and_variable_visibility): Make the
14655 redirection just on target that supports aliasing.
14656 Fix GNU coding style.
14658 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
14661 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
14663 * config/arm/rtems.h: Likewise.
14664 * config/bfin/rtems.h: Likewise.
14665 * config/i386/rtemself.h: Likewise.
14666 * config/lm32/rtems.h: Likewise.
14667 * config/m32c/rtems.h: Likewise.
14668 * config/m68k/rtemself.h: Likewise.
14669 * config/microblaze/rtems.h: Likewise.
14670 * config/mips/rtems.h: Likewise.
14671 * config/moxie/rtems.h: Likewise.
14672 * config/nios2/rtems.h: Likewise.
14673 * config/powerpcspe/rtems.h: Likewise.
14674 * config/rs6000/rtems.h: Likewise.
14675 * config/rtems.h: Likewise.
14676 * config/sh/rtems.h: Likewise.
14677 * config/sh/rtemself.h: Likewise.
14678 * config/sparc/rtemself.h: Likewise.
14680 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
14683 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
14684 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
14685 * tree-ssa-structalias.c (alias_get_name): Same.
14687 2017-07-25 Bin Cheng <bin.cheng@arm.com>
14690 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
14691 instructions if no du chain is found.
14693 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
14695 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
14697 2017-07-25 Richard Biener <rguenther@suse.de>
14699 PR middle-end/81505
14700 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
14703 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
14705 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
14706 upper-regs options.
14707 (ISA_2_7_MASKS_SERVER): Likewise.
14708 (ISA_3_0_MASKS_IEEE): Likewise.
14709 (OTHER_P8_VECTOR_MASKS): Likewise.
14710 (OTHER_VSX_VECTOR_MASKS): Likewise.
14711 (POWERPC_MASKS): Likewise.
14712 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
14713 duplicate list of options.
14714 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
14715 explicit -mupper-regs options.
14716 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
14717 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
14718 alias for -mupper-regs-df.
14719 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
14720 (rs6000_init_hard_regno_mode_ok): Likewise.
14721 (rs6000_option_override_internal): Likewise.
14722 (rs6000_opt_masks): Likewise.
14723 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
14724 options in terms of whether -mvsx or -mpower8-vector was used.
14725 (TARGET_UPPER_REGS_DI): Likewise.
14726 (TARGET_UPPER_REGS_SF): Likewise.
14727 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
14728 -mupper-regs-* options.
14730 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
14732 * passes.c (emergency_dump_function): Print some empty lines and a
14733 header before the RTL dump.
14735 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
14737 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
14739 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
14742 * config/aarch64/aarch64.c (aarch64_classify_symbol):
14743 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
14745 2017-07-24 Carl Love <cel@us.ibm.com>
14747 * config/rs6000/rs6000-c.c: Add support for built-in functions
14748 vector float vec_extract_fp32_from_shorth (vector unsigned short);
14749 vector float vec_extract_fp32_from_shortl (vector unsigned short);
14750 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
14751 vec_extract_fp_from_shortl): Add defines for the two builtins.
14752 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
14753 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
14755 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
14756 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
14757 * doc/extend.texi: Update the built-in documentation file for the
14758 new built-in function.
14760 2017-07-24 Jakub Jelinek <jakub@redhat.com>
14763 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
14765 * doc/generic.texi: Likewise.
14766 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
14767 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
14769 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
14771 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
14772 (aarch64_mls_elt_merge<mode>): Likewise.
14774 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
14776 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
14777 having __cxa_atexit.
14779 2017-07-23 Michael Collison <michael.collison@arm.com>
14781 * config/arm/arm.c (arm_option_override): Deprecate
14782 use of -mstructure-size-boundary.
14783 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
14784 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
14786 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14789 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
14790 Reduce cost estimate for direct moves.
14792 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
14795 * config/i386/i386.c (ix86_option_override_internal): Disable
14796 BMI, BMI2 and TBM instructions for -m16.
14798 2017-07-21 Carl Love <cel@us.ibm.com>
14800 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14801 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14802 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14803 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14804 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14805 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
14806 VMULOSW): New enum "unspec" values.
14807 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
14808 altivec_vmulosw): New patterns.
14809 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14810 VMULOSW): Add definitions.
14812 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
14814 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
14815 (qdf24xx): Likewise.
14816 * config/aarch64/aarch64-options-extensions.def (rdma); New.
14817 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
14818 (AARCH64_FL_V8_1): Renumber.
14819 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
14820 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
14821 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
14822 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
14823 rdma to feature modifiers list.
14825 2017-07-21 Yury Gribov <tetra2005@gmail.com>
14827 PR middle-end/56727
14828 * ipa-visibility (function_and_variable_visibility): Convert
14829 recursive PLT call to direct call if appropriate.
14831 2017-07-21 Andrew Pinski <apinski@cavium.com>
14833 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
14834 operand 1 to see if the types precision matches.
14835 * fold-const.c (operand_equal_p): Likewise.
14837 2017-07-21 Richard Biener <rguenther@suse.de>
14839 PR tree-optimization/81303
14840 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
14841 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
14842 (vect_peeling_hash_get_lowest_cost): Adjust.
14843 (vect_enhance_data_refs_alignment): Likewise. Use
14844 vect_get_peeling_costs_all_drs to compute the penalty for no
14845 peeling to match up costs.
14847 2017-07-21 Richard Biener <rguenther@suse.de>
14849 PR tree-optimization/81500
14850 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
14851 we didn't identify a reduction path.
14853 2017-07-21 Tom de Vries <tom@codesourcery.com>
14854 Cesar Philippidis <cesar@codesourcery.com>
14856 PR gcov-profile/81442
14857 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
14860 2017-07-21 Tom de Vries <tom@codesourcery.com>
14863 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
14865 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
14866 nvptx_override_options_after_change.
14868 2017-07-21 Ulrich Drepper <drepper@redhat.com>
14870 * dwarf2out.c (output_file_names): Avoid double testing for
14871 dwarf_version >= 5.
14873 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
14875 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
14877 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
14879 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
14881 (try_crossjump_to_edge): Do not punt on partitioned functions.
14883 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
14885 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
14886 Put all BBs reachable only via paths crossing cold region to cold
14888 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
14890 2016-07-21 Richard Biener <rguenther@suse.de>
14892 PR tree-optimization/81303
14893 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
14894 into account prologue and epilogue iterations when raising
14895 min_profitable_iters to sth at least covering one vector iteration.
14897 2017-07-21 Tamar Christina <tamar.christina@arm.com>
14899 * config/arm/arm.c (arm_test_cpu_arch_dat):
14902 2017-07-20 Nathan Sidwell <nathan@acm.org>
14904 Remove TYPE_METHODS.
14905 * tree.h (TYPE_METHODS): Delete.
14906 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
14907 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
14908 (dbxout_type_methods): Scan TYPE_FIELDS.
14909 (dbxout_type): Don't check TYPE_METHODS here.
14910 * function.c (use_register_for_decl): Always ignore register for
14911 class types when not optimizing.
14912 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
14913 * tree.c (free_lang_data_in_type): Stitch out member functions and
14914 templates from TYPE_FIELDS.
14915 (build_distinct_type_copy, verify_type_variant,
14916 verify_type): Member fns are on TYPE_FIELDS.
14917 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
14918 * tree-pretty-print.c (dump_generic_node): Likewise.
14920 2017-07-20 Jakub Jelinek <jakub@redhat.com>
14923 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
14924 V2TImode and V4TImode.
14925 (ix86_expand_vector_extract): Likewise.
14926 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
14927 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
14928 (ssescalarmode): Handle V4TImode and V2TImode.
14929 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
14930 (*vec_extractv2ti, *vec_extractv4ti): New insns.
14931 (VEXTRACTI128_MODE): New mode iterator.
14932 (splitter for *vec_extractv?ti first element): New.
14933 (VEC_INIT_MODE): New mode iterator.
14934 (vec_init<mode>): Consolidate 3 expanders into one using
14935 VEC_INIT_MODE mode iterator.
14937 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
14939 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
14940 non_spilled_static_chain_regno_p.
14942 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
14944 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
14946 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
14948 * bb-reorder.c (connect_traces): Allow copying of blocks within
14951 2017-07-20 Richard Biener <rguenther@suse.de>
14953 * gimple.h (gimple_phi_result): Add gphi * overload.
14954 (gimple_phi_result_ptr): Likewise.
14955 (gimple_phi_arg): Likewise. Adjust index assert to only
14956 allow actual argument accesses rather than all slots available
14958 (gimple_phi_arg_def): Add gphi * overload.
14959 * tree-phinodes.c (make_phi_node): Initialize only actual
14961 (resize_phi_node): Clear memory not covered by old node,
14962 do not initialize excess argument slots.
14963 (reserve_phi_args_for_new_edge): Initialize new argument slot
14966 2017-07-20 Bin Cheng <bin.cheng@arm.com>
14968 PR tree-optimization/81388
14970 2016-07-21 Bin Cheng <bin.cheng@arm.com>
14972 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
14973 by removing computation of may_be_zero.
14975 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
14976 Tom de Vries <tom@codesourcery.com>
14978 PR middle-end/81030
14979 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
14980 when gimple level profile disagrees with what RTL expander did.
14982 2017-07-20 Richard Biener <rguenther@suse.de>
14984 PR tree-optimization/61171
14985 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
14986 (vect_analyze_stmt): Add slp instance parameter.
14987 (vectorizable_reduction): Likewise.
14988 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
14989 (vect_is_simple_reduction): Deal with chains not detected
14990 as SLP reduction chain, specifically not properly associated
14991 chains containing a mix of plus/minus.
14992 (get_reduction_op): Remove.
14993 (get_initial_defs_for_reduction): Simplify, pass in whether
14994 this is a reduction chain, pass in the SLP node for the PHIs.
14995 (vect_create_epilog_for_reduction): Get the SLP instance as
14997 (vectorizable_reduction): Get the SLP instance as arg.
14998 During analysis remember the SLP node with the PHIs in the
14999 instance. Simplify getting at the vectorized reduction PHIs.
15000 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
15001 through SLP instance.
15002 (vect_slp_analyze_operations): Likewise.
15003 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
15004 (vect_transform_stmt): Likewise.
15006 2017-07-20 Tom de Vries <tom@codesourcery.com>
15008 PR tree-optimization/81489
15009 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
15010 read of phi arg location to before loop that modifies phi.
15012 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15014 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
15017 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15019 PR middle-end/81331
15020 * except.c (execute): Fix ordering issue.
15022 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
15024 PR rtl-optimization/81423
15025 * combine.c (make_compound_operation_int): Don't try to optimize
15026 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
15028 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
15030 PR rtl-optimization/81423
15031 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
15032 with a constant that is -1 in the truncated to mode.
15034 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15036 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
15037 (determine_unlikely_bbs): ... here.
15038 * predict.h (propagate_unlikely_bbs_forward): Declare.
15039 * cfgexpand.c (pass_expand::execute): Use it.
15040 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
15042 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
15043 propagate_unlikely_bbs_forward.
15045 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15047 PR middle-end/81331
15048 * except.c (maybe_add_nop_after_section_switch): New function.
15051 2017-07-19 Tom de Vries <tom@codesourcery.com>
15053 * gimple.h (gimple_phi_set_arg): Make assert more strict.
15055 2017-07-19 Tom de Vries <tom@codesourcery.com>
15057 * gimple.h (gimple_phi_arg): Make assert more strict.
15059 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
15061 * config.gcc (powerpc*-*-*): Add mmintrin.h.
15062 * config/rs6000/mmintrin.h: New file.
15063 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
15065 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15067 PR tree-optimization/81346
15068 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
15070 2017-07-19 Tom de Vries <tom@codesourcery.com>
15072 * config/nvptx/nvptx.md (VECIM): Add V2DI.
15074 2017-07-19 Tom de Vries <tom@codesourcery.com>
15076 * config/nvptx/nvptx-modes.def: Add V2DImode.
15077 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
15078 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
15079 (nvptx_output_mov_insn): Handle lack of mov.b128.
15080 (nvptx_print_operand): Handle 'H' and 'L' codes.
15081 (nvptx_vector_mode_supported): Allow V2DImode.
15082 (nvptx_preferred_simd_mode): New function.
15083 (nvptx_data_alignment): New function.
15084 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
15085 nvptx_preferred_simd_mode.
15086 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
15088 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
15090 2017-07-19 Tom de Vries <tom@codesourcery.com>
15092 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
15093 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
15094 (nvptx_vector_mode_supported): New function. Allow V2SImode.
15095 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
15096 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
15097 (mov<VECIM>_insn): New define_insn.
15098 (define_expand "mov<VECIM>): New define_expand.
15100 2017-07-19 Tom de Vries <tom@codesourcery.com>
15102 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
15104 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15106 PR tree-optimization/81346
15107 * fold-const.h (fold_div_compare, range_check_type): Declare.
15108 * fold-const.c (range_check_type): New function.
15109 (build_range_check): Use range_check_type.
15110 (fold_div_compare): No longer static, rewritten into
15111 a match.pd helper function.
15112 (fold_comparison): Don't call fold_div_compare here.
15113 * match.pd (X / C1 op C2): New optimization using fold_div_compare
15114 as helper function.
15116 2017-07-19 Nathan Sidwell <nathan@acm.org>
15118 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
15119 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
15120 * tree.c (find_decls_types_r, verify_type): Use
15121 TYPE_{MIN,MAX}_VALUE_RAW.
15122 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
15123 (hash_tree): Likewise.
15124 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
15126 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
15129 2017-07-18 Tom de Vries <tom@codesourcery.com>
15131 PR middle-end/81464
15132 * omp-expand.c (expand_omp_for_static_chunk): Handle
15133 equal-argument loop exit phi.
15135 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
15138 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
15139 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
15140 operand 2 predicate.
15141 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
15142 operand 2 predicate.
15143 (ror,rol -> rorx splitters): Use const_int_operand as
15144 operand 2 predicate.
15146 2017-06-18 Richard Biener <rguenther@suse.de>
15148 PR tree-optimization/81410
15149 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
15150 the gap in the ! slp_perm SLP case after each group.
15152 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15154 PR middle-end/81463
15155 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
15158 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15160 PR middle-end/81462
15161 * predict.c (set_even_probabilities): Cleanup; do not affect
15162 probabilities that are already known.
15163 (combine_predictions_for_bb): Call even when count is set.
15165 2017-07-18 Nathan Sidwell <nathan@acm.org>
15167 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
15170 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15173 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
15174 optimization for loop niter analysis.
15176 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
15179 * config/avr/avr.c (avr_optimize_casesi): Don't use
15180 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
15182 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
15184 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
15185 body_cost_vec from _vect_peel_extended_info.
15186 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
15187 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
15190 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15192 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
15194 2017-07-18 Richard Biener <rguenther@suse.de>
15196 PR tree-optimization/80620
15197 PR tree-optimization/81403
15198 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
15199 info when re-using a VN table entry.
15201 2017-07-18 Richard Biener <rguenther@suse.de>
15203 PR tree-optimization/81418
15204 * tree-vect-loop.c (vectorizable_reduction): Properly compute
15205 vectype_in. Verify that with lane-reducing reduction operations
15206 we have a single def-use cycle.
15208 2017-07-17 Carl Love <cel@us.ibm.com>
15210 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
15212 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15213 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15214 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15215 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15216 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15217 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15218 VMULOSW): New enum "unspec" values.
15219 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
15220 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
15221 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15222 altivec_vmulosw): New patterns.
15223 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15224 VMULOSW): Add definitions.
15225 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
15227 * config/alpha/alpha.c: Include predict.h.
15229 2017-07-17 Yury Gribov <tetra2005@gmail.com>
15231 * tree-vrp.c (compare_assert_loc): Fix comparison function
15232 to return predictable results.
15234 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15236 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
15238 (subdi3): Likewise.
15239 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
15240 * doc/invoke.texi (mexpand-adddi): Update text.
15242 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15244 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
15245 that also clobbers the CC register. The old expand code is moved
15247 (*arc_clzsi2): ... here.
15248 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
15249 the CC register. The old expand code is moved to ...
15250 (arc_ctzsi2): ... here.
15252 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15254 * config/arc/arc.opt (mindexed-loads): Use initial value
15255 TARGET_INDEXED_LOADS_DEFAULT.
15256 (mauto-modify-reg): Use initial value
15257 TARGET_AUTO_MODIFY_REG_DEFAULT.
15258 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15259 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15260 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15261 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15263 2017-07-17 Martin Liska <mliska@suse.cz>
15266 * opts.c (finish_options): Do not allow -fgnu-tm
15267 w/ -fsanitize={kernel-,}address. Say sorry.
15269 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15272 * tree-loop-distribution.c (classify_partition): Only assert on
15273 numer of iterations.
15274 (merge_dep_scc_partitions): Delete prameter. Update function call.
15275 (distribute_loop): Remove code handling loop with unknown niters.
15276 (pass_loop_distribution::execute): Skip loop with unknown niters.
15278 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15281 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
15282 function sort_partitions_by_post_order.
15284 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15286 PR tree-optimization/81374
15287 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
15288 the max index of basic blocks, rather than number of basic blocks.
15290 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15292 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
15294 (arc_legitimate_pic_operand_p): Likewise.
15295 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
15297 (arc_needs_pcl_p): Likewise.
15298 (arc_legitimate_pc_offset_p): Likewise.
15299 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
15300 function is also used in constrains.md.
15301 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
15302 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
15303 PLUS. Only return true/false in known cases, otherwise assert.
15304 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
15305 is already called in arc_legitimate_constant_p.
15306 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
15308 (LEGITIMATE_PIC_OPERAND_P): Use
15309 arc_raw_symbolic_reference_mentioned_p function.
15310 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
15315 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15316 Andrew Burgess <andrew.burgess@embecosm.com>
15318 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
15319 (arc_return_address_register): New function.
15320 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
15321 (arc_handle_fndecl_attribute): Add naked attribute.
15322 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15323 (TARGET_WARN_FUNC_RETURN): Likewise.
15324 (arc_allocate_stack_slots_for_args): New function.
15325 (arc_warn_func_return): Likewise.
15326 (machine_function): Change type fn_type.
15327 (arc_compute_function_type): Consider new naked function type,
15328 change function return type.
15329 (arc_must_save_register): Adapt to handle new
15330 arc_compute_function_type's return type.
15331 (arc_expand_prologue): Likewise.
15332 (arc_expand_epilogue): Likewise.
15333 (arc_return_address_regs): Delete.
15334 (arc_return_address_register): New function.
15335 (arc_epilogue_uses): Use above function.
15336 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
15337 (arc_function_type): Change encoding, add naked type.
15338 (ARC_INTERRUPT_P): Change to handle the new encoding.
15339 (ARC_FAST_INTERRUPT_P): Likewise.
15340 (ARC_NORMAL_P): Define.
15341 (ARC_NAKED_P): Likewise.
15342 (arc_compute_function_type): Delete prototype.
15343 * config/arc/arc.md (in_ret_delay_slot): Use
15344 arc_return_address_register function.
15345 (simple_return): Likewise.
15346 (p_return_i): Likewise.
15348 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15350 PR tree-optimization/81428
15351 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
15352 can't be built for those types.
15354 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
15356 Remove stuff dead since r239246.
15358 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
15359 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
15360 (avr_inform_devices): Remove dead stuff.
15362 2017-07-17 Tamar Christina <tamar.christina@arm.com>
15364 * config/arm/arm_neon.h: Fix softp typo.
15366 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15368 PR tree-optimization/81365
15369 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
15370 aggregate moves onto bb predecessor edges, make sure there are no
15371 loads that could alias the lhs in between the start of bb and the
15374 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
15377 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
15378 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
15379 [LSHIFTRT, outer_code = TRUNCATE]: Same.
15381 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15383 PR tree-optimization/81396
15384 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
15385 (init_symbolic_number): Initialize it to 1.
15386 (perform_symbolic_merge): Add n_ops from both operands into the new
15388 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
15389 without base_addr as useless if they need more than one operation.
15390 (bswap_replace): Handle !bswap case for NULL base_addr.
15392 2017-07-17 Tom de Vries <tom@codesourcery.com>
15395 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
15398 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
15400 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
15401 conditional builtin define __FIX_LEON3FT_B2BST.
15403 2017-07-17 Daniel Cederman <cederman@gaisler.com>
15405 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
15406 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
15409 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
15411 PR rtl-optimization/81424
15412 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
15413 to remove potential trapping from operands if -fnon-call-exceptions.
15415 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15417 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
15418 profile_proability for scalling.
15419 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
15421 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15423 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
15425 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15427 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
15428 fixpoint arithmetics.
15430 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15432 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
15433 fixpoint arithmetics.
15435 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15437 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
15438 fixpoint arithmetics.
15440 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15442 * profile-count.h (profile_probability::from_reg_br_prob_note,
15443 profile_probability::to_reg_br_prob_note): New functions.
15444 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
15445 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
15446 * predict.c (probability_reliable_p): Update.
15447 (edge_probability_reliable_p): Update.
15448 (br_prob_note_reliable_p): Update.
15449 (invert_br_probabilities): Update.
15450 (add_reg_br_prob_note): New function.
15451 (combine_predictions_for_insn): Update.
15452 * asan.c (asan_clear_shadow): Update.
15453 * cfgbuild.c (compute_outgoing_frequencies): Update.
15454 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
15455 (update_br_prob_note): Update.
15456 (rtl_verify_edges): Update.
15457 (purge_dead_edges): Update.
15458 (fixup_reorder_chain): Update.
15459 * emit-rtl.c (try_split): Update.
15460 * ifcvt.c (cond_exec_process_insns): Update.
15461 (cond_exec_process_if_block): Update.
15462 (dead_or_predicable): Update.
15463 * internal-fn.c (expand_addsub_overflow): Update.
15464 (expand_neg_overflow): Update.
15465 (expand_mul_overflow): Update.
15466 * loop-doloop.c (doloop_modify): Update.
15467 * loop-unroll.c (compare_and_jump_seq): Update.
15468 * optabs.c (emit_cmp_and_jump_insn_1): Update.
15469 * predict.h: Update.
15470 * reorg.c (mostly_true_jump): Update.
15472 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
15473 * config/alpha/alpha.c (emit_unlikely_jump): Update.
15474 * config/arc/arc.c: (emit_unlikely_jump): Update.
15475 * config/arm/arm.c: (emit_unlikely_jump): Update.
15476 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
15477 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
15478 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
15479 (ix86_print_operand): Update.
15480 (ix86_split_fp_branch): Update.
15481 (predict_jump): Update.
15482 * config/ia64/ia64.c (ia64_print_operand): Update.
15483 * config/mmix/mmix.c (mmix_print_operand): Update.
15484 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
15485 (rs6000_expand_split_stack_prologue): Update.
15486 * config/rs6000/rs6000.c: Update.
15487 * config/s390/s390.c (s390_expand_vec_strlen): Update.
15488 (s390_expand_vec_movstr): Update.
15489 (s390_expand_cs_tdsi): Update.
15490 (s390_expand_split_stack_prologue): Update.
15491 * config/sh/sh.c (sh_print_operand): Update.
15492 (expand_cbranchsi4): Update.
15493 (expand_cbranchdi4): Update.
15494 * config/sparc/sparc.c (output_v9branch): Update.
15495 * config/spu/spu.c (get_branch_target): Update.
15496 (ea_load_store_inline): Update.
15497 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
15498 * config/tilepro/tilepro.c: Update.
15500 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
15502 * gimplify.c (mostly_copy_tree_r): Revert latest change.
15503 (gimplify_save_expr): Likewise.
15505 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15507 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
15509 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15511 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
15513 * timevar.def (TV_IPA_FNSUMMARY): Define.
15515 2017-07-16 Daniel Cederman <cederman@gaisler.com>
15517 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
15518 to back store errata sensitive sequence from being generated.
15519 (sqrtdf2_fix): Likewise.
15521 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15523 * tree-ssa-threadupdate.c (compute_path_counts,
15524 update_joiner_offpath_counts): Use profile_probability.
15526 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
15529 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15531 * config/arm/arm-c.c (arm_cpu_builtins): Define
15532 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
15534 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
15536 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15537 array entries to represent __ieee128 versions of the
15538 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
15539 scalar_extract_sig, and scalar_insert_exp built-in functions.
15540 (altivec_resolve_overloaded_builtin): Add special case handling
15541 for the __builtin_scalar_insert_exp function, as represented by
15542 the P9V_BUILTIN_VEC_VSIEDP constant.
15543 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
15544 exponent support for __ieee128 argument.
15545 (VSESQP): Add scalar extract signature support for __ieee128
15547 (VSTDCNQP): Add scalar test negative support for __ieee128
15549 (VSIEQP): Add scalar insert exponent support for __int128 argument
15550 with __ieee128 result.
15551 (VSIEQPF): Add scalar insert exponent support for __ieee128
15552 argument with __ieee128 result.
15553 (VSTDCQP): Add scalar test data class support for __ieee128
15555 (VSTDCNQP): Add overload support for scalar test negative with
15556 __ieee128 argument.
15557 (VSTDCQP): Add overload support for scalar test data class
15558 __ieee128 argument.
15559 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
15560 UNSPEC_VSX_SXSIGDP.
15561 (UNSPEC_VSX_SIEXPQP): New constant.
15562 (xsxexpqp): New insn for VSX scalar extract exponent quad
15564 (xsxsigqp): New insn for VSX scalar extract significand quad
15566 (xsiexpqpf): New insn for VSX scalar insert exponent quad
15567 precision with floating point argument.
15568 (xststdcqp): New expand for VSX scalar test data class quad
15570 (xststdcnegqp): New expand for VSX scalar test negative quad
15572 (xststdcqp): New insn to match expansions for VSX scalar test data
15573 class quad precision and VSX scalar test negative quad precision.
15574 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
15575 special case operand checking to enforce that second operand of
15576 VSX scalar test data class with quad precision argument is a 7-bit
15578 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
15579 prototypes and descriptions of __ieee128 versions of
15580 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
15581 scalar_test_data_class, and scalar_test_neg built-in functions.
15583 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15585 PR tree-optimization/81162
15586 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
15587 replace a negate with an add.
15589 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
15591 * doc/invoke.texi (arm/-mcpu): Document +crypto.
15593 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15595 * config/arm/arm-c.c (arm_cpu_builtins): Define
15596 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
15598 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15600 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
15601 (armv8-r): Set ARM Cortex-R52 as default CPU.
15602 * config/arm/arm-tables.opt: Regenerate.
15603 * config/arm/arm-tune.md: Regenerate.
15604 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
15606 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
15607 extension for -mcpu=cortex-r52.
15609 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15611 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
15612 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
15613 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
15614 (fp-armv8): Define it as FP_ARMv8 only.
15615 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
15616 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
15618 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
15619 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
15620 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
15621 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
15622 than TARGET_FPU_ARMV8.
15623 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
15624 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
15625 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
15627 * config/arm/neon.md (neon_vrint): Likewise.
15628 (neon_vcvt): Likewise.
15629 (neon_<fmaxmin_op><mode>): Likewise.
15630 (<fmaxmin><mode>3): Likewise.
15631 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
15632 * config/arm/predicates.md (arm_cond_move_operator): Check against
15633 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
15635 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
15637 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
15638 to top of function.
15640 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15642 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
15643 loop in comment with memset.
15645 2017-07-14 Martin Liska <mliska@suse.cz>
15647 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
15648 * dwarf2out.c (is_java): Remove the function.
15649 (output_pubname): Remove usage of the function.
15650 (lower_bound_default): Remove usage of DW_LANG_Java.
15651 (gen_compile_unit_die): Likewise.
15652 * gcc.c: Remove compiler defaults for .java and .zip files.
15653 * gimple-expr.c (remove_suffix): Change as there's no longer
15654 extension than 4-letter one.
15655 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
15656 (gimplify_save_expr): Likewise.
15657 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
15658 as it's possible even for other languages than Java.
15659 * langhooks.h (struct lang_hooks): Remove Java from a comment.
15660 * lto-opts.c (lto_write_options): Remove reference to Java.
15661 * opts.c (strip_off_ending): Update file extension handling.
15662 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
15663 * tree-eh.c (lower_resx): Likewise.
15664 * tree.c (free_lang_data_in_type): Remove dead code.
15665 (find_decls_types_r): Likewise.
15666 (build_common_builtin_nodes): Remove Java from a comment.
15667 (verify_type): Remove dead code.
15668 * varasm.c (assemble_external): Remove Java from a comment.
15670 2017-07-14 Martin Liska <mliska@suse.cz>
15672 * opts.c (finish_options): Add quotes.
15673 (common_handle_option): Likewise.
15675 2017-07-14 Martin Liska <mliska@suse.cz>
15677 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
15678 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
15679 Remove N_SO_PASCAL.
15680 * dwarf2out.c (lower_bound_default): Do not handle
15682 (gen_compile_unit_die): Likewise.
15683 * gcc.c: Remove default extension binding for GNU Pascal.
15684 * stmt.c: Remove Pascal language from a comment.
15685 * xcoffout.c: Likewise.
15687 2017-07-13 David Malcolm <dmalcolm@redhat.com>
15690 * diagnostic-show-locus.c (fixit_cmp): New function.
15691 (layout::layout): Sort m_fixit_hints.
15692 (column_range::column_range): Assert that the values are valid.
15693 (struct char_span): New struct.
15694 (correction::overwrite): New method.
15695 (struct source_line): New struct.
15696 (line_corrections::add_hint): Add assertions. Reimplement memcpy
15697 calls in terms of classes source_line and char_span, and
15698 correction::overwrite.
15699 (selftest::test_overlapped_fixit_printing_2): New function.
15700 (selftest::diagnostic_show_locus_c_tests): Call it.
15702 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
15704 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
15705 early if there is no lhs.
15707 2017-07-13 Martin Liska <mliska@suse.cz>
15709 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
15710 (gen_reference_type_die): Likewise.
15711 * stor-layout.c: Remove Pascal-related comment.
15713 2017-07-13 Martin Liska <mliska@suse.cz>
15715 * opts.c (finish_options): Add quotes to error messages.
15716 (parse_sanitizer_options): Likewise.
15718 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
15720 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
15722 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
15724 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
15726 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
15728 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
15730 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
15732 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
15735 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
15736 provides the hardware capability bits, define the macro
15737 __BUILTIN_CPU_SUPPORTS__.
15738 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
15739 if GLIBC does not provide the hardware capability bits. Add a
15740 gcc_unreachable call if the built-in cpu function is neither
15741 __builtin_cpu_is nor __builtin_cpu_supports.
15742 (rs6000_get_function_versions_dispatcher): Change the warning
15743 that an old GLIBC is used which does not export the capability
15744 bits to be an error.
15745 * doc/extend.texi (target_clones attribute): Document the
15746 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
15747 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
15748 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
15749 the macros defined by GCC if the newer GLIBC is available.
15751 2017-07-12 Jeff Law <law@redhat.com>
15753 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
15754 remaining includes slightly.
15755 * config/riscv/riscv-builtins.c: Include profile-count.h.
15757 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
15760 * config/avr/avr.c (avr_set_current_function): In diagnostic
15761 messages: Quote keywords and (parts of) identifiers.
15762 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
15765 2017-07-12 Carl Love <cel@us.ibm.com>
15767 * config/rs6000/rs6000-c.c: Add support for built-in functions
15768 vector bool char vec_revb (vector bool char);
15769 vector bool short vec_revb (vector short char);
15770 vector bool int vec_revb (vector bool int);
15771 vector bool long long vec_revb (vector bool long long);
15772 * doc/extend.texi: Update the built-in documentation file for the
15773 new built-in functions.
15775 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15777 * config/s390/s390.md: Remove movcc splitter.
15779 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15781 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
15782 load/store on condition.
15784 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
15787 * config/avr/avr.c (avr_encode_section_info)
15788 [progmem && !TREE_READONLY]: Error if progmem object needs
15791 2017-07-11 Michael Collison <michael.collison@arm.com>
15793 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
15796 2017-07-11 Carl Love <cel@us.ibm.com>
15798 * config/rs6000/rs6000-c.c: Add support for builtins
15799 vector unsigned int vec_parity_lsbb (vector signed int);
15800 vector unsigned int vec_parity_lsbb (vector unsigned int);
15801 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
15802 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
15803 vector unsigned long long vec_parity_lsbb (vector signed long long);
15804 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
15805 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
15806 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
15807 * doc/extend.texi: Update the built-in documentation file for the
15808 new built-in functions.
15810 2017-07-11 David Malcolm <dmalcolm@redhat.com>
15812 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
15813 (layout::m_primary_loc): New field.
15814 (layout::layout): Initialize new field. Move location filtering
15815 logic from here to...
15816 (layout::maybe_add_location_range): ...this new method. Add
15817 support for filtering to just the lines already specified by other
15819 (layout::will_show_line_p): New method.
15820 (gcc_rich_location::add_location_if_nearby): New method.
15821 (selftest::test_add_location_if_nearby): New test function.
15822 (selftest::diagnostic_show_locus_c_tests): Call it.
15823 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
15826 2017-07-11 Tom de Vries <tom@codesourcery.com>
15828 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
15829 (bb_first_real_insn): New function.
15830 (nvptx_single): Add extra initialization of broadcasted condition
15833 2017-07-11 Nathan Sidwell <nathan@acm.org>
15835 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
15837 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
15839 * doc/extend.texi (AVR Function Attributes): Remove weblink to
15840 Binutils doc as TEXI will mess them up.
15841 * doc/invoke.texi (AVR Options): Same here.
15843 2017-07-11 Daniel Cederman <cederman@gaisler.com>
15845 * config/sparc/sparc.opt (mfix-ut700): New option.
15846 (mfix-gr712rc): Likewise.
15847 (sparc_fix_b2bst): New variable.
15848 * doc/invoke.texi (SPARC options): Document them.
15849 (ARM options): Fix warnings.
15850 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
15851 instructions to prevent sequences that can trigger the store-store
15852 errata for certain LEON3FT processors.
15853 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
15854 (sparc_option_override): Set sparc_fix_b2bst appropriately.
15855 * config/sparc/sparc.md (fix_b2bst): New attribute.
15856 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
15858 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
15861 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
15863 (*rsqrtsf2_sse): Ditto.
15865 (div<mode>3): Macroize insn from divdf3 and divsf3
15866 using MODEF mode iterator.
15868 2017-07-10 Martin Sebor <msebor@redhat.com>
15870 PR tree-optimization/80397
15871 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
15872 instead of testing for equality to INTEGER_TYPE.
15874 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
15876 * config.gcc: Remove uclibc from arc target spec.
15878 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
15880 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
15882 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15885 * lto-wrapper.c (remove_option): New function.
15886 (merge_and_complain): Merge PIC/PIE options more realistically.
15888 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
15890 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
15894 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
15895 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
15896 * config.in: Regenerate.
15897 * configure: Regenerate.
15898 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
15899 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
15900 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
15901 (TARGET_GASISR_PROLOGUES): ...target mask.
15902 * common/config/avr/avr-common.c
15903 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
15904 Set -mgas-isr-prologues.
15905 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
15906 INSERT_PASS_BEFORE for it.
15907 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
15908 * config/avr/avr.c (avr_option_override)
15909 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
15910 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
15911 (avr_attribute_table) <no_gccisr>: Add new function attribute.
15912 (avr_set_current_function) <is_no_gccisr>: Init machine field.
15913 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
15915 (make_avr_pass_pre_proep): New function.
15916 (emit_push_sfr) <treg>: Add argument to function and use it
15917 instead of TMP_REG.
15918 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
15919 and set machine->gasisr.yes.
15920 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
15921 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
15922 __gcc_isr.n_pushed to .L__stack_usage.
15923 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
15924 (avr_asm_final_postscan_insn): ...this new static function.
15925 * config/avr/avr.h (machine_function)
15926 <is_no_gccisr, use_L__stack_usage>: New fields.
15927 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
15928 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
15929 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
15930 (gasisr, *gasisr): New expander and insn.
15931 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
15932 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
15933 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
15935 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
15937 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
15940 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
15942 Move jump-tables out of .text again.
15945 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
15946 (ASM_OUTPUT_ADDR_VEC): New function.
15947 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
15948 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
15949 INSN_ADDRESSes as asm comment.
15950 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
15951 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
15952 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
15953 * config/avr/avr.md (*tablejump): Adjust comment.
15954 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
15955 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
15957 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
15958 (avr_output_addr_vec): New proto.
15959 (avr_log_t) <insn_addresses>: New field.
15961 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
15964 * config/i386/i386.c (ix86_function_arg_advance): Set
15965 outgoing_args_on_stack to true if there are outgoing arguments
15967 (ix86_function_arg): Likewise.
15968 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
15969 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
15970 * config/i386/i386.h (machine_function): Add
15971 outgoing_args_on_stack.
15973 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
15975 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
15976 supporting pthreds.
15977 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
15979 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
15981 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
15982 (REAL_H): Remove $(MACHMODE_H).
15983 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
15985 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
15986 $(MACHMODE_H) and double-int.h.
15987 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
15989 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
15992 2017-07-07 Andrew Pinski <apinski@cavium.com>
15994 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
15995 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
15997 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15999 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
16000 Add warning if GCC was not configured to link against a GLIBC that
16001 exports the hardware capability bits.
16002 (make_resolver_func): Make resolver function private and not a
16003 COMDAT function. Create the name with clone_function_name instead
16004 of make_unique_name.
16007 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
16008 correct operand in doing the split.
16010 2017-07-07 Carl Love <cel@us.ibm.com>
16012 * config/rs6000/rs6000-c: Add support for built-in function
16013 vector unsigned short vec_pack_to_short_fp32 (vector float,
16015 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
16016 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
16017 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
16018 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
16019 (convert_4f32_8i16): Add define_expand.
16020 * doc/extend.texi: Update the built-in documentation file for the
16021 new built-in function.
16023 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16025 * config/sparc/m8.md: New file.
16026 * config/sparc/sparc.md: Include m8.md.
16028 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16030 * config/sparc/sparc.opt: New option -mvis4b.
16031 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
16032 (sparc_option_override): Handle VIS4B.
16033 (enum sparc_builtins): Define
16034 SPARC_BUILTIN_DICTUNPACK{8,16,32},
16035 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
16036 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
16037 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
16038 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
16039 (check_constant_argument): New function.
16040 (sparc_vis_init_builtins): Define builtins
16041 __builtin_vis_dictunpack{8,16,32},
16042 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
16043 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
16044 __builtin_vis_fpcmpde{8,16,32}shl and
16045 __builtin_vis_fpcmpur{8,16,32}shl.
16046 (sparc_expand_builtin): Check that the constant operands to
16047 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
16048 constant and in range.
16049 * config/sparc/sparc-c.c (sparc_target_macros): Handle
16051 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
16052 (SPARC_IMM5_P): Likewise.
16053 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
16054 (enabled): Handle vis4b.
16055 (UNSPEC_DICTUNPACK): New unspec.
16056 (UNSPEC_FPCMPSHL): Likewise.
16057 (UNSPEC_FPUCMPSHL): Likewise.
16058 (UNSPEC_FPCMPDESHL): Likewise.
16059 (UNSPEC_FPCMPURSHL): Likewise.
16060 (cpu_feature): New CPU feature `vis4b'.
16061 (dictunpack{8,16,32}): New insns.
16062 (FPCSMODE): New mode iterator.
16063 (fpcscond): New code iterator.
16064 (fpcsucond): Likewise.
16065 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
16066 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
16067 (fpcmpde{8,16,32}{si,di}shl): Likewise.
16068 (fpcmpur{8,16,32}{si,di}shl): Likewise.
16069 * config/sparc/constraints.md: Define constraints `q' for unsigned
16070 2-bit integer constants and `t' for unsigned 5-bit integer
16072 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
16074 (imm5_operand_dictunpack16): Likewise.
16075 (imm5_operand_dictunpack32): Likewise.
16076 (imm2_operand): Likewise.
16077 * doc/invoke.texi (SPARC Options): Document -mvis4b.
16078 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
16079 ditunpack* and fpcmp*shl builtins.
16081 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16083 * config.gcc: Handle m8 in --with-{cpu,tune} options.
16084 * config.in: Add HAVE_AS_SPARC6 define.
16085 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
16087 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
16089 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
16090 (CPP_CPU_SPEC): Handle m8.
16091 (ASM_CPU_SPEC): Likewise.
16092 * config/sparc/sparc-opts.h (enum processor_type): Add
16094 * config/sparc/sparc.c (m8_costs): New struct.
16095 (sparc_option_override): Handle TARGET_CPU_m8.
16096 (sparc32_initialize_trampoline): Likewise.
16097 (sparc64_initialize_trampoline): Likewise.
16098 (sparc_issue_rate): Likewise.
16099 (sparc_register_move_cost): Likewise.
16100 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
16101 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
16102 (ASM_CPU64_DEFAULT_SPEC): Likewise.
16103 (CPP_CPU_SPEC): Handle M8.
16104 (ASM_CPU_SPEC): Likewise.
16105 (AS_M8_FLAG): Define.
16106 * config/sparc/sparc.md: Add m8 to the cpu attribute.
16107 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
16108 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
16110 * configure: Regenerate.
16111 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
16114 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16116 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
16118 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
16119 ("*movdi_insn_sp32"): Do not set v3pipe.
16120 ("*movsi_insn"): Likewise.
16121 ("*movdi_insn_sp64"): Likewise.
16122 ("*movsf_insn"): Likewise.
16123 ("*movdf_insn_sp32"): Likewise.
16124 ("*movdf_insn_sp64"): Likewise.
16125 ("*zero_extendsidi2_insn_sp64"): Likewise.
16126 ("*sign_extendsidi2_insn"): Likewise.
16127 ("*mov<VM32:mode>_insn"): Likewise.
16128 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16129 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16130 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16131 ("<vlop:code><VL:mode>3"): Likewise.
16132 ("*not_<vlop:code><VL:mode>3"): Likewise.
16133 ("*nand<VL:mode>_vis"): Likewise.
16134 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
16135 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
16136 ("one_cmpl<VL:mode>2"): Likewise.
16137 ("faligndata<VM64:mode>_vis"): Likewise.
16138 ("alignaddrsi_vis"): Likewise.
16139 ("alignaddrdi_vis"): Likweise.
16140 ("alignaddrlsi_vis"): Likewise.
16141 ("alignaddrldi_vis"): Likewise.
16142 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16143 ("bmaskdi_vis"): Likewise.
16144 ("bmasksi_vis"): Likewise.
16145 ("bshuffle<VM64:mode>_vis"): Likewise.
16146 ("cmask8<P:mode>_vis"): Likewise.
16147 ("cmask16<P:mode>_vis"): Likewise.
16148 ("cmask32<P:mode>_vis"): Likewise.
16149 ("pdistn<P:mode>_vis"): Likewise.
16150 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16152 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16154 * config/sparc/sparc.md ("subtype"): New insn attribute.
16155 ("*wrgsr_sp64"): Set insn subtype.
16156 ("*rdgsr_sp64"): Likewise.
16157 ("alignaddrsi_vis"): Likewise.
16158 ("alignaddrdi_vis"): Likewise.
16159 ("alignaddrlsi_vis"): Likewise.
16160 ("alignaddrldi_vis"): Likewise.
16161 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16162 ("fexpand_vis"): Likewise.
16163 ("fpmerge_vis"): Likewise.
16164 ("faligndata<VM64:mode>_vis"): Likewise.
16165 ("bshuffle<VM64:mode>_vis"): Likewise.
16166 ("cmask8<P:mode>_vis"): Likewise.
16167 ("cmask16<P:mode>_vis"): Likewise.
16168 ("cmask32<P:mode>_vis"): Likewise.
16169 ("fchksm16_vis"): Likewise.
16170 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
16171 ("fmean16_vis"): Likewise.
16172 ("fp<plusminus_insn>64_vis"): Likewise.
16173 ("<plusminus_insn>v8qi3"): Likewise.
16174 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16175 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
16176 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
16177 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
16178 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
16179 ("*movqi_insn"): Likewise.
16180 ("*movhi_insn"): Likewise.
16181 ("*movsi_insn"): Likewise.
16182 ("movsi_pic_gotdata_op"): Likewise.
16183 ("*movdi_insn_sp32"): Likewise.
16184 ("*movdi_insn_sp64"): Likewise.
16185 ("movdi_pic_gotdata_op"): Likewise.
16186 ("*movsf_insn"): Likewise.
16187 ("*movdf_insn_sp32"): Likewise.
16188 ("*movdf_insn_sp64"): Likewise.
16189 ("*zero_extendhisi2_insn"): Likewise.
16190 ("*zero_extendqihi2_insn"): Likewise.
16191 ("*zero_extendqisi2_insn"): Likewise.
16192 ("*zero_extendqidi2_insn"): Likewise.
16193 ("*zero_extendhidi2_insn"): Likewise.
16194 ("*zero_extendsidi2_insn_sp64"): Likewise.
16195 ("ldfsr"): Likewise.
16196 ("prefetch_64"): Likewise.
16197 ("prefetch_32"): Likewise.
16198 ("tie_ld32"): Likewise.
16199 ("tie_ld64"): Likewise.
16200 ("*tldo_ldub_sp32"): Likewise.
16201 ("*tldo_ldub1_sp32"): Likewise.
16202 ("*tldo_ldub2_sp32"): Likewise.
16203 ("*tldo_ldub_sp64"): Likewise.
16204 ("*tldo_ldub1_sp64"): Likewise.
16205 ("*tldo_ldub2_sp64"): Likewise.
16206 ("*tldo_ldub3_sp64"): Likewise.
16207 ("*tldo_lduh_sp32"): Likewise.
16208 ("*tldo_lduh1_sp32"): Likewise.
16209 ("*tldo_lduh_sp64"): Likewise.
16210 ("*tldo_lduh1_sp64"): Likewise.
16211 ("*tldo_lduh2_sp64"): Likewise.
16212 ("*tldo_lduw_sp32"): Likewise.
16213 ("*tldo_lduw_sp64"): Likewise.
16214 ("*tldo_lduw1_sp64"): Likewise.
16215 ("*tldo_ldx_sp64"): Likewise.
16216 ("*mov<VM32:mode>_insn"): Likewise.
16217 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16218 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16220 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16222 * config/sparc/sparc.md ("type"): New insn type viscmp.
16223 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
16225 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
16226 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
16227 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16228 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
16230 ("n7_vis_logical_11cycle"): Likewise.
16231 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
16232 * config/sparc/niagara2.md ("niag3_vis": Likewise.
16233 * config/sparc/niagara.md ("niag_vis"): Likewise.
16234 * config/sparc/ultra3.md ("us3_fga"): Likewise.
16235 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
16237 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16239 * config/sparc/sparc.md: New instruction type `bmask'.
16240 (bmaskdi_vis): Use the `bmask' type.
16241 (bmasksi_vis): Likewise.
16242 * config/sparc/ultra3.md (us3_array): Likewise.
16243 * config/sparc/niagara7.md (n7_array): Likewise.
16244 * config/sparc/niagara4.md (n4_array): Likewise.
16245 * config/sparc/niagara2.md (niag2_vis): Likewise.
16246 (niag3_vis): Likewise.
16247 * config/sparc/niagara.md (niag_vis): Likewise.
16249 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16251 * ipa-comdats.c: Remove optimize check from gate.
16252 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
16253 for functions not optimized.
16254 (ipa_fn_summary_read): Skip optimize check.
16255 (ipa_fn_summary_write): Likewise.
16256 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
16258 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
16260 (can_inline_edge_p): Check flag_pcc_struct_return for match.
16261 (check_callers): Give up on caller which is not optimized.
16262 (inline_small_functions): Likewise.
16263 (ipa_inline): Do not give up when not optimizing.
16264 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
16265 away unoptimizes cdtors.
16266 (whole_program_function_and_variable_visibility): Do
16267 ipa_discover_readonly_nonaddressable_vars in LTO mode.
16268 * ipa.c (process_references): Do not check optimize.
16269 (symbol_table::remove_unreachable_nodes): Update optimize check.
16270 (set_writeonly_bit): Update optimize check.
16271 (pass_ipa_cdtor_merge::gate): Do not check optimize.
16272 (pass_ipa_single_use::gate): Remove.
16274 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16276 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
16277 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
16278 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
16279 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
16280 permute_load, permute_store, adjust_extract, adjust_splat,
16281 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
16282 replace_swap_with_copy, dump_swap_insn_table,
16283 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
16284 recombine_lvx_pattern, recombine_stvx_pattern,
16285 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
16286 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
16287 to file rs6000-p8swap.c.
16288 * config/rs6000/rs6000-p8swap.c: New file.
16289 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
16290 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
16291 and rs6000*-*-* targets.
16293 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16295 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
16297 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16299 * lto-wrapper.c (merge_and_complain): Do not merge
16300 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
16301 fsigned_zeros, ftrapping_math, fwrapv.
16302 (append_compiler_options): Do not track these options.
16303 (append_linker_options): Likewie
16305 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16307 * cgraphunit.c (cgraph_node::finalize_function): When
16308 !flag_toplevel_reorde set no_reorder flag.
16309 (varpool_node::finalize_decl): Likewise.
16310 (symbol_table::compile): Drop no toplevel reorder path.
16312 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16314 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
16315 edges; zero probability is not better than uninitialized.
16317 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
16319 * asan.h (asan_sanitize_allocas_p): Declare.
16320 * asan.c (asan_sanitize_allocas_p): New function.
16321 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
16322 (handle_builtin_alloca): Likewise.
16323 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
16324 if !asan_sanitize_allocas_p.
16325 * params.def (asan-instrument-allocas): Add new option.
16326 * params.h (ASAN_PROTECT_ALLOCAS): Define.
16327 * opts.c (common_handle_option): Disable allocas sanitization for
16330 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
16332 * asan.c: Include gimple-fold.h.
16333 (get_last_alloca_addr): New function.
16334 (handle_builtin_stackrestore): Likewise.
16335 (handle_builtin_alloca): Likewise.
16336 (asan_emit_allocas_unpoison): Likewise.
16337 (get_mem_refs_of_builtin_call): Add new parameter, remove const
16338 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
16339 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
16340 (instrument_builtin_call): Pass gimple iterator to
16341 get_mem_refs_of_builtin_call.
16342 (last_alloca_addr): New global.
16343 * asan.h (asan_emit_allocas_unpoison): Declare.
16344 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
16345 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
16346 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
16347 if function calls alloca.
16348 * gimple-fold.c (replace_call_with_value): Remove static keyword.
16349 * gimple-fold.h (replace_call_with_value): Declare.
16350 * internal-fn.c: Include asan.h.
16351 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
16352 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
16354 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16356 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
16357 (C_SELFTEST_FLAGS): New.
16358 (CPP_SELFTEST_FLAGS): New.
16359 (SELFTEST_DEPS): New, from deps of s-selftest.
16360 (C_SELFTEST_DEPS): New, from deps of s-selftest.
16361 (CPP_SELFTEST_DEPS): New.
16362 (selftest): Add dependency on s-selftest-c++.
16363 (s-selftest): Rename to...
16364 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
16365 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
16366 than SELFTEST_FLAGS.
16367 (selftest-gdb): Rename to...
16368 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
16370 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
16371 (selftest-valgrind): Rename to...
16372 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
16374 (selftest-valgrind): Reintroduce as an alias for
16375 selftest-c-valgrind.
16376 (s-selftest-c++): New.
16377 (selftest-c++-gdb): New.
16378 (selftest-c++-valgrind): New.
16380 2017-07-06 Olivier Hainque <hainque@adacore.com>
16382 * gcc.c (process_command): When deciding if undefined variables
16383 should be ignored when processing specs, accept "gcc -v" as well.
16385 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16387 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
16388 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
16390 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
16392 * config/arm/arm-cpus.in (armv8-r): Add new entry.
16393 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
16394 * config/arm/arm-tables.opt: Regenerate.
16395 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
16397 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
16399 2017-07-06 Carl Love <cel@us.ibm.com>
16401 * ChangeLog: Clean up from mid air collision
16403 2017-07-06 Carl Love <cel@us.ibm.com>
16405 * config/rs6000/rs6000-c.c: Add support for built-in functions
16406 vector signed int vec_subc (vector signed int, vector signed int);
16407 vector signed __int128 vec_subc (vector signed __int128,
16408 vector signed __int128);
16409 vector unsigned __int128 vec_subc (vector unsigned __int128,
16410 vector unsigned __int128);
16411 vector signed int vec_sube (vector signed int, vector signed int,
16412 vector signed int);
16413 vector unsigned int vec_sube (vector unsigned int,
16414 vector unsigned int,
16415 vector unsigned int);
16416 vector signed __int128 vec_sube (vector signed __int128,
16417 vector signed __int128,
16418 vector signed__int128);
16419 vector unsigned __int128 vec_sube (vector unsigned __int128,
16420 vector unsigned __int128,
16421 vector unsigned __int128);
16422 vector signed int vec_subec (vector signed int, vector signed int,
16423 vector signed int);
16424 vector unsigned int vec_subec (vector unsigned int,
16425 vector unsigned int,
16426 vector unsigned int);
16427 vector signed __int128 vec_subec (vector signed __int128,
16428 vector signed __int128,
16429 vector signed__int128);
16430 vector unsigned __int128 vec_subec (vector unsigned __int128,
16431 vector unsigned __int128,
16432 vector unsigned __int128);
16433 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
16434 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
16435 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
16436 BU_ALTIVEC_OVERLOAD_X definitions.
16437 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
16438 * doc/extend.texi: Update the built-in documentation file for the new
16439 built-in functions.
16441 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16444 * diagnostic-show-locus.c (layout::layout): Use start and finish
16445 spelling location for the start and finish of each range.
16446 * genmatch.c (linemap_client_expand_location_to_spelling_point):
16447 Add unused aspect param.
16448 * input.c (expand_location_1): Add "aspect" param, and use it
16449 to access the correct part of the location.
16450 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
16452 (expand_location_to_spelling_point): Likewise.
16453 (linemap_client_expand_location_to_spelling_point): Add "aspect"
16454 param, and pass it to expand_location_1.
16456 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
16458 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
16459 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
16460 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
16461 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
16462 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
16463 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
16464 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
16465 _mm_maskz_getmant_ss): New intrinsics.
16466 (__builtin_ia32_getexpss128_mask): Changed to ...
16467 __builtin_ia32_getexpss128_round ... this.
16468 (__builtin_ia32_getexpsd128_mask): Changed to ...
16469 __builtin_ia32_getexpsd128_round ... this.
16470 * config/i386/i386-builtin-types.def
16471 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
16472 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
16473 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
16474 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
16475 __builtin_ia32_getmantss_mask_round): New builtins.
16476 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
16477 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
16478 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
16479 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
16480 * config/i386/sse.md
16481 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
16482 avx512f_sgetexp<mode><mask_scalar_name>
16483 <round_saeonly_scalar_name> ... this.
16484 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
16485 %0, %1, %2<round_saeonly_op3>}): Changed to ...
16486 vgetexp<ssescalarmodesuffix>
16487 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16488 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
16489 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
16490 avx512f_vgetmant<mode><mask_scalar_name>
16491 <round_saeonly_scalar_name> ... this.
16492 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
16493 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
16494 vgetmant<ssescalarmodesuffix>
16495 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
16496 %0<mask_scalar_operand4>, %1, %2
16497 <round_saeonly_scalar_mask_op4>, %3} ... this.
16498 * config/i386/subst.md (mask_scalar_operand4,
16499 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
16500 round_saeonly_scalar_nimm_predicate): New subst attributes.
16502 2017-07-06 Julia Koval <julia.koval@intel.com>
16504 * config/i386/i386.c (ix86_erase_embedded_rounding):
16505 Remove code for old rounding pattern.
16507 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
16509 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
16511 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
16513 * doc/sourcebuild.texi (Test Directives, Variants of
16514 dg-require-support): Add documentation for dg-require-stack-check.
16516 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
16518 * config/i386/subst.md (mask_scalar, round_scalar,
16519 round_saeonly_scalar): New meta-templates.
16520 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
16521 round_scalar_mask_operand3, round_scalar_mask_op3,
16522 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
16523 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
16524 round_saeonly_scalar_constraint,
16525 round_saeonly_scalar_prefix): New subst attribute.
16526 * config/i386/sse.md
16527 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
16528 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
16529 <round_scalar_name> ... this.
16530 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
16531 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
16532 <round_scalar_name> ... this.
16533 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
16534 <sse>_vm<code><mode>3<mask_scalar_name>
16535 <round_saeonly_scalar_name> ... this.
16536 (v<plusminus_mnemonic><ssescalarmodesuffix>
16537 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
16538 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
16539 v<plusminus_mnemonic><ssescalarmodesuffix>
16540 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16541 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
16542 (v<multdiv_mnemonic><ssescalarmodesuffix>
16543 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
16544 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
16545 v<multdiv_mnemonic><ssescalarmodesuffix>
16546 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16547 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
16548 (v<maxmin_float><ssescalarmodesuffix>
16549 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
16550 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
16551 v<maxmin_float><ssescalarmodesuffix>
16552 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16553 %0<mask_scalar_operand3>, %1, %<iptr>2
16554 <round_saeonly_scalar_mask_op3>} ... this.
16556 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
16558 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
16559 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
16561 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
16562 Alan Hayward <alan.hayward@arm.com>
16563 David Sherwood <david.sherwood@arm.com>
16565 * combine.c (simplify_if_then_else): Remove "enum" before
16567 * compare-elim.c (can_eliminate_compare): Likewise.
16568 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
16570 (aarch64_lookup_simd_builtin_type): Likewise.
16571 (aarch64_simd_builtin_type): Likewise.
16572 (aarch64_init_simd_builtin_types): Likewise.
16573 (aarch64_simd_expand_args): Likewise.
16574 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
16576 (aarch64_reverse_mask): Likewise.
16577 (aarch64_simd_emit_reg_reg_move): Likewise.
16578 (aarch64_gen_adjusted_ldpstp): Likewise.
16579 (aarch64_ccmp_mode_to_code): Likewise.
16580 (aarch64_operands_ok_for_ldpstp): Likewise.
16581 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16582 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
16584 (aarch64_min_divisions_for_recip_mul): Likewise.
16585 (aarch64_reassociation_width): Likewise.
16586 (aarch64_get_condition_code_1): Likewise.
16587 (aarch64_simd_emit_reg_reg_move): Likewise.
16588 (aarch64_simd_attr_length_rglist): Likewise.
16589 (aarch64_reverse_mask): Likewise.
16590 (aarch64_operands_ok_for_ldpstp): Likewise.
16591 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16592 (aarch64_gen_adjusted_ldpstp): Likewise.
16593 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
16595 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
16596 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
16597 (arm_lookup_simd_builtin_type): Likewise.
16598 (arm_simd_builtin_type): Likewise.
16599 (arm_init_simd_builtin_types): Likewise.
16600 (arm_expand_builtin_args): Likewise.
16601 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
16602 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
16603 (ft32_setup_incoming_varargs): Likewise.
16604 (ft32_function_arg): Likewise.
16605 (ft32_function_arg_advance): Likewise.
16606 (ft32_pass_by_reference): Likewise.
16607 (ft32_arg_partial_bytes): Likewise.
16608 (ft32_valid_pointer_mode): Likewise.
16609 (ft32_addr_space_pointer_mode): Likewise.
16610 (ft32_addr_space_legitimate_address_p): Likewise.
16611 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
16613 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
16614 (ix86_emit_outlined_ms2sysv_restore): Likewise.
16615 (iamcu_alignment): Likewise.
16616 (canonicalize_vector_int_perm): Likewise.
16617 (ix86_noce_conversion_profitable_p): Likewise.
16618 (ix86_mpx_bound_mode): Likewise.
16619 (ix86_operands_ok_for_move_multiple): Likewise.
16620 * config/microblaze/microblaze-protos.h
16621 (microblaze_expand_conditional_branch_reg): Likewise.
16622 * config/microblaze/microblaze.c
16623 (microblaze_expand_conditional_branch_reg): Likewise.
16624 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
16626 (rs6000_reassociation_width): Likewise.
16627 (rs6000_invalid_binary_op): Likewise.
16628 (fusion_p9_p): Likewise.
16629 (emit_fusion_p9_load): Likewise.
16630 (emit_fusion_p9_store): Likewise.
16631 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
16633 (riscv_hard_regno_mode_ok_p): Likewise.
16634 (riscv_address_insns): Likewise.
16635 (riscv_split_symbol): Likewise.
16636 (riscv_legitimize_move): Likewise.
16637 (riscv_function_value): Likewise.
16638 (riscv_hard_regno_nregs): Likewise.
16639 (riscv_expand_builtin): Likewise.
16640 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
16641 (riscv_build_integer): Likewise.
16642 (riscv_split_integer): Likewise.
16643 (riscv_legitimate_constant_p): Likewise.
16644 (riscv_cannot_force_const_mem): Likewise.
16645 (riscv_regno_mode_ok_for_base_p): Likewise.
16646 (riscv_valid_base_register_p): Likewise.
16647 (riscv_valid_offset_p): Likewise.
16648 (riscv_valid_lo_sum_p): Likewise.
16649 (riscv_classify_address): Likewise.
16650 (riscv_legitimate_address_p): Likewise.
16651 (riscv_address_insns): Likewise.
16652 (riscv_load_store_insns): Likewise.
16653 (riscv_force_binary): Likewise.
16654 (riscv_split_symbol): Likewise.
16655 (riscv_force_address): Likewise.
16656 (riscv_legitimize_address): Likewise.
16657 (riscv_move_integer): Likewise.
16658 (riscv_legitimize_const_move): Likewise.
16659 (riscv_legitimize_move): Likewise.
16660 (riscv_address_cost): Likewise.
16661 (riscv_subword): Likewise.
16662 (riscv_output_move): Likewise.
16663 (riscv_canonicalize_int_order_test): Likewise.
16664 (riscv_emit_int_order_test): Likewise.
16665 (riscv_function_arg_boundary): Likewise.
16666 (riscv_pass_mode_in_fpr_p): Likewise.
16667 (riscv_pass_fpr_single): Likewise.
16668 (riscv_pass_fpr_pair): Likewise.
16669 (riscv_get_arg_info): Likewise.
16670 (riscv_function_arg): Likewise.
16671 (riscv_function_arg_advance): Likewise.
16672 (riscv_arg_partial_bytes): Likewise.
16673 (riscv_function_value): Likewise.
16674 (riscv_pass_by_reference): Likewise.
16675 (riscv_setup_incoming_varargs): Likewise.
16676 (riscv_print_operand): Likewise.
16677 (riscv_elf_select_rtx_section): Likewise.
16678 (riscv_save_restore_reg): Likewise.
16679 (riscv_for_each_saved_reg): Likewise.
16680 (riscv_register_move_cost): Likewise.
16681 (riscv_hard_regno_mode_ok_p): Likewise.
16682 (riscv_hard_regno_nregs): Likewise.
16683 (riscv_class_max_nregs): Likewise.
16684 (riscv_memory_move_cost): Likewise.
16685 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
16686 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
16687 (rl78_addr_space_address_mode): Likewise.
16688 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16690 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
16691 (rs6000_reassociation_width): Likewise.
16692 (rs6000_invalid_binary_op): Likewise.
16693 (fusion_p9_p): Likewise.
16694 (emit_fusion_p9_load): Likewise.
16695 (emit_fusion_p9_store): Likewise.
16696 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
16697 (ok_for_simple_move_operands): Likewise.
16698 (ok_for_simple_move_strict_operands): Likewise.
16699 (ok_for_simple_arith_logic_operands): Likewise.
16700 (visium_legitimize_reload_address): Likewise.
16701 (visium_select_cc_mode): Likewise.
16702 (output_cbranch): Likewise.
16703 (visium_split_double_move): Likewise.
16704 (visium_expand_copysign): Likewise.
16705 (visium_expand_int_cstore): Likewise.
16706 (visium_expand_fp_cstore): Likewise.
16707 * config/visium/visium.c (visium_pass_by_reference): Likewise.
16708 (visium_function_arg): Likewise.
16709 (visium_function_arg_advance): Likewise.
16710 (visium_libcall_value): Likewise.
16711 (visium_setup_incoming_varargs): Likewise.
16712 (visium_legitimate_constant_p): Likewise.
16713 (visium_legitimate_address_p): Likewise.
16714 (visium_legitimize_address): Likewise.
16715 (visium_secondary_reload): Likewise.
16716 (visium_register_move_cost): Likewise.
16717 (visium_memory_move_cost): Likewise.
16718 (prepare_move_operands): Likewise.
16719 (ok_for_simple_move_operands): Likewise.
16720 (ok_for_simple_move_strict_operands): Likewise.
16721 (ok_for_simple_arith_logic_operands): Likewise.
16722 (visium_function_value_1): Likewise.
16723 (rtx_ok_for_offset_p): Likewise.
16724 (visium_legitimize_reload_address): Likewise.
16725 (visium_split_double_move): Likewise.
16726 (visium_expand_copysign): Likewise.
16727 (visium_expand_int_cstore): Likewise.
16728 (visium_expand_fp_cstore): Likewise.
16729 (visium_split_cstore): Likewise.
16730 (visium_select_cc_mode): Likewise.
16731 (visium_split_cbranch): Likewise.
16732 (output_cbranch): Likewise.
16733 (visium_print_operand_address): Likewise.
16734 * expmed.c (flip_storage_order): Likewise.
16735 * expmed.h (emit_cstore): Likewise.
16736 (flip_storage_order): Likewise.
16737 * genrecog.c (validate_pattern): Likewise.
16738 * hsa-gen.c (gen_hsa_addr): Likewise.
16739 * internal-fn.c (expand_arith_overflow): Likewise.
16740 * ira-color.c (allocno_copy_cost_saving): Likewise.
16741 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16742 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
16743 (process_invariant_for_inheritance): Likewise.
16744 * lra-eliminations.c (move_plus_up): Likewise.
16745 * omp-low.c (lower_oacc_reductions): Likewise.
16746 * simplify-rtx.c (simplify_subreg): Likewise.
16747 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
16748 (TARGET_CHKP_BOUND_MODE): Likewise..
16749 * targhooks.c (default_chkp_bound_mode): Likewise.
16750 (default_setup_incoming_vararg_bounds): Likewise.
16751 * targhooks.h (default_chkp_bound_mode): Likewise.
16752 (default_setup_incoming_vararg_bounds): Likewise.
16753 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
16754 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
16755 (have_whole_vector_shift): Likewise.
16756 * tree-vect-stmts.c (vectorizable_load): Likewise.
16757 * doc/tm.texi: Regenerate.
16759 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
16761 Graceful degrade if Binutils PR21472 is not available.
16764 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
16765 .rodata in flash test fails.
16766 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
16767 * confgure: Regenerate.
16768 * config.in: Regenerate.
16769 * config/avr/avr.c (avr_asm_named_section)
16770 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
16771 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
16772 (avr_asm_init_sections): Same.
16774 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16776 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
16777 (fma<VH:mode>4_intrinsic): Likewise.
16778 (*fmsub<VCVTF:mode>4): Likewise.
16779 (*fmsub<VH:mode>4_intrinsic): Likewise.
16781 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
16784 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
16785 Don't depend on "optimize > 0".
16786 (out_movhi_r_mr, out_movqi_mr_r): Same.
16787 (out_movhi_mr_r, out_movqi_r_mr): Same.
16788 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
16789 io_address_operand on "optimize > 0".
16791 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16793 * tree-loop-distribution.c: Add general explanantion on the pass.
16794 (generate_loops_for_partition): Mark distributed loop.
16795 (pg_add_dependence_edges): New parameter. Handle alias data
16796 dependence specially and record it in the parameter if asked.
16797 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
16798 (init_partition_graph_vertices, add_partition_graph_edge): New.
16799 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
16800 (free_partition_graph_vdata, build_partition_graph): New.
16801 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
16802 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
16803 (data_ref_segment_size, latch_dominated_by_data_ref): New.
16804 (compute_alias_check_pairs, version_loop_by_alias_check): New.
16805 (version_for_distribution_p, finalize_partitions): New.
16806 (distribute_loop): Handle alias data dependence specially. Factor
16807 out loop fusion code as functions and call these functions.
16809 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16811 * tree-loop-distribution.c (classify_partition): New parameter and
16812 better handle reduction statement.
16813 (rdg_build_partitions): Revise comment.
16814 (distribute_loop): Compute statements in all partitions and pass it
16815 to classify_partition.
16817 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16819 * tree-loop-distribution.c (enum partition_type): New.
16820 (struct partition): New field type.
16821 (partition_merge_into): Add parameter. Update partition type.
16822 (data_dep_in_cycle_p, update_type_for_merge): New functions.
16823 (build_rdg_partition_for_vertex): Compute partition type.
16824 (rdg_build_partitions): Dump partition type.
16825 (distribute_loop): Update calls to partition_merge_into.
16827 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16829 * tree-loop-distribution.c (struct ddr_hasher): New.
16830 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
16832 (classify_partition): Call get_data_dependence.
16833 (pg_add_dependence_edges): Ditto.
16834 (distribute_loop): Release data dependence hash table.
16836 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16838 * tree-loop-distribution.c (ref_base_address): Delete.
16839 (similar_memory_accesses): Rename ...
16840 (share_memory_accesses): ... to this. Check if partitions access
16841 the same memory reference.
16842 (distribute_loop): Call share_memory_accesses.
16844 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16846 * tree-loop-distribution.c (struct partition): New field recording
16847 its data reference.
16848 (partition_alloc, partition_free): Init and release data refs.
16849 (partition_merge_into): Merge data refs.
16850 (build_rdg_partition_for_vertex): Collect data refs for partition.
16851 (pg_add_dependence_edges): Change parameters from vector to bitmap.
16853 (distribute_loop): Remve data refs from vertice data of partition
16856 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16858 * tree-loop-distribution.c (params.h): Include header file.
16859 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
16860 (datarefs_vec): New global var.
16861 (create_rdg_vertices): Use datarefs_vec directly.
16862 (free_rdg): Don't free data references.
16863 (build_rdg): Update use. Don't free data references.
16864 (distribute_loop): Compute global variable for data references.
16865 Bail out if there are too many data references.
16867 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16869 * tree-loop-distribution.c (loop_nest): New global var.
16870 (build_rdg): Use loop directly, rather than loop nest.
16871 (pg_add_dependence_edges): Remove loop nest parameter. Use global
16873 (distribute_loop): Compute global variable loop nest. Update use.
16875 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16877 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
16878 (partition_merge_into): New parameter. Dump reason for fusion.
16879 (distribute_loop): Update use of partition_merge_into.
16881 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16883 * tree-loop-distribution.c (bb_top_order_index): New.
16884 (bb_top_order_index_size, bb_top_order_cmp): New.
16885 (stmts_from_loop): Use topological order.
16886 (pass_loop_distribution::execute): Compute and release topological
16887 order for basic blocks.
16889 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16891 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
16894 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16896 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
16897 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
16898 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
16899 * internal-fn.def (LOOP_DIST_ALIAS): New.
16900 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
16901 (fold_loop_internal_call): ... this.
16902 (vect_loop_dist_alias_call): New function.
16903 (set_uid_loop_bbs): Call fold_loop_internal_call.
16904 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
16907 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
16910 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
16911 Require dead FLAGS_REG at the beginning of a peephole.
16913 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
16916 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
16917 arguments in the call to __builtin_ia32_sbb_u32.
16918 (_subborrow_u64): Swap _X and _Y arguments in the call to
16919 __builtin_ia32_sbb_u64.
16921 2017-07-04 Jakub Jelinek <jakub@redhat.com>
16924 * tree-vrp.c (compare_assert_loc): Turn into a function template
16925 with stable template parameter. Only test if a->e is NULL,
16926 !a->e == !b->e has been verified already. Use e == NULL or
16927 e != NULL instead of e or ! e tests. If stable is true, don't use
16928 iterative_hash_expr, on the other side allow a or b or both NULL
16929 and sort the NULLs last.
16930 (process_assert_insertions): Sort using compare_assert_loc<false>
16931 instead of compare_assert_loc, later sort using
16932 compare_assert_loc<true> before calling process_assert_insertions_for
16933 in a loop. Use break instead of continue once seen NULL pointer.
16935 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
16937 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
16938 Cortex-R7 and Cortex-R8 processors.
16940 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
16942 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
16943 uninitialized while src is not.
16945 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
16947 * common/config/arm/arm-common.c: Adjust include path for
16949 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
16950 (arm-cpu.h): Create in build directory. Adjust dependency rules.
16951 (arm-cpu-data.h): Likewise.
16952 (arm-cpu-cdata.h): Likewise.
16953 * config/arm/arm-cpu.h: Delete.
16954 * config/arm/arm-cpu-cdata.h: Delete.
16955 * config/arm/arm-cpu-data.h: Delete.
16957 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
16959 * config/arm/arm-cpus.in (cortex-a55): New.
16960 (cortex-a75): Likewise.
16961 (cortex-a75.cortex-a55): Likewise.
16962 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
16964 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
16965 * config/arm/arm-cpu-cdata.h: Regenerate.
16966 * config/arm/arm-cpu-data.h: Regenerate.
16967 * config/arm/arm-cpu.h: Regenerate.
16968 * config/arm/arm-tables.opt: Regenerate.
16969 * config/arm/arm-tune.md: Regenerate.
16971 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
16973 * haifa-sched.c (sched_create_recovery_edges): Update profile.
16975 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
16977 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
16980 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
16982 PR tree-optimization/81292
16983 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
16984 full_string_p, also call adjust_related_strinfos if the adjustment
16985 is simple, otherwise invalidate related strinfos.
16987 2017-07-04 Martin Liska <mliska@suse.cz>
16990 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
16991 newly created variable as DECL_IGNORED_P.
16993 2017-07-04 Martin Liska <mliska@suse.cz>
16996 * ipa-inline.c (inline_small_functions):
16997 Use xstrdup_for_dump.
16999 2017-07-04 Tom de Vries <tom@codesourcery.com>
17001 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
17003 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
17006 * config/darwin.c (darwin_function_switched_text_sections):
17009 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
17011 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
17013 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17015 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
17017 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17019 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
17020 min_profitable_iters, and th as inclusive lower bounds.
17021 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
17022 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
17023 for min_profitable_iters and min_profitable_estimate.
17024 (vect_transform_loop): Treat th as an inclusive lower bound.
17025 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
17027 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
17030 * config/darwin.c (darwin_function_switched_text_sections):
17031 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
17032 in two pieces, and suppress the use of buf.
17034 2017-07-03 Nathan Sidwell <nathan@acm.org>
17036 * hash-table.h (hash_table_mod1): Fix indentation.
17038 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17040 PR middle-end/81290
17041 * predict.c (force_edge_cold): Be more careful about propagation
17043 * profile-count.h (profile_probability::guessed,
17044 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
17046 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
17048 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
17050 * doc/invoke.texi (rcpc architecture extension): Document it.
17052 2017-07-03 Richard Biener <rguenther@suse.de>
17054 PR tree-optimization/60510
17055 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
17056 the scalar reduction PHI and use it.
17057 (vectorizable_reduction): Properly guard the single_defuse_cycle
17058 path for non-SLP reduction chains where we cannot use it.
17059 Rework reduc_def/index and vector type deduction. Rework
17060 vector operand gathering during reduction op code-gen.
17061 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
17062 chains dissolve the chain and leave it to non-SLP reduction
17065 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17067 * tree-data-ref.h (dr_alignment): Declare.
17068 * tree-data-ref.c (dr_alignment): New function.
17069 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
17070 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
17072 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
17074 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17076 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
17077 and base_misalignment fields.
17078 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
17079 * tree-data-ref.c: Include builtins.h.
17080 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
17081 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
17082 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
17083 * tree-vect-data-refs.c: Include tree-cfg.h.
17084 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
17085 fields instead of calculating an alignment here.
17086 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
17087 innermost_loop_behavior fields.
17089 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17091 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
17093 (DR_STEP_ALIGNMENT): New macro.
17094 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
17095 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
17096 (create_data_ref): Print it.
17097 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
17098 to tell whether the step preserves vector (mis)alignment.
17099 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17100 Move the check for an integer step and generalise to all INTEGER_CST.
17101 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
17102 Print the outer step alignment.
17104 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17106 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
17107 with offset_alignment.
17108 (DR_ALIGNED_TO): Delete.
17109 (DR_OFFSET_ALIGNMENT): New macro.
17110 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
17111 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
17112 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
17113 (create_data_ref): Likewise.
17114 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17115 (vect_analyze_data_refs): Likewise.
17116 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
17117 creating dummy innermost behavior.
17119 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17121 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
17122 with a "innermost_loop_behavior *" and refeence tree.
17123 * tree-data-ref.c (dr_analyze_innermost): Likewise.
17124 (create_data_ref): Update call accordingly.
17125 * tree-predcom.c (find_looparound_phi): Likewise.
17127 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17129 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
17130 fields with dr_wrt_vec_loop.
17131 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
17132 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
17133 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
17134 (vect_dr_behavior): New function.
17135 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17136 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
17137 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
17138 track whether the step preserves the misalignment.
17139 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17140 Use vect_dr_behavior.
17141 (vect_setup_realignment): Update call accordingly.
17142 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
17143 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
17144 call to vect_create_addr_base_for_vector_ref.
17145 (vect_create_cond_for_align_checks): Likewise.
17146 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
17147 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
17148 (vect_recog_mask_conversion_pattern): Likewise.
17149 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
17150 (new_stmt_vec_info): Remove redundant zeroing.
17152 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17154 * common/config/arm/arm-common.c (arm_be8_option): New function.
17155 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
17156 (ISA_ARMv6): Add isa_bit_be8.
17157 * config/arm/arm.h (arm_be8_option): Add prototype.
17158 (BE8_SPEC_FUNCTION): New define.
17159 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
17160 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
17161 (mlittle-endian): Similarly.
17162 (mbe8, mbe32): New options.
17163 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
17164 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
17166 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17168 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
17170 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17172 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
17173 (cleanup_tree_cfg_bb): Use it.
17174 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
17176 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
17178 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17181 * loop-doloop.c (add_test): Update profile.
17183 2017-07-03 Martin Liska <mliska@suse.cz>
17186 * sanopt.c (rewrite_usage_of_param): New function.
17187 (sanitize_rewrite_addressable_params): Likewise.
17188 (pass_sanopt::execute): Call rewrite_usage_of_param.
17190 2017-07-03 Richard Biener <rguenther@suse.de>
17192 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
17193 back to using VIEW_CONVERT_EXPR.
17195 2017-07-03 Martin Liska <mliska@suse.cz>
17198 * doc/extend.texi: Document when a resolver function is
17199 generated for target_clones.
17201 2017-07-03 Martin Liska <mliska@suse.cz>
17203 * asan.c (asan_emit_stack_protection): Unpoison just red zones
17204 and shadow memory of auto variables which are subject of
17205 use-after-scope sanitization.
17206 (asan_expand_mark_ifn): Add do set only when is_poison.
17208 2016-07-03 Richard Biener <rguenther@suse.de>
17210 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
17212 (vect_force_simple_reduction): Record reduction def -> phi mapping.
17213 (vectorizable_reduction): Perform reduction PHI creation when
17214 visiting a reduction PHI and adjust and simplify code generation
17215 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
17216 (vect_transform_loop): Visit reduction PHIs.
17217 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
17218 defs into the SLP tree.
17219 (vect_build_slp_tree): Reduction defs terminate the recursion.
17220 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
17222 (vect_get_vec_defs_for_stmt_copy): Export.
17223 (vect_get_vec_defs): Likewise.
17224 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
17226 (vect_get_vec_defs_for_stmt_copy): Declare.
17227 (vect_get_vec_defs): Likewise.
17229 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17231 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
17232 parameter with a "loop" parameter and use it instead of the
17233 loop containing DR_STMT. Don't check simple_iv when doing
17234 BB analysis. Describe the two analysis modes in the comment.
17236 2017-07-03 Tom de Vries <tom@codesourcery.com>
17238 PR tree-optimization/69468
17239 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
17240 (find_same_succ_bb): Handle ignore_edge_flags.
17242 2017-07-03 Tom de Vries <tom@codesourcery.com>
17244 PR tree-optimization/81192
17245 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
17247 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
17249 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
17251 2017-07-03 Tom de Vries <tom@codesourcery.com>
17253 PR tree-optimization/81192
17254 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
17255 BB_SAME_SUCC (bb) == NULL.
17257 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17259 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
17262 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17264 * dumpfile.c: Include profile-count.h
17265 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
17267 (insert_cond_bb): Update profile.
17268 * tree-cfg.h (insert_cond_bb): Update prototype.
17269 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
17270 * tree-dump.c: Do not include tree-cfg.
17272 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17274 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
17276 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17278 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
17281 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17283 * tree-complex.c (expand_complex_div_wide): update profile.
17285 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17286 Alan Hayward <alan.hayward@arm.com>
17287 David Sherwood <david.sherwood@arm.com>
17289 * Makefile.in (MACHMODE_H): Remove insn-modes.h
17290 (CORETYPES_H): New define.
17291 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
17292 (insn-modes-inline.h, s-modes-inline-h): New rules.
17293 (generated_files): Add insn-modes-inline.h.
17294 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
17295 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
17296 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
17297 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
17298 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
17299 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
17300 (build/gencodes.o, build/genconditions.o): Likewise.
17301 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
17302 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
17303 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
17304 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
17305 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
17306 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
17307 * coretypes.h: Include everything up to real.h for generators.
17308 Include insn-modes.h first. Include wide-int-print.h after
17309 wide-int.h. Include insn-modes-inline.h and then machmode.h.
17310 * machmode.h: Don't include insn-modes.h here.
17311 * function-tests.c: Remove includes of signop.h, machmode.h,
17312 double-int.h and wide-int.h.
17314 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
17316 * optc-save-gen.awk: Likewise.
17317 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
17318 * godump.c: Remove include of wide-int-print.h.
17319 * pretty-print.h: Likewise.
17320 * wide-int-print.cc: Likewise.
17321 * wide-int.cc: Likewise.
17322 * hash-map-tests.c: Remove include of signop.h.
17323 * hash-set-tests.c: Likewise.
17324 * rtl-tests.c: Likewise.
17325 * mkconfig.sh: Remove include of machmode.h.
17326 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
17328 (emit_insn_modes_inline_h): ...this new function. Emit the code
17329 into an insn-modes-inline.h header file, adding appropriate
17330 include guards and end comments.
17331 (emit_insn_modes_c_header): Remove include of machmode.h.
17332 (emit_min_insn_modes_c_header): Include coretypes.h rather than
17334 (main): Handle -i flag and call emit_insn_modes_inline_h when
17337 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17339 * tree-ssa-strlen.c (strinfo): Rename the length field to
17340 nonzero_chars. Add a full_string_p field.
17341 (compare_nonzero_chars, zero_length_string_p): New functions.
17342 (get_addr_stridx): Add an offset_out parameter.
17343 Use compare_nonzero_chars.
17344 (get_stridx): Update accordingly. Use compare_nonzero_chars.
17345 (new_strinfo): Update after above changes to strinfo.
17346 (set_endptr_and_length): Set full_string_p.
17347 (get_string_length): Update after above changes to strinfo.
17348 (unshare_strinfo): Update call to new_strinfo.
17349 (maybe_invalidate): Likewise.
17350 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
17351 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
17352 as a uhwi instead of an shwi. Update after above changes to
17353 strinfo and new_strinfo.
17354 (zero_length_string): Assert that chainsi contains full strings.
17355 Use zero_length_string_p. Update call to new_strinfo.
17356 (adjust_related_strinfos): Update after above changes to strinfo.
17357 Copy full_string_p from origsi.
17358 (adjust_last_stmt): Use zero_length_string_p.
17359 (handle_builtin_strlen): Update after above changes to strinfo and
17360 new_strinfo. Install the lhs as the string length if the previous
17361 entry didn't describe a full string.
17362 (handle_builtin_strchr): Update after above changes to strinfo
17364 (handle_builtin_strcpy): Likewise.
17365 (handle_builtin_strcat): Likewise.
17366 (handle_builtin_malloc): Likewise.
17367 (handle_pointer_plus): Likewise.
17368 (handle_builtin_memcpy): Likewise. Track nonzero characters
17369 that aren't necessarily followed by a nul terminator.
17370 (handle_char_store): Likewise.
17372 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17374 PR tree-optimization/80769
17375 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
17376 for malloc and calloc. Document the new invariant that all related
17377 strinfos have delayed lengths or none do.
17378 (verify_related_strinfos): Move earlier in file.
17379 (set_endptr_and_length): New function, split out from...
17380 (get_string_length): ...here. Also set the lengths of related
17382 (zero_length_string): Assert that chainsi has known (rather than
17384 (adjust_related_strinfos): Likewise.
17386 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17388 PR tree-optimization/81136
17389 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
17390 assert that two references with the same misalignment have the same
17391 compile-time misalignment if those compile-time misalignments
17394 2017-07-01 Andi Kleen <ak@linux.intel.com>
17396 * print-tree.c (print_node): Print all attributes.
17398 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17400 * cfg.c (scale_bbs_frequencies): New function.
17401 * cfg.h (scale_bbs_frequencies): Declare it.
17402 * cfgloopanal.c (single_likely_exit): Cleanup.
17403 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
17405 (scale_loop_profile): Likewise.
17406 (loop_version): Likewise.
17407 (create_empty_loop_on_edge): Update.
17408 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
17409 scale_loop_frequencies, scale_loop_profile, loopify,
17410 loop_version): Update prototypes.
17411 * modulo-sched.c (sms_schedule): Update.
17412 * predict.c (unlikely_executed_edge_p): Also check probability.
17413 (probably_never_executed_edge_p): Fix typo.
17414 * tree-if-conv.c (version_loop_for_if_conversion): Update.
17415 * tree-parloops.c (gen_parallel_loop): Update.
17416 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
17417 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
17418 * tree-ssa-loop-split.c (split_loop): Update.
17419 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
17420 * tree-vect-loop-manip.c (vect_do_peeling): Update.
17421 (vect_loop_versioning): Update.
17422 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
17424 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17426 * trans-mem.c (split_bb_make_tm_edge): Update profile.
17428 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17430 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
17431 to keep profile consistent.
17433 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17435 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
17436 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
17437 * profile-count.h (max_safe_multiplier): Make unsigned.
17438 (profile_count::guessed_zero): New.
17440 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17442 * bb-reorder.c (fix_up_crossing_landing_pad,
17443 fix_crossing_conditional_branches): Use make_single_succ_edge
17444 to keep profile consistent.
17446 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17448 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
17451 2017-07-01 Jakub Jelinek <jakub@redhat.com>
17454 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
17455 the right scopes, make sure cond_jump isn't preserved between multiple
17456 iterations. Search for fallthru edge whenever there are 3+ edges and
17457 use find_fallthru_edge for it.
17459 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17461 Patch by Alexander Monakov <amonakov@ispras.ru>
17462 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
17463 probabilities consistently.
17465 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17467 * pa.c (pa_expand_compare_and_swap_loop): Update call of
17468 emit_cmp_and_jump_insns.
17470 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17473 * tree-inline.c (expand_call_inline): Combine profile statuses.
17475 2017-06-30 Andrew Pinski <apinski@cavium.com>
17477 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
17478 fold_stmt returned true.
17480 2017-06-30 Nathan Sidwell <nathan@acm.org>
17482 * ggc.h (empty_string): Delete.
17483 * cfgexpand.c (expand_asm_stmt): Use plain "".
17484 * optabs.c (expand_asm_memory_barrier): Likewise.
17485 * stringpool.c (empty_string): Delete.
17486 (digit_vector, digit_string): Delete.
17487 (ggc_alloc_string): Use plain "", don't optimize single digit
17488 strings. Use ggc_alloc_atomic.
17490 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
17492 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
17493 comparison set and one other set, use the cost of the non-comparison
17496 2017-06-30 Nathan Sidwell <nathan@acm.org>
17498 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
17501 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
17503 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
17504 loops. Remove now unneeded calls to gimple_switch_set_label() that
17505 just set removed labels to NULL_TREE.
17507 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
17509 * tree-ssanames.c (set_range_info_raw): Abstract from ...
17510 (set_range_info): ...here. Only call set_range_info_raw if domain
17512 (set_nonzero_bits): Call set_range_info_raw.
17513 * tree-ssanames.h (set_range_info_raw): New.
17515 2017-06-30 Jakub Jelinek <jakub@redhat.com>
17518 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
17519 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
17520 of nonimmediate_operand and <store_mask_constraint> instead of m
17521 for the input operand. For V8FI iterator, always split if input
17522 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
17523 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
17524 <store_mask_predicate> instead of register_operand and
17525 <store_mask_constraint> instead of v for the input operand. Make
17526 sure both operands aren't MEMs for if not <mask_applied>.
17528 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
17530 * lto-wrapper.c (copy_file) Close both file descriptors before
17533 2017-06-30 Martin Liska <mliska@suse.cz>
17536 * multiple_target.c (create_dispatcher_calls): Make ifunc
17537 also for function that don't have calls or are not referenced.
17539 2017-06-30 Richard Biener <rguenther@suse.de>
17541 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
17542 analyze the first scalar stmt. Move vector type computation
17543 for the BB case here from ...
17544 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
17545 live operation processing in the SLP case properly.
17547 2017-06-30 Richard Biener <rguenther@suse.de>
17549 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
17551 2017-06-30 Martin Liska <mliska@suse.cz>
17554 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
17555 before BUILT_IN_UNWIND_RESUME when ASAN is used.
17557 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
17559 * doc/invoke.texi (AArch64): Add missing options and remove redundant
17562 2017-06-30 Richard Biener <rguenther@suse.de>
17564 PR tree-optimization/81249
17565 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
17566 condition reduction result to original scalar type.
17568 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17570 * profile-count.h (enum profile_quality): Fix typos and whitespace
17573 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17575 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
17576 type for branch probabilities.
17578 2017-06-29 Julian Brown <julian@codesourcery.com>
17579 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17581 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
17582 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
17583 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
17584 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
17586 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17588 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
17589 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
17590 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
17591 CC usage from generic code to here.
17592 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
17593 CC usage into the target macros.
17595 2017-06-29 Maya Rashish <coypu@sdf.org>
17597 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
17600 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17602 * arm/arm-builtins.c: Include profile-count.h
17603 * except.c (sjlj_emit_function_enter): Use
17604 profile_probability::unlikely.
17606 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17608 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
17609 and tocrel_offset be pointer args rather than implicitly using
17611 (legitimate_constant_pool_address_p, rs6000_emit_move,
17612 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
17613 tocrel_offset and use in toc_relative_expr_p call.
17614 (print_operand, print_operand_address): Use static tocrel_base_oac
17615 and tocrel_offset_oac.
17616 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
17619 2017-06-29 Maya Rashish <coypu@sdf.org>
17621 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
17623 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
17625 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
17626 objects, take into account only the alignment of 'op0' and 'mode1' if
17629 2017-06-29 Steve Ellcey <sellcey@cavium.com>
17631 * ccmp.c (ccmp_tree_comparison_p): New function.
17632 (ccmp_candidate_p): Update to use above function.
17633 (get_compare_parts): New function.
17634 (expand_ccmp_next): Update to use new functions.
17635 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
17637 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
17638 take mode as argument.
17639 * ccmp.h (expand_ccmp_expr): Add mode as argument.
17640 * expr.c (expand_expr_real_1): Pass mode as argument.
17642 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
17644 * combine.c (combine_instructions): Print insns to dump_file, together
17647 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17649 * asan.c (asan_emit_stack_protection): Update.
17650 (create_cond_insert_point): Update.
17651 * auto-profile.c (afdo_propagate_circuit): Update.
17652 * basic-block.h (struct edge_def): Turn probability to
17653 profile_probability.
17654 (EDGE_FREQUENCY): Update.
17655 * bb-reorder.c (find_traces_1_round): Update.
17656 (better_edge_p): Update.
17657 (sanitize_hot_paths): Update.
17658 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
17659 (make_single_succ_edge): Update.
17660 (check_bb_profile): Update.
17661 (dump_edge_info): Update.
17662 (update_bb_profile_for_threading): Update.
17663 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
17664 probabilitycount to 0.
17665 * cfgbuild.c (compute_outgoing_frequencies): Update.
17666 * cfgcleanup.c (try_forward_edges): Update.
17667 (outgoing_edges_match): Update.
17668 (try_crossjump_to_edge): Update.
17669 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
17670 (expand_gimple_tailcall): Update.
17671 (construct_init_block): Use make_single_succ_edge.
17672 (construct_exit_block): Use make_single_succ_edge.
17673 * cfghooks.c (verify_flow_info): Update.
17674 (redirect_edge_succ_nodup): Update.
17675 (split_edge): Update.
17676 (account_profile_record): Update.
17677 * cfgloopanal.c (single_likely_exit): Update.
17678 * cfgloopmanip.c (scale_loop_profile): Update.
17679 (set_zero_probability): Remove.
17680 (duplicate_loop_to_header_edge): Update.
17681 * cfgloopmanip.h (loop_version): Update prototype.
17682 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
17683 (force_nonfallthru_and_redirect): Update.
17684 (update_br_prob_note): Update.
17685 (rtl_verify_edges): Update.
17686 (purge_dead_edges): Update.
17687 (rtl_lv_add_condition_to_bb): Update.
17688 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
17689 * cgraphunit.c (init_lowered_empty_function): Update.
17690 (cgraph_node::expand_thunk): Update.
17691 * cilk-common.c: Include profile-count.h
17692 * dojump.c (inv): Remove.
17693 (jumpifnot): Update.
17694 (jumpifnot_1): Update.
17695 (do_jump_1): Update.
17697 (do_jump_by_parts_greater_rtx): Update.
17698 (do_compare_rtx_and_jump): Update.
17699 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
17700 do_jump_1. do_compare_rtx_and_jump): Update prototype.
17701 * dwarf2cfi.c: Include profile-count.h
17702 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
17703 (sjlj_emit_dispatch_table): Likewise.
17704 * explow.c: Include profile-count.h
17705 * expmed.c (emit_store_flag_force): Update.
17706 (do_cmp_and_jump): Update.
17707 * expr.c (compare_by_pieces_d::generate): Update.
17708 (compare_by_pieces_d::finish_mode): Update.
17709 (emit_block_move_via_loop): Update.
17710 (store_expr_with_bounds): Update.
17711 (store_constructor): Update.
17712 (expand_expr_real_2): Update.
17713 (expand_expr_real_1): Update.
17714 * expr.h (try_casesi, try_tablejump): Update prototypes.
17715 * gimple-pretty-print.c (dump_probability): Update.
17716 (dump_profile): New.
17717 (dump_gimple_label): Update.
17718 (dump_gimple_bb_header): Update.
17719 * graph.c (draw_cfg_node_succ_edges): Update.
17720 * hsa-gen.c (convert_switch_statements): Update.
17721 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
17722 (find_if_case_1): Update.
17723 (find_if_case_2): Update.
17724 * internal-fn.c (expand_arith_overflow_result_store): Update.
17725 (expand_addsub_overflow): Update.
17726 (expand_neg_overflow): Update.
17727 (expand_mul_overflow): Update.
17728 (expand_vector_ubsan_overflow): Update.
17729 * ipa-cp.c (good_cloning_opportunity_p): Update.
17730 * ipa-split.c (split_function): Use make_single_succ_edge.
17731 * ipa-utils.c (ipa_merge_profiles): Update.
17732 * loop-doloop.c (add_test): Update.
17733 (doloop_modify): Update.
17734 * loop-unroll.c (compare_and_jump_seq): Update.
17735 (unroll_loop_runtime_iterations): Update.
17736 * lra-constraints.c (lra_inheritance): Update.
17737 * lto-streamer-in.c (input_cfg): Update.
17738 * lto-streamer-out.c (output_cfg): Update.
17739 * mcf.c (adjust_cfg_counts): Update.
17740 * modulo-sched.c (sms_schedule): Update.
17741 * omp-expand.c (expand_omp_for_init_counts): Update.
17742 (extract_omp_for_update_vars): Update.
17743 (expand_omp_ordered_sink): Update.
17744 (expand_omp_for_ordered_loops): Update.
17745 (expand_omp_for_generic): Update.
17746 (expand_omp_for_static_nochunk): Update.
17747 (expand_omp_for_static_chunk): Update.
17748 (expand_cilk_for): Update.
17749 (expand_omp_simd): Update.
17750 (expand_omp_taskloop_for_outer): Update.
17751 (expand_omp_taskloop_for_inner): Update.
17752 * omp-simd-clone.c (simd_clone_adjust): Update.
17753 * optabs.c (expand_doubleword_shift): Update.
17754 (expand_abs): Update.
17755 (emit_cmp_and_jump_insn_1): Update.
17756 (expand_compare_and_swap_loop): Update.
17757 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
17758 * predict.c (predictable_edge_p): Update.
17759 (edge_probability_reliable_p): Update.
17760 (set_even_probabilities): Update.
17761 (combine_predictions_for_insn): Update.
17762 (combine_predictions_for_bb): Update.
17763 (propagate_freq): Update.
17764 (estimate_bb_frequencies): Update.
17765 (force_edge_cold): Update.
17766 * profile-count.c (profile_count::dump): Add missing space into dump.
17767 (profile_count::debug): Add newline.
17768 (profile_count::differs_from_p): Explicitly convert to unsigned.
17769 (profile_count::stream_in): Update.
17770 (profile_probability::dump): New member function.
17771 (profile_probability::debug): New member function.
17772 (profile_probability::differs_from_p): New member function.
17773 (profile_probability::differs_lot_from_p): New member function.
17774 (profile_probability::stream_in): New member function.
17775 (profile_probability::stream_out): New member function.
17776 * profile-count.h (profile_count_quality): Rename to ...
17777 (profile_quality): ... this one.
17778 (profile_probability): New.
17779 (profile_count): Update.
17780 * profile.c (compute_branch_probabilities): Update.
17781 * recog.c (peep2_attempt): Update.
17782 * sched-ebb.c (schedule_ebbs): Update.
17783 * sched-rgn.c (find_single_block_region): Update.
17784 (compute_dom_prob_ps): Update.
17785 (schedule_region): Update.
17786 * sel-sched-ir.c (compute_succs_info): Update.
17787 * stmt.c (struct case_node): Update.
17788 (do_jump_if_equal): Update.
17789 (get_outgoing_edge_probs): Update.
17790 (conditional_probability): Update.
17791 (emit_case_dispatch_table): Update.
17792 (expand_case): Update.
17793 (expand_sjlj_dispatch_table): Update.
17794 (emit_case_nodes): Update.
17795 * targhooks.c: Update.
17796 * tracer.c (better_p): Update.
17797 (find_best_successor): Update.
17798 * trans-mem.c (expand_transaction): Update.
17799 * tree-call-cdce.c: Update.
17800 * tree-cfg.c (gimple_split_edge): Upate.
17801 (move_sese_region_to_fn): Upate.
17802 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
17803 * tree-eh.c (lower_resx): Upate.
17804 (cleanup_empty_eh_move_lp): Upate.
17805 * tree-if-conv.c (version_loop_for_if_conversion): Update.
17806 * tree-inline.c (copy_edges_for_bb): Update.
17807 (copy_cfg_body): Update.
17808 * tree-parloops.c (gen_parallel_loop): Update.
17809 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
17810 (gimple_gen_time_profiler): Update.
17811 * tree-ssa-dce.c (remove_dead_stmt): Update.
17812 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
17813 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
17814 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
17815 (unloop_loops): Update.
17816 (try_peel_loop): Update.
17817 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
17818 * tree-ssa-loop-split.c (connect_loops): Update.
17819 (split_loop): Update.
17820 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
17821 (hoist_guard): Update.
17822 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
17823 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
17824 (value_replacement): Update.
17825 * tree-ssa-reassoc.c (branch_fixup): Update.
17826 * tree-ssa-tail-merge.c (replace_block_by): Update.
17827 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
17828 (create_edge_and_update_destination_phis): Update.
17829 (compute_path_counts): Update.
17830 (recompute_probabilities): Update.
17831 (update_joiner_offpath_counts): Update.
17832 (freqs_to_counts_path): Update.
17833 (duplicate_thread_path): Update.
17834 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
17835 (struct switch_conv_info): Update.
17836 (gen_inbound_check): Update.
17837 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
17838 (vect_do_peeling): Update.
17839 (vect_loop_versioning): Update.
17840 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
17841 (optimize_mask_stores): Update.
17842 * ubsan.c (ubsan_expand_null_ifn): Update.
17843 * value-prof.c (gimple_divmod_fixed_value): Update.
17844 (gimple_divmod_fixed_value_transform): Update.
17845 (gimple_mod_pow2): Update.
17846 (gimple_mod_pow2_value_transform): Update.
17847 (gimple_mod_subtract): Update.
17848 (gimple_mod_subtract_transform): Update.
17849 (gimple_ic): Update.
17850 (gimple_stringop_fixed_value): Update.
17851 (gimple_stringops_transform): Update.
17852 * value-prof.h: Update.
17854 2017-06-29 Carl Love <cel@us.ibm.com>
17856 * config/rs6000/rs6000-c.c: Add support for built-in functions
17857 vector signed int vec_signed (vector float);
17858 vector signed long long vec_signed (vector double);
17859 vector signed int vec_signed2 (vector double, vector double);
17860 vector signed int vec_signede (vector double);
17861 vector signed int vec_signedo (vector double);
17862 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
17863 instruction generator.
17864 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
17865 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
17866 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
17868 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
17869 vunsignede_v2df): Add define_expands.
17870 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
17871 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
17872 VEC_UNSIGNEDO): Add definitions.
17873 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
17874 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
17875 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
17876 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
17877 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
17878 * config/rs6000/altivec.h (vec_signed, vec_signed2,
17879 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
17880 vec_unsignede, vec_unsignedo): Add builtin defines.
17881 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
17883 * doc/extend.texi: Update the built-in documentation file for the
17884 new built-in functions.
17886 2017-06-29 Richard Biener <rguenther@suse.de>
17888 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
17889 reduction chains to LOOP_VINFO_REDUCTIONS.
17890 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
17891 SLP reductions after processing reduction chains.
17893 2017-06-29 Nathan Sidwell <nathan@acm.org>
17895 * builtins.c (fold_builtin_FUNCTION): Use
17896 lang_hooks.decl_printable_name.
17898 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
17900 PR middle-end/81194
17901 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
17902 with only one label.
17903 * stmt.c (expand_case): Assert NCASES is greater than one.
17905 2017-06-29 Richard Biener <rguenther@suse.de>
17907 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
17909 (group_case_labels): Likewise.
17910 (find_taken_edge): Push sanity checking on val to workers...
17911 (find_taken_edge_cond_expr): ... here
17912 (find_taken_edge_switch_expr): ... and here, handle cases
17913 with just a default label.
17914 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
17915 (group_case_labels): Likewise.
17916 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
17917 group_case_labels does anything cleanup the CFG again.
17919 2017-06-29 Bin Cheng <bin.cheng@arm.com>
17921 PR tree-optimization/81196
17922 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
17923 exit condition comparing two IVs.
17925 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
17927 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
17928 profile to the dummy entry at the end of the list of architectures.
17929 * config/arm/arm-cpu-cdata.h: Regenerated.
17931 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17932 Michael Collison <michael.collison@arm.com>
17935 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
17937 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
17938 (*aarch64_reg_<mode>3_minus_mask): New pattern.
17939 (*aarch64_<optab>_reg_di3_mask2): New pattern.
17940 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
17941 of shift when the shift amount is masked with constant equal to
17942 the size of the mode.
17943 * config/aarch64/predicates.md (subreg_lowpart_operator): New
17946 2017-06-29 Martin Liska <mliska@suse.cz>
17948 * config/i386/i386.opt: Change range from [1,5] to [0,5].
17950 2017-06-29 Yury Gribov <tetra2005@gmail.com>
17953 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
17956 (edge_growth_cache_entry::edge_growth_cache_entry): New
17958 (reset_edge_growth_cache): Update to use constructor.
17960 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17962 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
17963 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
17964 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
17966 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
17968 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
17969 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
17971 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
17973 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
17974 (*-linux-uclibc*): Add t-uclibc tmake_file.
17975 * config/t-musl: New.
17976 * config/t-uclibc: New.
17978 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
17980 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
17982 (gen_comm_data): Emit architectural setting of arch_prof.
17983 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
17985 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
17986 (armv8-m.base, armv8-m.main): Likewise.
17987 * arm-protos.h (arm_build_target): Add profile field.
17988 (arch_option): Likewise.
17989 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
17991 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
17992 arm_active_target.profile.
17994 2017-06-28 Richard Biener <rguenther@suse.de>
17996 PR middle-end/81227
17997 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
17998 TYPE_OVERFLOW_WRAPS.
17999 * match.pd (negate_expr_p): Likewise.
18000 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
18001 fold_build2, not fold_binary.
18003 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
18005 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
18006 Convert memory address to Pmode.
18007 (aarch64_print_operand): Assert MEM operands are always Pmode.
18009 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
18012 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
18013 Remove redundant if.
18014 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
18015 * config/arm/aarch-common-protos.h
18016 (aarch_forward_to_shift_is_not_shifted_re): Remove.
18017 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
18019 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
18022 * multiple_target.c (create_dispatcher_calls): Set the default
18023 clone to be static, not public.
18025 2017-06-28 Richard Biener <rguenther@suse.de>
18027 * tree-vect-loop.c (vectorizable_reduction): Move special
18028 cond reduction IV var creation ...
18029 (vect_create_epilog_for_reduction): ... here. Remove induction_index
18030 parameter. Use STMT_VINFO_VECTYPE.
18031 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
18034 2017-06-28 Martin Liska <mliska@suse.cz>
18037 * ipa-visibility.c (non_local_p): Handle visibility.
18039 2017-06-28 Martin Liska <mliska@suse.cz>
18042 * common.opt: Add IntegerRange to various options.
18043 * opt-functions.awk (integer_range_info): New function.
18044 * optc-gen.awk: Add integer_range_info to cl_options struct.
18045 * opts-common.c (decode_cmdline_option): Handle
18046 CL_ERR_INT_RANGE_ARG.
18047 (cmdline_handle_error): Likewise.
18048 * opts.c (print_filtered_help): Show valid interval in
18049 when --help is provided.
18050 * opts.h (struct cl_option): Add range_min and range_max fields.
18051 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
18053 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
18055 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
18056 (x * C EQ/NE y * C): New transformation.
18058 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
18060 * genmultilib (combination_space): Accept '+' in option names.
18062 2017-06-28 Martin Liska <mliska@suse.cz>
18065 * asan.c (instrument_derefs): Bail out inner references
18066 that are hard register variables.
18068 2017-06-28 Jakub Jelinek <jakub@redhat.com>
18071 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
18072 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
18074 2017-06-28 Richard Biener <rguenther@suse.de>
18076 * tree-vectorizer.h (vect_get_vec_defs): Remove.
18077 (vect_get_slp_defs): Adjust.
18078 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
18080 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
18082 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
18083 get_initial_defs_for_reduction instead of vect_get_vec_defs.
18084 (vectorizable_reduction): Adjust.
18085 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
18087 (vect_get_slp_defs): Likewise.
18088 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
18089 (vectorizable_bswap): Adjust.
18090 (vectorizable_call): Likewise.
18091 (vectorizable_conversion): Likewise.
18092 (vectorizable_assignment): Likewise.
18093 (vectorizable_shift): Likewise.
18094 (vectorizable_operation): Likewise.
18095 (vectorizable_store): Likewise.
18096 (vectorizable_condition): Likewise.
18097 (vectorizable_comparison): Likewise.
18099 2017-06-28 Michael Collison <michael.collison@arm.com>
18102 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
18104 (arm_gen_movmemqi): Removed unused variable 'i'.
18105 Convert 'for' loop into 'while' loop.
18106 (arm_expand_prologue): Remove last unnecessary set of insn.
18107 (thumb_pop): Remove unused variable 'pushed_words'.
18108 (thumb_exit): Remove last unnecessary set of regs_to_pop.
18110 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18112 * config/s390/predicates.md: Use s390_rel_address_ok_p.
18113 * config/s390/s390-protos.h: Add prototype of
18114 s390_rel_address_ok_p.
18115 * config/s390/s390.c (s390_got_symbol): New function.
18116 (s390_rel_address_ok_p): New function.
18117 (legitimize_pic_address): Use s390_rel_address_ok_p.
18118 (s390_load_got): Use s390_got_symbol.
18119 (s390_option_override): Issue error if
18120 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
18121 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
18123 * config/s390/s390.opt: New option mpic-data-is-text-relative.
18125 2017-06-27 Andrew Pinski <apinski@cavium.com>
18127 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
18128 (X * copysign (1.0, X)): New pattern.
18129 (X * copysign (1.0, -X)): New pattern.
18130 (copysign (-1.0, CST)): New pattern.
18132 2017-06-27 Joseph Myers <joseph@codesourcery.com>
18134 * genmultilib (combination_space): Remove variable.
18135 Validate reuse rules against regular expression for any sequence
18136 of multilib options in any order.
18138 2017-06-27 Michael Collison <michael.collison@arm.com>
18140 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
18141 call aarch64_split_simd_combine.
18142 * (aarch64_combine_internal<mode>): Delete pattern.
18143 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
18144 Allow register and subreg operands.
18146 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18148 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
18149 specific need, just fallback on defaults.
18150 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
18152 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18153 Olivier Hainque <hainque@adacore.com>
18155 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
18157 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
18158 targets. Pick a default if no particular attempt applied.
18159 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
18162 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18164 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
18165 (x86_64-wrs-vxworks7): Likewise.
18167 2017-06-27 Marek Polacek <polacek@redhat.com>
18170 * ubsan.c (instrument_null): Check get_base_address's result for null.
18172 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18174 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
18176 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18178 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
18179 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
18180 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
18181 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
18182 New function types.
18183 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
18184 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
18185 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
18186 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
18187 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
18188 BUILT_IN_FEUPDATEENV): New builtins.
18189 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
18190 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
18191 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
18192 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
18194 (builtin_structptr_types): Adjust size.
18195 * tree.c (builtin_structptr_types): Add four entries.
18197 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18198 Olivier Hainque <hainque@adacore.com>
18200 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
18201 (TLS_SYM): New local macro, forcing reference to __tls__ on
18202 link command lines for VxWorks 7 RTPs, triggering initialization
18204 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
18205 OS features TLS support, true for RTPs on VxWorks 7.
18206 * config/vxworks.c (vxworks_override_options): Setup emutls
18209 2017-06-27 Jakub Jelinek <jakub@redhat.com>
18211 * predict.c (test_prediction_value_range): Use -1U instead of -1
18212 to avoid narrowing conversion warning.
18213 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
18214 to avoid narrowing conversion warning.
18215 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
18217 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
18219 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18221 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
18222 64bit configurations.
18223 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
18224 (SIZE_TYPE): Likewise.
18225 * config/vxworks.c (vxworks_emutls_var_fields): Use
18226 long_unsigned_type_node instead of unsigned_type_node as the offset
18227 field type, which is "pointer" mode in emutls.c.
18229 2017-06-27 Jakub Jelinek <jakub@redhat.com>
18232 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
18234 PR middle-end/81207
18235 * gimple-fold.c (replace_call_with_call_and_fold): Handle
18236 gimple_vuse copying separately from gimple_vdef copying.
18238 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18240 * value-prof.c (free_hist): Remove call to memset and the enclosing if
18243 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18244 Olivier Hainque <hainque@adacore.com>
18246 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
18247 for all vxworks7 targets.
18248 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
18249 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
18250 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
18251 variations for VX6/VX7 and 32/64bits later on in ...
18252 (VXWORKS_LIB_SPEC): Leverage new macros.
18253 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
18254 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
18256 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18258 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
18259 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
18261 2017-06-26 Carl Love <cel@us.ibm.com>
18263 * config/rs6000/rs6000-c.c: Add support for built-in functions
18264 vector bool char vec_reve (vector bool char);
18265 vector signed char vec_reve (vector signed char);
18266 vector unsigned char vec_reve (vector unsigned char);
18267 vector bool int vec_reve (vector bool int);
18268 vector signed int vec_reve (vector signed int);
18269 vector unsigned int vec_reve (vector unsigned int);
18270 vector bool long long vec_reve (vector bool long long);
18271 vector signed long long vec_reve (vector signed long long);
18272 vector unsigned long long vec_reve (vector unsigned long long);
18273 vector bool short vec_reve (vector bool short);
18274 vector signed short vec_reve (vector signed short);
18275 vector double vec_reve (vector double);
18276 vector float vec_reve (vector float);
18277 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
18278 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
18279 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
18280 (altivec_vreve): New pattern.
18281 * config/rs6000/altivec.h (vec_reve): New define.
18282 * doc/extend.texi (vec_rev): Update the built-in documentation file
18283 for the new built-in functions.
18285 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18287 PR tree-optimization/71815
18288 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
18290 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
18292 (slsr_process_phi): Likewise.
18293 (replace_uncond_cands_and_profitable_phis): Don't replace a
18294 multiply candidate with a stride of 1 (copy or cast).
18295 (phi_incr_cost): Call uses_consumed_by_stmt rather than
18297 (lowest_cost_path): Likewise.
18298 (total_savings): Likewise.
18300 2017-06-26 Richard Biener <rguenther@suse.de>
18303 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
18304 Use def_builtin_pure for all gather builtins.
18306 2017-06-26 Richard Biener <rguenther@suse.de>
18308 PR tree-optimization/81203
18309 * tree-tailcall.c (find_tail_calls): Do not move stmts into
18310 non-dominating BBs.
18312 2017-06-26 Marek Polacek <polacek@redhat.com>
18315 * doc/invoke.texi: Document -Wmultistatement-macros.
18317 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
18319 * doc/sourcebuild.texi (ARM-specific attributes): Document new
18320 arm_neon_ok_no_float_abi effective target.
18322 2017-06-26 Richard Biener <rguenther@suse.de>
18324 PR tree-optimization/80928
18325 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
18326 (copy_bbs): Set BB_DUPLICATED flag early.
18327 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
18329 (execute_on_shrinking_pred): Likewise.
18330 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
18331 BB_DUPLICATED blocks.
18332 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
18333 iterate over all PHIs considering removal of *gsi.
18335 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
18337 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
18340 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18342 * config/rs6000/rs6000-string.c: (expand_block_clear,
18343 do_load_for_compare, select_block_compare_mode,
18344 compute_current_alignment, expand_block_compare,
18345 expand_strncmp_align_check, expand_strn_compare,
18346 expand_block_move, rs6000_output_load_multiple)
18347 Move functions related to string/block move/compare
18348 to a separate file.
18349 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
18350 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
18351 for this function which is now used in two files.
18352 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
18353 * config.gcc: Add rs6000-string.o to extra_objs for
18354 targets powerpc*-*-* and rs6000*-*-*.
18356 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
18359 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
18360 32-bit, since indexed is not valid for DImode.
18361 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
18362 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
18363 (define_peephole2 for Altivec d-form load): Add 32-bit support.
18364 (define_peephole2 for Altivec d-form store): Likewise.
18367 * multiple_target.c (create_dispatcher_calls): Only create the
18368 dispatcher call if the function is the default clone of a
18369 versioned function.
18371 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
18373 PR middle-end/80902
18374 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
18375 a call, force the call to not be a tail call.
18377 2017-06-23 Jeff Law <law@redhat.com>
18379 * doc/contrib.texi: Add entry for Steven Pemberton's work on
18382 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
18384 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
18385 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
18386 handling for early expansion of vector shifts (sl,sr,sra,rl).
18387 (builtin_function_type): Add vector shift right instructions
18388 to the unsigned argument list.
18390 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
18392 rtl-optimizatoin/79286
18393 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
18394 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
18395 trap. PIC register plus a const unspec without offset can never trap.
18397 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
18399 * tree.h (builtin_structptr_type): New type.
18400 (builtin_structptr_types): Declare new array.
18401 * tree.c (builtin_structptr_types): New array.
18402 (free_lang_data, build_common_tree_nodes): Use it.
18404 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
18407 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
18410 2017-06-22 Matt Turner <mattst88@gmail.com>
18412 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
18413 Lake models to skylake case. Assume skylake for unknown
18414 models with clflushopt.
18416 2017-06-22 Jeff Law <law@redhat.com>
18418 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
18419 frame sizes that do not satisfy aarch64_uimm12_shift.
18421 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
18423 * profile-count.h (apply_probability,
18424 apply_scale, probability_in): Fix checks for zero.
18426 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18428 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
18429 * doc/cppdiropts.texi (-I @var{dir}): Document it.
18431 2016-06-22 Richard Biener <rguenther@suse.de>
18433 * tree-vect-loop.c (vect_model_reduction_cost): Handle
18434 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
18435 REDUC_MAX_EXPR support.
18436 (vectorizable_reduction): Likewise.
18437 (vect_create_epilog_for_reduction): Likewise.
18439 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
18441 * match.pd (A / (1 << B) -> A >> B): New.
18442 * generic-match-head.c: Include optabs-tree.h.
18443 * gimple-match-head.c: Likewise.
18444 * optabs-tree.h (target_supports_op_p): New.
18445 * optabs-tree.c (target_supports_op_p): New.
18447 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18449 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
18450 $gcc_cv_ld --help output.
18451 (gcc_cv_ld_demangle): Likewise.
18452 (gcc_cv_ld_eh_frame_hdr): Likewise.
18453 (gcc_cv_ld_pie): Likewise.
18454 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
18455 (gcc_cv_ld_buildid): Likewise.
18456 (gcc_cv_ld_sysroot): Likewise.
18457 (ld_bndplt_support): Likewise.
18458 (ld_pushpopstate_support): Likewise.
18459 * configure: Regenerate.
18460 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
18462 2017-06-21 Jakub Jelinek <jakub@redhat.com>
18465 * config/i386/sse.md (round<mode>2): Renumber match_dup and
18466 operands indexes to avoid gap between operands and match_dups.
18468 2017-06-21 Andrew Pinski <apinski@cavium.com>
18470 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
18471 Increment Arith_shift and Arith_shift_reg by 1.
18472 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
18474 * config/aarch64/aarch64.c (thunderx_tunings): Enable
18475 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
18476 (aarch64_strip_extend): Add new argument and test for it.
18477 (aarch64_cheap_mult_shift_p): New function.
18478 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
18479 add a cost if it is true.
18480 Update calls to aarch64_strip_extend.
18481 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
18483 2017-06-21 Andrew Pinski <apinski@cavium.com>
18485 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
18487 (thunderxt88): Likewise.
18488 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
18489 (thunderx_prefetch_tune): New variable.
18490 (thunderx2t99_prefetch_tune): Update for the correct values.
18491 (thunderxt88_tunings): New variable.
18492 (thunderx_tunings): Use thunderx_prefetch_tune instead of
18493 generic_prefetch_tune.
18494 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
18496 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18498 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
18499 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
18500 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
18501 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
18502 (aarch64_atomic_cas<mode>, GPI): Likewise.
18504 2017-06-21 Martin Liska <mliska@suse.cz>
18506 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
18507 statements on cold and hot labels.
18508 * predict.c (tree_estimate_probability_bb): Remove the
18509 prediction from this place.
18511 2017-06-21 Martin Liska <mliska@suse.cz>
18513 PR tree-optimization/79489
18514 * gimplify.c (maybe_add_early_return_predict_stmt): New
18516 (gimplify_return_expr): Call the function.
18517 * predict.c (tree_estimate_probability_bb): Remove handling
18519 * predict.def: Update comment about early return predictor.
18520 * gimple-predict.h (is_gimple_predict): New function.
18521 * predict.def: Change default value of early return to 66.
18522 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
18524 * passes.def: Put pass_strip_predict_hints to the beginning of
18527 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
18529 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
18530 FUNCTION_DECL declarations.
18531 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
18533 (dwaf2out_decl): Likewise.
18534 * godump.c (go_early_global_decl): Skip call to the real debug hook
18535 for FUNCTION_DECL declarations.
18536 * passes.c (rest_of_decl_compilation): Skip call to the
18537 early_global_decl debug hook for FUNCTION_DECL declarations, unless
18538 -fdump-go-spec is passed.
18540 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
18542 * config/i386/i386.c (struct builtin_isa): New field pure_p.
18543 Reorder for compactness.
18544 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
18545 (def_builtin_pure, def_builtin_pure2): New functions.
18546 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
18548 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
18550 * match.pd (nop_convert): New predicate.
18551 ((A +- CST1) +- CST2): Allow some NOP conversions.
18553 2017-06-21 Jakub Jelinek <jakub@redhat.com>
18556 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
18557 with ctors/dtors if GOVD_SHARED is set.
18559 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
18561 * config/aarch64/aarch64.md (movti_aarch64):
18562 Emit mov rather than orr.
18563 (movtf_aarch64): Likewise.
18564 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
18565 Emit mov rather than orr.
18567 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
18569 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
18570 Swap alternatives, make integer dup more expensive.
18572 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
18574 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
18575 Return true for non-tls symbols.
18577 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
18579 * config/aarch64/aarch64-cores.def (cortex-a55): New.
18580 (cortex-a75): Likewise.
18581 (cortex-a75.cortex-a55): Likewise.
18582 * config/aarch64/aarch64-tune.md: Regenerate.
18583 * doc/invoke.texi (-mtune): Document new values for -mtune.
18585 2017-06-21 Tom de Vries <tom@codesourcery.com>
18587 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
18588 stack_size feature.
18589 (Effective-Target Keywords, Other attributes): Suggest using
18590 dg-add-options stack_size feature to get stack limit in stack_size
18591 effective target documentation.
18593 2017-06-21 Julian Brown <julian@codesourcery.com>
18594 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18596 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
18597 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
18598 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
18600 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
18601 attribute type list for neon_multiply.
18602 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
18603 attribute type list for neon_multiply.
18604 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
18605 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
18606 attribute type list for neon_multiply.
18607 * config/arm/types.md (crypto_pmull): Add.
18608 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
18609 attribute type list.
18611 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
18613 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
18616 2017-06-20 Jakub Jelinek <jakub@redhat.com>
18618 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
18619 to make sure not to dereference a NULL cost_classes_ptr pointer.
18621 2017-06-20 Carl Love <cel@us.ibm.com>
18623 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18624 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
18625 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
18626 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
18627 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
18628 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
18629 VMULOSW): New enum "unspec" values.
18630 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
18631 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
18632 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
18633 altivec_vmulosw): New patterns.
18634 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
18635 VMULOSW): Add definitions.
18637 2017-06-20 Julia Koval <julia.koval@intel.com>
18639 * config/i386/i386.c: Fix rounding expand for new pattern.
18640 * config/i386/subst.md: Fix pattern (parallel -> unspec).
18642 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
18644 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
18645 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
18647 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
18649 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
18652 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
18654 * config/aarch64/aarch64-cores.def: Rearrange to sort by
18655 architecture, then by implementer ID.
18656 * config/aarch64/aarch64-tune.md: Regenerate.
18658 2017-06-20 Richard Biener <rguenther@suse.de>
18660 PR middle-end/81097
18661 * fold-const.c (split_tree): Fold to type before negating.
18663 2017-06-20 David Malcolm <dmalcolm@redhat.com>
18665 * diagnostic-show-locus.c
18666 (selftest::test_fixit_deletion_affecting_newline): New function.
18667 (selftest::diagnostic_show_locus_c_tests): Call it.
18669 2017-06-20 Andreas Schwab <schwab@suse.de>
18672 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
18675 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18677 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
18678 __ARM_FEATURE_COPROC according to support.
18680 2017-06-20 Jakub Jelinek <jakub@redhat.com>
18682 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
18683 Rewritten to avoid overflow for > 32-bit pointers.
18686 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
18687 by removing enum keyword.
18688 (ubsan_type_descriptor): Likewise. Formatting fix.
18691 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
18692 splitter): Require TARGET_SSE2 in the condition.
18694 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
18697 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
18698 for doing vector set of SFmode on ISA 3.0.
18699 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
18700 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
18702 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
18703 SFmode value into a V4SF variable that was extracted from another
18704 V4SF variable without converting the element to double precision
18705 and back to single precision vector format.
18706 (vsx_insert_extract_v4sf_p9_2): Likewise.
18708 2017-06-19 Jakub Jelinek <jakub@redhat.com>
18710 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
18711 in UWHI to avoid undefined overflow.
18714 * ubsan.h (enum ubsan_encode_value_phase): New.
18715 (ubsan_encode_value): Change second argument to
18716 enum ubsan_encode_value_phase with default value of
18717 UBSAN_ENCODE_VALUE_GENERIC.
18718 * ubsan.c (ubsan_encode_value): Change second argument to
18719 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
18720 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
18721 create_tmp_var_raw instead of create_tmp_var and use a
18723 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
18724 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
18725 ubsan_encode_value callers.
18728 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
18729 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
18730 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
18732 2017-06-19 Richard Biener <rguenther@suse.de>
18734 PR middle-end/81118
18735 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
18736 estimates if we changed anything.
18738 2017-06-19 Richard Biener <rguenther@suse.de>
18740 PR tree-optimization/80887
18741 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
18742 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
18743 simplified lookups, then reset mprts_hook.
18744 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
18746 (try_to_simplify): Likewise.
18748 2017-06-19 Martin Liska <mliska@suse.cz>
18751 * gimplify.c (gimplify_switch_expr):
18752 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
18754 2017-06-19 Martin Liska <mliska@suse.cz>
18756 * doc/install.texi: Document that PGO runs in 4 stages.
18758 2017-06-19 Martin Liska <mliska@suse.cz>
18761 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
18762 to dispatcher function name.
18763 * multiple_target.c (replace_function_decl): New function.
18764 (create_dispatcher_calls): Redirect both edges and references.
18766 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
18768 * profile-count.c (profile_count::dump): Dump quality.
18769 (profile_count::differs_from_p): Update for unsigned val.
18770 * profile-count.h (profile_count_quality): New enum.
18771 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
18773 2017-06-19 Richard Biener <rguenther@suse.de>
18775 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
18776 struct function as arg.
18777 (estimate_numbers_of_iterations): Export overload with loop arg.
18778 (free_numbers_of_iterations_estimates_loop): Use an overload of
18779 free_numbers_of_iterations_estimates instead.
18780 * tree-cfg.c (remove_bb): Adjust.
18781 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
18782 * tree-parloops.c (gen_parallel_loop): Likewise.
18783 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
18785 (tree_unroll_loops_completely): Likewise.
18786 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
18787 Use an overload instead and export.
18788 (estimated_loop_iterations): Adjust.
18789 (max_loop_iterations): Likewise.
18790 (likely_max_loop_iterations): Likewise.
18791 (estimate_numbers_of_iterations): Take struct function as arg
18793 (loop_exits_before_overflow): Adjust.
18794 (free_numbers_of_iterations_estimates_loop): Use an overload.
18795 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
18796 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
18798 2017-06-19 Richard Biener <rguenther@suse.de>
18801 * ipa-prop.c (find_constructor_constant_at_offset): Handle
18802 RANGE_EXPR conservatively.
18804 2017-06-16 Carl Love <cel@us.ibm.com>
18806 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18807 definitions for vec_float, vec_float2, vec_floato,
18808 vec_floate built-ins.
18809 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
18810 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
18812 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
18813 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
18814 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
18815 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
18816 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
18817 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
18818 vec_floato): Add builtin defines.
18819 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
18820 Update the built-in documentation file for the new built-in
18823 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18825 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
18826 (mthumb): Mark as the negative of -marm.
18828 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18830 * doc/invoke.texi (ARM Options, -mcpu): Document supported
18832 (ARM Options, -mtune): Document that this accepts the same
18833 extension options as -mcpu.
18834 (ARM Options, -mfpu): Document addition of -mfpu=auto.
18836 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18838 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
18839 permitted extensions.
18841 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18843 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
18844 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
18845 (armv8-m.main): Add option +nodsp.
18846 * config/arm/arm-cpu-cdata.h: Regenerated.
18848 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18850 * config/arm/t-fuchsia: New file.
18851 * config.gcc (arm*-*-fuchsia*): Use it.
18853 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18855 * config/arm/t-symbian: Rewrite for new option infrastructure.
18857 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18859 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
18860 (MULTILIB_REQUIRED): Likewise.
18862 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18864 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
18865 (MULTILIB_RESUE): Likewise.
18866 (MULTILIB_MATCHES): Likewise.
18867 (MULTLIB_REQUIRED): Likewise.
18869 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18871 * config/arm/t-rtems: Rewrite for new option framework.
18873 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18875 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
18876 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
18877 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
18878 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
18879 * config/arm/t-multilib: ... here.
18880 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
18881 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
18882 armv7-a and armv8*-a when A-profile libraries have not been built.
18883 * config/arm/t-rmprofile: Rewrite.
18885 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18887 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
18888 with a backslash. Remove the backslash after substituting unescaped
18890 * doc/fragments.texi (MULTILIB_REUSE): Document it.
18892 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18894 * config.gcc: (arm*-*-*): When building a-profile libraries, force
18895 the driver to pass through the default setting of -mfloat-abi.
18896 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
18898 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
18899 (all_feat_combs): New rule.
18900 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
18902 * config/arm/t-aprofile: Rewrite.
18904 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18906 * config/arm/arm.h (FPUTYPE_AUTO): Define.
18907 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
18908 fpu is not specified by the user/command-line.
18909 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
18910 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
18911 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
18912 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
18913 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
18914 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
18916 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18918 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
18919 * config/arm/t-arm-elf: Rewritten.
18921 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18923 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
18924 have some floating-point instructions.
18925 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
18926 (TARGET_MAYBE_HARD_FLOAT): New macro.
18927 * config/arm/arm-builtins.c (arm_init_builtins): Use
18928 TARGET_MAYBE_HARD_FLOAT.
18929 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
18931 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18933 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
18934 (configargs.h): Include it.
18935 (arm_print_hint_for_fpu_option): New function.
18936 (arm_parse_fpu_option): New function.
18937 (candidate_extension): New class.
18938 (arm_canon_for_multilib): New function.
18939 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
18940 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
18941 (ARCH_CANONICAL_SPECS): New macro.
18942 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
18944 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18946 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
18947 are set after handling multilib fragments. Set target_cpu_default2
18950 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18952 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
18954 (arm*-*-*): Set target_cpu_default2 to a quoted string.
18955 * config/arm/parsecpu.awk (check_cpu): Validate any extension
18957 (check_arch): Likewise.
18958 * config/arm/arm.c (arm_configure_build_target): Handle
18959 TARGET_CPU_DEFAULT being a string constant. Scan any feature
18960 options in the default.
18962 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18964 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
18965 when an option is an alias of another.
18966 * config/arm/parsecpu.awk (optalias): New parser token.
18967 (gen_comm_data): Mark non-alias options as such. Emit entries
18968 for extension aliases.
18969 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
18970 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
18971 (armv6kz, armv6zk, armv6t2): Likewise.
18972 (armv7): Make vfpv3-d16 an alias.
18973 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
18975 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
18976 Sort in canonical order.
18977 (armv8-a): Sort in canonical order.
18978 (armv8.1-a, armv8.2-a): Likewise.
18979 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
18981 (cortex-a9): Sort in canonical order.
18982 * config/arm/arm.c (selftests.h): Include it.
18983 (arm_test_cpu_arch_data): New function.
18984 (arm_run_self_tests): New function.
18985 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
18986 (targetm): Move declaration to the end of the file.
18987 * arm-cpu-cdata.h: Regenerated.
18989 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18991 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
18992 call to target_mode_check describing the type of option passed.
18993 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
18994 (arm_target_thumb_only): Use arm_parse_arch_option_name or
18995 arm_parse_cpu_option_name to match parameters against list of
18997 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
18998 arm_arch_core_flags data structure.
18999 * config/arm/arm-cpu_cdata.h: Regenerated.
19001 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19003 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
19005 (arm_print_hint_for_cpu_option): Likewise.
19006 (arm_print_hint_for_arch_option): Likewise.
19007 (arm_parse_cpu_option_name): Likewise.
19008 (arm_parse_arch_option_name): Likewise.
19009 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
19010 of entries in the all_fpus list.
19011 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
19012 (arm_parse_cpu_option_name): Declare.
19013 (arm_parse_arch_option_name): Declare.
19014 (arm_parse_option_features): Declare.
19015 (arm_intialize_isa): Declare.
19016 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
19018 (gen_comm_data): ... here. Make definitions non-static.
19019 * config/arm/arm-cpu-data.h: Regenerated.
19020 * config/arm/arm-cpu-cdata.h: Regenerated.
19022 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19024 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
19025 (cpu_arch_extension): New structure.
19026 (cpu_arch_option, arch_option, cpu_option): New structures.
19027 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
19028 architecture types.
19029 (gen_data): Generate new format data tables.
19030 * config/arm/arm.c (cpu_tune): New structure.
19031 (cpu_option, processors): Delete.
19032 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
19033 (arm_print_hint_for_cpu_option): ... this and ...
19034 (arm_print_hint_for_arch_option): ... this.
19035 (arm_parse_arch_cpu_name): Delete. Replace with ...
19036 (arm_parse_cpu_option_name): ... this and ...
19037 (arm_parse_arch_option_name): ... this.
19038 (arm_unrecognized_feature): Change type of target parameter to
19040 (arm_parse_arch_cpu_features): Delete. Replace with ...
19041 (arm_parse_option_features): ... this.
19042 (arm_configure_build_target): Rework to use new configuration data
19044 (arm_print_tune_info): Rework for new configuration data tables.
19045 * config/arm/arm-cpu-data.h: Regenerated.
19046 * config/arm/arm-cpu.h: Regenerated.
19048 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19050 * Makefile.in (OBJS): Move sbitmap.o from here ...
19051 (OBJS-libcommon): ... to here.
19053 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19055 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
19056 (ISA_ALL_CRYPTO): New macro.
19057 (ISA_ALL_SIMD): New macro
19058 (ISA_ALL_FP): New macro.
19059 * config/arm/arm.c (fpu_bitlist): Update initializer.
19060 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
19062 (arm9e): Add fpu. Add option for nofp
19063 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
19064 (arm926ej-s, arm1026ej-s): Likewise.
19065 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
19066 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
19067 neon-fp16, neon-vfpv4, nofp and nosimd.
19068 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
19069 (cortex-a8): Add fpu. Add option for nofp.
19070 (cortex-a9): Add fpu. Add options for nosimd and nofp.
19071 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
19072 (cortex-r4f): Add fpu.
19073 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
19074 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
19076 (cortex-r8): Likewise.
19077 (cortex-m4): Add fpu. Add option for nofp.
19078 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
19079 (cortex-a17.cortex-a7): Likewise.
19080 (cortex-a32): Add fpu. Add options for crypto and nofp.
19081 (cortex-a35, cortex-a53): Likewise.
19082 (cortex-a57): Add fpu. Add option for crypto.
19083 (cortex-a72, cortex-a73): Likewise.
19084 (exynos-m1): Likewise.
19085 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
19086 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
19087 (cortex-m33): Add fpu. Add option for nofp.
19088 * config/arm/arm-cpu-cdata.h: Regenerated
19089 * config/arm/arm-cpu-data.h: Regenerated.
19091 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19093 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
19094 (armv5te, armv5tej): Likewise.
19095 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
19096 (armv7): Add options fp and vfpv3-d16.
19097 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
19098 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
19100 (armv7ve): Likewise.
19101 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
19102 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
19103 (armv8-a): Add nocrypto option.
19104 (armv8.1-a, armv8.2-a): Likewise.
19105 (armv8-m.main): add options fp, fp.dp and nofp.
19107 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19109 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
19111 (armv8-a+crc): Delete.
19112 (armv8.1-a): Add options simd, crypto and nofp.
19113 (armv8.2-a): Add options fp16, simd, crypto and nofp.
19114 (armv8.2-a+fp16): Delete.
19115 (armv8-m.main): Add option dsp.
19116 (armv8-m.main+dsp): Delete.
19117 (cortex-a8): Add fpu. Add nofp option.
19118 (cortex-a9): Add fpu. Add nofp and nosimd options.
19119 * config/arm/parsecpu.awk (gen_data): Generate option tables and
19120 link to main cpu and architecture data structures.
19121 (gen_comm_data): Only put isa attributes from the main architecture
19123 (option): New statement for architecture and CPU entries.
19124 * arm.c (struct cpu_option): New structure.
19125 (struct processors): Add entry for options.
19126 (arm_unrecognized_feature): New function.
19127 (arm_parse_arch_cpu_name): Ignore any characters after the first
19129 (arm_parse_arch_cpu_feature): New function.
19130 (arm_configure_build_target): Separate out any CPU and architecture
19131 features and parse separately. Don't error out if -mfpu=auto is
19132 used with only an architecture string.
19133 (arm_print_asm_arch_directives): New function.
19134 (arm_file_start): Call it.
19135 * config/arm/arm-cpu-cdata.h: Regenerated.
19136 * config/arm/arm-cpu-data.h: Likewise.
19137 * config/arm/arm-tables.opt: Likewise.
19139 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19141 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
19142 assembler when it is not -mfpu=auto.
19144 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19146 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
19147 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
19148 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
19149 (ASM_CPU_SPEC): Rewrite.
19150 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
19151 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
19152 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
19153 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
19154 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
19155 copied string is NUL-terminated. Also strip any characters prefixed
19157 (arm_rewrite_selected_arch): New function.
19158 (arm_rewrite_march): New function.
19160 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19162 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
19163 (x_arm_cpu_string, x_arm_tune_string): Likewise.
19164 (march, mcpu, mtune): Convert to string-based options.
19165 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
19166 (arm_parse_arch_cpu_name): New function.
19167 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
19168 identify selected architecture or CPU.
19169 (arm_option_save): New function.
19170 (TARGET_OPTION_SAVE): Redefine.
19171 (arm_option_restore): Restore string options.
19172 (arm_option_print): Print string options.
19174 2017-06-16 Martin Sebor <msebor@redhat.com>
19176 PR tree-optimization/80933
19177 PR tree-optimization/80934
19178 * builtins.c (fold_builtin_3): Do not handle bcmp here.
19179 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
19180 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
19181 (gimple_fold_builtin): Call them.
19183 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19185 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
19186 as unlikely; update profile.
19188 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19190 * predict.c (force_edge_cold): Handle declaring edges impossible
19193 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19195 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
19197 (try_unroll_loop_completely): Fix reporting.
19199 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19201 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
19203 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
19206 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
19207 if given a non-constant argument for an intrinsic which requires a
19210 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19212 * profile.c (compare_freqs): New function.
19213 (branch_prob): Sort edge list.
19214 (find_spanning_tree): Assume that the list is priority sorted.
19216 2017-06-16 Richard Biener <rguenther@suse.de>
19218 PR tree-optimization/81090
19219 * passes.def (pass_record_bounds): Remove.
19220 * tree-pass.h (make_pass_record_bounds): Likewise.
19221 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
19222 make_pass_record_bounds): Likewise.
19223 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
19224 not free niter estimates at the beginning but at the end.
19225 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
19227 2017-06-16 Richard Biener <rguenther@suse.de>
19229 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
19230 initializer to workaround ICE in host GCC 4.8.
19232 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19234 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
19236 (clone_inlined_nodes): Update.
19238 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19240 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
19241 prefetch settings, and enable prefetching by default at -O3.
19243 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19245 * config/aarch64/aarch64.c (aarch64_override_options_internal):
19246 Set flag_prefetch_loop_arrays according to tuning data.
19248 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19250 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
19251 New tune structure.
19252 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
19253 [Unrelated to main purpose of the patch] Place the pointer field last
19254 to enable type checking errors when tune structure are wrongly merged.
19255 * config/aarch64/aarch64.c (generic_prefetch_tune,)
19256 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
19257 (thunderx2t99_prefetch_tune): New tune constants.
19258 (tune_params *_tunings): Update all tunings (no functional change).
19259 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
19260 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
19261 from tunings structures.
19263 2017-06-16 Jakub Jelinek <jakub@redhat.com>
19266 * ubsan.c (instrument_null): Add T argument, use it instead
19267 of computing it based on IS_LHS.
19268 (instrument_object_size): Likewise.
19269 (pass_ubsan::execute): Adjust instrument_null and
19270 instrument_object_size callers to pass gimple_get_lhs or
19271 gimple_assign_rhs1 result to it. Use instrument_null instead of
19272 calling get_base_address and instrument_mem_ref. Handle
19273 aggregate call arguments for object-size sanitization.
19275 2017-06-16 Yury Gribov <tetra2005@gmail.com>
19277 PR tree-optimization/81089
19278 * tree-vrp.c (is_masked_range_test): Validate operands of
19281 2017-06-15 Martin Sebor <msebor@redhat.com>
19284 * dumpfile.c (dump_register): Avoid calling memset to initialize
19285 a class with a default ctor.
19286 * gcc.c (struct compiler): Remove const qualification.
19287 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
19288 * hash-table.h: Ditto.
19289 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
19291 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
19292 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
19294 * params.h (struct param_info): Make struct members non-const.
19295 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
19296 with default initialization.
19297 * vec.h (vec_copy_construct, vec_default_construct): New helper
19299 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
19300 with vec_copy_construct.
19301 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
19302 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
19303 * doc/invoke.texi (-Wclass-memaccess): Document.
19305 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19307 * emit-rtl.h (is_leaf): Update comment about local
19308 register allocator.
19310 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19313 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
19314 for a variable to have a section before checking if the section has a
19316 Set section to.persistent if persistent attribute is set.
19317 Warn if .persistent attribute is used on an automatic variable.
19319 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
19321 PR rtl-optimization/80474
19322 * reorg.c (update_block): Do not ignore instructions in a delay slot.
19324 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
19326 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
19329 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
19331 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
19332 (casesi): Emit bounds checking as RTL.
19333 (casesi_internal_mips16_<mode>): Remove bounds checking.
19335 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
19337 * config/xtensa/xtensa.c (xtensa_option_override): Append
19338 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
19339 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
19340 xtensa_doloop_hooks): Define unconditionally.
19341 (xtensa_reorg_loops): Only call reorg_loops in the presence of
19343 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
19344 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
19345 for it in xtensa_option_override.
19346 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
19347 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
19349 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
19351 * doc/cppopts.texi: Document '-' special value to -MF.
19353 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
19355 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
19356 (cortex_a53_fconst): Likewise.
19357 (cortex_a53_fpmul): Likewise.
19358 (cortex_a53_f_load_64): Likewise.
19359 (cortex_a53_f_load_many): Likewise.
19360 (cortex_a53_advsimd_alu): Likewise.
19361 (cortex_a53_advsimd_alu_q): Likewise.
19362 (cortex_a53_advsimd_mul): Likewise.
19363 (cortex_a53_advsimd_mul_q): Likewise.
19364 (fpmac bypass): Add new bypass for fpmac-fpmac case.
19365 Add missing fmul, r2f_cvt and fconst cases.
19367 2017-06-14 Richard Biener <rguenther@suse.de>
19369 PR middle-end/81088
19370 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
19372 (fold_binary_loc): When associating do not treat pre-existing
19373 TREE_OVERFLOW on literal constants as a reason to allow
19374 TREE_OVERFLOW on associated literal constants.
19376 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19378 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
19379 (MASK_FEATURES): New macro.
19380 * config/sparc/sparc.c (sparc_option_override): Remove the special
19381 handling of -mfpu and generalize it to all MASK_FEATURES switches.
19383 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19385 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
19386 a division of 0 if non-call exceptions are enabled.
19388 2017-06-14 Andrew Pinski <apinski@cavium.com>
19389 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19392 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
19393 Improve vector initialization code gen for only variable case.
19395 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19397 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
19399 2017-06-14 Richard Biener <rguenther@suse.de>
19401 PR tree-optimization/81083
19402 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
19405 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19407 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
19408 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
19409 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
19410 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
19411 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
19412 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
19414 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19416 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
19417 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
19419 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19421 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
19423 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19425 * config/rs6000/t-rtems: Don't handle SPE.
19427 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19429 * config/rs6000/t-linux: Don't handle SPE.
19431 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19433 * config/rs6000/eabispe.h: Delete file.
19435 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19437 * config/rs6000/t-spe: Delete file.
19439 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19441 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
19442 (rs6000_legitimate_offset_address_p): Return false for anything in
19443 V2SImode or V2SFmode.
19445 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19447 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
19448 except V2SF and V2SI. Rearrange the vector modes, and add comments.
19449 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
19451 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
19452 (rs6000_legitimate_offset_address_p): Ditto.
19453 (rs6000_emit_move): Ditto.
19454 (rs6000_init_builtins): Remove V4HI_type_node.
19456 2017-06-13 Martin Liska <mliska@suse.cz>
19459 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
19460 (gate_asan): Likewise.
19461 * asan.h (asan_no_sanitize_address_p): Remove the function.
19462 (sanitize_flags_p): New function.
19463 * builtins.def: Fix coding style.
19464 * common.opt: Use renamed enum value.
19465 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
19466 * doc/extend.texi: Document no_sanitize attribute.
19467 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
19468 to SANITIZE_UNDEFINED_NONDEFAULT.
19469 * gcc.c (sanitize_spec_function): Use the renamed enum value.
19470 * gimple-fold.c (optimize_atomic_compare_exchange_p):
19471 Use sanitize_flags_p.
19472 * gimplify.c (gimplify_function_tree): Likewise.
19473 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
19474 * opts.c (parse_no_sanitize_attribute): New function.
19475 (common_handle_option): Use renamed enum value.
19476 * opts.h (parse_no_sanitize_attribute): Declare.
19477 * tree.c (sanitize_flags_p): New function.
19478 * tree.h: Declared here.
19479 * tsan.c: Use sanitize_flags_p.
19480 * ubsan.c (ubsan_expand_null_ifn): Likewise.
19481 (instrument_mem_ref): Likewise.
19482 (instrument_bool_enum_load): Likewise.
19483 (do_ubsan_in_current_function): Remove the function.
19484 (pass_ubsan::execute): Use sanitize_flags_p.
19485 * ubsan.h: Remove do_ubsan_in_current_function
19486 * tree-cfg.c (print_no_sanitize_attr_value): New function.
19487 (dump_function_to_file): Use it here.
19489 2017-06-13 Martin Jambor <mjambor@suse.cz>
19491 PR tree-optimization/80803
19492 PR tree-optimization/81063
19493 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
19494 (propagate_subaccesses_across_link): Enqueue subtree whenever
19495 necessary instead of relying on the caller.
19497 2017-06-13 Martin Jambor <mjambor@suse.cz>
19499 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
19500 that have a first_link.
19501 (sort_and_splice_var_accesses): Do not check first_link before
19503 (subtree_mark_written_and_enqueue): Likewise.
19504 (propagate_all_subaccesses): Likewise and do not stop at first
19505 parent with a first_link.
19507 2017-06-13 Martin Jambor <mjambor@suse.cz>
19509 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
19512 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19514 * match.pd: New pattern.
19516 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19518 * tree-vrp.c (is_masked_range_test): New function.
19519 (register_edge_assert_for): Determine ranges for
19522 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19524 PR tree-optimization/67328
19525 * fold-const.c (maskable_range_p): New function.
19526 (build_range_check): Generate bittests if possible.
19528 2017-06-13 Martin Liska <mliska@suse.cz>
19530 * gimple-pretty-print.c (dump_probability): Add new argument.
19531 (dump_edge_probability): Dump both probability and count.
19532 (dump_gimple_label): Likewise.
19533 (dump_gimple_bb_header): Likewise.
19535 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
19538 * config/avr/avr-devices.c: Fix indentation.
19539 * config/avr/gen-avr-mmcu-specs.c: Dito.
19541 2017-06-13 Richard Biener <rguenther@suse.de>
19543 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
19544 instead get vector type from stmt_info.
19545 (vectorizable_reduction): Adjust. Remove dead code.
19547 2017-06-13 Richard Biener <rguenther@suse.de>
19549 PR middle-end/81065
19550 * fold-const.c (extract_muldiv_1): Remove bogus distribution
19551 case of C * (x * C2 + C3).
19552 (fold_addr_of_array_ref_difference): Properly fold index difference.
19554 2017-06-12 David S. Miller <davem@davemloft.net>
19557 * config/sparc/sparc.md (return expander): Emit frame blockage if
19558 function uses alloca.
19560 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
19562 * combine.c (make_field_assignment): Check len rather than the mode
19563 precision when calling force_to_mode.
19565 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
19567 Support multilibs and devices that see flash in RAM address range.
19570 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
19571 (avr_mcu_t) <flash_pm_offset>: New field.
19572 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
19573 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
19574 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
19575 (AVR_TINY_PM_OFFSET): Remove macro.
19576 * config/avr/avr.opt (-mshort-calls): New option.
19577 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
19578 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
19579 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
19580 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
19581 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
19582 instead of avr_arch->have_jmp_call.
19583 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
19584 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
19585 avr_arch->flash_pm_offset to define.
19586 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
19587 new field flash_pm_offset. Add entry for avrxmega3.
19588 (avr_texinfo): Add entry for avrxmega3.
19589 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
19590 attiny212, attiny214,
19591 attiny412, attiny414, attiny416, attiny417,
19592 attiny814, attiny816, attiny817,
19593 attiny1614, attiny1616, attiny1617,
19594 attiny3214, attiny3216, attiny3217.
19595 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
19596 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
19597 (avr_print_operand_address) [AVR_TINY]: Same.
19598 (avr_asm_init_sections) <readonly_data_section>: Only patch
19599 callback if avr_arch->flash_pm_offset = 0.
19600 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
19601 for rodata if avr_arch->flash_pm_offset != 0.
19602 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
19603 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
19604 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
19605 (m_options): Append opt_rcall.
19606 (m_dirnames): Append dir_rcall.
19607 * config/avr/t-multilib: Regenerate.
19609 * configure.ac [target=avr]: Check whether avrxmega3 default
19610 linker description file works as needed.
19611 * configure: Regenerate.
19612 * doc/avr-mmcu.texi: Regenerate.
19613 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
19614 <__AVR_ARCH__>: Document avrxmega3 and 103.
19615 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
19616 <__AVR_SHORT_CALLS__>: Document it.
19617 <__AVR_PM_BASE_ADDRESS__>: Document it.
19618 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
19619 (AVR Variable Attributes) <progmem>: Document this is
19620 not needed for avrxmega3.
19621 (AVR Named Address Spaces) <__flash>: Dito.
19623 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
19625 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
19627 2017-06-12 Doug Rupp <rupp@adacore.com>
19629 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
19630 Append vxworks-stdint.h to the tm_file list.
19631 * config/vxworks-stdint.h: New file.
19633 2017-06-12 Martin Liska <mliska@suse.cz>
19635 PR tree-optimization/81041
19636 * tree-profile.c (gimple_gen_ic_func_profiler):
19637 Create an extra BB in profile-generate
19638 (gimple_gen_time_profiler): Likewise.
19640 2017-06-12 Jakub Jelinek <jakub@redhat.com>
19642 PR tree-optimization/81003
19643 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
19644 (update_range_test): Use it instead of force_gimple_operand_gsi.
19646 2017-06-12 Richard Biener <rguenther@suse.de>
19648 PR tree-optimization/81053
19649 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
19650 with backedge value not defined in loop. Simplify def stmt
19653 2017-06-11 Tom de Vries <tom@codesourcery.com>
19656 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
19658 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
19659 nvptx_cannot_force_const_mem.
19661 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19663 * opts.c (finish_options): Move test for flag_split_stack after
19664 it has been initialized.
19666 2017-06-11 Jason Merrill <jason@redhat.com>
19668 * tree.h (id_equal): New.
19669 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
19670 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
19671 instead of strcmp of IDENTIFIER_POINTER.
19673 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19675 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
19676 (mark_all_inlined_calls_cdtor): Fix formating.
19677 (inline_transform): Rescale profile before inlining.
19679 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19681 * cgraph.h (cgraph_edge::clone): Update prototype.
19682 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
19683 (cgraph_node::create_clone): Update.
19684 (cgraph_node::create_version_clone): Update.
19685 * tree-inline.c (copy_bb): Update.
19686 (expand_call_inline): Update.
19688 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
19690 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
19691 factored out from ...
19692 (rs6000_emit_prologue): ... here.
19694 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
19696 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
19697 factored out from ...
19698 (rs6000_emit_prologue): ... here.
19700 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19702 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
19704 (handle_missing_profiles): Fix computation of tp_first_run.
19705 (counts_to_freqs): Do not touch freqs when count is 0.
19707 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19709 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
19712 2017-06-10 Tom de Vries <tom@codesourcery.com>
19714 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
19715 attributes): Document signal effective target.
19717 2017-06-10 Tom de Vries <tom@codesourcery.com>
19719 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
19720 Document effective target stack_size.
19722 2017-06-09 David Malcolm <dmalcolm@redhat.com>
19724 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
19725 to the edit_context if they can be auto-applied.
19727 2017-06-9 Ian Lance Taylor <iant@golang.org>
19729 * opts.c (finish_options): If -fsplit-stack, disable implicit
19730 -forder-blocks-and-partition.
19731 * doc/invoke.texi (Optimize Options): Document that when using
19732 -fsplit-stack -forder-blocks-and-partition is not implicitly
19735 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
19737 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
19738 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
19739 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
19740 * builtins.def (abort, trap, unreachable): Declare cold.
19741 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
19742 * tree-core.h (ECF_COLD): New.
19743 * tree.c (set_call_expr_flags): Handle ECF_COLD.
19744 (build_common_builtin_nodes): Mark unreachable and abort as cold.
19746 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
19748 * predict.c (unlikely_executed_stmt_p): Cleanup.
19750 2017-06-09 Richard Biener <rguenther@suse.de>
19752 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
19753 model if the ref is always written to.
19755 2017-06-09 Tamar Christina <tamar.christina@arm.com>
19757 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
19759 2017-06-09 Tamar Christina <tamar.christina@arm.com>
19761 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
19764 2017-06-09 Tom de Vries <tom@codesourcery.com>
19767 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
19768 "target cannot support label values" when encountering LABEL_REF.
19770 2017-06-09 Martin Liska <mliska@suse.cz>
19772 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
19773 (gimple_gen_ic_func_profiler): Emit direct comparison
19774 of __gcov_indirect_call_callee with NULL.
19775 (gimple_gen_time_profiler): Change probability from
19776 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
19778 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
19780 * profile.c (edge_gcov_counts): Turn to pointer.
19781 (compute_branch_probabilities, compute_branch_probabilities): Update.
19782 (branch_prob): Do not clear edge_gcov_count.
19783 * profile.h (edge_gcov_counts): Turn to pointer.
19784 (edge_gcov_count): Update.
19786 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
19788 * gimple.h (gimple_check_failed): Mark cold.
19790 2017-06-09 Richard Biener <rguenther@suse.de>
19792 PR tree-optimization/66623
19793 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
19794 refactor check_reduction into two parts, properly computing
19795 whether we have to check reduction validity for outer loop
19798 2017-06-09 Richard Biener <rguenther@suse.de>
19800 PR tree-optimization/79483
19801 * graphite-scop-detection.c (order): New global.
19802 (get_order): Compute bb to order mapping that satisfies code
19803 generation constraints.
19804 (cmp_pbbs): New helper.
19805 (build_scops): Start domwalk at entry block, sort generated
19808 2017-06-09 Richard Biener <rguenther@suse.de>
19810 PR middle-end/81007
19811 * ipa-polymorphic-call.c
19812 (ipa_polymorphic_call_context::restrict_to_inner_class):
19813 Skip FIELD_DECLs with error_mark_node type.
19814 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
19817 2017-06-09 Martin Liska <mliska@suse.cz>
19819 * predict.c (struct branch_predictor): New struct.
19820 (test_prediction_value_range): New test.
19821 (predict_c_tests): New function.
19822 * selftest-run-tests.c (selftest::run_tests): Run the function.
19823 * selftest.h: Declare new tests.
19825 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
19828 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
19829 gen_add3_insn did not fail.
19830 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
19831 r0, construct that number in a temporary reg and add that reg to r0.
19832 If asked to put the result in r0 as well, fail.
19834 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
19836 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
19837 for early expansion of vec_eqv.
19839 2017-06-08 Jakub Jelinek <jakub@redhat.com>
19841 PR middle-end/81005
19842 * ubsan.c (instrument_null): Avoid pointless code temporary.
19843 (pass_ubsan::execute): Instrument aggregate arguments of calls.
19845 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
19849 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
19852 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
19855 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19857 * predict.c (unlikely_executed_edge_p): Move ahead.
19858 (probably_never_executed_edge_p): Use it.
19860 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
19862 PR middle-end/79988
19863 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
19864 gimple_call_builtin_p call.
19866 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19868 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
19869 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
19870 rtl_check_failed_type2, rtl_check_failed_code1,
19871 rtl_check_failed_code2, rtl_check_failed_code_mode,
19872 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
19873 rtvec_check_failed_bounds, rtl_check_failed_flag,
19874 _fatal_insn_not_found, _fatal_insn): Likewise.
19875 * tree.h (tree_contains_struct_check_failed,
19876 tree_check_failed, tree_not_check_failed,
19877 tree_class_check_failed, tree_range_check_failed,
19878 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
19879 tree_vec_elt_check_failed, phi_node_elt_check_failed,
19880 tree_operand_check_failed, omp_clause_check_failed,
19881 omp_clause_operand_check_failed, omp_clause_range_check_failed):
19884 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19886 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
19887 flag_branch_probabilities.
19888 * ipa-inline.c (edge_badness): Likewise.
19889 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
19890 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
19891 * predict.c (maybe_hot_frequency_p): Likewise.
19892 (probably_never_executed): Likewise.
19893 * sched-ebb.c (schedule_ebbs): Likewise.
19894 * sched-rgn.c (find_single_block_region): Likewise.
19895 * tracer.c (tail_duplicate): Likewise.
19897 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19899 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
19900 longer requires x_flag_profile_use.
19902 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19904 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
19905 instead of flag_reorder_blocks_and_partition.
19906 * dbxout.c (dbxout_function_end): Likewise.
19907 * dwarf2out.c (gen_subprogram_die): Likewise.
19908 * haifa-sched.c (sched_create_recovery_edges): Likewise.
19909 * hw-doloop.c (reorg_loops): Likewise.
19910 * varasm.c (assemble_start_function,
19911 assemble_end_function): Likewise.
19912 (decide_function_section): Do not check for
19913 flag_reorder_blocks_and_partition.
19915 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
19917 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
19919 (chkp_get_hard_register_fake_addr_expr): Ditto.
19920 (chkp_build_addr_expr): Add check for hard reg case.
19921 (chkp_parse_array_and_component_ref): Ditto.
19922 (chkp_find_bounds_1): Ditto.
19923 (chkp_process_stmt): Don't generate bounds store for
19926 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19928 * predict.c (maybe_hot_bb_p): Do not check profile status.
19929 (maybe_hot_edge_p): Likewise.
19930 (probably_never_executed): Check for zero counts even if profile
19932 (unlikely_executed_edge_p): New function.
19933 (unlikely_executed_stmt_p): New function.
19934 (unlikely_executed_bb_p): New function.
19935 (set_even_probabilities): Use unlikely predicates.
19936 (combine_predictions_for_bb): Likewise.
19937 (predict_paths_for_bb): Likewise.
19938 (predict_paths_leading_to_edge): Likewise.
19939 (determine_unlikely_bbs): New function.
19940 (estimate_bb_frequencies): Use it.
19941 (compute_function_frequency): Use zero counts even if profile is
19943 * profile-count.h: Fix typo.
19945 2017-08-08 Julia Koval <julia.koval@intel.com>
19947 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
19948 _mm512_mask_cvtsepi16_storeu_epi8,
19949 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
19950 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
19951 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
19952 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
19953 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
19954 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
19955 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
19956 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
19957 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
19958 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
19959 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
19960 __builtin_ia32_pmovuswb256mem_mask,
19961 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
19962 __builtin_ia32_pmovwb512mem_mask): New builtins.
19964 2017-08-08 Julia Koval <julia.koval@intel.com>
19966 PR target/73350,80862
19967 * config/i386/subst.md (round): Fix round pattern.
19968 * config/i386/i386.c (ix86_erase_embedded_rounding):
19969 Fix erasing rounding for the fixed pattern.
19971 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19973 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
19975 2017-06-08 Martin Liska <mliska@suse.cz>
19977 PR gcov-profile/80911
19978 * gcov.c (block_info::block_info): New constructor.
19980 2017-06-07 Carl Love <cel@us.ibm.com>
19982 * config/rs6000/rs6000-c: The return type of the following
19983 built-in functions was implemented as int not long long. Fix sign
19984 of return value for the unsigned version of vec_mulo and vec_mule.
19985 vector unsigned long long vec_bperm (vector unsigned long long,
19986 vector unsigned char)
19987 vector signed long long vec_mule (vector signed int,
19989 vector unsigned long long vec_mule (vector unsigned int,
19990 vector unsigned int)
19991 vector signed long long vec_mulo (vector signed int,
19993 vector unsigned long long vec_mulo (vector unsigned int,
19994 vector unsigned int)
19995 * doc/extend.texi: Fix the documentation for the built-in
19998 2017-06-07 Carl Love <cel@us.ibm.com>
20001 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
20004 2017-06-07 Carl Love <cel@us.ibm.com>
20006 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
20007 support, Generate doublehv for signed int/float for BE case only.
20009 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
20011 * doc/invoke.texi (mcx16): Rewrite.
20013 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20015 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
20016 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
20017 *mov<mode>_softfloat, and an anonymous splitter): Use
20018 nonimmediate_operand instead of rs6000_nonimmediate_operand.
20020 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20022 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
20024 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20025 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
20026 (rs6000_debug_reg_global): Adjust.
20027 (rs6000_init_hard_regno_mode_ok): Adjust.
20028 (rs6000_dbx_register_number): Adjust.
20029 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
20030 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20031 Remove SPE_ACC and SPEFSCR.
20032 (REG_ALLOC_ORDER): Ditto.
20033 (FRAME_POINTER_REGNUM): Change to 111.
20034 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
20035 (REG_CLASS_NAMES): Ditto.
20036 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
20037 (REGISTER_NAMES): Ditto.
20038 (ADDITIONAL_REG_NAMES): Ditto.
20039 (rs6000_reg_names): Ditto.
20040 * config/rs6000/rs6000.md: Renumber some register number
20043 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20045 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
20047 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20048 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
20050 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
20051 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20052 Delete the SPE high registers.
20053 (REG_ALLOC_ORDER): Ditto.
20054 (enum reg_class): Remove SPE_HIGH_REGS.
20055 (REG_CLASS_NAMES): Ditto.
20056 (REG_CLASS_CONTENTS): Delete the SPE high registers.
20057 (REGISTER_NAMES): Ditto.
20058 (rs6000_reg_names): Ditto.
20059 * doc/tm.texi.in: Remove SPE as example.
20060 * doc/tm.texi: Regenerate.
20062 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20064 * config/rs6000/8540.md (ppc8540_brinc): Delete.
20065 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
20066 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
20067 * config/rs6000/rs6000.md (type): Remove "brinc".
20069 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20071 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
20072 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
20073 * config/rs6000/linuxspe.h: Delete file.
20074 * config/rs6000/rs6000.md: Don't include spe.md.
20075 * config/rs6000/spe.h: Delete file.
20076 * config/rs6000/spe.md: Delete file.
20077 * config/rs6000/t-rs6000: Remove spe.md.
20079 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20081 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
20082 (reg_or_none500mem_operand): Delete.
20083 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
20084 instead of reg_or_none500mem_operand.
20086 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20088 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
20089 handling of SPE flags.
20090 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
20092 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20094 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
20096 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
20097 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
20098 paired_divv2sf3): Similar.
20099 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
20100 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
20101 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
20103 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
20104 Rename the paired_* instruction patterns.
20105 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
20107 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
20108 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
20109 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
20110 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
20111 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
20112 PAIRED_VECTOR_MODE.
20113 (struct machine_function): Delete field spe_insn_chain_scanned_p.
20114 (spe_func_has_64bit_regs_p): Delete.
20115 (spe_expand_predicate_builtin): Delete.
20116 (spe_expand_evsel_builtin): Delete.
20117 (TARGET_DWARF_REGISTER_SPAN): Do not define.
20118 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
20119 (invalid_e500_subreg): Delete.
20120 (rs6000_legitimize_address): Always force_reg op2 as well, for
20121 paired single memory accesses.
20122 (rs6000_member_type_forces_blk): Delete.
20123 (rs6000_spe_function_arg): Delete.
20124 (rs6000_expand_unop_builtin): Delete SPE handling.
20125 (rs6000_expand_binop_builtin): Ditto.
20126 (spe_expand_stv_builtin): Delete.
20127 (bdesc_2arg_spe): Delete.
20128 (spe_expand_builtin): Delete.
20129 (spe_expand_predicate_builtin): Delete.
20130 (spe_expand_evsel_builtin): Delete.
20131 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
20132 (spe_init_builtins): Delete.
20133 (spe_func_has_64bit_regs_p): Delete.
20134 (savres_routine_name): Delete "info" parameter. Adjust callers.
20135 (rs6000_emit_stack_reset): Ditto.
20136 (rs6000_dwarf_register_span): Delete.
20137 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
20138 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
20139 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
20141 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
20143 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
20144 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
20145 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
20146 mulv2sf3, divv2sf3): Delete expanders.
20148 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20150 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
20152 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20154 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
20155 * config/rs6000/rs6000.c: Ditto.
20157 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20159 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
20160 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
20161 comparison_operator.
20163 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20165 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
20166 * config/rs6000/rs6000.opt: Ditto.
20167 * config/rs6000/t-rtems: Ditto.
20169 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20171 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
20172 TARGET_E500_SINGLE by 0, simplify.
20173 * config/rs6000/rs6000.c: Ditto.
20174 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
20175 (spe_build_register_parallel): Delete.
20176 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
20177 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
20178 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
20179 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
20180 (E500_CONVERT): Delete.
20181 * config/rs6000/spe.md: Remove many patterns and all define_constants.
20183 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20185 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
20186 * config/rs6000/dfp.md: Ditto.
20187 (negdd2, *negdd2_fpr): Merge.
20188 (absdd2, *absdd2_fpr): Merge.
20189 (negtd2, *negtd2_fpr): Merge.
20190 (abstd2, *abstd2_fpr): Merge.
20191 * config/rs6000/e500.h: Delete file.
20192 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
20193 TARGET_FPRS by 1 and simplify.
20194 * config/rs6000/rs6000-c.c: Ditto.
20195 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
20196 TARGET_DF_SPE by 0.
20197 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
20199 * config/rs6000/rs6000.md: Ditto.
20200 (floatdidf2, *floatdidf2_fpr): Merge.
20201 (move_from_CR_gt_bit): Delete.
20202 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
20203 (E500_CR_IOR_COMPARE): Delete.
20204 (All patterns that require !TARGET_FPRS): Delete.
20205 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
20207 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20209 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
20211 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20213 * graphds.c (add_edge): Intitialize edge's attached data.
20214 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
20215 pointer parameter. Call pointed function on each edge during
20216 graph traversing. Skip traversing the edge when the function
20218 (graphds_dfs, graphds_scc): Ditto.
20219 (for_each_edge): New parameter. Pass the new parameter to callback
20221 * graphds.h (skip_edge_callback): New function pointer type.
20222 (graphds_dfs, graphds_scc): New function pointer parameter.
20223 (graphds_edge_callback, for_each_edge): New parameter.
20225 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20227 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
20228 out code checking if runtime alias check is possible to below ...
20229 Call the new function.
20230 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
20231 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
20233 2017-06-07 Marek Polacek <polacek@redhat.com>
20236 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
20237 TYPE_OVERFLOW_WRAPS check.
20239 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20241 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
20242 if versioning is required.
20243 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
20244 peeling with the check for versioning.
20246 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20248 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
20249 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
20250 Set true to new parameter if new ssa variable is defined.
20251 (vect_gen_vector_loop_niters): Refactor. Set range information
20252 for the new vector loop bound variable.
20253 (vect_do_peeling): Ditto.
20255 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20257 * tree-affine.c (ssa.h): Include header file.
20258 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
20259 has wrapping overflow behavior.
20261 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20263 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
20265 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20267 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
20268 (tree_to_aff_combination): ... here.
20270 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20272 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
20273 reg_pressure model function.
20274 (ivopts_global_cost_for_size): Delete.
20275 (determine_set_costs, iv_ca_recount_cost): Call new model function
20276 ivopts_estimate_reg_pressure.
20278 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20280 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
20281 expensive than udiv. Remove floating point cases from mod.
20283 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20285 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
20286 Increase idiv cost.
20288 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20290 * config/aarch64/aarch64.md
20291 (copysignsf3): Fix mask generation.
20293 2017-06-07 Jakub Jelinek <jakub@redhat.com>
20295 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
20297 (class dump_manager): Add register_dumps method.
20298 * dumpfile.c: Include langhooks.h.
20299 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
20300 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
20301 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
20302 (dump_manager::dump_register): Start with 512 entries instead of 32.
20303 (dump_manager::register_dumps): New method.
20304 * toplev.c (general_init): Instead of invoking register_dumps
20305 langhook, invoke register_dumps method on the dump manager.
20306 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
20309 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
20311 * doc/md.texi: Clarify the restrictions on a define_insn condition.
20312 Say that # requires an associated define_split to exist, and that
20313 the define_split must be suitable for use after register allocation.
20315 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20317 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
20318 (compute_outgoing_frequencies): Also initialize zero counts.
20319 (find_many_sub_basic_blocks): Do not produce uninitialized profile
20320 around loops; preserve more of profile when nothing changes.
20322 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
20324 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
20326 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
20327 * config/arm/arm-cpu-cdata.h: Regenerate.
20328 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
20329 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
20330 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
20331 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
20332 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
20334 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
20336 * config/arm/t-rmprofile: Likewise.
20337 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
20339 2017-06-06 David S. Miller <davem@davemloft.net>
20342 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
20343 blockage if function uses alloca.
20345 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20347 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
20348 New "uid" fields to hold pretty-print IDs of group and ref.
20349 Memory references are now identified as <group_id>:<ref_id>
20350 instead of using [random] addresses.
20351 (dump_mem_details): Simplify, no functional change.
20352 (dump_mem_ref): Simplify and make output more concise.
20353 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
20354 (find_or_create_group): Initialize group uid.
20355 (record_ref): Initialize ref uid. Improve debug output.
20356 (prune_group_by_reuse, should_issue_prefetch_p,)
20357 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
20358 (mark_nontemporal_store, determine_loop_nest_reuse):
20359 Improve debug output.
20361 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20363 * dbgcnt.def (prefetch): New debug counter.
20364 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
20365 (schedule_prefetches): Stop issueing prefetches if debug counter
20368 2017-06-06 Tom de Vries <tom@codesourcery.com>
20370 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
20371 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
20373 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20375 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
20376 Use aarch64_reg_or_zero predicate for operand 4.
20377 (aarch64_compare_and_swap<mode> define_insn_and_split):
20378 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
20379 (aarch64_store_exclusive<mode>): Likewise for operand 2.
20381 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
20383 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
20384 (arm_compute_save_core_reg_mask): This.
20385 (thumb1_compute_save_reg_mask): Rename into ...
20386 (thumb1_compute_save_core_reg_mask): This.
20387 (arm_compute_save_reg0_reg12_mask): Adapt comment.
20388 (arm_compute_frame_layout): Likewise.
20390 2017-06-06 Richard Biener <rguenther@suse.de>
20392 PR tree-optimization/80974
20393 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
20394 keep or clear leaders SSA info.
20396 2017-06-06 Tom de Vries <tom@codesourcery.com>
20398 * config/nvptx/nvptx.c (split_mode_p): New function.
20399 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
20401 2017-06-06 Tom de Vries <tom@codesourcery.com>
20403 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
20405 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20408 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
20411 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20413 * shrink-wrap.c (handle_simple_exit): Update profile.
20414 (try_shrink_wrapping): Upate profile.
20416 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20418 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
20419 (tree_guess_outgoing_edge_probabilities): New.
20420 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
20421 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
20423 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20425 * ipa-split.c (split_function): Initialize return bb profile.
20427 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20429 * profile.c (compute_branch_probabilities): Also initialize
20430 EXIT_BLOCK profile.
20432 2017-06-06 Richard Biener <rguenther@suse.de>
20434 PR tree-optimization/80928
20435 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
20436 (vect_analyze_loop_operations): Properly guard analysis for
20438 (vect_transform_loop): Likewise.
20439 (vect_analyze_loop_2): Also reset SLP type on PHIs.
20440 (vect_model_induction_cost): Do not cost for pure SLP.
20441 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
20442 of induction in inner loop vectorization.
20443 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
20444 (vect_get_and_check_slp_defs): Handle vect_induction_def.
20445 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
20447 (vect_analyze_slp_cost_1): Cost induction.
20448 (vect_detect_hybrid_slp_stmts): Handle PHIs.
20449 (vect_get_slp_vect_defs): Likewise.
20450 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
20451 (vect_transform_stmt): Handle SLP reductions.
20452 * tree-vectorizer.h (vectorizable_induction): Adjust.
20454 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
20456 * config/rs6000/rs6000.c (make_resolver_func): Update
20457 init_lowered_empty_function call.
20459 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
20461 * doc/invoke.texi: Document the -fprofile-abs-path option.
20462 * common.opt (fprofile-abs-path): New option.
20463 * gcov-io.h (gcov_write_filename): Declare.
20464 * gcov-io.c (gcov_write_filename): New function.
20465 * coverage.c (coverage_begin_function): Use gcov_write_filename.
20466 * profile.c (output_location): Likewise.
20468 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20470 * shring-wrap.c: Revert accidental commit.
20472 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
20474 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
20476 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20478 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
20480 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
20481 profile in callgraph edge.
20482 * profile-count.h (apply_probability): If THIS is 0, then result is 0
20483 (apply_scale): Likewise.
20484 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
20485 Also scale profile when inlining function with zero profile.
20486 (initialize_cfun): Update exit block profile even when it is zero.
20487 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
20488 when profile is read.
20490 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
20492 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
20493 (CLONE_*): New constants to define the processors we can generate
20494 code for with the target_clone attribute.
20495 (rs6000_clone_map): New array to identify which clone processors
20496 the current program is running on.
20497 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
20498 target_clone attribute.
20499 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
20500 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
20501 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
20502 (cpu_expand_builtin): Add support for target_clone attribute.
20503 (rs6000_valid_attribute_p): Allow "default" attribute.
20504 (get_decl_name): New debug function to simplify printing the
20505 current function name in debugging statements.
20506 (rs6000_clone_priority): New functions to support the target_clone
20507 attribute, and be able to generate code to switch between ISA 2.05
20508 through ISA 3.0 (power6 through power9).
20509 (rs6000_compare_version_priority): Likewise.
20510 (rs6000_get_function_versions_dispatcher): Likewise.
20511 (make_resolver_func): Likewise.
20512 (add_condition_to_bb): Likewise.
20513 (dispatch_function_versions): Likewise.
20514 (rs6000_generate_version_dispatcher_body): Likewise.
20515 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
20516 (fusion_gpr_load_p): Fix a spacing issue.
20517 * doc/extend.texi (Common Function Attributes): Document that the
20518 PowerPC supports the target_clone attribute.
20520 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
20522 * config/arm/arm.h: explain F symbol found in description of ARM
20523 register allocation in its legend.
20525 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20527 * config/mips/frame-header-opt.c: Include profile-count.h.
20528 * config/riscv/riscv.c: Include profile-count.h
20530 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20532 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
20534 (sm_set_flag_if_changed): Add bbs field.
20535 (execute_sm_if_changed_flag_set): Pass BBS.
20536 (execute_sm): Update.
20538 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20540 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
20543 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20545 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
20546 (peephole2): New peephole2 to emit the above.
20547 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
20549 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20551 * config/aarch64/aarch64.c (define_peephole2 above
20552 *sub_<shift>_<mode>): New peephole.
20554 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
20556 * config/i386/i386.c (make_resolver_func): Update.
20557 * Makefile.in: Add profile-count.h and profile-count.o
20558 * auto-profile.c (afdo_indirect_call): Update to new API.
20559 (afdo_set_bb_count): Update.
20560 (afdo_propagate_edge): Update.
20561 (afdo_propagate_circuit): Update.
20562 (afdo_calculate_branch_prob): Update.
20563 (afdo_annotate_cfg): Update.
20564 * basic-block.h: Include profile-count.h
20565 (struct edge_def): Turn count to profile_count.
20566 (struct basic_block_def): Likewie.
20567 (REG_BR_PROB_BASE): Move to profile-count.h
20568 (RDIV): Move to profile-count.h
20569 * bb-reorder.c (max_entry_count): Turn to profile_count.
20570 (find_traces): Update.
20571 (rotate_loop):Update.
20572 (connect_traces):Update.
20573 (sanitize_hot_paths):Update.
20574 * bt-load.c (migrate_btr_defs): Update.
20575 * cfg.c (RDIV): Remove.
20576 (init_flow): Use alloc_block.
20577 (alloc_block): Uninitialize count.
20578 (unchecked_make_edge): Uninitialize count.
20579 (check_bb_profile): Update.
20580 (dump_edge_info): Update.
20581 (dump_bb_info): Update.
20582 (update_bb_profile_for_threading): Update.
20583 (scale_bbs_frequencies_int): Update.
20584 (scale_bbs_frequencies_gcov_type): Update.
20585 (scale_bbs_frequencies_profile_count): New.
20586 * cfg.h (update_bb_profile_for_threading): Update.
20587 (scale_bbs_frequencies_profile_count): Declare.
20588 * cfgbuild.c (compute_outgoing_frequencies): Update.
20589 (find_many_sub_basic_blocks): Update.
20590 * cfgcleanup.c (try_forward_edges): Update.
20591 (try_crossjump_to_edge): Update.
20592 * cfgexpand.c (expand_gimple_tailcall): Update.
20593 (construct_exit_block): Update.
20594 * cfghooks.c (verify_flow_info): Update.
20595 (dump_bb_for_graph): Update.
20596 (split_edge): Update.
20597 (make_forwarder_block): Update.
20598 (duplicate_block): Update.
20599 (account_profile_record): Update.
20600 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
20601 (get_estimated_loop_iterations): Update.
20602 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
20603 (single_likely_exit): Update.
20604 * cfgloopmanip.c (scale_loop_profile): Update.
20606 (set_zero_probability): Update.
20607 (lv_adjust_loop_entry_edge): Update.
20608 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
20609 (purge_dead_edges): Update.
20610 (rtl_account_profile_record): Update.
20611 * cgraph.c (cgraph_node::create): Uninitialize count.
20612 (symbol_table::create_edge): Uninitialize count.
20613 (cgraph_update_edges_for_call_stmt_node): Update.
20614 (cgraph_edge::dump_edge_flags): Update.
20615 (cgraph_node::dump): Update.
20616 (cgraph_edge::maybe_hot_p): Update.
20617 * cgraph.h: Include profile-count.h
20618 (create_clone), create_edge, create_indirect_edge): Update.
20619 (cgraph_node): Turn count to profile_count.
20620 (cgraph_edge0: Likewise.
20621 (make_speculative, clone): Update.
20622 (create_edge): Update.
20623 (init_lowered_empty_function): Update.
20624 * cgraphclones.c (cgraph_edge::clone): Update.
20625 (duplicate_thunk_for_node): Update.
20626 (cgraph_node::create_clone): Update.
20627 * cgraphunit.c (cgraph_node::analyze): Update.
20628 (cgraph_node::expand_thunk): Update.
20629 * final.c (dump_basic_block_info): Update.
20630 * gimple-streamer-in.c (input_bb): Update.
20631 * gimple-streamer-out.c (output_bb): Update.
20632 * graphite.c (print_global_statistics): Update.
20633 (print_graphite_scop_statistics): Update.
20634 * hsa-brig.c: Include basic-block.h.
20635 * hsa-dump.c: Include basic-block.h.
20636 * hsa-gen.c (T sum_slice): Update.
20637 (convert_switch_statements):Update.
20638 * hsa-regalloc.c: Include basic-block.h.
20639 * ipa-chkp.c (chkp_produce_thunks): Update.
20640 * ipa-cp.c (struct caller_statistics): Update.
20641 (init_caller_stats): Update.
20642 (gather_caller_stats): Update.
20643 (ipcp_cloning_candidate_p): Update.
20644 (good_cloning_opportunity_p): Update.
20645 (get_info_about_necessary_edges): Update.
20646 (dump_profile_updates): Update.
20647 (update_profiling_info): Update.
20648 (update_specialized_profile): Update.
20649 (perhaps_add_new_callers): Update.
20650 (decide_about_value): Update.
20651 (ipa_cp_c_finalize): Update.
20652 * ipa-devirt.c (struct odr_type_warn_count): Update.
20653 (struct decl_warn_count): Update.
20654 (struct final_warning_record): Update.
20655 (possible_polymorphic_call_targets): Update.
20656 (ipa_devirt): Update.
20657 * ipa-fnsummary.c (redirect_to_unreachable): Update.
20658 * ipa-icf.c (sem_function::merge): Update.
20659 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
20660 * ipa-inline.c (compute_uninlined_call_time): Update.
20661 (compute_inlined_call_time): Update.
20662 (want_inline_small_function_p): Update.
20663 (want_inline_self_recursive_call_p): Update.
20664 (edge_badness): Update.
20665 (lookup_recursive_calls): Update.
20666 (recursive_inlining): Update.
20667 (inline_small_functions): Update.
20668 (dump_overall_stats): Update.
20669 (dump_inline_stats): Update.
20670 * ipa-profile.c (ipa_profile_generate_summary): Update.
20671 (ipa_propagate_frequency): Update.
20672 (ipa_profile): Update.
20673 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
20674 * ipa-utils.c (ipa_merge_profiles): Update.
20675 * loop-doloop.c (doloop_modify): Update.
20676 * loop-unroll.c (report_unroll): Update.
20677 (unroll_loop_runtime_iterations): Update.
20678 * lto-cgraph.c (lto_output_edge): Update.
20679 (lto_output_node): Update.
20680 (input_node): Update.
20681 (input_edge): Update.
20682 (merge_profile_summaries): Update.
20683 * lto-streamer-in.c (input_cfg): Update.
20684 * lto-streamer-out.c (output_cfg): Update.
20685 * mcf.c (create_fixup_graph): Update.
20686 (adjust_cfg_counts): Update.
20687 (sum_edge_counts): Update.
20688 * modulo-sched.c (sms_schedule): Update.
20689 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
20690 * predict.c (maybe_hot_count_p): Update.
20691 (probably_never_executed): Update.
20692 (dump_prediction): Update.
20693 (combine_predictions_for_bb): Update.
20694 (propagate_freq): Update.
20695 (handle_missing_profiles): Update.
20696 (counts_to_freqs): Update.
20697 (rebuild_frequencies): Update.
20698 (force_edge_cold): Update.
20699 * predict.h: Include profile-count.h
20700 (maybe_hot_count_p, counts_to_freqs): UPdate.
20701 * print-rtl-function.c: Do not include cfg.h
20702 * print-rtl.c: Include basic-block.h
20703 * profile-count.c: New file.
20704 * profile-count.h: New file.
20705 * profile.c (is_edge_inconsistent): Update.
20706 (correct_negative_edge_counts): Update.
20707 (is_inconsistent): Update.
20708 (set_bb_counts): Update.
20709 (read_profile_edge_counts): Update.
20710 (compute_frequency_overlap): Update.
20711 (compute_branch_probabilities): Update; Initialize and deinitialize
20713 (branch_prob): Update.
20714 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
20715 (edge_gcov_count): New.
20716 (bb_gcov_count): New.
20717 * shrink-wrap.c (try_shrink_wrapping): Update.
20718 * tracer.c (better_p): Update.
20719 * trans-mem.c (expand_transaction): Update.
20720 (ipa_tm_insert_irr_call): Update.
20721 (ipa_tm_insert_gettmclone_call): Update.
20722 * tree-call-cdce.c: Update.
20723 * tree-cfg.c (gimple_duplicate_sese_region): Update.
20724 (gimple_duplicate_sese_tail): Update.
20725 (gimple_account_profile_record): Update.
20726 (execute_fixup_cfg): Update.
20727 * tree-inline.c (copy_bb): Update.
20728 (copy_edges_for_bb): Update.
20729 (initialize_cfun): Update.
20730 (freqs_to_counts): Update.
20731 (copy_cfg_body): Update.
20732 (expand_call_inline): Update.
20733 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
20734 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
20735 (try_unroll_loop_completely): Update.
20736 (try_peel_loop): Update.
20737 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
20738 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
20739 * tree-ssa-loop-split.c (connect_loops): Update.
20740 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
20741 * tree-ssa-reassoc.c (branch_fixup): Update.
20742 * tree-ssa-tail-merge.c (replace_block_by): Update.
20743 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
20744 (compute_path_counts): Update.
20745 (update_profile): Update.
20746 (recompute_probabilities): Update.
20747 (update_joiner_offpath_counts): Update.
20748 (estimated_freqs_path): Update.
20749 (freqs_to_counts_path): Update.
20750 (clear_counts_path): Update.
20751 (ssa_fix_duplicate_block_edges): Update.
20752 (duplicate_thread_path): Update.
20753 * tree-switch-conversion.c (case_bit_test_cmp): Update.
20754 (struct switch_conv_info): Update.
20755 * tree-tailcall.c (decrease_profile): Update.
20756 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
20757 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
20758 * value-prof.c (check_counter): Update.
20759 (gimple_divmod_fixed_value): Update.
20760 (gimple_mod_pow2): Update.
20761 (gimple_mod_subtract): Update.
20762 (gimple_ic_transform): Update.
20763 (gimple_stringop_fixed_value): Update.
20764 * value-prof.h (gimple_ic): Update.
20766 2017-06-02 Carl Love <cel@us.ibm.com>
20768 * config/rs6000/rs6000-c: Add support for built-in functions
20769 vector double vec_doublee (vector signed int);
20770 vector double vec_doublee (vector unsigned int);
20771 vector double vec_doublee (vector float);
20772 vector double vec_doubleh (vector signed int);
20773 vector double vec_doubleh (vector unsigned int);
20774 vector double vec_doubleh (vector float);
20775 vector double vec_doublel (vector signed int);
20776 vector double vec_doublel (vector unsigned int);
20777 vector double vec_doublel (vector float);
20778 vector double vec_doubleo (vector signed int);
20779 vector double vec_doubleo (vector unsigned int);
20780 vector double vec_doubleo (vector float);.
20781 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
20782 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
20784 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
20785 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
20786 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
20788 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
20789 vec_doublel, vec_doubleh.
20790 * doc/extend.texi: Update the built-in documentation file for the
20791 new built-in functions.
20793 2017-06-02 David Malcolm <dmalcolm@redhat.com>
20796 * ipa-inline-analysis.c (free_growth_caches): Set
20797 edge_removal_hook_holder to NULL after removing it.
20799 2017-06-02 Sudakshina Das <sudi.das@arm.com>
20801 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
20802 comparision with zero.
20804 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
20805 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
20806 for early expansion of vec_min and vec_max builtins.
20807 (builtin_function_type): Add min/max unsigned variants to those
20808 identified as having unsigned arguments.
20810 2017-06-02 Olivier Hainque <hainque@adacore.com>
20812 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
20814 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20816 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
20817 Use VALL_F16 iterator rather than VALL.
20819 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20821 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
20822 Emit CBNZ inside loop when doing a strong exchange and comparing
20823 against zero. Generate the CC flags after the loop.
20825 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
20827 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
20828 (dl_section_ref): New.
20829 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
20830 On AIX, append an expression to subtract the size of the
20831 section length to dl_section_ref.
20833 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
20835 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
20836 for early expansion of vector absolute builtins.
20838 2017-06-02 Richard Biener <rguenther@suse.de>
20840 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
20841 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
20843 2017-06-02 Richard Biener <rguenther@suse.de>
20845 PR tree-optimization/80948
20846 * tree-tailcall.c (find_tail_calls): Track stmts to move in
20847 stmt order as well.
20849 2017-06-02 Richard Biener <rguenther@suse.de>
20851 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
20853 * tree-vect-stmts.c (process_use): Do not mark backedge defs
20854 for inductions as relevant.
20856 2017-06-02 Richard Biener <rguenther@suse.de>
20858 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
20859 (vectorizable_induction): ... this. Remove dead code.
20861 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
20863 * builtins. (expand_builtin_alloca): Remove second parameter and
20864 infer its value from the first parameter instead.
20865 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
20867 2017-06-02 Jakub Jelinek <jakub@redhat.com>
20869 PR rtl-optimization/80903
20870 * loop-doloop.c (add_test): Unshare sequence.
20872 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
20874 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
20876 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
20878 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
20880 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
20881 xlogue_layout::get_instance, logue_layout::xlogue_layout,
20882 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
20883 (xlogue_layout::get_stub_rtx): Make static.
20884 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
20885 (xlogue_layout::compute_stub_managed_regs): Rename to...
20886 (xlogue_layout::count_stub_managed_regs): ...this.
20887 (xlogue_layout::is_stub_managed_reg): New function.
20888 (xlogue_layout::m_stub_names): Rename to...
20889 (xlogue_layout::s_stub_names): ...this, make static.
20890 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
20891 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
20892 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
20893 xlogue_layout::s_stub_names): Instantiate statics.
20894 (stub_managed_regs): Remove.
20895 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
20896 (disable_call_ms2sysv_xlogues): Rename to...
20897 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
20898 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
20900 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
20901 change after reload_completed.
20902 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
20904 (ix86_expand_prologue): Likewise.
20905 (ix86_expand_epilogue): Likewise.
20906 (ix86_expand_split_stack_prologue): Likewise.
20907 (ix86_compute_frame_layout): Remove frame parameter ...
20908 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
20909 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
20911 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
20912 (ix86_frame): Move from here ...
20913 * config/i386/i386.h (ix86_frame): ... to here.
20914 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
20915 complete ix86_frame data structure instead. Remove some_ld_name.
20917 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
20919 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
20920 symbols that hold a DECL_VALUE_EXPR.
20922 2017-06-01 Martin Jambor <mjambor@suse.cz>
20924 PR tree-optimization/80898
20925 * tree-sra.c (process_subtree_disqualification): Removed.
20926 (disqualify_candidate): Do not acll
20927 process_subtree_disqualification.
20928 (subtree_mark_written_and_enqueue): New function.
20929 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
20930 RHS has been disqualified and re-queue LHS if necessary. Apart
20931 from that, ignore disqualified RHS.
20933 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20935 * config/s390/s390.c (s390_emit_epilogue): Disable early return
20936 address fetch for z10 or later.
20938 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20940 * config/arc/arc.md (tst_movb): Add guard when splitting.
20942 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20944 * config/arc/arc.c (arc_can_eliminate): Test against
20945 arc_frame_pointer_needed.
20947 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20949 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
20950 to prevent store reordering.
20951 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
20952 (type): Add block type.
20953 (stack_tie): Define special instruction to be used in
20956 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20958 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
20959 constraint. It is not valid for the pattern.
20960 (noncommutative_binary_comparison): Likewise.
20962 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20964 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
20967 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20969 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
20970 be used by the reg-alloc.
20972 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20974 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
20975 reg-alloc when having mul64 or mul32x16 instructions.
20976 (mulsidi3): Likewise.
20977 (umulsidi3): Likewise.
20978 (mulsi32x16): New pattern.
20979 (mulsi64): Likewise.
20980 (mulsidi64): Likewise.
20981 (umulsidi64): Likewise.
20982 (MUL32x16_REG): Define.
20983 (mul64_600): Use MUL32x16_REG.
20984 (mac64_600): Likewise.
20985 (umul64_600): Likewise.
20986 (umac64_600): Likewise.
20988 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20990 * config/arc/arc.md (mulsi3_700): Make it commutative.
20992 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
20994 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
20996 (*sign_extendsidi2_insn): Likewise for movstosw.
20998 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21000 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
21001 the type of the input discriminant value. Convert the
21002 discriminant value of signedness vary.
21004 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
21006 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
21007 Add to -Wall section.
21009 2017-06-01 Richard Biener <rguenther@suse.de>
21011 PR middle-end/66313
21012 * fold-const.c (fold_plusminus_mult_expr): If the factored
21013 factor may be zero use a wrapping type for the inner operation.
21014 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
21015 and handle moved defs.
21016 (process_assignment): Properly guard the unary op case. Return a
21017 tri-state indicating that moving the stmt before the call may allow
21018 to continue. Pass through to_move.
21019 (find_tail_calls): Handle moving unrelated defs before
21022 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
21025 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
21026 splitter result in the canonical way.
21028 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
21030 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
21031 also for 32bit target. Update insn attributes.
21032 (zero-extendsidi2 splitter): Allow all registers for operand 1.
21034 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
21036 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
21037 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
21038 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
21039 (_mm_maskz_min_ss): New intrinsics.
21041 2017-05-31 Martin Liska <mliska@suse.cz>
21043 * tree-vect-loop.c (vect_create_epilog_for_reduction):
21044 Change comment style to one we normally use.
21045 (vectorizable_reduction): Likewise.
21046 (vectorizable_induction): Likewise.
21047 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
21048 (vectorizable_call): Likewise.
21049 (vectorizable_simd_clone_call): Likewise.
21050 (vectorizable_conversion): Likewise.
21051 (vectorizable_assignment): Likewise.
21052 (vectorizable_shift): Likewise.
21053 (vectorizable_operation): Likewise.
21054 (vectorizable_store): Likewise.
21055 (vectorizable_load): Likewise.
21056 * tree-vectorizer.h: Likewise.
21058 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
21060 * passes.c (emergency_dump_function): New.
21061 * tree-pass.h (emergency_dump_function): Declare.
21062 * plugin.c (plugins_internal_error_function): Remove.
21063 * plugin.h (plugins_internal_error_function): Remove declaration.
21064 * toplev.c (internal_error_function): New static function. Use it...
21065 (general_init): ...here.
21067 2017-05-31 Graham Markall <graham.markall@embecosm.com>
21069 * config/arc/arc.c (arc_print_operand): Handle constant operands.
21070 (arc_rtx_costs): Add costs for new patterns.
21071 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
21072 * config/arc/predicates.md: Add _1_2_3_operand predicate.
21074 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21076 * tree-ssa-strlen.c (get_next_strinfo): New function.
21077 (get_stridx_plus_constant): Use it.
21078 (zero_length_string): Likewise.
21079 (adjust_related_strinfos): Likewise.
21080 (adjust_last_stmt): Likewise.
21082 2017-05-31 Richard Biener <rguenther@suse.de>
21085 * config/i386/i386.c (ix86_expand_builtin): Remove assert
21086 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
21088 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21090 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
21091 loop_vinfo argument and use of dependence distance vectors.
21092 Check instead whether the two references differ only in their
21093 initial value and assume that they have the same alignment if the
21094 difference is a multiple of the vector alignment.
21095 (vect_analyze_data_refs_alignment): Update call accordingly.
21097 2017-05-31 Martin Liska <mliska@suse.cz>
21100 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
21102 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21104 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
21105 (create_intersect_range_checks): Move from ...
21106 * tree-data-ref.c (create_intersect_range_checks_index)
21107 (create_intersect_range_checks): ... to here.
21108 (create_runtime_alias_checks): New function factored from ...
21109 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
21110 here. Call above function.
21111 * tree-data-ref.h (create_runtime_alias_checks): New function.
21113 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21115 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
21116 segment length for dr_b and compute it in wide_int.
21118 2017-05-31 Richard Biener <rguenther@suse.de>
21120 PR tree-optimization/80906
21121 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
21122 and pass through iv_map.
21123 (copy_bb_and_scalar_dependences): Adjust.
21124 (translate_pending_phi_nodes): Likewise.
21125 (copy_loop_close_phi_args): Handle code-generating IVs instead
21128 2017-05-30 David Malcolm <dmalcolm@redhat.com>
21130 * diagnostic-color.c (color_dict): Add "type-diff".
21131 (parse_gcc_colors): Update comment.
21132 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
21133 -fdiagnostics-show-template-tree and -fno-elide-type.
21134 (GCC_COLORS): Add type-diff to example.
21136 (-fdiagnostics-show-template-tree): New.
21137 (-fno-elide-type): New.
21138 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
21139 the pp_format_decoder callback. Call any m_format_postprocessor's
21141 (pretty_printer::pretty_printer): Initialize
21142 m_format_postprocessor.
21143 (pretty_printer::~pretty_printer): Delete any
21144 m_format_postprocessor.
21145 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
21146 (class format_postprocessor): New class.
21147 (struct pretty_printer::format_decoder): Document the new parameters.
21148 (struct pretty_printer::m_format_postprocessor): New field.
21149 * tree-diagnostic.c (default_tree_printer): Update for new
21150 bool and const char ** params.
21151 * tree-diagnostic.h (default_tree_printer): Likewise.
21153 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
21155 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
21156 (lwa_operand): Delete rs6000_gen_cell_microcode test.
21157 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21158 rs6000_gen_cell_microcode code.
21159 (rs6000_final_prescan_insn): Delete.
21160 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
21161 "warn-cell-microcode" entries.
21162 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
21163 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
21164 throughout. Change cc_reg_not_micro_cr0_operand to
21165 cc_reg_not_cr0_operand throughout.
21166 (*extendhi<mode>2_noload): Delete.
21167 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
21168 (mwarn-cell-microcode): Delete.
21169 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
21170 -mgen-cell-microcode and -mwarn-cell-microcode.
21172 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
21175 * config/i386/constraints.md (Yd): New constraint.
21177 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
21178 and (?Yd, r) alternatives. Update insn attributes.
21179 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
21180 and (?*Yd, r) alternatives. Update insn attributes.
21181 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
21183 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
21185 * gimplify.c (gimplify_modify_expr): Don't create a
21186 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
21189 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
21191 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
21193 2017-05-30 Richard Biener <rguenther@suse.de>
21195 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
21196 and reduc_def fields.
21197 (STMT_VINFO_REDUC_TYPE): New define.
21198 (STMT_VINFO_REDUC_DEF): Likewise.
21199 (vect_force_simple_reduction): Adjust prototype.
21200 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
21201 (vect_is_simple_reduction): Remove check_reduction argument.
21202 (vect_force_simple_reduction): Adjust and set
21203 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21204 (vectorizable_reduction): Do not re-do reduction analysis
21205 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21206 * tree-parloops.c (gather_scalar_reductions): Adjust.
21208 2017-05-30 Richard Biener <rguenther@suse.de>
21210 PR middle-end/80901
21211 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
21214 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
21216 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
21217 Introduce unknown_misalignment parameter and remove vf.
21218 (vect_peeling_hash_get_lowest_cost):
21219 Pass unknown_misalignment parameter.
21220 (vect_enhance_data_refs_alignment):
21221 Fix unsupportable data ref treatment.
21223 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21225 * tree-vect-data-refs.c (vect_get_data_access_cost):
21226 Workaround for SLP handling.
21227 (vect_enhance_data_refs_alignment):
21228 Compute costs for doing no peeling at all, compare to the best
21229 peeling costs so far and avoid peeling if cheaper.
21231 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21233 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
21234 Return peeling info and set costs to zero for unlimited cost
21236 (vect_enhance_data_refs_alignment): Also inspect all datarefs
21237 with unknown misalignment. Compute and costs for unknown
21238 misalignment, compare them to the costs for known misalignment
21239 and choose the cheapest for peeling.
21241 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21243 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
21244 (vect_get_peeling_costs_all_drs): Create function.
21245 (vect_peeling_hash_get_lowest_cost):
21246 Use vect_get_peeling_costs_all_drs.
21247 (vect_peeling_supportable): Create function.
21248 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
21250 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21252 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
21253 DR_HAS_NEGATIVE_STEP.
21254 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
21255 (vect_enhance_data_refs_alignment): Use.
21256 (vect_duplicate_ssa_name_ptr_info): Use.
21257 * tree-vectorizer.h (dr_misalignment): Use.
21258 (known_alignment_for_access_p): Use.
21260 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21263 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
21265 (has_section_name): New function.
21267 2017-05-30 Martin Liska <mliska@suse.cz>
21270 * auto-profile.c (get_function_decl_from_block): Fix
21273 2017-05-30 Richard Biener <rguenther@suse.de>
21275 PR middle-end/80876
21276 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
21278 2017-05-30 Martin Liska <mliska@suse.cz>
21280 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
21281 * dumpfile.h (struct dump_file_info): Remove ctors.
21283 2017-05-30 Martin Liska <mliska@suse.cz>
21285 * predict.def: Fix GNU coding style.
21287 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21289 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
21290 Mark 'to' argument with ATTRIBUTE_UNUSED.
21292 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21294 * config/xtensa/xtensa.c (xtensa_emit_call): Use
21295 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
21296 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
21299 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
21301 * doc/install.texi (Options specification): Restore entry of
21302 --enable-sjlj-exceptions.
21304 2017-05-27 Michael Eager <eager@eagercon.com>
21307 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
21309 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
21311 * config/microblaze/microblaze.h
21312 (FIXED_REGISTERS): Update in macro.
21313 (CALL_USED_REGISTERS): Update in macro.
21315 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
21317 * doc/install.texi: Add links to macOS binary distributions.
21319 2017-05-27 Jakub Jelinek <jakub@redhat.com>
21323 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21325 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
21327 2017-05-26 Martin Liska <mliska@suse.cz>
21329 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
21331 2017-05-26 Martin Liska <mliska@suse.cz>
21333 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
21334 always leading ';; '.
21335 (dump_bb_info): Likewise.
21336 (brief_dump_cfg): Likewise.
21337 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
21338 * dumpfile.c: Remove usage of TDF_VERBOSE.
21339 * dumpfile.h (enum dump_kind): Likewise.
21340 (dump_gimple_bb_header): Do not use TDF_COMMENT.
21341 * print-tree.c (debug_verbose): Remove.
21342 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
21343 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
21344 * tree-diagnostic.c (default_tree_printer): Replace
21345 TDF_DIAGNOSTIC with TDF_SLIM.
21347 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21349 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
21350 in parameter loop, rather than loop_vinfo.
21351 (create_intersect_range_checks): Ditto.
21352 (vect_create_cond_for_alias_checks): Update call to above functions.
21354 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21356 PR tree-optimization/80815
21357 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
21358 for merging runtime alias checks. Handle negative DR_STEPs.
21360 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21362 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
21364 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
21365 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
21366 out code pruning runtime alias checks.
21367 * tree-data-ref.c (prune_runtime_alias_test_list): New function
21368 factored out from above.
21369 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
21371 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
21373 (prune_runtime_alias_test_list): New decalaration.
21375 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21377 * tree-vect-data-refs.c (compare_tree): Rename and move ...
21378 * tree-data-ref.c (data_ref_compare_tree): ... to here.
21379 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
21380 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
21381 (operator==, comp_dr_with_seg_len_pair): Ditto.
21382 (vect_prune_runtime_alias_test_list): Ditto.
21384 2017-05-26 Martin Liska <mliska@suse.cz>
21387 * params.def: Bound partial-inlining-entry-probability param.
21389 2017-05-26 Marek Polacek <polacek@redhat.com>
21392 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
21395 2017-05-26 Richard Biener <rguenther@suse.de>
21397 PR tree-optimization/80842
21398 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
21401 2017-05-26 Richard Biener <rguenther@suse.de>
21403 PR tree-optimization/80844
21404 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
21406 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
21408 * doc/md.texi (Machine Constraints): Update x86 family
21409 machine constraints section to match 'config/i386/constraints.md'.
21411 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
21413 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
21415 2017-05-25 Nathan Sidwell <nathan@acm.org>
21417 * doc/invoke.texi (--enable-languages): Update documentation.
21419 2017-05-25 Martin Liska <mliska@suse.cz>
21421 * dumpfile.c: Add TDF_FOLDING.
21422 * dumpfile.h (enum dump_kind): Likewise.
21423 * genmatch.c (dt_simplify::gen_1): Use it.
21425 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21427 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
21429 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21431 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
21432 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
21434 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21436 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
21437 * match.pd (X == C): Rewrite it here.
21438 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
21439 with_certain_nonzero_bits2): New predicates.
21440 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
21442 2017-05-24 Nathan Sidwell <nathan@acm.org>
21444 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
21447 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
21450 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
21452 * config/powerpcspe: New port. Files are copied from the rs6000
21453 port, with "rs6000" in filenames replaced by "powerpcspe".
21455 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
21457 PR rtl-optimization/80754
21458 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
21460 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
21462 * config/sparc/sparc.md (length): Return the correct value for -mflat
21463 sibcalls to match output_sibcall.
21465 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
21469 * config/rs6000/rs6000.c (struct machine_function): Add new field
21471 (rs6000_get_separate_components): Init that field, use it.
21472 (rs6000_components_for_bb): Use the field.
21474 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
21476 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
21478 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
21480 PR middle-end/80823
21481 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
21483 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21486 * config/s390/s390.c (s390_check_qrst_address): Check incoming
21487 address against address_operand predicate.
21488 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
21490 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
21492 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
21493 parameters passed indirectly.
21495 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
21497 * config/i386/i386.md (*movdi_internal): Remove SSE4
21498 alternative 18 (?r, *v). Update insn attributes.
21499 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
21500 Update insn attributes.
21501 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
21502 Update insn attributes.
21503 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
21504 alternative 1 (r, v). Remove isa attribute.
21505 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
21506 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
21507 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
21509 2017-05-23 Tom de Vries <tom@codesourcery.com>
21511 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
21514 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21516 * cgraphunit.c (symbol_table::process_new_functions): Update.
21517 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
21518 (inline_generate_summary): Rename to ...
21519 (ipa_fn_summary_generate): ... this one.
21520 (inline_read_summary): Rename to ...
21521 (ipa_fn_summary_read): ... this one.
21522 (inline_write_summary): Rename to ...
21523 (ipa_fn_summary_write): ... this one.
21524 (inline_free_summary): Rename to ...
21525 (ipa_free_fn_summary): ... this one.
21526 (pass_data_local_fn_summary, pass_local_fn_summary,
21527 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
21528 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
21529 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
21530 make_pass_ipa_fn_summary): New.
21531 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
21532 inline_write_summary, inline_free_summary): Remove.
21533 (ipa_free_fn_summary) : New.
21534 * ipa-inline.c (ipa_inline): Update.
21535 (pass_ipa_inline): Do not generate summaries.
21536 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
21538 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
21539 and add pass_ipa_fn_summary.
21540 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
21542 (make_pass_inline_parameters): Remove.
21544 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
21546 * omp-low.c (struct omp_context): Remove "default_kind" member.
21549 * omp-offload.c (execute_oacc_device_lower): Remove the
21550 parallelism dimensions function attributes for unparallelized
21551 OpenACC kernels constructs.
21553 2017-05-23 Martin Liska <mliska@suse.cz>
21555 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
21557 (cgraph_edge::make_speculative): Likewise.
21558 (cgraph_edge::resolve_speculation): Likewise.
21559 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
21560 (cgraph_node::dump): Likewise.
21561 * cgraph.h: Likewise.
21562 * cgraphunit.c (analyze_functions): Likewise.
21563 (symbol_table::compile): Likewise.
21564 * ipa-cp.c (print_all_lattices): Likewise.
21565 (determine_versionability): Likewise.
21566 (initialize_node_lattices): Likewise.
21567 (ipcp_verify_propagated_values): Likewise.
21568 (estimate_local_effects): Likewise.
21569 (update_profiling_info): Likewise.
21570 (create_specialized_node): Likewise.
21571 (perhaps_add_new_callers): Likewise.
21572 (decide_about_value): Likewise.
21573 (decide_whether_version_node): Likewise.
21574 (identify_dead_nodes): Likewise.
21575 (ipcp_store_bits_results): Likewise.
21576 * ipa-devirt.c (dump_targets): Likewise.
21577 (ipa_devirt): Likewise.
21578 * ipa-icf.c (sem_item::dump): Likewise.
21579 (sem_function::equals): Likewise.
21580 (sem_variable::equals): Likewise.
21581 (sem_item_optimizer::read_section): Likewise.
21582 (sem_item_optimizer::execute): Likewise.
21583 (congruence_class::dump): Likewise.
21584 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
21585 (dump_inline_summary): Likewise.
21586 (estimate_node_size_and_time): Likewise.
21587 (inline_analyze_function): Likewise.
21588 * ipa-inline-transform.c (inline_call): Likewise.
21589 * ipa-inline.c (report_inline_failed_reason): Likewise.
21590 (want_early_inline_function_p): Likewise.
21591 (edge_badness): Likewise.
21592 (update_edge_key): Likewise.
21593 (inline_small_functions): Likewise.
21594 * ipa-profile.c (ipa_profile): Likewise.
21595 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
21596 (ipa_make_edge_direct_to_target): Likewise.
21597 (remove_described_reference): Likewise.
21598 (ipa_impossible_devirt_target): Likewise.
21599 (propagate_controlled_uses): Likewise.
21600 (ipa_print_node_params): Likewise.
21601 (ipcp_transform_function): Likewise.
21602 * ipa-pure-const.c (pure_const_read_summary): Likewise.
21603 (propagate_pure_const): Likewise.
21604 * ipa-reference.c (generate_summary): Likewise.
21605 (read_write_all_from_decl): Likewise.
21606 (propagate): Likewise.
21607 (ipa_reference_read_optimization_summary): Likewise.
21608 * ipa-utils.c (ipa_merge_profiles): Likewise.
21609 * ipa.c (walk_polymorphic_call_targets): Likewise.
21610 (symbol_table::remove_unreachable_nodes): Likewise.
21611 (ipa_single_use): Likewise.
21612 * passes.c (execute_todo): Likewise.
21613 * predict.c (drop_profile): Likewise.
21614 * symtab.c (symtab_node::get_dump_name): New function.
21615 (symtab_node::dump_name): Likewise.
21616 (symtab_node::dump_asm_name): Likewise.
21617 (symtab_node::dump_references): Likewise.
21618 (symtab_node::dump_referring): Likewise.
21619 (symtab_node::dump_base): Likewise.
21620 (symtab_node::debug_symtab): Likewise.
21621 * tree-sra.c (convert_callers_for_node): Likewise.
21622 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
21623 * value-prof.c (init_node_map): Likewise.
21625 2017-05-23 Martin Liska <mliska@suse.cz>
21627 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
21628 and symtab_node::debug_symtab to symbol_table::debug.
21629 * cgraphunit.c (analyze_functions): Use the renamed function.
21630 (symbol_table::compile): Likewise.
21631 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
21632 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
21633 * passes.c (execute_todo): Likewise.
21634 * symtab.c (symbol_table::dump): New function.
21635 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
21637 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21639 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
21640 that nonconst implies exec.
21642 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21644 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
21645 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
21646 (inline_edge_summary_vec): Turn into ...
21647 (ipa_call_summaries): ... this one.
21648 (redirect_to_unreachable, edge_set_predicate,
21649 evaluate_properties_for_edge, inline_summary_alloc,
21650 reset_ipa_call_summary, reset_inline_summary,
21651 inline_summary_t::duplicate): Update.
21652 (inline_edge_duplication_hook): Turn to ...
21653 (ipa_call_summary_t::duplicate): ... this one.
21654 (inline_edge_removal_hook): Turn to ...
21655 (ipa_call_summary_t::remove): ... this one.
21656 (dump_inline_edge_summary): Turn to ...
21657 (dump_ipa_call_summary): ... this one.
21658 (estimate_function_body_sizes): Update.
21659 (inline_update_callee_summaries): Update.
21660 (remap_edge_change_prob): Update.
21661 (remap_edge_summaries): Update.
21662 (inline_merge_summary): Update.
21663 (do_estimate_edge_time): Update.
21664 (inline_generate_summary): Update.
21665 (inline_read_section): Update.
21666 (inline_read_summary): Update.
21667 (inline_free_summary): Update.
21668 * ipa-inline.c (can_inline_edge_p): Update.
21669 (compute_inlined_call_time): Update.
21670 (want_inline_small_function_p): Update.
21671 (edge_badness): Update.
21672 (early_inliner): Update.
21673 * ipa-inline.h (inline_edge_summary): Turn to ...
21674 (ipa_call_summary): ... this one.
21675 (ipa_call_summary_t): New class.
21676 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
21677 (ipa_call_summaries): New.
21678 (inline_edge_summary): Remove.
21679 (estimate_edge_growth): Update.
21680 * ipa-profile.c (ipa_propagate_frequency_1): Update.
21681 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21682 * ipa-split.c (execute_split_functions): Update.
21683 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
21685 2017-05-23 Tom de Vries <tom@codesourcery.com>
21687 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
21688 attributes): Document rdrand effective target.
21690 2017-05-23 Tom de Vries <tom@codesourcery.com>
21692 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
21693 attributes): Sort alphabetically.
21695 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
21697 * config/avr/genmultilib.awk: Use gsub instead of gensub.
21699 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
21702 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
21703 V2DF/V2DI splat into two separate patterns, one that handles
21704 registers, and the other that only handles memory. Drop support
21705 for splatting from a GPR on ISA 2.07 and then splitting the
21706 splat into direct move and splat.
21707 (vsx_splat_<mode>_reg): Likewise.
21708 (vsx_splat_<mode>_mem): Likewise.
21710 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
21712 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
21714 2017-05-22 Jakub Jelinek <jakub@redhat.com>
21716 PR middle-end/80809
21717 * omp-low.c (finish_taskreg_remap): New function.
21718 (finish_taskreg_scan): If unit size of ctx->record_type
21719 is non-constant, unshare the size expression and replace
21720 decls in it with possible outer var refs.
21722 PR middle-end/80809
21723 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
21724 GOVD_SHARED rather than GOVD_PRIVATE with it.
21725 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
21726 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
21728 PR middle-end/80853
21729 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
21730 as last argument to build_outer_var_ref for pointer bases of array
21731 section reductions.
21733 2017-05-19 Martin Sebor <msebor@redhat.com>
21735 * print-tree.c (print_node): Print DECL_READ_P flag.
21737 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21739 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
21740 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
21741 * cgraph.c: Likewise.
21742 * cgraphunit.c: Likewise.
21743 * gengtype.c: Likewise.
21744 * ipa-cp.c: Likewise.
21745 * ipa-devirt.c: Likewise.
21746 * ipa-icf.c: Likewise.
21747 * ipa-predicate.c: Likewise.
21748 * ipa-profile.c: Likewise.
21749 * ipa-prop.c: Likewise.
21750 * ipa-split.c: Likewise.
21752 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
21753 edge_predicate_pool, dump_inline_hints,
21754 inline_summary::account_size_time, redirect_to_unreachable,
21755 edge_set_predicate, set_hint_predicate,
21756 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
21757 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
21758 inline_summary_t::remove, remap_hint_predicate_after_duplication,
21759 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
21760 ipa_call_summary_t::remove, initialize_growth_caches,
21761 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
21762 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
21763 mark_modified, unmodified_parm_1, unmodified_parm,
21764 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
21765 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
21766 compute_bb_predicates, will_be_nonconstant_expr_predicate,
21767 will_be_nonconstant_predicate, record_modified_bb_info,
21768 get_minimal_bb, record_modified, param_change_prob,
21769 phi_result_unknown_predicate, predicate_for_phi_result,
21770 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
21771 estimate_function_body_sizes, compute_inline_parameters,
21772 compute_inline_parameters_for_curren, pass_data_inline_parameters,
21773 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
21774 inline_update_callee_summaries, remap_edge_change_prob,
21775 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
21776 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
21777 inline_analyze_function, inline_summary_t::insert,
21778 inline_generate_summary, read_ipa_call_summary, inline_read_section,
21779 inline_read_summary, write_ipa_call_summary, inline_write_summary,
21780 inline_free_summary): Move to ipa-fnsummary.h
21781 (predicate_t): Remove.
21782 * ipa-fnsummary.c: New file.
21783 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
21784 (enum inline_hints_vals, inline_hints, agg_position_info,
21785 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
21786 inline_summaries, ipa_call_summary, ipa_call_summary_t,
21787 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
21788 dump_inline_summary, dump_inline_hints, inline_generate_summary,
21789 inline_read_summary, inline_write_summary, inline_free_summary,
21790 inline_analyze_function, initialize_inline_failed,
21791 inline_merge_summary, inline_update_overall_summary,
21792 compute_inline_parameters): Move to ipa-fnsummary.h
21793 * ipa-fnsummary.h: New file.
21794 * ipa-inline-transform.h: Include ipa-inline.h.
21795 * ipa-inline.c: LIkewise.
21797 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21799 * ipa-inline.c (edge_badness): Use inlined_time instead of
21800 inline_summaries->get.
21802 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21804 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
21806 2017-05-22 Nathan Sidwell <nathan@acm.org>
21808 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
21809 (fdump-lang): Document 'raw' option.
21810 * dumpfile.h (TDI_tu): Delete.
21811 * dumpfile.c (dump_files): Remove translation-unit.
21812 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
21814 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
21816 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
21817 command option from $(AWK) call.
21818 * config/avr/genmultilib.awk: Simplify and rewrite so that it
21819 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
21820 [FORMAT]: Remove handling of variable.
21821 * config/avr/t-multilib: Regenerate.
21823 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21825 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
21827 (dump_inline_summary): Do not print self_time.
21828 (estimate_function_body_sizes): Do not set self_time.
21829 (compute_inline_parameters): Likewise.
21830 (inline_read_section, inline_write_summary): Do not stream self_time.
21831 * ipa-inline.h (inline_summary): Drop self_time.
21833 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21835 * ipa-inline-analysis.c (account_size_time): Rename to ...
21836 (inline_summary::account_size_time): ... this one.
21837 (reset_ipa_call_summary): Turn to ...
21838 (ipa_call_summary::reset): ... this one.
21839 (reset_inline_summary): Turn to ...
21840 (inline_summary::reset): ... this one.
21841 (inline_summary_t::remove): Update.
21842 (inline_summary_t::duplicate): Update.
21843 (ipa_call_summary_t::remove): Update.
21844 (dump_inline_summary): Update.
21845 (estimate_function_body_sizes): Update.
21846 (compute_inline_parameters): Update.
21847 (estimate_node_size_and_time): Update.
21848 (inline_merge_summary): Update.
21849 (inline_update_overall_summary): Update.
21850 (inline_read_section): Update.
21851 (inline_write_summary): Update.
21852 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
21853 add account_size_time and reset member functions.
21854 (ipa_call_summary): Add reset function.
21855 * ipa-predicate.h (predicate::operator &): Constify.
21857 2017-05-22 Richard Biener <rguenther@suse.de>
21859 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
21861 2017-05-19 Jason Merrill <jason@redhat.com>
21863 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
21865 2017-05-19 Marek Polacek <polacek@redhat.com>
21868 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
21869 TYPE_OVERFLOW_WRAPS checks.
21871 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
21873 * tree-core.h (enum omp_clause_default_kind): Add
21874 "OMP_CLAUSE_DEFAULT_PRESENT".
21875 * tree-pretty-print.c (dump_omp_clause): Handle it.
21876 * gimplify.c (enum gimplify_omp_var_data): Add
21877 "GOVD_MAP_FORCE_PRESENT".
21878 (gimplify_adjust_omp_clauses_1): Map it to
21879 "GOMP_MAP_FORCE_PRESENT".
21880 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
21882 * gimplify.c (oacc_default_clause): Clarify.
21884 2017-05-19 Nathan Sidwell <nathan@acm.org>
21886 LANG_HOOK_REGISTER_DUMPS
21887 * toplev.c (general_init): Call register dump lang hook.
21888 * doc/invoke.texi: Document -fdump-lang option family.
21889 * dumpfile.c (dump_files): Remove class dump here.
21890 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
21891 * dumpfile.h (tree_dump_index): Remove TDI_class.
21892 * langhooks-def.h (lhd_register_dumps): Declare.
21893 (LANG_HOOKS_REGISTER_DUMPS): Define.
21894 (LANG_HOOKS_INITIALIZER): Add it.
21895 * langhooks.c (lhd_register_dumps): Define.
21896 * langhooks.h (struct lang_hooks): Add register_dumps.
21898 2017-05-19 Nathan Sidwell <nathan@acm.org>
21900 * context.h (context::set_passes): New.
21901 * context.c (context::context): Do not create pass manager.
21902 * toplev.c (general_init): Create pass manager here.
21904 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
21906 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
21907 use this splitter if two add or or instructions would also work for
21908 the constant we want to generate.
21910 2017-05-19 Richard Biener <rguenther@suse.de>
21913 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
21914 predicate evaluation.
21916 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
21918 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
21920 * ipa-inline.c (want_inline_small_function_p): Do not cast to
21923 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
21925 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
21926 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
21927 (inline_edge_summary_vec): Turn into ...
21928 (ipa_call_summaries): ... this one.
21929 (redirect_to_unreachable, edge_set_predicate,
21930 evaluate_properties_for_edge, inline_summary_alloc,
21931 reset_ipa_call_summary, reset_inline_summary,
21932 inline_summary_t::duplicate): Update.
21933 (inline_edge_duplication_hook): Turn to ...
21934 (ipa_call_summary_t::duplicate): ... this one.
21935 (inline_edge_removal_hook): Turn to ...
21936 (ipa_call_summary_t::remove): ... this one.
21937 (dump_inline_edge_summary): Turn to ...
21938 (dump_ipa_call_summary): ... this one.
21939 (estimate_function_body_sizes): Update.
21940 (inline_update_callee_summaries): Update.
21941 (remap_edge_change_prob): Update.
21942 (remap_edge_summaries): Update.
21943 (inline_merge_summary): Update.
21944 (do_estimate_edge_time): Update.
21945 (inline_generate_summary): Update.
21946 (inline_read_section): Update.
21947 (inline_read_summary): Update.
21948 (inline_free_summary): Update.
21949 * ipa-inline.c (can_inline_edge_p): Update.
21950 (compute_inlined_call_time): Update.
21951 (want_inline_small_function_p): Update.
21952 (edge_badness): Update.
21953 (early_inliner): Update.
21954 * ipa-inline.h (inline_edge_summary): Turn to ...
21955 (ipa_call_summary): ... this one.
21956 (ipa_call_summary_t): New class.
21957 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
21958 (ipa_call_summaries): New.
21959 (inline_edge_summary): Remove.
21960 (estimate_edge_growth): Update.
21961 * ipa-profile.c (ipa_propagate_frequency_1): Update.
21962 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21963 * ipa-split.c (execute_split_functions): Update.
21964 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
21966 2017-05-19 Richard Biener <rguenther@suse.de>
21968 PR middle-end/80764
21969 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
21971 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
21973 * config/rs6000/rs6000.c (struct machine_function): Add field
21974 fpr_is_wrapped_separately.
21975 (rs6000_get_separate_components): Use 64 components. Handle the
21976 new FPR components.
21977 (rs6000_components_for_bb): Handle the FPR components.
21978 (rs6000_emit_prologue_components): Handle the FPR components.
21979 (rs6000_emit_epilogue_components): Handle the FPR components.
21980 (rs6000_set_handled_components): Handle the FPR components.
21981 (rs6000_emit_prologue): Don't output prologue code for those FPRs
21982 that are already separately shrink-wrapped.
21983 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
21984 that are already separately shrink-wrapped.
21986 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
21989 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
21992 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
21993 (define_peephole2 for Altivec d-form load): Add peepholes to catch
21994 cases where the register allocator uses a move and an offsettable
21995 memory operation to/from a FPR register on ISA 2.06/2.07.
21996 (define_peephole2 for Altivec d-form store): Likewise.
21998 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
22001 * config/i386/mmx.md (*mov<mode>_internal): Enable
22002 alternatives 11, 12, 13 and 14 also for 32bit targets.
22003 Remove alternatives 15, 16, 17 and 18.
22004 * config/i386/sse.md (vec_concatv2di): Change
22005 alternative (!x, *y) to (x, ?!*Yn).
22007 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
22009 * dumpfile.h (enum dump_kind): Remove stray comma.
22011 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22013 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
22014 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
22015 predicate::num_conditions
22016 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
22017 (CHANGED): turn into predicate::changed.
22018 (agg_position_info): Move to ipa-predicate.h
22019 (add_condition, predicate::add_clause, predicate::operator &=,
22020 predicate::or_with, predicate::evaluate, predicate::probability,
22021 dump_condition, dump_clause, predicate::dump,
22022 predicate::remap_after_duplication, predicate::remap_after_inlining,
22023 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
22024 (evaluate_conditions_for_known_args): Update.
22025 (set_cond_stmt_execution_predicate): Update.
22026 * ipa-inline.h: Include ipa-predicate.h
22027 (condition, inline_param_summary, conditions, agg_position_info,
22028 predicate): Move to ipa-predicate.h
22029 * ipa-predicate.c: New file.
22030 * ipa-predicate.h: New file.
22032 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
22034 * final.c (leaf_function_p): Check we are not in a sequence.
22036 2017-05-18 Martin Liska <mliska@suse.cz>
22038 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
22039 * dumpfile.c (dump_register): Use new enum dump_kind.
22040 (get_dump_file_name): Likewise.
22041 (dump_enable_all): Likewise.
22042 (dump_switch_p_1): Likewise.
22043 (enable_rtl_dump_file): Remove usage of TDF_RTL.
22044 * dumpfile.h (enum dump_kind): New enum type.
22045 (struct dump_file_info): Create constructor and
22046 format fields and comments.
22047 * passes.c (pass_manager::register_one_dump_file):
22049 * statistics.c (statistics_early_init): Likewise.
22050 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
22051 TDF_TREE with TDF_SLIM.
22052 (gather_memory_references_ref): Likewise.
22054 2017-05-18 Martin Liska <mliska@suse.cz>
22056 * vec.h (struct vnull): Use it.
22058 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22060 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
22061 (true_predicate, false_predicate, true_predicate_p,
22062 false_predicate_p): Remove.
22063 (single_cond_predicate, not_inlined_predicate): Turn to member function
22065 (add_condition): Update.
22066 (add_clause): Turn to...
22067 (predicate::add_clause): ... this one; update; allow passing NULL
22069 (and_predicates): Turn to ...
22070 (predicate::operator &=): ... this one.
22071 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
22072 (or_predicates): Turn to ...
22073 (predicate::or_with): ... this one.
22074 (evaluate_predicate): Turn to ...
22075 (predicate::evaluate): ... this one.
22076 (predicate_probability): Turn to ...
22077 (predicate::probability): ... this one.
22078 (dump_condition): Update.
22079 (dump_predicate): Turn to ...
22080 (predicate::dump): ... this one.
22081 (account_size_time): Update.
22082 (edge_set_predicate): Update.
22083 (set_hint_predicate): UPdate.
22084 (evaluate_conditions_for_known_args): Update.
22085 (evaluate_properties_for_edge): Update.
22086 (remap_predicate_after_duplication): Turn to...
22087 (predicate::remap_after_duplication): ... this one.
22088 (remap_hint_predicate_after_duplication): Update.
22089 (inline_summary_t::duplicate): UPdate.
22090 (dump_inline_edge_summary): Update.
22091 (dump_inline_summary): Update.
22092 (set_cond_stmt_execution_predicate): Update.
22093 (set_switch_stmt_execution_predicate): Update.
22094 (compute_bb_predicates): Update.
22095 (will_be_nonconstant_expr_predicate): Update.
22096 (will_be_nonconstant_predicate): Update.
22097 (phi_result_unknown_predicate): Update.
22098 (predicate_for_phi_result): Update.
22099 (array_index_predicate): Update.
22100 (estimate_function_body_sizes): Update.
22101 (estimate_node_size_and_time): Update.
22102 (estimate_ipcp_clone_size_and_time): Update.
22103 (remap_predicate): Rename to ...
22104 (predicate::remap_after_inlining): ... this one.
22105 (remap_hint_predicate): Update.
22106 (inline_merge_summary): Update.
22107 (inline_update_overall_summary): Update.
22108 (estimate_size_after_inlining): Update.
22109 (read_predicate): Rename to ...
22110 (predicate::stream_in): ... this one.
22111 (read_inline_edge_summary): Update.
22112 (write_predicate): Rename to ...
22113 (predicate::stream_out): ... this one.
22114 (write_inline_edge_summary): Update.
22115 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
22116 (clause_t): Turn to uint32_t
22117 (predicate): Turn to class; implement constructor and operators
22119 (size_time_entry): Update.
22120 (inline_summary): Update.
22121 (inline_edge_summary): Update.
22123 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
22125 * fold-const.c (fold_binary_loc): Move transformation...
22126 * match.pd (C - X CMP X): ... here.
22128 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
22130 * config/sparc/sparc.c (sparc_option_override): Set function
22131 alignment for -mcpu=niagara7 to 64 to match the I$ line.
22132 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
22134 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
22136 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
22138 2017-05-18 Marek Polacek <polacek@redhat.com>
22141 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
22142 (pass_ubsan::execute): Call gimple_assign_single_p instead of
22143 gimple_assign_load_p.
22145 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
22147 PR middle-end/80692
22148 * real.c (do_compare): Give decimal_do_compare preference over
22149 comparing just the signs.
22151 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
22153 * doc/md.texi (Canonicalization of Instructions): Describe the
22154 canonical form of instructions that inherently set a condition
22157 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
22159 PR middle-end/80775
22160 * tree-cfg.c: Move deletion of unreachable case statements to after
22161 the merging of consecutive case labels.
22163 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
22165 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
22166 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
22167 restoring of callee-saved registers.
22169 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
22171 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
22172 * config/visium/visium.c (single_set_and_flags): Likewise.
22173 * config/visium/visium.md (Substitutions): Likewise.
22175 2017-05-17 Martin Liska <mliska@suse.cz>
22177 * cfg.c: Introduce dump_flags_t type and
22178 use it instead of int type.
22180 * cfghooks.c: Likewise.
22181 * cfghooks.h (struct cfg_hooks): Likewise.
22182 * cfgrtl.c: Likewise.
22183 * cfgrtl.h: Likewise.
22184 * cgraph.c (cgraph_node::get_body): Likewise.
22185 * coretypes.h: Likewise.
22186 * domwalk.c: Likewise.
22187 * domwalk.h: Likewise.
22188 * dumpfile.c (struct dump_option_value_info): Likewise.
22189 (dump_enable_all): Likewise.
22190 (dump_switch_p_1): Likewise.
22191 (opt_info_switch_p): Likewise.
22192 * dumpfile.h (enum tree_dump_index): Likewise.
22193 (struct dump_file_info): Likewise.
22194 * genemit.c: Likewise.
22195 * generic-match-head.c: Likewise.
22196 * gengtype.c (open_base_files): Likewise.
22197 * gimple-pretty-print.c: Likewise.
22198 * gimple-pretty-print.h: Likewise.
22199 * graph.c (print_graph_cfg): Likewise.
22200 * graphite-scop-detection.c (dot_all_sese): Likewise.
22201 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
22202 * loop-unroll.c (report_unroll): Likewise.
22203 * passes.c (pass_manager::register_one_dump_file): Likewise.
22204 * print-tree.c: Likewise.
22205 * statistics.c: Likewise.
22206 * tree-cfg.c: Likewise.
22207 * tree-cfg.h: Likewise.
22208 * tree-dfa.c: Likewise.
22209 * tree-dfa.h: Likewise.
22210 * tree-dump.c (dump_function): Likewise.
22211 * tree-dump.h (struct dump_info): Likewise.
22212 * tree-pretty-print.c: Likewise.
22213 * tree-pretty-print.h: Likewise.
22214 * tree-ssa-live.c: Likewise.
22215 * tree-ssa-live.h: Likewise.
22216 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
22217 * tree-vect-loop.c: Likewise.
22218 * tree-vect-slp.c: Likewise.
22220 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
22221 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22223 PR tree-optimization/80457
22224 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
22225 of all arguments to a statement as scalar_to_vec operations.
22226 (vectorizable_call): Adjust call to vect_model_simple_cost for
22228 (vectorizable_conversion): Likewise.
22229 (vectorizable_assignment): Likewise.
22230 (vectorizable_shift): Likewise.
22231 (vectorizable_operation): Likewise.
22232 (vectorizable_comparison): Likewise.
22233 (vect_is_simple_cond): Record the def types for operands.
22234 (vectorizable_condition): Likewise, call vect_model_simple_cost.
22235 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
22236 for statement argument count.
22238 2017-05-16 Carl Love <cel@us.ibm.com>
22240 * config/rs6000/rs6000-c: Add support for built-in functions
22241 vector unsigned long long vec_bperm (vector unsigned long long,
22242 vector unsigned char)
22243 vector signed long long vec_mule (vector signed int,
22245 vector unsigned long long vec_mule (vector unsigned int,
22246 vector unsigned int)
22247 vector signed long long vec_mulo (vector signed int,
22249 vector unsigned long long vec_mulo (vector unsigned int,
22250 vector unsigned int)
22251 vector signed char vec_sldw (vector signed char,
22252 vector signed char,
22254 vector unsigned char vec_sldw (vector unsigned char,
22255 vector unsigned char,
22257 vector signed short vec_sldw (vector signed short,
22258 vector signed short,
22260 vector unsigned short vec_sldw (vector unsigned short,
22261 vector unsigned short,
22263 vector signed int vec_sldw (vector signed int,
22266 vector unsigned int vec_sldw (vector unsigned int,
22267 vector unsigned int,
22269 vector signed long long vec_sldw (vector signed long long,
22270 vector signed long long,
22272 vector unsigned long long vec_sldw (vector unsigned long long,
22273 vector unsigned long long,
22275 * config/rs6000/rs6000-c: Add support for built-in functions
22276 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
22277 * config/rs6000/altivec.h: Add defintion for vec_sldw.
22278 * doc/extend.texi: Update the built-in documentation for the
22279 new built-in functions.
22281 2017-05-16 Marek Polacek <polacek@redhat.com>
22285 * tree.c (save_expr): Don't fold the expression.
22287 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
22289 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
22290 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
22291 and (?*y,m). Update insn attributes.
22293 2017-05-16 Martin Liska <mliska@suse.cz>
22295 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
22296 flags argument of print_gimple_stmt, print_gimple_expr,
22297 print_generic_stmt and print_generic_expr.
22298 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
22299 * coretypes.h: Likewise.
22300 * except.c (dump_eh_tree): Likewise.
22301 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
22302 * gimple-pretty-print.h: Likewise.
22303 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
22304 (backprop::push_to_worklist): Likewise.
22305 (backprop::pop_from_worklist): Likewise.
22306 (backprop::process_use): Likewise.
22307 (backprop::intersect_uses): Likewise.
22308 (note_replacement): Likewise.
22309 * gimple-ssa-store-merging.c
22310 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
22311 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
22312 (pass_store_merging::execute): Likewise.
22313 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
22314 (ssa_base_cand_dump_callback): Likewise.
22315 (dump_incr_vec): Likewise.
22316 (replace_refs): Likewise.
22317 (replace_mult_candidate): Likewise.
22318 (create_add_on_incoming_edge): Likewise.
22319 (create_phi_basis): Likewise.
22320 (insert_initializers): Likewise.
22321 (all_phi_incrs_profitable): Likewise.
22322 (introduce_cast_before_cand): Likewise.
22323 (replace_one_candidate): Likewise.
22324 * gimplify.c (gimplify_expr): Likewise.
22325 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
22326 (set_rename): Likewise.
22327 (rename_uses): Likewise.
22328 (copy_loop_phi_nodes): Likewise.
22329 (add_close_phis_to_merge_points): Likewise.
22330 (copy_loop_close_phi_args): Likewise.
22331 (copy_cond_phi_args): Likewise.
22332 (graphite_copy_stmts_from_block): Likewise.
22333 (translate_pending_phi_nodes): Likewise.
22334 * graphite-poly.c (print_pdr): Likewise.
22335 (dump_gbb_cases): Likewise.
22336 (dump_gbb_conditions): Likewise.
22337 (print_scop_params): Likewise.
22338 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
22339 (build_cross_bb_scalars_use): Likewise.
22340 (gather_bbs::before_dom_children): Likewise.
22341 * hsa-dump.c (dump_hsa_immed): Likewise.
22342 * ipa-cp.c (print_ipcp_constant_value): Likewise.
22343 (get_replacement_map): Likewise.
22344 * ipa-inline-analysis.c (dump_condition): Likewise.
22345 (estimate_function_body_sizes): Likewise.
22346 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
22347 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
22348 * ipa-prop.c (ipa_dump_param): Likewise.
22349 (ipa_print_node_jump_functions_for_edge): Likewise.
22350 (ipa_modify_call_arguments): Likewise.
22351 (ipa_modify_expr): Likewise.
22352 (ipa_dump_param_adjustments): Likewise.
22353 (ipa_dump_agg_replacement_values): Likewise.
22354 (ipcp_modif_dom_walker::before_dom_children): Likewise.
22355 * ipa-pure-const.c (check_stmt): Likewise.
22356 (pass_nothrow::execute): Likewise.
22357 * ipa-split.c (execute_split_functions): Likewise.
22358 * omp-offload.c (dump_oacc_loop_part): Likewise.
22359 (dump_oacc_loop): Likewise.
22360 * trans-mem.c (tm_log_emit): Likewise.
22361 (tm_memopt_accumulate_memops): Likewise.
22362 (dump_tm_memopt_set): Likewise.
22363 (dump_tm_memopt_transform): Likewise.
22364 * tree-cfg.c (gimple_verify_flow_info): Likewise.
22365 (print_loop): Likewise.
22366 * tree-chkp-opt.c (chkp_print_addr): Likewise.
22367 (chkp_gather_checks_info): Likewise.
22368 (chkp_get_check_result): Likewise.
22369 (chkp_remove_check_if_pass): Likewise.
22370 (chkp_use_outer_bounds_if_possible): Likewise.
22371 (chkp_reduce_bounds_lifetime): Likewise.
22372 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
22373 (chkp_mark_completed_bounds): Likewise.
22374 (chkp_register_incomplete_bounds): Likewise.
22375 (chkp_mark_invalid_bounds): Likewise.
22376 (chkp_maybe_copy_and_register_bounds): Likewise.
22377 (chkp_build_returned_bound): Likewise.
22378 (chkp_get_bound_for_parm): Likewise.
22379 (chkp_build_bndldx): Likewise.
22380 (chkp_get_bounds_by_definition): Likewise.
22381 (chkp_generate_extern_var_bounds): Likewise.
22382 (chkp_get_bounds_for_decl_addr): Likewise.
22383 * tree-chrec.c (chrec_apply): Likewise.
22384 * tree-data-ref.c (dump_data_reference): Likewise.
22385 (dump_subscript): Likewise.
22386 (dump_data_dependence_relation): Likewise.
22387 (analyze_overlapping_iterations): Likewise.
22388 * tree-inline.c (expand_call_inline): Likewise.
22389 (tree_function_versioning): Likewise.
22390 * tree-into-ssa.c (dump_defs_stack): Likewise.
22391 (dump_currdefs): Likewise.
22392 (dump_names_replaced_by): Likewise.
22393 (dump_update_ssa): Likewise.
22394 (update_ssa): Likewise.
22395 * tree-object-size.c (pass_object_sizes::execute): Likewise.
22396 * tree-parloops.c (build_new_reduction): Likewise.
22397 (try_create_reduction_list): Likewise.
22398 (ref_conflicts_with_region): Likewise.
22399 (oacc_entry_exit_ok_1): Likewise.
22400 (oacc_entry_exit_single_gang): Likewise.
22401 * tree-pretty-print.h: Likewise.
22402 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
22403 (get_scalar_evolution): Likewise.
22404 (add_to_evolution): Likewise.
22405 (get_loop_exit_condition): Likewise.
22406 (analyze_evolution_in_loop): Likewise.
22407 (analyze_initial_condition): Likewise.
22408 (analyze_scalar_evolution): Likewise.
22409 (instantiate_scev): Likewise.
22410 (number_of_latch_executions): Likewise.
22411 (gather_chrec_stats): Likewise.
22412 (final_value_replacement_loop): Likewise.
22413 (scev_const_prop): Likewise.
22414 * tree-sra.c (dump_access): Likewise.
22415 (disqualify_candidate): Likewise.
22416 (create_access): Likewise.
22417 (reject): Likewise.
22418 (maybe_add_sra_candidate): Likewise.
22419 (create_access_replacement): Likewise.
22420 (analyze_access_subtree): Likewise.
22421 (analyze_all_variable_accesses): Likewise.
22422 (sra_modify_assign): Likewise.
22423 (initialize_constant_pool_replacements): Likewise.
22424 (find_param_candidates): Likewise.
22425 (decide_one_param_reduction): Likewise.
22426 (replace_removed_params_ssa_names): Likewise.
22427 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
22428 * tree-ssa-copy.c (dump_copy_of): Likewise.
22429 (copy_prop_visit_cond_stmt): Likewise.
22430 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
22431 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
22432 (record_equivalences_from_stmt): Likewise.
22433 * tree-ssa-dse.c (compute_trims): Likewise.
22434 (delete_dead_call): Likewise.
22435 (delete_dead_assignment): Likewise.
22436 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
22437 (forward_propagate_into_cond): Likewise.
22438 (pass_forwprop::execute): Likewise.
22439 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
22440 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
22442 (move_computations_worker): Likewise.
22443 (execute_sm): Likewise.
22444 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
22445 (remove_exits_and_undefined_stmts): Likewise.
22446 (remove_redundant_iv_tests): Likewise.
22447 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
22448 (adjust_iv_update_pos): Likewise.
22449 * tree-ssa-math-opts.c (bswap_replace): Likewise.
22450 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
22451 (value_replacement): Likewise.
22452 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
22453 * tree-ssa-pre.c (print_pre_expr): Likewise.
22454 (get_representative_for): Likewise.
22455 (create_expression_by_pieces): Likewise.
22456 (insert_into_preds_of_block): Likewise.
22457 (eliminate_insert): Likewise.
22458 (eliminate_dom_walker::before_dom_children): Likewise.
22459 (eliminate): Likewise.
22460 (remove_dead_inserted_code): Likewise.
22461 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
22462 * tree-ssa-reassoc.c (get_rank): Likewise.
22463 (eliminate_duplicate_pair): Likewise.
22464 (eliminate_plus_minus_pair): Likewise.
22465 (eliminate_not_pairs): Likewise.
22466 (undistribute_ops_list): Likewise.
22467 (eliminate_redundant_comparison): Likewise.
22468 (update_range_test): Likewise.
22469 (optimize_range_tests_var_bound): Likewise.
22470 (optimize_vec_cond_expr): Likewise.
22471 (rewrite_expr_tree): Likewise.
22472 (rewrite_expr_tree_parallel): Likewise.
22473 (linearize_expr): Likewise.
22474 (break_up_subtract): Likewise.
22475 (linearize_expr_tree): Likewise.
22476 (attempt_builtin_powi): Likewise.
22477 (attempt_builtin_copysign): Likewise.
22478 (transform_stmt_to_copy): Likewise.
22479 (transform_stmt_to_multiply): Likewise.
22480 (dump_ops_vector): Likewise.
22481 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
22482 (print_scc): Likewise.
22483 (set_ssa_val_to): Likewise.
22484 (visit_reference_op_store): Likewise.
22485 (visit_use): Likewise.
22486 (sccvn_dom_walker::before_dom_children): Likewise.
22487 (run_scc_vn): Likewise.
22488 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
22490 (expr_hash_elt::print): Likewise.
22491 (const_and_copies::pop_to_marker): Likewise.
22492 (const_and_copies::record_const_or_copy_raw): Likewise.
22493 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
22494 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
22495 (dump_predicates): Likewise.
22496 (find_uninit_use): Likewise.
22497 (warn_uninitialized_phi): Likewise.
22498 (pass_late_warn_uninitialized::execute): Likewise.
22499 * tree-ssa.c (verify_vssa): Likewise.
22500 (verify_ssa): Likewise.
22501 (maybe_optimize_var): Likewise.
22502 * tree-vrp.c (dump_value_range): Likewise.
22503 (dump_all_value_ranges): Likewise.
22504 (dump_asserts_for): Likewise.
22505 (register_edge_assert_for_2): Likewise.
22506 (vrp_visit_cond_stmt): Likewise.
22507 (vrp_visit_switch_stmt): Likewise.
22508 (vrp_visit_stmt): Likewise.
22509 (vrp_visit_phi_node): Likewise.
22510 (simplify_cond_using_ranges_1): Likewise.
22511 (fold_predicate_in): Likewise.
22512 (evrp_dom_walker::before_dom_children): Likewise.
22513 (evrp_dom_walker::push_value_range): Likewise.
22514 (evrp_dom_walker::pop_value_range): Likewise.
22515 (execute_early_vrp): Likewise.
22517 2017-05-16 Richard Biener <rguenther@suse.de>
22519 * dwarf2out.c (loc_list_from_tree_1): Do not create
22520 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
22522 2017-05-16 Richard Biener <rguenther@suse.de>
22524 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
22526 (note_variable_value_in_expr): If we resolved the decl ref
22527 do not push to the stack.
22529 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
22531 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
22532 operations in fast-math mode.
22533 (vaddq_f16): Likewise.
22534 (vmul_f16): Likewise.
22535 (vmulq_f16): Likewise.
22536 (vsub_f16): Likewise.
22537 (vsubq_f16): Likewise.
22538 * config/arm/neon.md (add<mode>3): New.
22540 (fma:<VH:mode>3): New. Also remove outdated comment.
22543 2017-05-16 Martin Liska <mliska@suse.cz>
22547 * ipa-devirt.c (warn_types_mismatch): Fix typo.
22548 (odr_types_equivalent_p): Likewise.
22550 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
22552 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
22554 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
22557 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
22558 non-interunit SSE move alternatives with '?'.
22559 (zero-extendsidi peephole2): New peephole to skip intermediate
22560 general register in SSE zero-extend sequence.
22562 2017-05-15 Jeff Law <law@redhat.com>
22564 * reorg.c (relax_delay_slots): Create a new variable to hold
22565 the temporary target rather than clobbering TARGET_LABEL.
22567 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
22568 missing argument to extract_bit_field call.
22569 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
22571 2017-05-15 Martin Liska <mliska@suse.cz>
22574 * gcc.c (process_command): Do not allow empty argument of -o option.
22576 2017-05-15 Renlin Li <renlin.li@arm.com>
22578 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
22579 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
22580 * config/aarch64/constraints.md (Usf): Add long call check.
22581 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
22582 (call_value): Likewise.
22583 (sibcall): Likewise.
22584 (sibcall_value): Likewise.
22586 (call_value_insn): New.
22587 (sibcall_insn): Update rtx pattern.
22588 (sibcall_value_insn): Likewise.
22589 (call_internal): Remove.
22590 (call_value_internal): Likewise.
22591 (sibcall_internal): Likewise.
22592 (sibcall_value_internal): Likewise.
22593 (call_reg): Likewise.
22594 (call_symbol): Likewise.
22595 (call_value_reg): Likewise.
22596 (call_value_symbol): Likewise.
22598 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
22601 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
22603 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
22605 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
22606 compatible with CCGOCmode and with CCZmode.
22608 2017-05-14 Martin Sebor <msebor@redhat.com>
22610 PR middle-end/77671
22611 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
22612 (gimple_fold_builtin_snprintf): Same.
22613 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
22614 (gimple_fold_builtin_snprintf): Same.
22615 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
22616 of character types.
22617 (is_call_safe): New function.
22618 (try_substitute_return_value): Call it.
22619 (try_simplify_call): New function.
22620 (pass_sprintf_length::handle_gimple_call): Call it.
22622 2017-05-14 Martin Sebor <msebor@redhat.com>
22624 PR middle-end/80669
22625 * builtins.c (expand_builtin_stpncpy): Simplify.
22627 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
22629 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
22630 * config/i386/i386.h
22631 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
22632 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
22633 (struct machine_function): Add new members call_ms2sysv,
22634 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
22635 (struct machine_frame_state): New fields sp_realigned and
22636 sp_realigned_offset.
22637 * config/i386/i386.c
22638 (enum xlogue_stub): New enum.
22639 (enum xlogue_stub_sets): New enum.
22640 (class xlogue_layout): New class.
22641 (struct ix86_frame): New fields stack_realign_allocate_offset,
22642 stack_realign_offset and outlined_save_offset. Modify comments to
22643 detail stack layout when using out-of-line stubs.
22644 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
22645 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
22646 -mcall-ms2sysv-xlogues.
22647 (stub_managed_regs): New static variable.
22648 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
22649 registers managed by out-of-line stub.
22650 (disable_call_ms2sysv_xlogues): New function.
22651 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
22652 m->call_ms2sysv when appropriate and compute frame layout for
22654 (sp_valid_at, fp_valid_at): New inline functions.
22655 (choose_basereg): New function.
22656 (choose_baseaddr): Add align parameter, use choose_basereg and modify
22658 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
22659 Use align parameter of choose_baseaddr to generated aligned SSE movs
22661 (pro_epilogue_adjust_stack): Modify to track
22662 machine_frame_state::sp_realigned.
22663 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
22664 (ix86_nsaved_sseregs): Likewise.
22665 (ix86_emit_save_regs): Likewise.
22666 (ix86_emit_save_regs_using_mov): Likewise.
22667 (ix86_emit_save_sse_regs_using_mov): Likewise.
22668 (get_scratch_register_on_entry): Likewise.
22669 (gen_frame_set): New function.
22670 (gen_frame_load): Likewise.
22671 (gen_frame_store): Likewise.
22672 (emit_outlined_ms2sysv_save): Likewise.
22673 (emit_outlined_ms2sysv_restore): Likewise.
22674 (ix86_expand_prologue): Modify stack re-alignment code and call
22675 emit_outlined_ms2sysv_save when appropriate.
22676 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
22677 parameter rtx_insn *insn, which allows the function to be used to only
22678 generate the notes.
22679 (ix86_expand_epilogue): Modify validity checks of frame and stack
22680 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
22681 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
22682 * config/i386/predicates.md
22683 (save_multiple): New predicate.
22684 (restore_multiple): Likewise.
22685 * config/i386/sse.md
22686 (save_multiple<mode>): New pattern.
22687 (save_multiple_realign<mode>): Likewise.
22688 (restore_multiple<mode>): Likewise.
22689 (restore_multiple_and_return<mode>): Likewise.
22690 (restore_multiple_leave_return<mode>): Likewise.
22691 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
22693 2017-05-14 Julia Koval <julia.koval@intel.com>
22695 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
22696 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
22697 (__builtin_ia32_xsetbv): New builtins.
22698 * config/i386/i386.c (ix86_expand_special_args_builtin):
22700 (ix86_expand_builtin): Special expand for new intrinsics.
22701 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
22702 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
22703 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
22705 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22707 * cfganal.c (inverted_post_order_compute): Change argument type
22709 * cfganal.h (inverted_post_order_compute): Adjust prototype.
22710 * df-core.c (rest_of_handle_df_initialize): Adjust.
22711 (rest_of_handle_df_finish): Likewise.
22712 (df_analyze_1): Likewise.
22713 (df_analyze): Likewise.
22714 (loop_inverted_post_order_compute): Change argument to be a vec *.
22715 (df_analyze_loop): Adjust.
22716 (df_get_n_blocks): Likewise.
22717 (df_get_postorder): Likewise.
22718 * df.h (struct df_d): Change field to be a vec.
22719 * lcm.c (compute_laterin): Adjust.
22720 (compute_available): Likewise.
22721 * lra-lives.c (lra_create_live_ranges_1): Likewise.
22722 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
22723 * tree-ssa-pre.c (compute_antic): Likewise.
22725 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22727 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
22728 (depth_first_search::depth_first_search): Change structure init
22729 function to this constructor.
22730 (depth_first_search::add_bb): Rename function to this member.
22731 (depth_first_search::execute): Likewise.
22732 (flow_dfs_compute_reverse_finish): Adjust.
22734 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22736 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
22737 (longest_simple_path): Likewise.
22738 * shrink-wrap.c (spread_components): Likewise.
22739 (disqualify_problematic_components): Likewise.
22740 (emit_common_heads_for_components): Likewise.
22741 (emit_common_tails_for_components): Likewise.
22742 (insert_prologue_epilogue_for_components): Likewise.
22744 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22746 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
22749 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22751 * df-core.c (df_set_blocks): Start using auto_bitmap.
22752 (df_compact_blocks): Likewise.
22753 * df-problems.c (df_rd_confluence_n): Likewise.
22754 * df-scan.c (df_insn_rescan_all): Likewise.
22755 (df_process_deferred_rescans): Likewise.
22756 (df_update_entry_block_defs): Likewise.
22757 (df_update_exit_block_uses): Likewise.
22758 (df_entry_block_bitmap_verify): Likewise.
22759 (df_exit_block_bitmap_verify): Likewise.
22760 (df_scan_verify): Likewise.
22761 * lra-constraints.c (lra_constraints): Likewise.
22762 (undo_optional_reloads): Likewise.
22763 (lra_undo_inheritance): Likewise.
22764 * lra-remat.c (calculate_gen_cands): Likewise.
22765 (do_remat): Likewise.
22766 * lra-spills.c (assign_spill_hard_regs): Likewise.
22767 (spill_pseudos): Likewise.
22768 * tree-ssa-pre.c (bitmap_set_and): Likewise.
22769 (bitmap_set_subtract_values): Likewise.
22771 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22773 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
22774 management with auto_bitmap.
22775 (fix_inter_tick): Likewise.
22776 (fix_recovery_deps): Likewise.
22777 * ira.c (add_store_equivs): Likewise.
22778 (find_moveable_pseudos): Likewise.
22779 (split_live_ranges_for_shrink_wrap): Likewise.
22780 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
22781 (rtx_reuse_manager::seen_def_p): Likewise.
22782 (rtx_reuse_manager::set_seen_def): Likewise.
22783 * print-rtl.h (class rtx_reuse_manager): Likewise.
22785 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22787 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
22789 (migrate_btr_def): Likewise.
22790 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
22791 * df-core.c (loop_post_order_compute): Likewise.
22792 (loop_inverted_post_order_compute): Likewise.
22793 * hsa-common.h: Likewise.
22794 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
22795 * init-regs.c (initialize_uninitialized_regs): Likewise.
22796 * ipa-inline.c (resolve_noninline_speculation): Likewise.
22797 (inline_small_functions): Likewise.
22798 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
22799 * ira.c (combine_and_move_insns): Likewise.
22800 (build_insn_chain): Likewise.
22801 * loop-invariant.c (find_invariants): Likewise.
22802 * lower-subreg.c (propagate_pseudo_copies): Likewise.
22803 * predict.c (tree_predict_by_opcode): Likewise.
22804 (predict_paths_leading_to): Likewise.
22805 (predict_paths_leading_to_edge): Likewise.
22806 (estimate_loops_at_level): Likewise.
22807 (estimate_loops): Likewise.
22808 * shrink-wrap.c (try_shrink_wrapping): Likewise.
22809 (spread_components): Likewise.
22810 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
22811 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
22812 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
22813 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
22814 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
22815 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
22816 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
22817 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
22818 (mark_threaded_blocks): Likewise.
22819 (thread_through_all_blocks): Likewise.
22820 * tree-ssa.c (verify_ssa): Likewise.
22821 (execute_update_addresses_taken): Likewise.
22822 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
22824 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22826 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
22828 (post_order_compute): Likewise.
22829 (inverted_post_order_compute): Likewise.
22830 (pre_and_rev_post_order_compute_fn): Likewise.
22832 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22834 * genrecog.c (int_set::int_set): Explicitly construct our
22835 auto_vec base class.
22836 * vec.h (auto_vec::auto_vec): New constructor.
22838 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22840 * bitmap.h (class auto_bitmap): New constructor taking
22841 bitmap_obstack * argument.
22843 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22845 * bitmap.h (class auto_bitmap): Change type of m_bits to
22846 bitmap_head, and adjust ctor / dtor and member operators.
22848 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
22850 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
22851 when returned register mode doesn't match original mode.
22853 2017-05-12 Jeff Law <law@redhat.com>
22854 Jakub Jelinek <jakub@redhat.com>
22856 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
22857 we look for cc setter after the compare-elim changes.
22858 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
22859 within the vector to match what compare-elim now expects.
22860 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
22861 (xorsi3_flags, one_cmplsi2_flags): Likewise.
22863 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
22864 after the compare-elim changes.
22865 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
22866 the vector to match what compare-elim now expects.
22867 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
22868 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
22869 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
22870 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
22871 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
22873 * config/visium/visium.c (single_set_and_flags): Fix where
22874 we look for cc setter after the compare-elim changes.
22875 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
22876 with the vector to match what compare-elim now expects.
22877 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
22878 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
22879 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
22880 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
22881 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
22882 (neg<mode>2_insn_set_overflow): Likewise.
22884 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
22886 PR middle-end/79794
22887 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
22888 maybe_expand_insn call, set ops[0].target. If still set after call,
22889 set alt_rtl. Add extra arg to recursive calls.
22890 (extract_bit_field): Add alt_rtl argument. Pass to
22892 * expmed.h (extract_bit_field): Fix prototype.
22893 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
22894 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
22895 to extract_bit_field_calls.
22896 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
22897 Pass alt_rtl to extract_bit_field calls.
22898 * calls.c (store_unaligned_arguments_into_psuedos)
22899 load_register_parameters): Pass extra NULL to extract_bit_field calls.
22900 * optabs.c (maybe_legitimize_operand): Clear op->target when call
22902 * optabs.h (struct expand_operand): Add target bitfield.
22904 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
22906 * compare-elim.c (try_eliminate_compare): Canonicalize
22907 operation with embedded compare to
22908 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
22909 (set (reg) (operation)].
22911 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
22913 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
22916 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
22917 cost of adding a carry flag for ADC instruction.
22918 [case MINUS]: Ignore the cost of subtracting a carry flag
22919 for SBB instruction.
22921 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
22923 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
22925 * config/rs6000/bmiintrin.h: New file.
22926 * config/rs6000/bmi2intrin.h: New file.
22927 * config/rs6000/x86intrin.h: New file.
22929 2017-05-12 Jeff Law <law@redhat.com>
22931 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
22934 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
22936 PR middle-end/80707
22937 * tree-cfg.c: Remove cfg edges of unreachable case statements.
22939 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
22941 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
22942 early expansion of vector divide builtins.
22943 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
22944 builtins identified as having unsigned arguments.
22946 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
22948 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
22949 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
22950 expansion of vector logical operations (and, andc, or, xor,
22953 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
22955 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
22956 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
22958 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
22960 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
22961 early GIMPLE expansion of vector multiplies.
22963 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22965 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
22966 TARGET_HAVE_MOVT conditional.
22967 (movt splitter): Likewise.
22969 2017-05-12 Richard Biener <rguenther@suse.de>
22971 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
22972 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
22973 Fold all stmts not inplace.
22975 2017-05-12 Richard Biener <rguenther@suse.de>
22977 PR tree-optimization/80713
22978 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
22979 inserted_exprs bit for not removed stmts.
22981 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
22983 PR middle-end/69921
22984 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
22985 parallelized" attribute for parallelized OpenACC kernels.
22986 * omp-offload.c (execute_oacc_device_lower): Use it.
22988 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
22989 Set "oacc kernels" attribute.
22990 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
22991 parameter. Adjust all users.
22992 (oacc_fn_attrib_kernels_p): Remove function.
22993 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
22994 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
22995 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
22996 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
22997 assert "oacc kernels" attribute is set.
22999 2017-05-11 Carl Love <cel@us.ibm.com>
23001 * config/rs6000/rs6000-c: Add support for built-in functions
23002 vector unsigned char vec_popcnt (vector signed char)
23003 vector unsigned char vec_popcnt (vector unsigned char)
23004 vector unsigned short vec_popcnt (vector signed short)
23005 vector unsigned short vec_popcnt (vector unsigned short)
23006 vector unsigned int vec_popcnt (vector signed int)
23007 vector unsigned int vec_popcnt (vector unsigned int)
23008 vector unsigned long long vec_popcnt (vector signed long long)
23009 vector unsigned long long vec_popcnt (vector unsigned long long)
23010 vector signed long long vec_slo (vector signed long long,
23011 vector signed char)
23012 vector signed long long vec_slo (vector signed long long,
23013 vector unsigned char)
23014 vector unsigned long long vec_slo (vector unsigned long long,
23015 vector signed char)
23016 vector unsigned long long vec_slo (vector unsigned long long,
23017 vector unsigned char)
23018 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
23019 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
23020 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
23021 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
23022 * doc/extend.texi: Update the built-in documentation file for the
23023 new built-in functions.
23025 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
23027 * attribs.h (sorted_attr_string): Move machine independent
23028 functions for target clone support from the i386 port to common
23029 code. Rename ix86_function_versions to common_function_versions.
23030 Rename make_name to make_unique_name.
23031 (common_function_versions): Likewise.
23032 (make_unique_name): Likewise.
23033 (make_dispatcher_decl): Likewise.
23034 (is_function_default_version): Likewise.
23035 * attribs.c (attr_strcmp): Likewise.
23036 (sorted_attr_string): Likewise.
23037 (common_function_versions): Likewise.
23038 (make_unique_name): Likewise.
23039 (make_dispatcher_decl): Likewise.
23040 (is_function_default_version): Likewise.
23041 * config/i386/i386.c (attr_strcmp): Likewise.
23042 (sorted_attr_string): Likewise.
23043 (ix86_function_versions): Likewise.
23044 (make_name): Likewise.
23045 (make_dispatcher_decl): Likewise.
23046 (is_function_default_version): Likewise.
23047 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
23049 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23052 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23053 Account for direct move costs for vec_construct of integer
23056 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
23059 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
23060 (UNSPEC_STX_ATOMIC): Ditto.
23061 (loaddi_via_sse): New insn.
23062 (storedi_via_sse): Ditto.
23063 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
23064 Update corresponding peephole2 patterns.
23065 (atomic_storedi_fpu): Ditto.
23067 2017-05-11 Julia Koval <julia.koval@intel.com>
23069 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
23070 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
23072 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
23073 (__builtin_ia32_rsqrt14ss_mask): New builtins.
23074 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
23076 2017-05-11 Nathan Sidwell <nathan@acm.org>
23078 * graphite-poly.c: Include dumpfile.h.
23080 * dumpfle.h (dump_function): Declare here ...
23081 * tree-dump.h (dump_function): ... not here.
23082 * dumpfile.c: #include tree-cfg.h.
23083 (dump_function): Move here from ...
23084 * tree-dump.c (dump_function): ... here.
23085 * gimplify.c: #include splay-tree.h, not tree-dump.h.
23086 * graphite-poly.c: Don't include tree-dump.h.
23087 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
23088 * print-tree.c: Likewise.
23089 * stor-layout.c: Likewise.
23090 * tree-nested.c: Likewise.
23092 * dumpfile.c (dump_start): Use TDF_FLAGS.
23093 (dump_enable_all): Fix TDF_KIND check thinko.
23095 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
23097 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23098 array entries to represent two legal parameterizations of the
23099 overloaded __builtin_cmpb function, as represented by the
23100 P6_OV_BUILTIN_CMPB constant.
23101 (altivec_resolve_overloaded_builtin): Add special case handling
23102 for the __builtin_cmpb function, as represented by the
23103 P6_OV_BUILTIN_CMPB constant.
23104 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
23105 (BU_P6_64BIT_2): New macro.
23106 (BU_P6_OVERLOAD_2): New macro
23107 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
23108 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
23109 (CMPB): Add overload support to represent both 32-bit and 64-bit
23110 compare-bytes function.
23111 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
23112 support for TARGET_CMPB.
23113 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
23114 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23115 documentation of the __builtin_cmpb overloaded built-in function.
23117 2017-05-11 Richard Biener <rguenther@suse.de>
23119 PR tree-optimization/80705
23120 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
23121 bases are not vectorizable.
23123 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23125 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
23126 when counting register pressure.
23128 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23130 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
23131 (iv_ca_more_deps): Renamed to ...
23132 (iv_ca_compare_deps): ... this.
23133 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
23135 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23137 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
23139 (determine_group_iv_costs): ... here.
23140 (find_inv_vars_cb): Record inv var if it's not recorded before.
23142 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23144 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
23145 (get_shiftadd_cost): Ditto.
23147 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23149 * tree-ssa-address.c: Include header file.
23150 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
23152 (add_to_parts): Refactor.
23153 (addr_to_parts): New parameter. Update use of move_hint_to_base.
23154 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
23157 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23159 PR tree-optimization/53090
23160 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
23162 (extract_cond_operands): Detect condition with IV on both sides
23163 and return COMP_IV_EXPR_2.
23164 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
23165 (rewrite_use_compare): Simplify by removing call to function
23166 extract_cond_operands.
23168 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23170 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
23171 (extract_cond_operands): Detect condition comparing against non-
23172 invariant bound and return appropriate enum value.
23173 (find_interesting_uses_cond): Update use of extract_cond_operands.
23174 Handle its return value accordingly.
23175 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
23177 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23179 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
23180 nonlinear iv_use computation in loop invariant sensitive way.
23182 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23184 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
23185 (find_iv_candidates): Call relate_compare_use_with_all_cands.
23187 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23189 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
23190 (dump_cand): Support iv_cand.inv_exprs.
23191 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
23193 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
23196 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23198 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
23200 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
23201 as local function. Include necessary header files.
23202 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
23204 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23206 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
23208 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23210 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
23211 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
23212 RSHIFT_EXPR and BIT_NOT_EXPR.
23214 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23216 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
23217 (adjust_setup_cost): New parameter supporting round up adjustment.
23218 (struct address_cost_data): Delete.
23219 (force_expr_to_var_cost): Don't bound cost with spill_cost.
23220 (split_address_cost, ptr_difference_cost): Delete.
23221 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
23222 (struct ainc_cost_data): New struct.
23223 (get_address_cost_ainc): New function.
23224 (get_address_cost, get_computation_cost): Reimplement.
23225 (determine_group_iv_cost_address): Record inv_expr for all uses of
23227 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
23228 (iv_ca_has_deps): Reimplemented to ...
23229 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
23231 (iv_ca_extend): Call iv_ca_more_deps.
23233 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23235 * tree-ssa-address.c (struct mem_address): Move to header file.
23236 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
23237 * tree-ssa-address.h (struct mem_address): Move from C file.
23238 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
23240 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23242 * tree-affine.h (aff_combination_type): New interface.
23243 (aff_combination_zero_p): Remove static.
23244 (aff_combination_const_p): New interface.
23245 (aff_combination_singleton_var_p): New interfaces.
23247 2017-05-11 Richard Biener <rguenther@suse.de>
23249 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23250 Skip unreachable blocks and destinations.
23251 (eliminate): Move stmt removal and fixup ...
23252 (fini_eliminate): ... here. Skip inserted exprs.
23253 (pass_pre::execute): Move fini_pre after fini_eliminate.
23254 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
23255 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
23256 PRE to get rid of dead code that has invalid SSA form and
23257 split critical edges again.
23259 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23261 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
23263 2017-05-11 Richard Biener <rguenther@suse.de>
23265 * passes.c (execute_function_todo): Verify loops if they are
23266 said to be up-to-date.
23267 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
23268 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
23270 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
23273 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
23274 handle calling assemble_external ourself.
23277 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
23278 modes with zero size. Enhance comment.
23280 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23282 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
23283 built-ins for vec_xl and vec_xst with short and char pointer
23286 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
23288 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
23289 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
23290 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
23291 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
23292 (_mm_maskz_min_round_ss): New intrinsics.
23293 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
23294 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
23295 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
23296 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
23297 (__builtin_ia32_minss_mask_round): New builtins.
23298 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23299 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23300 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
23302 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
23303 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
23305 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
23308 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
23310 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
23311 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
23312 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
23313 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
23314 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
23315 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
23316 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
23317 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23318 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
23319 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
23320 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
23321 (__builtin_ia32_mulss_mask_round): New builtins.
23322 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23323 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23324 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
23326 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
23327 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
23329 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
23332 2017-05-10 Julia Koval <julia.koval@intel.com>
23334 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
23335 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
23336 (_mm256_setr_m128i): New intrinsics.
23338 2017-05-10 Julia Koval <julia.koval@intel.com>
23340 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
23341 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
23342 (_mm_maskz_rcp14_ss): New intrinsics.
23343 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
23344 (__builtin_ia32_rcp14ss_mask): New builtins.
23345 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
23347 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
23349 PR tree-optimization/51513
23350 * tree-cfg.c (gimple_seq_unreachable_p): New function.
23351 (assert_unreachable_fallthru_edge_p): Use it.
23352 (group_case_labels_stmt): Likewise.
23353 * tree-cfg.h: Prototype it.
23354 * stmt.c: Include cfghooks.h and tree-cfg.h.
23355 (emit_case_dispatch_table) <gap_label>: New local variable.
23356 Use it to fill dispatch table gaps.
23357 Test for default_label before updating probabilities.
23358 (expand_case) <default_label>: Remove unneeded initialization.
23359 Test for unreachable default case statement and remove its edge.
23360 Set default_label accordingly.
23361 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
23363 2017-05-10 Carl Love <cel@us.ibm.com>
23365 * config/rs6000/rs6000-c: Add support for built-in functions
23366 vector signed char vec_neg (vector signed char)
23367 vector signed short int vec_neg (vector short int)
23368 vector signed int vec_neg (vector signed int)
23369 vector signed long long vec_neg (vector signed long long)
23370 vector float vec_neg (vector float)
23371 vector double vec_neg (vector double)
23372 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
23374 * config/rs6000/altivec.h: Add define for vec_neg
23375 * doc/extend.texi: Update the built-in documentation for the
23376 new built-in functions.
23378 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23380 PR tree-optimization/77644
23381 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
23383 2017-05-10 Nathan Sidwell <nathan@acm.org>
23385 * dumpfile.h (TDI_lang_all): New.
23386 (TDF_KIND): New. Renumber others
23387 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
23389 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
23391 (get_dump_file_name): Adjust suffix generation.
23392 (dump_enable_all): Use TDF_KIND.
23393 * doc/invoke.texi (-fdump-lang-all): Document.
23395 * dumpfile.h: Tabify.
23397 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
23400 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
23401 Move member access before delete.
23403 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
23405 * tree-inline.c (expand_call_inline): Split block at stmt
23408 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
23411 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
23412 are now unused after splitting mov{sf,sd}_hardfloat.
23413 (f32_lr2): Likewise.
23414 (f32_lm): Likewise.
23415 (f32_lm2): Likewise.
23416 (f32_li): Likewise.
23417 (f32_li2): Likewise.
23418 (f32_lv): Likewise.
23419 (f32_sr): Likewise.
23420 (f32_sr2): Likewise.
23421 (f32_sm): Likewise.
23422 (f32_sm2): Likewise.
23423 (f32_si): Likewise.
23424 (f32_si2): Likewise.
23425 (f32_sv): Likewise.
23426 (f32_dm): Likewise.
23427 (f32_vsx): Likewise.
23428 (f32_av): Likewise.
23429 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
23430 For movsf, order stores so the VSX stores occur before the GPR
23431 store which encourages the register allocator to use a traditional
23432 FPR instead of a GPR. For movsd, order the stores so that the GPR
23433 store comes before the VSX stores to allow the power6 to work.
23434 This is due to the power6 not having a 32-bit integer store
23435 instruction from a FPR.
23436 (movsf_hardfloat): Likewise.
23437 (movsd_hardfloat): Likewise.
23439 2017-05-09 Martin Sebor <msebor@redhat.com>
23441 PR translation/80280
23442 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
23445 PR translation/80280
23446 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
23447 data member added in r247778.
23448 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
23450 2017-05-09 Nathan Sidwell <nathan@acm.org>
23452 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
23454 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
23457 2017-05-09 Marek Polacek <polacek@redhat.com>
23459 * doc/invoke.texi: Fix typo.
23461 2017-05-09 Richard Biener <rguenther@suse.de>
23463 * tree-vrp.c (vrp_val_is_max): Adjust comment.
23464 (vrp_val_is_min): Likewise.
23465 (set_value_range_to_value): Likewise.
23466 (set_value_range_to_nonnegative): Likewise.
23467 (gimple_assign_nonzero_p): Likewise.
23468 (gimple_stmt_nonzero_p): Likewise.
23469 (vrp_int_const_binop): Likewise. Remove unreachable case.
23470 (adjust_range_with_scev): Adjust comments.
23471 (compare_range_with_value): Likewise.
23472 (extract_range_from_phi_node): Likewise.
23473 (test_for_singularity): Likewise.
23475 2017-05-09 Richard Biener <rguenther@suse.de>
23477 * tree-vrp.c (get_single_symbol): Add assert that we don't
23478 get overflowed constants as invariant part.
23479 (compare_values_warnv): Add comment before the TREE_NO_WARNING
23480 checks. Use wi::cmp instead of recursing for integer constants.
23481 (compare_values): Just ignore whether we assumed undefined
23482 overflow instead of failing the compare.
23483 (extract_range_for_var_from_comparison_expr): Add comment before the
23484 TREE_NO_WARNING sets.
23485 (test_for_singularity): Likewise.
23486 (extract_range_from_comparison): Do not disable optimization
23487 when we assumed undefined overflow.
23488 (extract_range_basic): Remove init of unused var.
23490 2017-05-09 Richard Biener <rguenther@suse.de>
23492 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
23493 (extract_range_from_multiplicative_op_1): Adjust.
23494 (extract_range_from_binary_expr_1): Use int_const_binop.
23496 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
23499 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
23500 rs6000_store_data_bypass_p in seven define_bypass directives and
23501 in several comments.
23502 * config/rs6000/rs6000-protos.h: Add prototype for
23503 rs6000_store_data_bypass_p function.
23504 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
23505 function implements slightly different (rs6000-specific) semantics
23506 than store_data_bypass_p, returning false rather than aborting
23507 with assertion error when arguments do not satisfy the
23508 requirements of store data bypass.
23509 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
23510 rs6000_store_data_bypass_p.
23512 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
23514 * config/xtensa/xtensa-protos.h
23515 (xtensa_initial_elimination_offset): New declaration.
23516 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
23517 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
23518 macro definition, add case for FRAME_POINTER_REGNUM when
23519 FRAME_GROWS_DOWNWARD.
23520 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
23521 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
23522 xtensa_initial_elimination_offset.
23524 2017-05-08 Nathan Sidwell <nathan@acm.org>
23526 * doc/invoke.texi: Alphabetize -fdump options.
23528 2017-05-08 Martin Sebor <msebor@redhat.com>
23530 PR translation/80280
23531 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
23533 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
23535 * target.def (compute_frame_layout): New optional target hook.
23536 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
23537 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
23538 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
23540 * reload1.c (verify_initial_elim_offsets): Likewise.
23541 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
23542 (use_simple_return_p): Call arm_compute_frame_layout if needed.
23543 (arm_get_frame_offsets): Split up into this ...
23544 (arm_compute_frame_layout): ... and this function.
23546 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
23548 * config/aarch64/constraints.md (Usa): New constraint.
23549 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
23551 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
23553 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
23554 with_multilib_list after it has been checked.
23556 2017-05-08 Richard Biener <rguenther@suse.de>
23558 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
23559 (bitmap_set_subtract_values): Likewise.
23561 2017-05-08 Richard Biener <rguenther@suse.de>
23563 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
23564 (gimple_assign_nonzero): ... this and remove strict_overflow_p
23566 (gimple_stmt_nonzero_warnv_p): Rename to ...
23567 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
23569 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
23570 (extract_range_basic): Adjust, do not disable propagation on
23571 strict overflow sensitive simplification.
23572 (vrp_visit_cond_stmt): Likewise.
23574 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
23576 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
23577 body size unconditionally.
23579 2017-05-07 Jeff Law <law@redhat.com>
23582 2017-05-06 Jeff Law <law@redhat.com>
23583 PR tree-optimization/78496
23584 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
23587 PR tree-optimization/78496
23588 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
23589 (simplify_stmt_using_ranges): Call it.
23590 (vrp_dom_walker::before_dom_children): Extract equivalences
23591 from an ASSERT_EXPR with an equality comparison against a
23594 2017-05-06 Jeff Law <law@redhat.com>
23596 PR tree-optimization/78496
23597 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
23600 PR tree-optimization/78496
23601 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
23602 (simplify_stmt_using_ranges): Call it.
23603 (vrp_dom_walker::before_dom_children): Extract equivalences
23604 from an ASSERT_EXPR with an equality comparison against a
23607 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
23609 * lra-constraints.c (lra_copy_reg_equiv): New function.
23610 (split_reg): Use it to copy equivalence information from the
23611 original register to the spill register.
23613 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
23615 PR rtl-optimization/75964
23616 * simplify-rtx.c (simplify_const_relational_operation): Remove
23617 invalid handling of comparisons of integer ABS.
23619 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
23621 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
23622 initialize to zero.
23623 (init_regs): Remove declaration.
23624 (function_arg_advance_32): Initialize error_p as boolean variable.
23626 2017-05-05 Nathan Sidwell <nathan@acm.org>
23628 * store-motion.c (remove_reachable_equiv_notes): Reformat long
23629 lines. Use for (;;).
23631 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23633 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
23634 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
23635 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
23636 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
23637 VF=2 that require versioning.
23639 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23641 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
23644 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23646 * diagnostic.h (diagnostic_override_option_index): Convert from
23647 macro to inline function.
23649 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23651 * diagnostic.c (last_module_changed_p): New function.
23652 (set_last_module): New function.
23653 (diagnostic_report_current_module): Convert macro usage to
23654 the above functions.
23655 * diagnostic.h (diagnostic_context::last_module): Strengthen
23656 from const line_map * to const line_map_ordinary *.
23657 (diagnostic_last_module_changed): Delete macro.
23658 (diagnostic_set_last_module): Delete macro.
23660 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23662 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
23663 with diagnostic_report_diagnostic.
23664 (diagnostic_n_impl_richloc): Likewise.
23665 * diagnostic.h (report_diagnostic): Delete macro.
23666 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
23667 with diagnostic_report_diagnostic.
23668 * substring-locations.c (format_warning_va): Likewise.
23670 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23672 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
23673 save/restor of format_spec. Move option-printing code to...
23674 (print_option_information): ...this new function, and
23675 reimplement by simply printing to the pretty_printer,
23676 rather than appending to the format string.
23678 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23680 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
23681 handling logic into...
23682 (update_effective_level_from_pragmas): ...this new function.
23684 2017-05-04 Andrew Waterman <andrew@sifive.com>
23686 * config/riscv/riscv.opt (mstrict-align): New option.
23687 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
23688 (SLOW_UNALIGNED_ACCESS): Define.
23689 (riscv_slow_unaligned_access): Declare.
23690 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
23692 (riscv_slow_unaligned_access): New variable.
23693 (rocket_tune_info): Set slow_unaligned_access to true.
23694 (optimize_size_tune_info): Set slow_unaligned_access to false.
23695 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
23696 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
23697 (riscv_option_override): Set riscv_slow_unaligned_access.
23698 * doc/invoke.texi: Add -mstrict-align to RISC-V.
23700 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
23702 * config/riscv/riscv.md: Unify indentation.
23704 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
23709 * config/rs6000/rs6000.md (u code attribute): Add FIX and
23711 (extendsi<mode>2): Add support for doing sign extension via
23712 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
23713 don't have ISA 3.0 instructions.
23714 (extendsi<mode>2 splitter): Likewise.
23715 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
23716 generate the normal insns since SImode can now go in vector
23717 registers. Disallow the special UNSPECs needed for previous
23718 machines to hide SImode being used. Add new insns
23719 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
23720 (fix_trunc<mode>si2_stfiwx): Likewise.
23721 (fix_trunc<mode>si2_internal): Likewise.
23722 (fixuns_trunc<mode>si2): Likewise.
23723 (fixuns_trunc<mode>si2_stfiwx): Likewise.
23724 (fctiw<u>z_<mode>_smallint): Likewise.
23725 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
23726 of floating point to 32-bit integer from doing a direct move to
23727 the GPR registers to do a store.
23728 (fctiwz_<mode>): Break long line.
23730 2017-05-05 Bin Cheng <bin.cheng@arm.com>
23732 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
23733 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
23734 (addr_list, addr_offset_valid_p): New.
23735 (split_address_groups): Check offset validity with above function.
23736 (gt-tree-ssa-loop-ivopts.h): Include header file.
23738 2017-05-05 Nathan Sidwell <nathan@acm.org>
23740 * config.gcc (arm*-*-*): Add missing 'fi'.
23742 2017-05-05 Steve Ellcey <sellcey@cavium.com>
23744 * doc/invoke.texi (-fopt-info): Explicitly say order of options
23745 included in -fopt-info does not matter.
23746 * doc/optinfo.texi (-fopt-info): Fix description of default
23747 behavour. Explicitly say order of options included in -fopt-info
23750 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
23752 * config.gcc: Allow combinations of aprofile and rmprofile values for
23753 --with-multilib-list.
23754 * config/arm/t-multilib: New file.
23755 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
23756 variables. Remove setting of ISA and floating-point ABI in
23757 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
23758 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
23759 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
23760 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
23762 * config/arm/t-rmprofile: Likewise except for the matches changes.
23763 * doc/install.texi (--with-multilib-list): Document the combination of
23764 aprofile and rmprofile values and warn about pitfalls in doing that.
23766 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
23768 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
23769 (movdi_aarch64): Likewise.
23771 2017-05-05 Jakub Jelinek <jakub@redhat.com>
23773 PR tree-optimization/80632
23774 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
23776 (build_arrays): Initialize it for virtual phis.
23777 (fix_phi_nodes): Use it for virtual phis.
23779 PR tree-optimization/80558
23780 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
23781 [x, y] op z into [x op, y op z] for op & or | if conditions
23784 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
23785 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23788 * config/arm/arm.md (use_literal_pool): Remove.
23789 (64-bit immediate split): No longer takes cost into consideration
23790 if arm_disable_literal_pool is enabled.
23791 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
23792 used when arm_disable_literal_pool is enabled.
23793 (arm_max_const_double_inline_cost): Remove use of
23794 arm_disable_literal_pool.
23795 (push_minipool_fix): Add assert.
23796 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
23797 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
23798 (no_literal_pool_sf_immediate): New.
23800 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23802 PR tree-optimization/80613
23803 * tree-ssa-dce.c (propagate_necessity): Remove cases for
23804 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
23806 2017-05-05 Richard Biener <rguenther@suse.de>
23808 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
23810 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
23812 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
23813 of this flag from insn conditions due to removal from r247495.
23815 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
23817 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
23819 (arm_early_store_addr_dep_ptr): Likewise.
23820 * config/arm/aarch-common-protos.h
23821 (arm_early_load_addr_dep_ptr): Add prototype.
23822 (arm_early_store_addr_dep_ptr): Likewise.
23823 * config/arm/cortex-a53.md: Add new bypasses.
23825 2017-05-05 Jakub Jelinek <jakub@redhat.com>
23827 * tree.c (next_type_uid): Change type to unsigned.
23828 (type_hash_canon): Decrement back next_type_uid if
23829 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
23830 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
23833 2017-05-04 Martin Sebor <msebor@redhat.com>
23835 * builtins.c: Fix a trivial typo in a comment.
23837 PR middle-end/79234
23838 * builtins.c (check_sizes): Adjust to handle reading past the end.
23839 Avoid printing excessive upper bound of ranges. Use %E to print
23840 tree nodes instead of converting them to %wu.
23841 (expand_builtin_memchr): New function.
23842 (compute_dest_size): Rename...
23843 (compute_objsize): ...to this.
23844 (expand_builtin_memcpy): Adjust.
23845 (expand_builtin_mempcpy): Adjust.
23846 (expand_builtin_strcat): Adjust.
23847 (expand_builtin_strcpy): Adjust.
23848 (check_strncat_sizes): Adjust.
23849 (expand_builtin_strncat): Adjust.
23850 (expand_builtin_strncpy): Adjust and simplify.
23851 (expand_builtin_memset): Adjust.
23852 (expand_builtin_bzero): Adjust.
23853 (expand_builtin_memcmp): Adjust.
23854 (expand_builtin): Handle memcmp.
23855 (maybe_emit_chk_warning): Check strncat just once.
23857 2017-05-04 Martin Sebor <msebor@redhat.com>
23859 PR preprocessor/79214
23860 PR middle-end/79222
23861 PR middle-end/79223
23862 * builtins.c (check_sizes): Add inlining context and issue
23863 warnings even when -Wno-system-headers is set.
23864 (check_strncat_sizes): Same.
23865 (expand_builtin_strncat): Same.
23866 (expand_builtin_memmove): New function.
23867 (expand_builtin_stpncpy): Same.
23868 (expand_builtin): Handle memmove and stpncpy.
23870 2017-05-04 Bin Cheng <bin.cheng@arm.com>
23872 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
23873 which is not used any more.
23875 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
23877 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
23879 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
23881 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
23882 (cortexa53_tunings): Likewise.
23883 (cortexa57_tunings): Likewise.
23884 (cortexa72_tunings): Likewise.
23885 (cortexa73_tunings): Likewise.
23887 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
23889 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
23890 Set loop alignment to 8.
23892 2017-05-04 Martin Sebor <msebor@redhat.com>
23894 PR translation/80280
23895 * builtins.c (expand_builtin_object_size): Add missing quoting to
23896 %D and like directives.
23897 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
23898 (hsa_type_for_tree_type): Same.
23899 (verify_function_arguments): Same.
23900 * symtab.c (symbol_table::change_decl_assembler_name): Same.
23901 * varasm.c (get_section): Same.
23904 2017-05-04 Martin Sebor <msebor@redhat.com>
23906 PR translation/80280
23907 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
23909 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
23911 * config/aarch64/aarch64.c (generic_addrcost_table):
23912 Change HI/TI mode setting.
23914 2017-05-04 Martin Jambor <mjambor@suse.cz>
23916 PR tree-optimization/80622
23917 * tree-sra.c (comes_initialized_p): New function.
23918 (build_accesses_from_assign): Only set write lazily when
23919 comes_initialized_p is false.
23920 (analyze_access_subtree): Use comes_initialized_p.
23921 (propagate_subaccesses_across_link): Assert !comes_initialized_p
23922 instead of testing for PARM_DECL.
23924 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23926 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
23927 constraint on operand 0 to allow more general addressing modes.
23928 Adjust output template.
23929 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
23931 * config/aarch64/aarch64-protos.h
23932 (aarch64_address_valid_for_prefetch_p): Declare prototype.
23933 * config/aarch64/constraints.md (Dp): New address constraint.
23934 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
23937 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
23939 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
23940 update use of estimate_ipcp_clone_size_and_time.
23941 (estimate_local_effects): Update use of
23942 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
23943 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
23944 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
23945 Return nonspecialized time.
23947 2017-05-04 Richard Biener <rguenther@suse.de>
23949 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
23950 for the last VUSE which def dominates the PHI. Directly call
23952 (get_continuation_for_phi_1): Remove.
23954 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
23956 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
23957 to explain the use of truncating division. Cap the number of
23958 iterations to the maximum given by nb_iterations_upper_bound,
23961 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
23963 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
23964 * configure: Regenerate.
23965 * config.in: Regenerate.
23966 * config/i386/driver-mingw32.c: new file.
23967 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
23968 * config.host: Link driver-mingw32.o on MinGW host.
23969 * doc/install.texi: Document new --enable-mingw-wildcard configure
23972 2017-05-04 Marek Polacek <polacek@redhat.com>
23974 PR tree-optimization/80612
23975 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
23977 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23978 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
23980 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
23981 (movt splitter): Likewise.
23982 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
23983 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
23984 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
23985 block for Thumb-1 with MOVT.
23986 (thumb2_legitimate_address_p): Move code block ...
23987 (can_avoid_literal_pool_for_label_p): ... into this new function.
23988 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
23990 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
23991 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
23992 "M-profile targets with the MOVT instruction".
23994 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23996 * config/arm/arm-builtins.c (arm_init_builtins): Rename
23997 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
23998 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
24000 2017-05-04 Martin Liska <mliska@suse.cz>
24002 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
24003 variable cond_code.
24005 2017-05-04 Richard Biener <rguenther@suse.de>
24007 * tree.c (array_at_struct_end_p): Handle arrays at struct
24008 end with flexarrays more conservatively. Refactor and treat
24009 arrays of arrays or aggregates more strict. Fix
24010 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
24011 * tree.c (array_at_struct_end_p): Adjust prototype.
24012 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
24013 * gimple-fold.c (get_range_strlen): Likewise.
24014 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
24016 2017-05-04 Richard Biener <rguenther@suse.de>
24018 PR tree-optimization/31130
24019 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
24021 (supports_overflow_infinity): Likewise.
24022 (is_negative_overflow_infinity): Likewise.
24023 (is_positive_overflow_infinity): Likewise.
24024 (is_overflow_infinity): Likewise.
24025 (stmt_overflow_infinity): Likewise.
24026 (overflow_infinity_range_p): Likewise.
24027 (usable_range_p): Remove as always returning true.
24028 (make_overflow_infinity): Remove.
24029 (negative_overflow_infinity): Likewise.
24030 (positive_overflow_infinity): Likewise.
24031 (avoid_overflow_infinity): Likewise.
24032 (set_value_range): Adjust accordingly.
24033 (set_value_range_to_nonnegative): Likewise, remove now unused
24034 overflow_infinity arg.
24035 (vrp_operand_equal_p): Adjust.
24036 (update_value_range): Likewise.
24037 (range_int_cst_singleton_p): Likewise.
24038 (operand_less_p): Likewise.
24039 (compare_values_warnv): Likewise.
24040 (extract_range_for_var_from_comparison_expr): Likewise.
24041 (vrp_int_const_binop): Likewise.
24042 (zero_nonzero_bits_from_vr): Likewise.
24043 (extract_range_from_multiplicative_op_1): Likewise.
24044 (extract_range_from_binary_expr_1): Likewise.
24045 (extract_range_from_unary_expr): Likewise.
24046 (extract_range_from_comparison): Likewise.
24047 (extract_range_basic): Likewise.
24048 (adjust_range_with_scev): Likewise.
24049 (compare_ranges): Likewise.
24050 (compare_range_with_value): Likewise.
24051 (dump_value_range): Likewise.
24052 (test_for_singularity): Likewise, remove strict_overflow_p parameter
24054 (simplify_cond_using_ranges): Adjust.
24056 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
24058 * brig-builtins.def: Added a builtin for class_f64.
24059 * builtin-types.def: Added a builtin type needed by class_f64.
24061 2017-05-03 Jason Merrill <jason@redhat.com>
24063 * timevar.def: Add TV_CONSTEXPR.
24065 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24067 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
24069 2017-05-03 Martin Jambor <mjambor@suse.cz>
24071 * ipa-prop.c (ipa_update_after_lto_read): Removed.
24072 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
24073 * ipa-cp.c (ipcp_propagate_stage): Do not call
24074 ipa_update_after_lto_read.
24075 * ipa-inline.c (ipa_inline): Likewise.
24077 2017-05-03 Martin Jambor <mjambor@suse.cz>
24079 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
24080 tag. Added a default constructor and a destructor.
24081 (ipa_edge_args_sum_t): New class;
24082 (ipa_edge_args_sum): Declare.
24083 (ipa_edge_args_vector): Remove declaration.
24084 (IPA_EDGE_REF): Use ipa_edge_args_sum.
24085 (ipa_free_edge_args_substructures): Remove declaration.
24086 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
24087 (ipa_edge_args_info_available_for_edge_p): Likewise.
24088 * ipa-prop.c (ipa_edge_args_vector): Removed.
24089 (edge_removal_hook_holder): Likewise.
24090 (edge_duplication_hook_holder): Likewise.
24091 (ipa_edge_args_sum): New variable.
24092 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
24093 ipa_edge_args_vector.
24094 (ipa_free_edge_args_substructures): Likewise.
24095 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
24096 ipa_edge_args_vector.
24097 (ipa_edge_removal_hook): Turned into method
24098 ipa_edge_args_sum_t::remove.
24099 (ipa_edge_duplication_hook): Turned into method
24100 ipa_edge_args_sum_t::duplicate.
24101 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
24102 registering edge hooks.
24103 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
24104 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
24105 ipa_edge_args_sum instead of ipa_edge_args_vector.
24106 * ipa-profile.c (ipa_profile): Likewise.
24108 2017-05-03 Martin Jambor <mjambor@suse.cz>
24110 * symbol-summary.h (function_summary): New method exists.
24111 (function_summary::symtab_removal): Deallocate through release.
24112 (call_summary): New class.
24113 (gt_ggc_mx): New overload.
24114 (gt_pch_nx): Likewise.
24115 (gt_pch_nx): Likewise.
24117 2017-05-03 Jeff Law <law@redhat.com>
24119 PR tree-optimization/78496
24120 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
24121 from simplify_cond_using_ranges. Split off code to walk
24122 backwards through casts into ...
24123 (simplify_cond_using_ranges_2): New function.
24124 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
24125 (execute_vrp): After identifying jump threads, call
24126 simplify_cond_using_ranges_2.
24128 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
24131 * ipa-inline.h (inline_summary): Add ctor.
24132 (create_ggc): Do not use ggc_cleared_alloc.
24134 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
24135 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24137 * gcc.c (handle_braces): Support escaping in switch matching
24139 * doc/invoke.texi (Spec Files): Document it.
24140 Remove superfluous @code markup in items.
24142 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24144 * diagnostic-show-locus.c (struct column_range): New struct.
24145 (get_affected_columns): New function.
24146 (get_printed_columns): New function.
24147 (struct correction): New struct.
24148 (correction::ensure_capacity): New function.
24149 (correction::ensure_terminated): New function.
24150 (struct line_corrections): New struct.
24151 (line_corrections::~line_corrections): New dtor.
24152 (line_corrections::add_hint): New function.
24153 (layout::print_trailing_fixits): Reimplement in terms of the new
24155 (selftest::test_overlapped_fixit_printing): New function.
24156 (selftest::diagnostic_show_locus_c_tests): Call it.
24158 2017-05-03 Nathan Sidwell <nathan@acm.org>
24160 Canonicalize canonical type hashing
24161 * tree.h (type_hash_canon_hash): Declare.
24162 * tree.c (type_hash_list, attribute_hash_list): Move into
24163 type_hash_canon_hash.
24164 (build_type_attribute_qual_variant): Break out hash code calc into
24165 type_hash_canon_hash.
24166 (type_hash_canon_hash): New. Generic type hash computation.
24167 (build_range_type_1, build_array_type_1, build_function_type,
24168 build_method_type_directly, build_offset_type, build_complex_type,
24169 make_vector_type): Call it.
24171 2017-05-03 Richard Biener <rguenther@suse.de>
24173 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
24174 When all DRs have unknown misaligned do not always peel
24175 when there is a store but apply the same costing model as if
24176 there were only loads.
24178 2017-05-03 Richard Biener <rguenther@suse.de>
24181 PR tree-optimization/80492
24182 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
24183 compare_base_decls returning dont-know properly.
24185 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
24187 * config/arm/iterators.md (CCSI): New mode iterator.
24188 (arch): New mode attribute.
24189 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
24190 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
24191 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
24192 code iterator for success result mode.
24193 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
24194 the corresponding new insn generators.
24196 2017-05-03 Bin Cheng <bin.cheng@arm.com>
24199 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24200 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24202 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
24204 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
24205 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
24206 (DDR_A): Wrap DDR argument in brackets.
24207 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
24208 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
24209 (DDR_REVERSED_P): Likewise.
24211 2017-05-03 Jakub Jelinek <jakub@redhat.com>
24213 PR tree-optimization/79472
24214 * tree-switch-conversion.c (struct switch_conv_info): Add
24215 contiguous_range and default_case_nonstandard fields.
24216 (collect_switch_conv_info): Compute contiguous_range and
24217 default_case_nonstandard fields, don't clear final_bb if
24218 contiguous_range and only the default case doesn't have the required
24220 (check_all_empty_except_final): Set default_case_nonstandard instead
24221 of failing if contiguous_range and the default case doesn't have empty
24223 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
24224 and only the default case doesn't have the required constants. Skip
24226 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
24227 if default_case_nonstandard.
24228 (build_constructors): Build constant 1 just once. Assert that default
24229 values aren't inserted in between cases if contiguous_range. Skip
24231 (build_arrays): Skip virtual phis.
24232 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
24233 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
24234 Handle virtual phis.
24235 (gen_inbound_check): Handle default_case_nonstandard case.
24236 (process_switch): Adjust check_final_bb caller. Call
24237 gather_default_values with the first non-default case instead of
24238 default case if default_case_nonstandard.
24240 2017-05-02 Nathan Sidwell <nathan@acm.org>
24242 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
24243 check. Fix formatting.
24245 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
24247 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
24248 errors when comparing specialized and unspecialized times.
24250 2017-05-02 David Malcolm <dmalcolm@redhat.com>
24252 * diagnostic-show-locus.c
24253 (layout::should_print_annotation_line_p): Make private.
24254 (layout::print_annotation_line): Make private.
24255 (layout::annotation_line_showed_range_p): Make private.
24256 (layout::show_ruler): Make private.
24257 (layout::print_source_line): Make private. Pass in line and
24258 line_width, rather than calling location_get_source_line. Drop
24260 (layout::print_leading_fixits): New method.
24261 (layout::print_any_fixits): Rename to...
24262 (layout::print_trailing_fixits): ...this, and make private.
24263 Don't print newline fixits.
24264 (diagnostic_show_locus): Move logic for printing one row into...
24265 (layout::print_line): ...this new function. Move the
24266 location_get_source_line call and error-handling from
24267 print_source_line to here. Call print_leading_fixits, and rename
24268 print_any_fixits to print_trailing_fixits.
24269 (selftest::test_fixit_insert_containing_newline): Update now that
24270 newlines are partially supported.
24271 (selftest::test_fixit_insert_containing_newline_2): New test.
24272 (selftest::test_fixit_replace_containing_newline): Update comments.
24273 (selftest::diagnostic_show_locus_c_tests): Call the new test.
24274 * edit-context.c (class added_line): New class.
24275 (class edited_line): Describe newline handling in comment.
24276 (edited_line::actually_edited_p): New method.
24277 (edited_line::print_content): Delete redundant decl.
24278 (edited_line::m_predecessors): New field.
24279 (edited_file::print_content): Call edited_line::print_content.
24280 (edited_file::print_diff): Update to support newlines.
24281 (edited_file::print_diff_hunk): Likewise.
24282 (edited_file::print_run_of_changed_lines): New function.
24283 (edited_file::print_diff_line): Convert to...
24284 (print_diff_line): ...this.
24285 (edited_file::get_effective_line_count): New function.
24286 (edited_line::edited_line): Initialize new field m_predecessors.
24287 (edited_line::~edited_line): Clean up m_predecessors.
24288 (edited_line::apply_fixit): Handle newlines.
24289 (edited_line::get_effective_line_count): New function.
24290 (edited_line::print_content): New function.
24291 (edited_line::print_diff_lines): New function.
24292 (selftest::test_applying_fixits_insert_containing_newline): New
24294 (selftest::test_applying_fixits_replace_containing_newline): New
24296 (selftest::insert_line): New function.
24297 (selftest::test_applying_fixits_multiple_lines): Add example of
24299 (selftest::edit_context_c_tests): Call the new tests.
24301 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24303 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
24304 parameter cand. Update dump information.
24305 (get_computation_cost): Update uses.
24307 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24309 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
24310 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
24311 (get_computation_at, rewrite_use_address): Update use of
24312 get_computation_aff.
24314 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24316 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
24317 (get_computation): Delete.
24318 (get_computation_cost): Implement like get_computation_cost_at.
24319 Use get_computation_at.
24320 (get_computation_cost_at): Delete.
24321 (rewrite_use_nonlinear_expr): Use get_computation_at.
24322 (rewrite_use_compare, remove_unused_ivs): Ditto.
24324 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24326 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
24328 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24330 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
24331 (ivopts_global_cost_for_size): Rename parameter and update uses.
24332 (iv_ca_recount_cost): Update uses.
24333 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
24334 candidates seperately in n_invs and n_cands.
24335 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
24337 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24339 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
24340 (find_inv_vars_cb): New.
24341 (find_depends): Renamed to ...
24342 (find_inv_vars): ... this.
24343 (add_candidate_1, force_var_cost): Call find_inv_vars.
24344 (split_address_cost, determine_group_iv_cost_cond): Ditto.
24346 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24348 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
24349 inv_vars. Add inv_exprs.
24350 (struct iv_cand): Rename depends_on to inv_vars.
24351 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
24352 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
24353 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
24354 (dump_cand): Dump inv_vars.
24355 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
24356 (record_invariant, find_depends, add_candidate_1): Ditto.
24357 (set_group_iv_cost, force_var_cost): Ditto.
24358 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
24359 (get_computation_cost_at, get_computation_cost): Ditto.
24360 (determine_group_iv_cost_generic): Ditto.
24361 (determine_group_iv_cost_address): Ditto.
24362 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
24363 (determine_group_iv_costs): Ditto.
24364 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
24365 (iv_ca_set_remove_invariants): Renamed to ...
24366 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
24367 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
24368 (iv_ca_set_add_invariants): Renamed to ...
24369 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
24370 (iv_ca_set_cp): Use iv_ca_set_add_invs.
24371 (iv_ca_has_deps): Support inv_vars and inv_exprs.
24372 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
24373 (create_new_ivs): Remove useless dump.
24375 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24377 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
24379 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
24380 (iv_ca_set_no_cp, create_new_iv): Ditto.
24382 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24384 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24386 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
24388 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
24389 function tree_check2.
24391 2017-05-02 Martin Liska <mliska@suse.cz>
24393 * doc/gcov.texi: Add missing preposition.
24394 * gcov.c (function_info::function_info): Properly fill up
24395 all member variables.
24397 2017-05-02 Tamar Christina <tamar.christina@arm.com>
24399 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
24401 2017-05-02 Tamar Christina <tamar.christina@arm.com>
24403 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
24405 2017-05-02 Martin Liska <mliska@suse.cz>
24408 * lto-streamer-in.c (lto_read_tree_1): Remove
24409 LTO_STREAMER_DEBUG.
24410 * lto-streamer.c (struct tree_hash_entry): Likewise.
24411 (struct tree_entry_hasher): Likewise.
24412 (tree_entry_hasher::hash): Likewise.
24413 (tree_entry_hasher::equal): Likewise.
24414 (lto_streamer_init): Likewise.
24415 (lto_orig_address_map): Likewise.
24416 (lto_orig_address_get): Likewise.
24417 (lto_orig_address_remove): Likewise.
24418 * lto-streamer.h: Likewise.
24419 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
24420 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
24422 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
24424 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
24425 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
24426 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
24427 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
24428 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
24429 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
24430 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
24431 (mm_maskz_sub_ss): New intrinsics.
24432 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24433 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
24434 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
24435 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
24436 (__builtin_ia32_subss_mask_round): New builtins.
24437 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24438 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24439 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
24441 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
24442 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
24444 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
24447 2017-05-02 Martin Jambor <mjambor@suse.cz>
24449 PR tree-optimization/78687
24450 * tree-sra.c (access): New field parent.
24451 (process_subtree_disqualification): New function.
24452 (disqualify_candidate): Call it.
24453 (build_accesses_from_assign): Reset write flag if creating an
24455 (build_access_subtree): Fill in parent field and also prpagate
24456 down grp_write flag.
24457 (create_artificial_child_access): New parameter set_grp_write, set
24458 grp_write to its value.
24459 (propagate_subaccesses_across_link): Also propagate grp_write flag
24461 (propagate_all_subaccesses): Push the closest parent back to work
24462 queue if add_access_to_work_queue returned true.
24464 2017-05-02 Richard Biener <rguenther@suse.de>
24466 * common.opt (fstrict-overflow): Alias negative to fwrapv.
24467 * doc/invoke.texi (fstrict-overflow): Remove all traces of
24468 -fstrict-overflow documentation.
24469 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
24470 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
24471 flag_strict_overflow.
24472 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
24473 * lto-opts.c (lto_write_options): Do not stream it.
24474 * lto-wrapper.c (merge_and_complain): Do not handle it.
24475 * opts.c (default_options_table): Do not set -fstrict-overflow.
24476 (finish_options): Likewise do not clear it when sanitizing.
24477 * simplify-rtx.c (simplify_const_relational_operation): Do not
24478 test flag_strict_overflow.
24480 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
24482 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
24483 using enabled attribute.
24484 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
24485 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
24486 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
24487 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
24488 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
24489 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
24490 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
24491 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
24492 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
24493 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
24495 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
24497 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
24499 2017-05-02 Richard Biener <rguenther@suse.de>
24501 PR tree-optimization/80591
24503 2017-04-10 Richard Biener <rguenther@suse.de>
24505 * tree-ssa-structalias.c (find_func_aliases): Properly handle
24508 2017-05-02 Richard Biener <rguenther@suse.de>
24510 PR tree-optimization/80549
24511 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
24512 (cleanup_tree_cfg_noloop): Create forwarders to known loop
24513 headers if they do not have a preheader.
24515 2017-05-02 Martin Liska <mliska@suse.cz>
24518 * common.opt: Fix typo.
24519 * doc/invoke.texi: Likewise.
24521 2017-05-01 Jan Beulich <jbeulich@suse.com>
24523 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
24524 swapping, add (x,x,m,x,n) alternative.
24526 2017-05-01 Nathan Sidwell <nathan@acm.org>
24528 * calls.c (combine_pending_stack_adjustment_and_call): Remove
24529 unnecessary unadjusted_alignment check.
24531 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
24534 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
24536 * gimplify.c (gimplify_cilk_detach): New function.
24537 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
24538 * tree-core.h: Document EXPR_CILK_SPAWN.
24539 * tree.h (EXPR_CILK_SPAWN): Define.
24541 2017-05-01 David Malcolm <dmalcolm@redhat.com>
24543 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
24544 to use new fixit_hint representation, using the "replace" logic.
24545 (get_line_span_for_fixit_hint): Likewise.
24546 (layout::print_any_fixits): Likewise.
24547 (selftest::test_one_liner_many_fixits): Rename to...
24548 (selftest::test_one_liner_many_fixits_1): ...this, and update
24549 comment and expected output to reflect that the multiple fix-it
24550 hints are now consolidated into one insertion.
24551 (selftest::test_one_liner_many_fixits_2): New test.
24552 (selftest::test_diagnostic_show_locus_one_liner): Update for
24554 (selftest::test_fixit_consolidation): Update for fix-it API
24556 * diagnostic.c (print_parseable_fixits): Likewise.
24557 * edit-context.c (edited_line::m_line_events): Convert from
24558 auto_vec <line_event *> to auto_vec <line_event>.
24559 (class line_event): Convert from abstract base class to a concrete
24560 class, taking over the role of replace_event.
24561 (class insert_event): Delete.
24562 (class replace_event): Rename to class line_event. Convert to
24564 (edit_context::add_fixits): Reimplement.
24565 (edit_context::apply_insert): Delete.
24566 (edit_context::apply_replace): Rename to...
24567 (edit_context::apply_fixit): ...this. Convert to half-open range.
24568 (edited_file::apply_insert): Delete.
24569 (edited_file::apply_replace): Rename to...
24570 (edited_file::apply_fixit): ...this.
24571 (edited_line::~edited_line): Drop deletion of events.
24572 (edited_line::apply_insert): Delete.
24573 (edited_line::apply_replace): Rename to...
24574 (edited_line::apply_fixit): ...this. Convert to half-open range.
24575 Update for change to type of m_line_events.
24576 * edit-context.h (edit_context::apply_insert): Delete.
24577 (edit_context::apply_replace): Rename to...
24578 (edit_context::apply_fixit): ...this.
24580 2017-05-01 Martin Sebor <msebor@redhat.com>
24582 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
24585 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
24588 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
24589 __get_cpuid_max returns 0.
24590 (__get_cpuid_count): Ditto.
24592 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
24594 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
24595 replacement expression is another instance of one of its arguments.
24597 2017-05-01 Jakub Jelinek <jakub@redhat.com>
24600 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
24601 check for stack push/pop autoinc.
24602 * config/i386/i386.c (ix86_agi_dependent): Return false
24603 if the only reason why modified_in_p returned true is that
24604 addr is SP based and set_insn is a push or pop.
24606 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
24608 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
24611 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
24614 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
24615 (account_size_time): Use two predicates - exec_pred and
24617 (evaluate_conditions_for_known_args): Compute both clause and
24619 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
24620 (inline_summary_t::duplicate): Update.
24621 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
24623 (compute_inline_parameters): Likewise.
24624 (estimate_edge_size_and_time): Update caluclation of time.
24625 (estimate_node_size_and_time): Compute both time and nonspecialized
24627 (estimate_ipcp_clone_size_and_time): Update.
24628 (inline_merge_summary): Update.
24629 (do_estimate_edge_time): Update.
24630 (do_estimate_edge_size): Update.
24631 (do_estimate_edge_hints): Update.
24632 (inline_read_section, inline_write_summary): Stream both new predicates.
24633 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
24635 (compute_inlined_call_time): Cleanup.
24636 (big_speedup_p): Update.
24637 (edge_badness): Update.
24638 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
24639 (size_time_entry): Replace predicate by exec_predicate and
24640 nonconst_predicate.
24641 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
24642 (estimate_edge_time): Return also nonspec_time.
24643 (reset_edge_growth_cache): Update.
24645 2017-04-29 Jakub Jelinek <jakub@redhat.com>
24647 PR rtl-optimization/80491
24648 * ifcvt.c (noce_process_if_block): When looking for x setter
24649 with missing else_bb, don't check only the insn right before
24650 cond_earliest, but look for the last insn that x is modified in
24651 within the same bb.
24653 PR rtl-optimization/80491
24654 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
24656 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
24658 PR tree-optimization/80487
24659 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
24661 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24663 PR tree-optimization/79697
24664 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
24665 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
24666 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
24668 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
24669 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
24671 2017-04-28 Martin Sebor <msebor@redhat.com>
24673 PR tree-optimization/80523
24674 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
24675 (init_target_to_host_charmap, target_to_host, target_strtol10): New
24677 (maybe_warn, format_directive, parse_directive): Use new functions.
24678 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
24680 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
24682 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
24684 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
24686 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
24687 target_header_dir): Set correctly.
24688 * configure: Regenerated.
24689 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
24690 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
24691 instead of SYSTEM_HEADER_DIR.
24693 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
24695 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
24696 (estimate_local_effects): Likewise.
24697 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
24698 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
24699 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
24700 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
24701 do_estimate_edge_time, estimate_edge_time): Likewise.
24702 * ipa-inline-analysis.c (estimate_node_size_and_time,
24703 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
24704 (estimate_time_after_inlining): Remove.
24706 2017-04-28 Martin Liska <mliska@suse.cz>
24708 * doc/gcov.texi: Enhance documentation of gcov.
24710 2017-04-28 Martin Liska <mliska@suse.cz>
24712 * doc/gcov.texi: Sort options in alphabetic order.
24713 * doc/gcov-dump.texi: Likewise.
24714 * doc/gcov-tool.texi: Likewise.
24715 * gcov.c (print_usage): Likewise.
24716 * gcov-dump.c (print_usage): Likewise.
24717 * gcov-tool.c (print_merge_usage_message): Likewise.
24718 (print_rewrite_usage_message): Likewise.
24719 (print_overlap_usage_message): Likewise.
24721 2017-04-28 Martin Liska <mliska@suse.cz>
24723 PR gcov-profile/53915
24724 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
24726 2017-04-28 Martin Liska <mliska@suse.cz>
24728 PR gcov-profile/79891
24729 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
24730 is marked by compiler as living on a line.
24731 (get_cycles_count): Remove usage of the union.
24732 (output_intermediate_file): Likewise.
24733 (find_source): Fix GNU coding style.
24734 (accumulate_line_counts): Remove old non-all block mode.
24735 (output_lines): Remove usage of the union.
24736 * profile.c (output_location): Include all BBs, even if
24737 belonging to a same line (and file) as a previous BB.
24739 2017-04-28 Martin Liska <mliska@suse.cz>
24741 * gcov.c (process_args): Handle new argument 'w'.
24742 (read_graph_file): Assign ID to BBs.
24743 (output_branch_count): Display BB # if verbose flag is set.
24744 (output_lines): Likewise for arcs.
24745 (print_usage): Add '--verbose' option help.
24746 * doc/gcov.texi: Document --verbose (-w) option.
24748 2017-04-28 Martin Liska <mliska@suse.cz>
24750 * gcov.c (struct block_location_info): New struct.
24751 (process_file): Fill up the new structure.
24752 (read_graph_file): Replace usage of encoding by the newly added
24754 (add_line_counts): Likewise.
24755 (accumulate_line_counts): Remove usage of the union.
24756 (function_info::function_info): New function.
24757 (function_info::~function_info): Likewise.
24758 (process_file): Call delete instead of release_function.
24759 (release_function): Release the function.
24760 (release_structures): Call delete instead of release_function.
24761 (solve_flow_graph): Replace usage of num_blocks.
24762 (find_exception_blocks): Likewise.
24763 (output_lines): Fix GNU coding style.
24765 2017-04-28 Martin Liska <mliska@suse.cz>
24768 * coverage.c (coverage_remove_note_file): New function.
24769 * coverage.h: Declare the function.
24770 * toplev.c (finalize): Clean if an error has been seen.
24772 2017-04-28 Martin Liska <mliska@suse.cz>
24774 PR gcov-profile/80031
24775 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
24776 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
24777 * gcov.c (read_graph_file): Read just number of blocks.
24778 * profile.c (branch_prob): Do not stream 0 flags per a basic
24781 2017-04-28 Martin Liska <mliska@suse.cz>
24783 * gcov-dump.c (tag_*): Add new argument to declarations.
24784 (dump_gcov_file): Likewise.
24785 (tag_blocks): Add and use new argument depth.
24786 (tag_arcs): Likewise.
24787 (tag_lines): Likewise.
24788 (tag_counters): Likewise.
24789 (tag_summary): Likewise.
24790 (dump_working_sets): Use depth to do a proper indentation.
24792 2017-04-28 Jakub Jelinek <jakub@redhat.com>
24795 * cgraph.h (symtab_node::debug_symtab): No longer inline.
24796 * symtab.c (symtab_node::debug_symtab): Move definition here.
24798 2017-04-28 Richard Biener <rguenther@suse.de>
24800 * lto-streamer.h (LTO_major_version): Bump to 7.
24802 2017-04-28 Richard Biener <rguenther@suse.de>
24804 * tree-vrp.c (assert_info): New struct.
24805 (add_assert_info): New helper.
24806 (register_edge_assert_for_2): Refactor to add asserts to a vector
24808 (register_edge_assert_for_1): Likewise.
24809 (register_edge_assert_for): Likewise.
24810 (finish_register_edge_assert_for): New helper actually registering
24811 asserts where live on edge.
24812 (find_conditional_asserts): Adjust.
24813 (find_switch_asserts): Likewise.
24814 (evrp_dom_walker::try_find_new_range): Generalize.
24815 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
24817 2017-04-27 Marek Polacek <polacek@redhat.com>
24820 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
24821 arg10 and arg11 to itype.
24823 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
24825 * doc/extend.texi (Object Size Checking): Improve grammar.
24827 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
24830 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
24831 that the logic for permitting reciprocal estimates matches that
24834 2017-04-27 Jakub Jelinek <jakub@redhat.com>
24837 * tree.c (type_cache_hasher::equal): Only compare
24838 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
24839 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
24840 non-aggregate element types.
24841 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
24842 about the flag on ARRAY_TYPEs in the comment, formatting fix.
24844 2017-04-27 Richard Biener <rguenther@suse.de>
24846 PR middle-end/80533
24847 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
24848 stripping ARRAY_REFs from MEM_EXPR make sure we're not
24849 keeping a reference to a trailing array.
24851 2017-04-27 Richard Biener <rguenther@suse.de>
24853 PR middle-end/80539
24854 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
24855 being in loop-closed SSA form conservatively.
24856 (chrec_fold_multiply_poly_poly): Likewise.
24858 2017-04-27 Tamar Christina <tamar.christina@arm.com>
24860 PR middle-end/79665
24861 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
24862 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
24864 2017-04-27 Jakub Jelinek <jakub@redhat.com>
24867 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
24868 (aarch64_function_arg_alignment): Return unsigned int again, but still
24869 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
24870 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
24871 Don't emit -Wpsabi note.
24872 (aarch64_function_arg_boundary): Likewise.
24873 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
24876 2017-04-26 Nathan Sidwell <nathan@acm.org>
24878 * tree.h (crc32_unsigned_n): Declare.
24879 (crc32_unsigned, crc32_unsigned): Make inline.
24880 * tree.c (crc32_unsigned_bits): Replace with ...
24881 (crc32_unsigned_n): ... this.
24882 (crc32_unsigned, crc32_byte): Remove.
24883 (crc32_string): Remove unnecessary braces.
24885 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
24887 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
24888 * ipa-inline-analysis.c (MAX_TIME): Remove.
24889 (account_size_time): Use sreal for time.
24890 (dump_inline_summary): Update.
24891 (estimate_function_body_sizes): Update.
24892 (estimate_edge_size_and_time): Update.
24893 (estimate_calls_size_and_time): Update.
24894 (estimate_node_size_and_time): Update.
24895 (inline_merge_summary): Update.
24896 (inline_update_overall_summary): Update.
24897 (estimate_time_after_inlining): Update.
24898 (inline_read_section): Update.
24899 (inline_write_summary): Update.
24900 * ipa-inline.c (compute_uninlined_call_time): Update.
24901 (compute_inlined_call_time): Update.
24902 (recursive_inlining): Update.
24903 (inline_small_functions): Update.
24904 (dump_overall_stats): Update.
24905 * ipa-inline.h: Include sreal.h.
24906 (size_time_entry): Turn time to sreal.
24907 (inline_summary): Turn self_time nad time to sreal.
24909 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
24911 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
24913 (sreal::stream_out, sreal::stream_in): New.
24914 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
24916 2017-04-25 Jakub Jelinek <jakub@redhat.com>
24918 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
24921 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
24924 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
24926 (dimode_scalar_chain::compute_convert_gain): Ditto.
24927 (dimode_scalar_chain::make_vector_copies): Ditto.
24928 (dimode_scalar_chain::convert_reg): Ditto.
24929 (dimode_scalar_chain::convert_insn): Ditto.
24930 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
24931 (VI248_AVX512BW_1): New mode iterator.
24932 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
24933 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
24936 2017-04-25 Martin Sebor <msebor@redhat.com>
24938 PR tree-optimization/80497
24939 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
24940 constants are representable in HOST_WIDE_INT.
24941 (parse_directive): Ditto.
24943 2017-04-25 Martin Sebor <msebor@redhat.com>
24946 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
24947 (new_zero_array): Adjust signature.
24948 (dom_info::dom_init): Used unsigned rather that size_t.
24949 (dom_info::dom_info): Same.
24951 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24952 Jakub Jelinek <jakub@redhat.com>
24955 * config/arm/arm.c: Include gimple.h.
24956 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
24957 returns negative, increment ncrn only if it returned positive.
24958 (arm_needs_doubleword_align): Return int instead of bool,
24959 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
24960 members, but if there is any such non-FIELD_DECL
24961 > PARM_BOUNDARY aligned decl, return -1 instead of false.
24962 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
24963 returns negative, increment nregs only if it returned positive.
24964 (arm_setup_incoming_varargs): Likewise.
24965 (arm_function_arg_boundary): Emit -Wpsabi note if
24966 arm_needs_doubleword_align returns negative, return
24967 DOUBLEWORD_ALIGNMENT only if it returned positive.
24969 2017-04-25 Marek Polacek <polacek@redhat.com>
24972 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
24973 first argument to type.
24975 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
24978 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
24979 type checks to test for compatibility instead of equality.
24981 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24982 Jakub Jelinek <jakub@redhat.com>
24985 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
24987 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
24988 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
24989 the alignment computation, but return their maximum in warn_alignment.
24990 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
24991 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
24993 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
24994 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
24997 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24999 * config/arc/simdext.md (dmpyh): Fix typo.
25001 2017-04-25 Richard Biener <rguenther@suse.de>
25003 PR tree-optimization/80492
25004 * alias.c (compare_base_decls): Handle registers with asm
25005 specification conservatively.
25006 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
25007 compare_base_decls returning dont-know properly.
25009 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25011 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
25012 (legitimate_offset_address_p): New function.
25013 (arc_legitimate_address_p): Use above function.
25015 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25017 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
25019 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25021 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
25022 ACCH registers whenever they are available.
25024 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25026 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
25027 double regs fix when not used.
25029 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25031 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
25033 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
25034 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
25036 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25038 * config/arc/arc.c (arc_output_addsi): Check for h-register class
25039 when emitting short ADD instructions.
25041 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25043 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
25045 (cmpsi_cc_c_insn): Likewise.
25046 (cbranchsi4_scratch): Compute proper instruction length using
25047 compact_hreg_operand.
25048 * config/arc/predicates.md (compact_hreg_operand): New predicate.
25050 2017-04-25 Richard Biener <rguenther@suse.de>
25052 PR middle-end/80509
25053 * passes.c (pass_manager::pass_manager): Initialize
25054 m_name_to_pass_map.
25056 2017-04-25 Richard Biener <rguenther@suse.de>
25058 PR tree-optimization/79201
25059 * tree-ssa-sink.c (statement_sink_location): Handle calls.
25061 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25064 * config/s390/vector.md: Split MEM->GPR vector moves for
25065 non-s_operand addresses.
25067 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25070 * config/s390/predicates.md (reload_const_wide_int_operand): New
25072 * config/s390/s390.md ("movti"): Remove d/P alternative.
25073 ("movti_bigconst"): New pattern definition.
25075 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25078 * s390-protos.h (s390_expand_cs_hqi): Removed.
25079 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
25080 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
25081 modes as well as CCZ1mode and CCZmode.
25082 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
25083 signature of s390_emit_compare_and_swap.
25084 (s390_expand_cs_hqi): Likewise, make static.
25085 (s390_expand_cs_tdsi): Generate an explicit compare before trying
25086 compare-and-swap, in some cases.
25087 (s390_expand_cs): Wrapper function.
25088 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
25090 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
25091 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
25092 patterns for small and large integers. Forbid symref memory operands.
25093 Move expander to s390.c. Require cc register.
25094 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
25095 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
25096 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
25097 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
25098 symref memory operands. Remove CC mode and call s390_match_ccmode
25100 ("atomic_exchange<mode>"): Allow and implement all integer modes.
25102 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25104 * config/s390/s390.md (define_peephole2): New peephole to help
25105 combining the load-and-test pattern with volatile memory.
25107 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25109 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
25110 with CCZmode for TARGET_Z196.
25112 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25114 PR rtl-optimization/80501
25115 * combine.c (make_compound_operation_int): Set subreg_code to SET
25116 even for AND with mask of the sign bit of mode.
25118 PR rtl-optimization/80500
25119 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
25120 sum's initial value.
25122 2017-04-25 Julian Brown <julian@codesourcery.com>
25123 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25125 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
25127 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
25129 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
25131 2017-04-25 Julian Brown <julian@codesourcery.com>
25132 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25134 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
25135 (thunderx2t99_sha): New Reservation.
25137 2017-04-25 Julian Brown <julian@codesourcery.com>
25138 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25140 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
25141 type for 1-element load.
25143 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
25145 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
25147 2017-04-24 Martin Jambor <mjambor@suse.cz>
25149 PR tree-optimization/80293
25150 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
25151 char arrays not totally scalarizable if it is false.
25152 (analyze_all_variable_accesses): Pass correct value in the new
25153 parameter. Add a statistics counter.
25155 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
25157 PR middle-end/79931
25158 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
25160 2017-04-24 Richard Biener <rguenther@suse.de>
25162 PR tree-optimization/80494
25163 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
25164 out for complex types.
25166 2017-04-24 Richard Biener <rguenther@suse.de>
25168 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
25169 * tree-ssa-sccvn.c (print_scc): Print SCC size.
25170 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
25171 (DFS): Adjust and never fail.
25172 (sccvn_dom_walker::fail): Remove.
25173 (sccvn_dom_walker::before_dom_children): Adjust.
25174 (run_scc_vn): Likewise and never fail.
25175 * tree-ssa-pre.c (pass_pre::execute): Adjust.
25176 (pass_fre::execute): Likewise.
25178 2017-04-24 Richard Biener <rguenther@suse.de>
25180 PR tree-optimization/79725
25181 * tree-ssa-sink.c (statement_sink_location): Return whether
25182 failure reason was zero uses. Move that check later.
25183 (sink_code_in_bb): Deal with zero uses by removing the stmt
25186 2017-04-24 Richard Biener <rguenther@suse.de>
25189 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
25190 pointer-based references.
25192 2017-04-24 Richard Biener <rguenther@suse.de>
25195 * pass_manager.h (pass_manager::operator new): Remove.
25196 (pass_manager::operator delete): Likewise.
25197 * passes.c (pass_manager::operator new): Remove.
25198 (pass_manager::operator delete): Likewise.
25199 (pass_manager::pass_manager): Zero individual pass members.
25201 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
25204 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
25205 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
25206 Check "XEXP (src, 1)" operand here.
25207 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
25208 Check "XEXP (src, 1)" operand here.
25209 (dimode_scalar_chain::make_vector_copies): Detect count register
25210 of a shift instruction. Zero extend count register from QImode
25211 to DImode to satisfy vector shift pattern count operand predicate.
25212 Substitute vector shift count operand with a DImode copy.
25213 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
25216 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
25218 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
25219 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
25220 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
25221 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
25222 (UNSPEC_NOREX_MEM): Remove definition.
25224 2017-04-21 Richard Biener <rguenther@suse.de>
25226 PR tree-optimization/79547
25227 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
25228 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
25229 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
25230 without any constraints.
25232 2017-04-21 Richard Biener <rguenther@suse.de>
25234 PR tree-optimization/78847
25235 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
25237 2017-04-21 Richard Biener <rguenther@suse.de>
25239 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
25240 (build_distinct_type_copy): Likewise.
25241 (build_variant_type_copy): Likewise.
25242 * tree.c (build_qualified_type): Pass down mem-stat info.
25243 (build_distinct_type_copy): Likewise.
25244 (build_variant_type_copy): Likewise.
25246 2017-04-21 Richard Biener <rguenther@suse.de>
25248 PR tree-optimization/80237
25249 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
25251 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
25252 for a simplified result.
25254 2016-04-21 Richard Biener <rguenther@suse.de>
25256 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
25257 sth as strict as a simple_iv but a chrec without symbols and an
25258 operand defined in the loop we are peeling (and not some subloop).
25259 (propagate_constants_for_unrolling): Propagate all constants.
25261 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25264 * config/i386/i386.c (print_reg): Remove assert for disalowed
25265 regno values, call output_operand_lossage instead.
25267 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25270 * config/i386/constraints.md (Yc): New register constraint.
25271 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
25272 Use Yc constraint for alternative 2 of operand 0. Remove
25273 preferred_for_speed attribute.
25275 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
25277 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
25278 lastprivate clauses in SIMT case.
25280 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
25282 * doc/invoke.texi (-Wextra-semi): Document new warning option.
25284 2017-04-20 Richard Biener <rguenther@suse.de>
25286 PR tree-optimization/57796
25287 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
25288 as N scalar stores.
25289 (vect_model_load_cost): Cost gathers as N scalar loads.
25291 2017-04-20 Richard Biener <rguenther@suse.de>
25293 * ggc-page.c (ggc_allocated_p): Rename to ...
25294 (safe_lookup_page_table_entry): ... this and return the lookup
25296 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
25298 2017-04-20 Richard Biener <rguenther@suse.de>
25300 PR tree-optimization/80453
25301 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
25302 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
25303 from the conditions.
25304 (vn_phi_eq): Pass them down.
25305 (vn_phi_lookup): Record them.
25306 (vn_phi_insert): Likewise.
25308 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
25310 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
25311 uninitialized variable warning to avoid buffer overrun.
25313 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
25316 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
25317 is suppressed for '{ 0 }' in C.
25319 2017-04-20 Jakub Jelinek <jakub@redhat.com>
25321 * BASE-VER: Set to 8.0.0.
25323 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
25325 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
25326 priority .init_array and .fini_array section with SECTION_NOTYPE
25329 2017-04-20 Jakub Jelinek <jakub@redhat.com>
25331 PR middle-end/80423
25332 * tree.h (build_array_type): Add typeless_storage default argument.
25333 * tree.c (type_cache_hasher::equal): Also compare
25334 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
25335 (build_array_type): Add typeless_storage argument, set
25336 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
25338 (build_nonshared_array_type): Adjust build_array_type_1 caller.
25339 (build_array_type): Likewise. Add typeless_storage argument.
25341 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
25342 Jakub Jelinek <jakub@redhat.com>
25344 PR tree-optimization/80426
25345 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
25346 operation on symbolic operands, also compute the overflow for the
25347 invariant part when the operation degenerates into a negation.
25349 2017-04-19 Jakub Jelinek <jakub@redhat.com>
25352 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
25353 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
25356 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
25358 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
25361 * config/avr/avr.c (tree.h): Include it.
25362 (cgraph.h): Include it.
25363 (avr_encode_section_info): Don't warn for uninitialized progmem
25364 variable if it's just an alias.
25366 2017-04-19 Richard Biener <rguenther@suse.de>
25369 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
25370 when needed by AutoPGO.
25372 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
25375 * doc/lto.texi: Remove an extra 'that'.
25377 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
25379 PR rtl-optimization/80429
25380 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
25381 are only used in debug insns.
25383 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
25384 Vladimir Makarov <vmakarov@redhat.com>
25386 * config/sparc/predicates.md (input_operand): Add comment. Return
25387 true for any memory operand when LRA is in progress.
25388 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
25390 2017-04-18 Jeff Law <law@redhat.com>
25393 * mips.md ({return,simple_return}_internal): Do not overwrite
25396 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25398 PR tree-optimization/80443
25399 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
25400 instead of adding 1, subtract -1 and similarly instead of subtracting
25403 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
25405 PR rtl-optimization/80357
25406 * haifa-sched.c (tmp_bitmap): New variable.
25407 (model_recompute): Handle duplicate use records.
25408 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
25409 (free_global_sched_pressure_data): Free it.
25411 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
25414 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
25415 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25416 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25417 instead of SYSTEM_HEADER_DIR.
25419 2017-04-18 Jeff Law <law@redhat.com>
25421 PR middle-end/80422
25422 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
25423 predecessors after walking up the insn chain.
25425 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25428 * dwarf2out.c (modified_type_die): Try harder not to emit internal
25429 sizetype type into debug info.
25431 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
25434 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
25435 unneeded test for TARGET_UPPER_REGS_SF.
25436 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
25438 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25441 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
25442 instead of gsi_after_labels.
25444 2017-04-18 Jeff Law <law@redhat.com>
25446 * regcprop.c (maybe_mode_change): Avoid creating copies of the
25450 2017-04-13 Jeff Law <law@redhat.com>
25451 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
25452 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
25454 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
25457 * config/avr/avr.c (intl.h): Include it.
25458 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
25460 2017-04-18 Martin Liska <mliska@suse.cz>
25462 PR gcov-profile/78783
25463 * gcov-tool.c (gcov_output_files): Validate that destination
25464 file is either removed by the tool or by a user.
25466 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
25467 Guy Benyei <guybe@mellanox.com>
25469 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
25470 block, and do not negate it, the stored id is already negative.
25472 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
25474 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
25476 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
25479 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
25480 masks of options that should be turned off if the VSX vector
25481 options are turned off.
25482 (OTHER_P8_VECTOR_MASKS): Likewise.
25483 (OTHER_VSX_VECTOR_MASKS): Likewise.
25484 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
25485 rs6000_disable_incompatible_switches to validate no type switches
25487 (rs6000_incompatible_switch): New function to disallow turning on
25488 other vector options if -mno-vsx, -mno-power8-vector, or
25489 -mno-power9-vector are specified.
25491 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25493 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
25495 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25497 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
25498 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
25499 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
25500 (ARG_POINTER_CFA_OFFSET): Likewise.
25502 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25504 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
25505 conditions to take advantage of various optimizations.
25507 2017-04-13 Jeff Law <law@redhat.com>
25509 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
25510 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
25511 (zero_extendsidi2_dext): Likewise.
25513 2017-04-13 Jakub Jelinek <jakub@redhat.com>
25516 * fold-const.c (fold_ternary_loc): Revert
25517 use op0 instead of fold_convert_loc (loc, type, arg0) part of
25520 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
25522 PR rtl-optimization/80343
25523 * lra-remat.c (update_scratch_ops): Assign original hard reg to
25524 new scratch pseudo.
25526 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
25529 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
25530 to ubsan_encode_value.
25532 2017-04-13 Jeff Law <law@redhat.com>
25534 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
25535 appearing in DEBUG_INSNs.
25537 2017-04-13 Martin Liska <mliska@suse.cz>
25539 PR gcov-profile/80413
25540 * gcov-io.c (gcov_write_string): Copy to buffer just when
25541 allocated size is greater than zero.
25543 2017-04-13 Jakub Jelinek <jakub@redhat.com>
25546 * dwarf2out.c (decls_for_scope): Ignore declarations of
25547 current_function_decl in BLOCK_NONLOCALIZED_VARS.
25549 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
25552 * ipa-visibility.c (non_local_p): Fix typos.
25553 (localize_node): When localizing symbol in same comdat group,
25554 dissolve the group only when we know external symbols are going
25556 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
25558 2017-04-12 Jakub Jelinek <jakub@redhat.com>
25560 PR tree-optimization/79390
25561 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
25562 order does not result in usable sequence, retry with reversed operand
25568 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
25569 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
25570 op0 instead of fold_convert_loc (loc, type, arg0).
25572 2017-04-12 Jeff Law <law@redhat.com>
25574 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
25575 has a delay slot in the generated code.
25577 * config/cris/cris.md (cris_preferred_reload_class): Return
25578 GENNONACR_REGS rather than GENERAL_REGS.
25580 2017-04-12 Jakub Jelinek <jakub@redhat.com>
25583 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
25584 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
25585 signedness of the result type.
25587 2017-04-12 Richard Biener <rguenther@suse.de>
25588 Jeff Law <law@redhat.com>
25590 PR tree-optimization/80359
25591 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
25592 trim stores to TARGET_MEM_REFs.
25594 2017-04-12 Richard Biener <rguenther@suse.de>
25596 PR tree-optimization/79390
25597 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
25598 threading case even more.
25600 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
25603 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
25604 for quad_address_p for TImode, instead of just not indexed_address.
25606 2017-04-12 Richard Biener <rguenther@suse.de>
25607 Bernd Edlinger <bernd.edlinger@hotmail.de>
25609 PR middle-end/79671
25610 * alias.c (component_uses_parent_alias_set_from): Handle
25611 TYPE_TYPELESS_STORAGE.
25612 (get_alias_set): Likewise.
25613 * tree-core.h (tree_type_common): Add typeless_storage flag.
25614 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
25615 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
25616 for types containing members with TYPE_TYPELESS_STORAGE.
25617 (place_field): Likewise.
25618 (layout_type): Likewise for ARRAY_TYPE.
25619 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
25620 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
25621 TYPE_TYPELESS_STORAGE.
25622 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
25624 2017-04-12 Jakub Jelinek <jakub@redhat.com>
25627 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
25628 first argument to type.
25630 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25634 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
25635 CONST0_RTX (mode) rather than const0_rtx where appropriate.
25636 (rs6000_expand_binop_builtin): Likewise.
25637 (rs6000_expand_ternop_builtin): Likewise; also add missing
25638 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
25639 vshasigma built-ins.
25640 * doc/extend.texi: Document that vec_xxpermdi's third argument
25641 must be a constant.
25643 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
25645 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
25646 Use shift_const cost parameter when calculating gain of STV shifts.
25648 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
25650 PR rtl-optimization/70478
25651 * lra-constraints.c (process_alt_operands): Check memory for
25652 disfavoring memory insn operand.
25654 2017-04-11 Jakub Jelinek <jakub@redhat.com>
25656 PR middle-end/80100
25657 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
25658 left shift in unsigned HOST_WIDE_INT type.
25660 PR rtl-optimization/80385
25661 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
25662 (not (neg X)) into (plus X -1) for complex or non-integral modes.
25665 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
25666 if they have any depend clauses.
25668 2017-04-11 Martin Liska <mliska@suse.cz>
25671 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
25672 * ipa-split.c (split_function): Create a local comdat symbol
25673 if caller is in a comdat group.
25675 2017-04-11 Martin Liska <mliska@suse.cz>
25678 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
25681 2017-04-11 Martin Sebor <msebor@redhat.com>
25683 PR middle-end/80364
25684 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
25685 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
25687 (directive::set_width, directive::set_precision, format_character):
25689 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
25692 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
25695 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
25696 conflict, set target->arch_name instead of target->cpu_name.
25698 2017-04-11 Richard Biener <rguenther@suse.de>
25700 PR tree-optimization/80374
25701 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
25702 build_zero_cst, remove fold_convertible_p check again.
25704 2017-04-11 Martin Liska <mliska@suse.cz>
25707 * ubsan.c (instrument_object_size): Do not instrument register
25710 2017-04-11 Jakub Jelinek <jakub@redhat.com>
25713 * config/i386/i386-builtin-types.def
25714 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
25715 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
25716 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
25717 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
25718 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
25719 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
25720 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
25721 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
25722 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
25723 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
25724 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
25725 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
25726 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
25727 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
25728 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
25729 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
25730 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
25731 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
25732 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
25733 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
25734 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
25735 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
25736 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
25737 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
25738 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
25739 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
25740 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
25741 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
25742 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
25743 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
25744 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
25745 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
25746 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
25747 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
25748 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
25749 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
25750 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
25751 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
25752 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
25753 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
25754 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
25755 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
25756 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
25757 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
25758 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
25759 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
25761 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
25762 flag to second_arg_count, handle 4 argument function type _COUNT
25763 aliases, handle second_arg_count on second argument rather than last.
25765 2017-04-10 Jeff Law <law@redhat.com>
25767 PR tree-optimization/80374
25768 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
25769 record anything if we can not convert integer_zero_node to the
25772 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
25775 * config/rs6000/rs6000.c (rs6000_option_override_internal):
25776 Enhance special handling given to the TARGET_P9_MINMAX option in
25777 relation to certain other options.
25779 2017-04-10 Bin Cheng <bin.cheng@arm.com>
25781 PR tree-optimization/80153
25782 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
25783 remove POINTER_PLUS_EXPR's base part directly, rather than through
25786 2017-04-10 Richard Biener <rguenther@suse.de>
25787 Bin Cheng <bin.cheng@arm.com>
25789 PR tree-optimization/80153
25790 * tree-affine.c (aff_combination_to_tree): Get base pointer from
25791 the first element of pointer type aff_tree. Build result expr in
25793 (add_elt_to_tree): Convert to type unconditionally. Remove other
25794 fold_convert calls.
25795 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
25796 (rewrite_use_nonlinear_expr): Check invariant using iv information.
25798 2017-04-10 Richard Biener <rguenther@suse.de>
25800 * tree-ssa-structalias.c (find_func_aliases): Properly handle
25803 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
25805 PR rtl-optimization/70478
25806 * lra-constraints.c (curr_small_class_check): New.
25807 (update_and_check_small_class_inputs): New.
25808 (process_alt_operands): Update curr_small_class_check. Disfavor
25809 alternative insn memory operands. Check available regs for small
25812 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
25815 * config/mips/mips.opt (-mvirt): Update description.
25816 * doc/invoke.texi (-mvirt): Likewise.
25818 2017-04-10 Richard Biener <rguenther@suse.de>
25820 PR middle-end/80362
25821 * fold-const.c (fold_binary_loc): Look at unstripped ops when
25822 looking for NEGATE_EXPR in -A / -B to A / B folding.
25824 2017-04-10 Martin Liska <mliska@suse.cz>
25826 PR gcov-profile/80224
25827 * gcov.c (print_usage): Fix usage string.
25828 (get_gcov_intermediate_filename): Remove.
25829 (output_gcov_file): Use both for normal and intermediate format.
25830 (generate_results): Do not initialize special file for
25831 intermediate format.
25833 2017-04-10 Richard Biener <rguenther@suse.de>
25835 PR tree-optimization/80304
25836 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
25839 2017-04-10 Nathan Sidwell <nathan@acm.org>
25842 * config/rs6000/rs6000.c (rs6000_vector_type): New.
25843 (rs6000_init_builtins): Use it.
25845 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25847 * config/arm/arm.md (<mrc>): Add mode to SET source.
25848 (<mrrc>): Likewise.
25850 2017-04-10 Richard Biener <rguenther@suse.de>
25852 PR middle-end/80344
25853 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
25855 2017-04-10 Jakub Jelinek <jakub@redhat.com>
25858 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
25859 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
25860 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
25861 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
25862 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
25863 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
25864 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
25865 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
25866 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
25867 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
25868 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
25869 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
25870 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
25871 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
25872 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
25873 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
25874 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
25875 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
25876 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
25877 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
25878 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
25879 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
25880 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
25882 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
25884 PR rtl-optimization/70478
25885 * lra-constraints.c: Reverse the last patch.
25887 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
25889 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
25890 Add comment for WCHAR_T.
25892 2017-04-08 Martin Liska <mliska@suse.cz>
25895 2017-04-07 Martin Liska <mliska@suse.cz>
25898 * ipa-split.c (split_function): Add function part to a same comdat
25901 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25904 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
25906 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
25908 * rs6000/rs6000.c (vec_load_pendulum): Rename...
25909 (vec_pairing): ...to this.
25910 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
25911 (rs6000_sched_init): Adjust for name change.
25912 (struct rs6000_sched_context): Likewise.
25913 (rs6000_init_sched_context): Likewise.
25914 (rs6000_set_sched_context): Likewise.
25916 2017-04-07 Jakub Jelinek <jakub@redhat.com>
25922 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
25924 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
25925 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
25926 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
25928 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
25930 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
25932 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
25934 PR rtl-optimization/70703
25935 * ira-color.c (update_conflict_hard_regno_costs): Use
25936 int64_t instead of HOST_WIDE_INT.
25938 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
25940 PR rtl-optimization/70478
25941 * lra-constraints.c (process_alt_operands): Disfavor alternative
25942 insn memory operands.
25944 2017-04-07 Jeff Law <law@redhat.com>
25946 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
25947 CALL and NOTE_INSN_CALL_ARG_LOCATION.
25949 2017-04-07 Martin Liska <mliska@suse.cz>
25952 * config/aarch64/aarch64.c (aarch64_process_target_attr):
25953 Show error message instead of an ICE.
25955 2017-04-07 Martin Liska <mliska@suse.cz>
25958 * ipa-split.c (split_function): Add function part to a same comdat
25961 2017-04-07 Richard Biener <rguenther@suse.de>
25963 PR middle-end/80341
25964 * tree.c (get_unwidened): Also handle ! for_type case for
25966 * convert.c (do_narrow): Split out from ...
25967 (convert_to_integer_1): ... here. Do not pass final truncation
25968 type to get_unwidened for TRUNC_DIV_EXPR.
25970 2017-04-07 Richard Biener <rguenther@suse.de>
25972 * tree-affine.c (wide_int_ext_for_comb): Take type rather
25974 (aff_combination_const): Adjust.
25975 (aff_combination_scale): Likewise.
25976 (aff_combination_add_elt): Likewise.
25977 (aff_combination_add_cst): Likewise.
25978 (aff_combination_convert): Likewise.
25979 (add_elt_to_tree): Likewise. Remove unused argument.
25980 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
25982 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
25984 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
25986 * config/arm/arm.c (arm_default_short_enums): Use
25987 ARM_DEFAULT_SHORT_ENUMS.
25988 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
25990 2017-04-06 Jakub Jelinek <jakub@redhat.com>
25993 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
25994 members with redundant out-of-class redeclaration.
25996 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
25999 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
26000 * config/i386/i386.md (*zero_extendsidi2):
26001 Add (?*x,*x) and (?*v,*v) alternatives.
26003 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
26006 * config/i386/i386.c (ix86_expand_builtin)
26007 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
26008 mode from insn data. Convert operands to insn operand mode.
26009 Copy operands that don't satisfy insn predicate to a register.
26011 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
26013 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
26016 2017-04-06 Richard Biener <rguenther@suse.de>
26018 PR tree-optimization/80334
26019 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
26020 preserve alignment of accesses.
26022 2017-04-06 Richard Biener <rguenther@suse.de>
26024 PR tree-optimization/80262
26025 * tree-sra.c (build_ref_for_offset): Preserve address-space
26027 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
26028 Drop useless address-space information on MEM_REF offsets.
26030 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
26032 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
26034 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26036 PR rtl-optimization/70703
26037 * ira-color.c (update_conflict_hard_regno_costs): Use
26038 HOST_WIDE_INT instead of long.
26040 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
26043 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
26044 not defined for x86_64 target. Add -mmmx target option when __SSE2__
26046 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
26047 for x86_64 target. Handle -m3dnowa option.
26049 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26051 PR rtl-optimization/70703
26052 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
26053 (update_conflict_hard_regno_costs): Use long instead of unsigned
26054 arithmetic for cost calculation.
26056 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26057 Bernd Edlinger <bernd.edlinger@hotmail.de>
26060 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
26063 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
26066 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
26067 ptr_mode with Pmode throughout.
26068 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
26069 into probe_stack_range and use DImode.
26071 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
26074 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
26075 call_eh_return is true.
26077 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26079 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
26080 Initialize last_match_fntype_index.
26082 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26085 * tree-nvr.c: Include internal-fn.h.
26086 (pass_return_slot::execute): Ignore internal calls without
26089 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26090 Richard Biener <rguenther@suse.de>
26093 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
26094 captures used multiple times, except for the last use.
26095 * generic-match-head.c: Include gimplify.h.
26097 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26099 PR tree-optimization/79390
26100 * target.h (struct noce_if_info): Declare.
26101 * targhooks.h (default_noce_conversion_profitable_p): Declare.
26102 * target.def (noce_conversion_profitable_p): New target hook.
26103 * ifcvt.h (struct noce_if_info): New type, moved from ...
26104 * ifcvt.c (struct noce_if_info): ... here.
26105 (noce_conversion_profitable_p): Renamed to ...
26106 (default_noce_conversion_profitable_p): ... this. No longer
26108 (noce_try_store_flag_constants, noce_try_addcc,
26109 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
26110 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
26111 instead of noce_conversion_profitable_p.
26112 * config/i386/i386.c: Include ifcvt.h.
26113 (ix86_option_override_internal): Don't override
26114 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
26115 (ix86_noce_conversion_profitable_p): New function.
26116 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
26117 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
26118 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
26119 * doc/tm.texi: Regenerated.
26121 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26123 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
26126 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26129 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
26130 instructions for small multiply cores.
26132 2017-04-04 Jeff Law <law@redhat.com>
26134 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
26136 (mips_expand_vec_perm_const): Initialize elements in orig_perm
26137 that are not set by the loop over the elements.
26139 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26142 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
26143 int mode, convert_modes it to mode as unsigned, otherwise use
26144 lowpart_subreg to mode rather than SImode.
26145 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
26146 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
26147 Use DImode instead of SImode for the shift count operand.
26148 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
26151 2017-04-04 Richard Biener <rguenther@suse.de>
26153 PR middle-end/80281
26154 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
26155 arithmetic done for the negate or the plus. Simplify.
26156 (A - (-B) -> A + B): Likewise.
26157 * fold-const.c (split_tree): Make sure to not negate pointers.
26159 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
26161 PR rtl-optimization/60818
26162 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
26163 a compare of comparisons with the thing compared if this results
26164 in a different machine mode.
26166 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
26168 * alias.c (base_alias_check): Fix typo in comment.
26169 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
26170 * cgraphunit.c (symbol_table::compile): Likewise.
26171 * collect2.c (maybe_run_lto_and_relink): Likewise.
26172 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
26173 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
26174 * config/avr/avr.c (avr_map_op_t): Likewise.
26175 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
26176 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
26177 * config/epiphany/epiphany.md (movcc): Likewise.
26178 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
26179 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
26181 * config/mips/mips.c (mips_save_restore_reg): Likewise.
26182 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
26183 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
26184 * config/sh/sh.c (sh_rtx_costs): Likewise.
26185 * fold-const.c (fold_truth_andor): Likewise.
26186 * genautomata.c (collapse_flag): Likewise.
26187 * gengtype.h (struct type::u::s): Likewise.
26188 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
26189 * input.c (FORMAT_AMOUNT): Likewise.
26190 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
26191 (known_aggs_to_agg_replacement_list): Likewise.
26192 * ipa-inline-analysis.c: Likewise.
26193 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
26194 * ipa-polymorphic-call.c
26195 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
26196 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
26197 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
26199 * modulo-sched.c (apply_reg_moves): Likewise.
26200 * omp-expand.c (build_omp_regions_1): Likewise.
26201 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
26202 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
26203 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
26204 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
26205 * value-prof.c: Likewise.
26206 * var-tracking.c (val_reset): Likewise.
26208 2017-04-03 Richard Biener <rguenther@suse.de>
26210 PR tree-optimization/80275
26211 * fold-const.c (split_address_to_core_and_offset): Handle
26214 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
26216 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
26217 descriptors is at least equal to that of functions.
26219 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
26221 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
26223 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
26226 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
26227 (mov<IMOD4:mode>): New expander.
26228 (*mov<IMOD4:mode>_internal): New insn and split pattern.
26230 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
26232 PR rtl-optimization/79405
26233 * fwprop.c (propagations_left): New variable.
26234 (forward_propagate_into): Decrement it.
26235 (fwprop_init): Initialize it.
26236 (fw_prop): If the variable has reached zero, stop propagating.
26237 (fwprop_addr): Ditto.
26239 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26242 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
26243 a FUNCTION_DECL, pass it as decl instead of origin to
26246 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
26248 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
26251 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
26254 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
26255 TARGET_VSX_SMALL_INTEGER.
26257 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26259 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
26260 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
26262 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26264 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
26265 extraction from odd-numbered MSA register.
26267 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26269 PR middle-end/80173
26270 * expmed.c (store_bit_field_1): Don't attempt to create
26271 a word subreg out of hard registers wider than word if they
26272 have HARD_REGNO_NREGS of 1 for their mode.
26274 PR middle-end/80163
26275 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
26276 conversions to integer types wider than word and pointer.
26279 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
26280 (rtx_equal_for_cselib_p): Pass 0 to it.
26281 * cselib.c (cselib_hasher::equal): Likewise.
26282 (rtx_equal_for_cselib_1): Add depth argument. If depth
26283 is 128, don't look up VALUE locs and punt. Increment
26284 depth in recursive calls when walking VALUE locs.
26286 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
26288 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
26289 (make_gcov_file_name): Use the canonical path name for generating
26291 (read_line): Fix handling of files with ascii null bytes.
26293 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
26295 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
26296 to initialise a vector register instead
26297 of using a const_int.
26299 2017-03-30 Jakub Jelinek <jakub@redhat.com>
26301 PR translation/80189
26302 * gimplify.c (omp_default_clause): Use %qs instead of %s in
26303 diagnostic messages.
26305 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
26308 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
26309 (dfp_diex_<mode>): Update mode of operand 1.
26310 * doc/extend.texi (dxex, dxexq): Document change to return type.
26311 (diex, diexq): Document change to argument type.
26313 2017-03-30 Martin Jambor <mjambor@suse.cz>
26316 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
26317 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
26318 it reflects the signature changes performed at the callee side.
26319 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
26320 to cgraph_build_function_type_skip_args.
26321 (build_function_decl_skip_args): Adjust call to the above function.
26323 2017-03-30 Jakub Jelinek <jakub@redhat.com>
26326 * config/i386/sse.md
26327 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
26328 register as dest whenever it is a MEM not rtx_equal_p to the
26329 corresponding dup operand, and when forcing into reg move the
26330 reg into the memory afterwards.
26331 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
26332 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
26333 for the force_reg mode.
26334 (avx512vl_vextractf128<mode>): Use register as dest either
26335 always when a MEM, or when it is a MEM not rtx_equal_p to the
26336 corresponding dup operand, or even not when it is a CONST_VECTOR
26337 depending on the mode and lo vs. hi.
26338 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
26340 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
26341 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
26342 Likewise. Require that operands[2] is even.
26343 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
26344 Remove extraneous parens. Require that operands[2] is a multiple
26346 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
26347 operands[0] is a MEM if <mask_applied>, the predicates/constraints
26348 disallow memory then.
26350 2017-03-30 Richard Biener <rguenther@suse.de>
26352 PR tree-optimization/77498
26353 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
26354 to non-constants over backedges.
26356 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
26358 PR rtl-optimization/80233
26359 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
26360 as last_combined_insn. Do not test for BARRIER_P separately.
26362 2017-03-29 Andreas Schwab <schwab@suse.de>
26365 * calls.c (prepare_call_address): Convert funexp to Pmode before
26366 copying to temp reg.
26368 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26370 PR tree-optimization/80158
26371 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
26372 Handle possible future case of more than one alternate
26374 (replace_rhs_if_not_dup): Likewise.
26375 (replace_one_candidate): Likewise.
26377 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
26379 PR rtl-optimization/80193
26380 * ira.c (ira): Do not check allocation for LRA.
26382 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
26384 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
26385 (nvptx_output_simt_exit): Declare.
26386 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
26387 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
26388 (init_softstack_frame): Move initialization of crtl->is_leaf to...
26389 (nvptx_declare_function_name): ...here. Emit declaration of local
26390 memory space buffer for omp_simt_enter insn.
26391 (nvptx_output_unisimt_switch): New.
26392 (nvptx_output_softstack_switch): New.
26393 (nvptx_output_simt_enter): New.
26394 (nvptx_output_simt_exit): New.
26395 * config/nvptx/nvptx.h (struct machine_function): New fields
26396 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
26397 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
26398 (UNSPECV_SIMT_EXIT): Ditto.
26399 (omp_simt_enter_insn): New insn.
26400 (omp_simt_enter): New expansion.
26401 (omp_simt_exit): New insn.
26402 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
26404 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
26405 (expand_GOMP_SIMT_ENTER_ALLOC): New.
26406 (expand_GOMP_SIMT_EXIT): New.
26407 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
26408 (GOMP_SIMT_ENTER_ALLOC): Ditto.
26409 (GOMP_SIMT_EXIT): Ditto.
26410 * target-insns.def (omp_simt_enter): New insn.
26411 (omp_simt_exit): Ditto.
26412 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
26414 (lower_rec_simd_input_clauses): Implement SIMT privatization.
26415 (lower_rec_input_clauses): Likewise.
26416 (lower_lastprivate_clauses): Handle SIMT privatization.
26418 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
26419 (ompdevlow_adjust_simt_enter): New.
26420 (find_simtpriv_var_op): New.
26421 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
26422 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
26424 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
26425 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
26426 (copy_decl_for_dup_finish): Ditto.
26428 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
26430 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
26433 * config/i386/i386.c (ix86_option_override_internal): Always
26434 allow -mpreferred-stack-boundary=3 for 64-bit targets.
26436 2017-03-28 Bin Cheng <bin.cheng@arm.com>
26438 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
26440 2017-03-28 Bin Cheng <bin.cheng@arm.com>
26442 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
26443 mark new edge's irreducible flag accordign to it.
26444 (vect_do_peeling): Check loop preheader edge's irreducible flag
26445 and pass it to function slpeel_add_loop_guard.
26447 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
26449 PR tree-optimization/80218
26450 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
26451 Update block frequencies and counts.
26453 2017-03-28 Richard Biener <rguenther@suse.de>
26455 PR tree-optimization/78644
26456 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
26457 of a simplification result we may not use it at all.
26459 2017-03-28 Richard Biener <rguenther@suse.de>
26462 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
26463 without arguments, generate default definition of a SSA name.
26465 2017-03-28 Richard Biener <rguenther@suse.de>
26467 PR middle-end/80222
26468 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
26469 TYPE_REF_CAN_ALIAS_ALL references.
26470 * fold-const.c (fold_indirect_ref_1): Likewise.
26472 2017-03-28 Martin Liska <mliska@suse.cz>
26475 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
26476 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
26478 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
26479 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
26481 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
26482 (EXTRA_SPECS): Define.
26483 (SUBTARGET_EXTRA_SPECS): Likewise.
26484 (SUBTARGET_CPP_SPEC): Likewise.
26485 * config/arc/elf.h (EXTRA_SPECS): Renamed to
26486 SUBTARGET_EXTRA_SPECS.
26487 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
26489 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
26491 * config/arc/simdext.md (vst64_insn): Update pattern.
26492 (vld32wh_insn): Likewise.
26493 (vld32wl_insn): Likewise.
26494 (vld64_insn): Likewise.
26495 (vld32_insn): Likewise.
26497 2017-03-28 Marek Polacek <polacek@redhat.com>
26500 * fold-const.c (fold_comparison): Use protected_set_expr_location
26501 instead of SET_EXPR_LOCATION.
26503 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
26505 * tree.c (add_expr): Avoid name lookup warning.
26507 2017-03-27 Jeff Law <law@redhat.com>
26509 PR tree-optimization/80216
26510 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
26511 function name. Limit recursion depth.
26512 (record_temporary_equivalences): Corresponding changes.
26514 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
26516 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
26519 2017-03-27 Jakub Jelinek <jakub@redhat.com>
26522 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
26524 * cfgcleanup.c (reg_note_cfa_p): New array.
26525 (insns_have_identical_cfa_notes): New function.
26526 (old_insns_match_p): Don't cross-jump in between /f
26527 and non-/f instructions. If both i1 and i2 are frame related,
26528 verify all CFA notes, their order and content.
26530 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
26533 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
26534 HImode and SImode with zero extend to DImode to one insn.
26535 (bswap<mode>2_extenddi): Likewise.
26536 (bswapsi2_extenddi): Likewise.
26537 (bswaphi2_extendsi): Likewise.
26538 (bswaphi2): Combine bswap HImode and SImode into one insn.
26539 Separate memory insns from swapping register.
26540 (bswapsi2): Likewise.
26541 (bswap<mode>2): Likewise.
26542 (bswaphi2_internal): Delete, no longer used.
26543 (bswapsi2_internal): Likewise.
26544 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
26545 store, and gpr<-gpr swap insns.
26546 (bswap<mode>2_store): Likewise.
26547 (bswaphi2_reg): Register only splitter, combine with the splitter.
26548 (bswaphi2 splitter): Likewise.
26549 (bswapsi2_reg): Likewise.
26550 (bswapsi2 splitter): Likewise.
26551 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
26552 the insns into load, store, and register/register insns.
26553 (bswapdi2_ldbrx): Likewise.
26554 (bswapdi2_load): Likewise.
26555 (bswapdi2_store): Likewise.
26556 (bswapdi2_reg): Likewise.
26558 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
26560 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
26561 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
26563 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
26566 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
26568 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
26569 special handling for target option conflicts between dform
26570 options (-mpower9-dform, -mpower9-dform-vector,
26571 -mpower9-dform-scalar) and -mno-direct-move.
26573 2017-03-27 Richard Biener <rguenther@suse.de>
26575 PR tree-optimization/80181
26576 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
26578 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26580 * config/arc/predicates.md (move_double_src_operand): Replace the
26581 call to move_double_src_operand with a call to address_operand.
26583 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26585 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
26586 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
26587 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
26589 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26591 * config/arc/predicates.md (long_immediate_loadstore_operand):
26592 Consider scaled addresses cases.
26594 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26596 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
26597 restored when in interrupt.
26598 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
26599 doesn't have delay slot.
26601 2017-03-27 Richard Biener <rguenther@suse.de>
26604 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
26605 inlined thunk clones.
26607 2017-03-27 Jakub Jelinek <jakub@redhat.com>
26610 * asan.c (instrument_derefs): Copy over last operand from
26611 original COMPONENT_REF to the new COMPONENT_REF with
26612 DECL_BIT_FIELD_REPRESENTATIVE.
26613 * ubsan.c (instrument_object_size): Likewise.
26615 2017-03-27 Richard Biener <rguenther@suse.de>
26617 PR tree-optimization/80170
26618 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
26619 sure DR/SCEV didnt fold in constants we do not see when looking
26620 at the reference base alignment.
26622 2017-03-27 Richard Biener <rguenther@suse.de>
26624 PR middle-end/80171
26625 * gimple-fold.c (fold_ctor_reference): Properly guard against
26626 NULL return value from canonicalize_constructor_val.
26628 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
26631 * config/i386/i386.c (ix86_expand_builtin)
26632 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
26633 flags reg setting and flags reg using instructions.
26634 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
26635 clobbering instructions to zero extend op2.
26637 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
26639 * doc/install.texi (Configuration) <--with-aix-soname>:
26640 Update link to AIX ld.
26642 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
26644 PR rtl-optimization/80160
26645 PR rtl-optimization/80159
26646 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
26647 reg_alternate_class into account.
26649 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
26652 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
26653 to consider in curr_insn_transform.
26655 2017-03-24 Jakub Jelinek <jakub@redhat.com>
26657 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
26658 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
26659 and emit_mode_inner.
26661 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26663 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
26664 argument to the overloaded builtin variants. Use the new flag to
26665 deprecate certain builtin variants.
26666 * config/s390/s390-builtin-types.def: Add new builtin types.
26667 * config/s390/s390-builtins.h: Support new flags field for
26668 overloaded builtins.
26669 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
26670 (s390_macro_to_expand): Enable vector float data type.
26671 (s390_cpu_cpp_builtins_internal): Indicate support of the new
26672 builtins by incrementing the __VEC__ version number.
26673 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
26675 (s390_resolve_overloaded_builtin): Emit error messages depending
26676 on the builtin flags.
26677 * config/s390/s390.c (s390_expand_builtin): Support additional
26678 flags argument. Change error message to match the messages
26679 emitted in s390-c.c.
26680 * config/s390/s390.md: New UNSPEC_* constants.
26681 (op_type): Add new instruction types.
26682 * config/s390/vecintrin.h: Add new builtins and test data class
26684 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
26685 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
26686 (VEC_INEXACT, VEC_NOINEXACT): New constants.
26687 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
26688 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
26689 ("vec_mergel<mode>"): V_HW -> VEC_HW.
26691 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
26692 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
26693 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
26694 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
26696 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
26697 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
26698 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
26699 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
26701 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
26702 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
26703 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
26704 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
26705 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
26706 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
26707 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
26709 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
26710 ("vec_scatter_element<V_HW_4:mode>_DI")
26711 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
26712 ("vec_fpint<mode>", "vflls")
26713 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
26714 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
26715 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
26716 ("*vec_cmphe<mode>_cc"): ... these.
26718 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
26719 mode constant instead of magic value.
26721 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26723 * config/s390/s390.c (s390_expand_vec_compare): Support other
26724 vector floating point modes than just V2DF.
26725 (s390_expand_vcond): Likewise.
26726 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
26727 (s390_cannot_change_mode_class): Prevent mode changes between TF
26728 and V1TF in vector registers.
26729 * config/s390/s390.md (DF, SF): New mode attributes.
26730 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
26731 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
26732 SFmode support for VRs.
26733 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
26735 (VFT, VF_HW): New mode iterators.
26736 (vw, sdx): New mode attributes.
26737 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
26738 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
26739 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
26740 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
26741 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
26742 also the new vector floating point modes. Renaming to ...
26744 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
26745 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
26746 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
26747 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
26748 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
26749 ("vec_unordered<mode>"): ... these.
26751 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
26752 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
26753 ("*vec_extendv2df"): New insn definitions.
26755 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26757 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
26758 ("mulditi3_2", "*muldi3_sign"): New patterns.
26759 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
26760 rename the pattern definition.
26762 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26764 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
26766 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
26768 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26770 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
26771 instruction if possible.
26772 * config/s390/vector.md (vec_halfnumelts): New mode
26774 ("*vec_vllezlf<mode>"): New pattern.
26776 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26778 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
26779 ("popcountv4si2", "popcountv2di2"): Rename to ...
26780 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
26781 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
26783 ("popcount<mode>2_vxe"): New pattern.
26785 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26787 * common/config/s390/s390-common.c (processor_flags_table): Add
26789 * config.gcc: Add arch12.
26790 * config/s390/driver-native.c (s390_host_detect_local_cpu):
26791 Default to arch12 for unknown CPU model numbers.
26792 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
26793 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
26794 PROCESSOR_max sanity check.
26795 * config/s390/s390-opts.h (enum processor_type): Add
26797 * config/s390/s390.c (processor_table): Add arch12.
26798 (s390_expand_builtin): Add check for B_VXE flag.
26799 (s390_issue_rate): Add PROCESSOR_ARCH12.
26800 (s390_get_sched_attrmask): Likewise.
26801 (s390_get_unit_mask): Likewise.
26802 (s390_sched_score): Enable z13 scheduling for arch12.
26803 (s390_sched_reorder): Likewise.
26804 (s390_sched_variable_issue): Likewise.
26805 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
26807 (s390_tune_attr): Use z13 scheduling also for arch12.
26808 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
26809 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
26810 (TARGET_VXE_P): New macros.
26811 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
26812 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
26813 * config/s390/s390.opt: Add arch12 as processor_type.
26815 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26817 * config/s390/s390.md
26818 ("fixuns_truncdddi2", "fixuns_trunctddi2")
26819 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
26820 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
26822 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
26823 Rename expanders to ...
26825 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
26826 ("fixuns_truncdddi2_emu"): ... these.
26828 ("fixuns_trunc<mode>si2_emu"): New expander.
26830 ("*fixuns_truncdfdi2_z13"): Rename to ...
26831 ("*fixuns_truncdfdi2_vx"): ... this.
26833 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26835 * config/s390/2964.md: Remove the single element vector compare
26836 instructions which are no longer used.
26837 * config/s390/s390.c (s390_select_ccmode): Remove handling of
26839 (s390_canonicalize_comparison): Remove handling of DFmode
26841 (s390_expand_vec_compare_scalar): Remove function.
26842 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
26843 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
26845 ("*cmp<mode>_ccs"): Add wfcdb instruction.
26847 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26849 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
26851 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
26852 will anyway by matched by mov<mode>_64dfp.
26854 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26856 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
26857 vlef/vstef. Add missing operand to vleif.
26859 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26861 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
26862 pair for all vector types with 64 bit elements.
26863 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
26864 * config/s390/vector.md (V_HW_64): ... here.
26865 (V_128_NOSINGLE): New mode iterator.
26866 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
26867 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
26868 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
26869 ("*vec_load_pairv2di"): Change to ...
26870 ("*vec_load_pair<mode>"): ... this one.
26872 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26874 * config/s390/constraints.md: Add comments.
26875 (jKK): Reject element sizes > 8 bytes.
26876 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
26878 * config/s390/s390.md: Add the s_operand checks formerly in
26879 s390_split_ok_p to various splitters where they are still
26881 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
26882 for 128 bit vectors. Plus two splitters.
26884 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26886 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
26889 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26892 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
26893 error if the boundary argument is not constant.
26895 2017-03-24 Jakub Jelinek <jakub@redhat.com>
26897 PR rtl-optimization/80112
26898 * loop-doloop.c (doloop_condition_get): Don't check condition
26899 if cmp isn't SET with IF_THEN_ELSE src.
26901 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26903 PR tree-optimization/80158
26904 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
26905 replacing a candidate statement, also replace it for the
26906 candidate's alternate interpretation.
26907 (replace_rhs_if_not_dup): Likewise.
26908 (replace_one_candidate): Likewise.
26910 2017-03-24 Richard Biener <rguenther@suse.de>
26912 PR tree-optimization/80167
26913 * graphite-isl-ast-to-gimple.c
26914 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
26916 (translate_isl_ast_to_gimple::get_rename): Likewise.
26918 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
26920 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
26921 handling of certain combinations of target options, including the
26922 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
26923 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
26925 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26928 * config/arm/arm.md (*load_multiple): Add reload_completed to
26929 matching condition.
26931 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26932 Richard Biener <rguenth@suse.de>
26934 PR tree-optimization/79908
26935 PR tree-optimization/80136
26936 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
26937 been cast away, gimplify_and_add suffices.
26939 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
26941 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
26943 2017-03-23 Richard Biener <rguenther@suse.de>
26945 PR tree-optimization/80032
26946 * gimplify.c (gimple_push_cleanup): Forced unconditional
26947 cleanups still have to go to the conditional_cleanups
26950 2017-03-22 Jakub Jelinek <jakub@redhat.com>
26952 PR tree-optimization/80072
26953 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
26955 (next_operand_entry_id): Change type to unsigned int.
26956 (sort_by_operand_rank): Make sure to return the right return value
26957 even if unsigned fields are bigger than INT_MAX.
26958 (struct oecount): Change cnt and id type to unsigned int.
26959 (oecount_hasher::equal): Formatting fix.
26960 (oecount_cmp): Make sure to return the right return value
26961 even if unsigned fields are bigger than INT_MAX.
26962 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
26965 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
26966 TREE_READONLY on result if writing it more than once.
26969 * doc/invoke.texi (-fsanitize=thread): Document that with
26970 -fnon-call-exceptions atomics are not able to throw
26974 * tsan.c: Include tree-eh.h.
26975 (instrument_builtin_call): Call maybe_clean_eh_stmt or
26976 maybe_clean_or_replace_eh_stmt where needed.
26977 (instrument_memory_accesses): Add cfg_changed argument.
26978 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
26979 if it returned true.
26980 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
26982 PR rtl-optimization/63191
26983 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
26984 wrapper function, moved the whole old content into ...
26985 (ix86_delegitimize_address_1): ... this. New inline function.
26986 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
26987 true as last argument instead of ix86_delegitimize_address.
26989 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
26991 * config/aarch64/aarch64.c (generic_branch_cost): Copy
26992 cortexa57_branch_cost.
26994 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
26996 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
26998 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27001 * doc/md.texi (Constraints): Document wA constraint.
27002 * config/rs6000/constraints.md (wA): New.
27003 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
27004 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
27005 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
27006 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
27008 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
27011 * gimplify.c (is_oacc_declared): New function.
27012 (oacc_default_clause): Use it to set default flags for acc declared
27013 variables inside parallel regions.
27014 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
27015 declared variables.
27016 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
27017 declare attribute to any decl as necessary.
27019 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
27022 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
27023 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
27024 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
27025 (arm_arch_lpae): This.
27026 * config/arm/arm.c (arm_arch7ve): Rename into ...
27027 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
27028 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
27031 2017-03-22 Martin Liska <mliska@suse.cz>
27034 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
27035 error message instead of an ICE.
27037 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27039 * doc/extend.texi (6.11 Additional Floating Types): Revise.
27041 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
27043 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
27045 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27048 2017-03-21 Martin Sebor <msebor@redhat.com>
27050 * doc/extend.texi: Use "cannot" instead of "can't."
27051 * doc/hostconfig.texi: Same.
27052 * doc/install.texi: Same.
27053 * doc/invoke.texi: Same.
27054 * doc/loop.texi: Same.
27055 * doc/md.texi: Same.
27056 * doc/objc.texi: Same.
27057 * doc/rtl.texi: Same.
27058 * doc/tm.texi: Same.
27059 * doc/tm.texi.in: Same.
27060 * doc/trouble.texi: Same.
27062 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
27065 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
27066 (collect_checksum_attributes): Set it.
27067 (die_checksum_ordered): Use it.
27069 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27071 PR tree-optimization/79908
27072 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
27073 change: For a VA_ARG whose LHS has been cast away, use
27074 force_gimple_operand to construct the side effects.
27076 2017-03-21 David Malcolm <dmalcolm@redhat.com>
27078 PR translation/80001
27079 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
27080 more amenable to translation.
27081 (oacc_loop_auto_partitions): Likewise.
27083 2017-03-21 Marek Polacek <polacek@redhat.com>
27084 Martin Sebor <msebor@redhat.com>
27086 PR tree-optimization/80109
27087 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
27088 on INTEGRAL_TYPE_P.
27090 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27091 Segher Boessenkool <segher@kernel.crashing.org>
27094 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
27095 check reg_used_between_p between insn and one of succ or succ2
27096 depending on if succ is artificial insn not inserted into insn
27099 2017-03-21 Martin Liska <mliska@suse.cz>
27101 PR gcov-profile/80081
27102 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
27103 * doc/gcc.texi: Include gcov-dump stuff.
27104 * doc/gcov-dump.texi: New file.
27106 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
27108 PR rtl-optimization/79150
27109 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
27110 conditional jump, if the jump is the last insn of the loop.
27112 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27113 Richard Biener <rguenth@suse.de>
27115 PR tree-optimization/79908
27116 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27117 been cast away, use force_gimple_operand to construct the side
27120 2017-03-21 Martin Liska <mliska@suse.cz>
27122 PR libfortran/79956
27123 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
27126 2017-03-21 Brad Spengler <spender@grsecurity.net>
27129 * plugin.c (htab_hash_plugin): New function.
27130 (add_new_plugin): Use it and adjust.
27131 (parse_plugin_arg_opt): Adjust.
27132 (init_one_plugin): Likewise.
27134 2017-03-21 Richard Biener <rguenther@suse.de>
27136 PR tree-optimization/80032
27137 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
27138 if set force the cleanup to happen unconditionally.
27139 (gimplify_target_expr): Push inserted clobbers with force_uncond
27140 to avoid them being removed by control-dependent DCE.
27142 2017-03-21 Richard Biener <rguenther@suse.de>
27144 PR tree-optimization/80122
27145 * tree-inline.c (copy_bb): Do not expans va-arg packs or
27146 va_arg_pack_len when the inlined call stmt requires pack
27148 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
27150 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27153 * tsan.c (instrument_builtin_call): If the memory model argument
27154 is not a constant, assume it is valid.
27157 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
27160 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27162 PR rtl-optimization/79910
27163 * combine.c (can_combine_p): Do not allow combining an I0 or I1
27164 if its dest is used by an insn before I2 (other than the combined
27165 insns themselves, which are properly handled already).
27167 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27170 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27172 * combine.c (record_used_regs): New static function.
27173 (try_combine): Handle situations where there is an additional
27174 instruction between I2 and I3 which needs to have a LOG_LINK
27178 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27180 * combine.c (try_combine): Delete redundant i1 test. Call
27181 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27183 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27186 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
27187 alternatives 13/14.
27189 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27191 PR tree-optimization/80054
27192 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
27193 the optimization if a PHI or any of its arguments is not dominated
27194 by the candidate's basis. Use gphi* rather than gimple* as
27196 (replace_profitable_candidates): Clean up a gimple* variable that
27197 should be a gphi* variable.
27199 2017-03-20 Martin Sebor <msebor@redhat.com>
27202 * doc/extend.texi (attribute constructor): Document present limitation.
27204 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
27207 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
27208 __POWER9_VECTOR__ #ifdef control, change template definition to
27209 use Power9-specific built-in function.
27210 (vec_any_eq): Likewise.
27211 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
27212 to control outcomes from this test.
27213 (vector_ae_<mode>p): For VEC_F modes, likewise.
27215 2017-03-20 Ian Lance Taylor <iant@google.com>
27217 * config/i386/i386.c (ix86_function_regparm): Save an extra
27218 register for -fsplit-stack with DECL_STATIC_CHAIN.
27220 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
27223 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
27224 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
27226 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
27228 * config/riscv/riscv.c (riscv_print_operand): Use "fence
27230 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
27233 2017-03-20 Marek Polacek <polacek@redhat.com>
27236 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
27238 2017-03-20 Richard Biener <rguenther@suse.de>
27240 PR tree-optimization/80113
27241 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
27242 allocate extra SSA name for PHI def.
27243 (add_close_phis_to_outer_loops): Likewise.
27244 (add_close_phis_to_merge_points): Likewise.
27245 (copy_loop_close_phi_args): Likewise.
27246 (copy_cond_phi_nodes): Likewise.
27248 2017-03-20 Martin Liska <mliska@suse.cz>
27250 PR middle-end/79753
27251 * tree-chkp.c (chkp_build_returned_bound): Do not build
27252 returned bounds for a LHS that's not a BOUNDED_P type.
27254 2017-03-20 Martin Liska <mliska@suse.cz>
27258 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
27259 COMPLEX_CST and VECTOR_CST.
27261 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27264 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
27265 target operand. A new splitter adds the clobber statement in case
27266 the target operand is dead anyway.
27268 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
27270 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
27271 to age-old versions of binutils and glibc.
27273 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
27275 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
27277 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27279 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
27281 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27283 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
27284 requirement for binutils 2.13.
27286 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27288 * combine.c (try_combine): Delete redundant i1 test. Call
27289 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27291 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
27293 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
27294 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
27296 <riscv64-*-elf>: Re-arrange section
27297 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
27298 <riscv32-*-linux>: Likewise.
27299 <riscv64-*-elf>: Likewise
27300 <riscv64-*-linux>: Likewise.
27302 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
27305 * aarch64.opt(verbose-cost-dump): Fix typo.
27307 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
27310 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
27311 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
27313 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27315 * reload.c (find_reloads): When reloading a nonoffsettable address,
27316 use RELOAD_OTHER for it and its address reloads.
27318 PR rtl-optimization/79910
27319 * combine.c (record_used_regs): New static function.
27320 (try_combine): Handle situations where there is an additional
27321 instruction between I2 and I3 which needs to have a LOG_LINK
27324 2017-03-17 Jeff Law <law@redhat.com>
27326 PR tree-optimization/71437
27327 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
27328 conditional in the hash table first.
27329 (vrp_dom_walker::before_dom_children): Extract condition from
27330 ASSERT_EXPR. Record condition, its inverion and any implied
27331 conditions as well.
27333 2017-03-17 Marek Polacek <polacek@redhat.com>
27334 Markus Trippelsdorf <markus@trippelsdorf.de>
27336 PR tree-optimization/80079
27337 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
27340 2017-03-17 Richard Biener <rguenther@suse.de>
27342 PR middle-end/80075
27343 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
27344 Properly verify the LHS before the RHS possibly claims to be
27346 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
27349 2017-03-17 Martin Jambor <mjambor@suse.cz>
27351 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
27352 (List of -O2 options): Likewise.
27353 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
27356 2017-03-17 Tom de Vries <tom@codesourcery.com>
27358 * gcov-dump.c (print_usage): Print bug_report_url.
27360 2017-03-17 Richard Biener <rguenther@suse.de>
27362 PR middle-end/80050
27363 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
27364 (parser::peek): Likewise.
27366 2017-03-17 Richard Biener <rguenther@suse.de>
27368 PR tree-optimization/80048
27369 * sese.c (free_sese_info): Properly release rename_map and
27370 copied_bb_map elements.
27372 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
27374 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
27375 Add linked-list forward and backlinks. Insert on
27376 construction, remove on destruction.
27377 (class pass_store_merging): Add m_stores_head field.
27378 (pass_store_merging::terminate_and_process_all_chains):
27379 Iterate over m_stores_head list.
27380 (pass_store_merging::terminate_all_aliasing_chains):
27382 (pass_store_merging::execute): Check for debug stmts first.
27383 Push new chains onto the m_stores_head stack.
27385 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
27388 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
27389 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
27390 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
27392 2017-03-16 Jeff Law <law@redhat.com>
27394 PR tree-optimization/71437
27395 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
27396 member function. Implementation moved into after_dom_children
27397 member function and into the threader's thread_outgoing_edges
27399 (dom_opt_dom_walker::after_dom_children): Simplify by moving
27400 some code into new thread_outgoing_edges.
27401 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
27402 definition. Simplify marker handling (do it here). Assume we always
27403 have the available expression and the const/copies tables.
27404 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
27406 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
27407 * tree-vrp.c (equiv_stack): No longer file scoped.
27408 (vrp_dom_walker): New class.
27409 (vrp_dom_walker::before_dom_children): New member function.
27410 (vrp_dom_walker::after_dom_children): Likewise.
27411 (identify_jump_threads): Setup domwalker. Use it rather than
27412 walking edges in a random order by hand. Simplify setup/finalization.
27413 (finalize_jump_threads): Remove.
27414 (vrp_finalize): Do not call identify_jump_threads here.
27415 (execute_vrp): Do it here instead and call thread_through_all_blocks
27418 PR tree-optimization/71437
27419 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
27421 (simplify_stmt_for_jump_threading): Add basic_block argument. All
27423 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
27424 (dom_opt_dom_walker::thread_across_edge): Remove
27425 handle_dominating_asserts argument. All callers changed.
27426 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
27427 changes. Remove calls to lhs_of_dominating_assert. Other
27428 uses of handle_dominating_asserts turn into unconditional code
27429 (simplify_control_stmt_condition_1): Likewise.
27430 (simplify_control_stmt_condition): Likewise.
27431 (thread_through_normal_block, thread_across_edge): Likewise.
27432 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
27433 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
27434 object if it is not an SSA_NAME.
27435 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
27436 before calling into the VRP specific simplifiers.
27437 (identify_jump_threads): Remove handle_dominating_asserts
27440 2017-03-16 Jakub Jelinek <jakub@redhat.com>
27443 * tree-diagnostic.c (default_tree_printer): No longer static.
27444 * tree-diagnostic.h (default_tree_printer): New prototype.
27446 2017-03-16 Tamar Christina <tamar.christina@arm.com>
27448 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
27449 Change ins into fmov.
27451 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27453 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
27454 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
27455 Use h_con constraint for operand 1.
27456 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
27457 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
27459 2017-03-15 Jeff Law <law@redhat.com>
27461 PR tree-optimization/71437
27462 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
27463 (record_temporary_equivalences): Use it.
27465 PR tree-optimization/71437
27466 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
27467 tree-ssa-scopedtables.
27468 (lookup_avail_expr, build_and_record_new_cond): Likewise.
27469 (record_conditions, record_cond, vuse_eq): Likewise.
27470 (record_edge_info): Adjust to API tweak of record_conditions.
27471 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
27472 (record_temporary_equivalences, optimize_stmt): Likewise.
27473 (eliminate_redundant_computations): Likewise.
27474 (record_equivalences_from_stmt): Likewise.
27475 * tree-ssa-scopedtables.c: Include options.h and params.h.
27476 (vuse_eq): New function, moved from tree-ssa-dom.c
27477 (build_and_record_new_cond): Likewise.
27478 (record_conditions): Likewise. Accept vector of conditions rather
27479 than edge_equivalence structure for first argument.
27480 for the first argument.
27481 (avail_exprs_stack::lookup_avail_expr): New member function, moved
27482 from tree-ssa-dom.c.
27483 (avail_exprs_stack::record_cond): Likewise.
27484 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
27485 from tree-ssa-dom.c.
27486 (avail_exprs_stack): Add new member functions lookup_avail_expr
27488 (record_conditions): Declare.
27490 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
27493 * lra-constraints.c (process_alt_operands): Increase reject for
27494 reloading an input/output operand.
27496 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
27499 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
27500 insns to convert from signed/unsigned char/short to IEEE 128-bit
27502 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
27504 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
27507 * config/i386/i386.c (ix86_vector_duplicate_value): Create
27508 subreg of inner mode for values already in registers.
27510 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
27512 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
27513 iteration reg is used after the loop.
27515 2017-03-14 Martin Sebor <msebor@redhat.com>
27517 PR tree-optimization/79800
27518 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
27519 precision in negative-positive range.
27520 (format_floating): Call non-const overload with adjusted precision.
27522 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
27525 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
27528 2017-03-14 Martin Sebor <msebor@redhat.com>
27530 PR middle-end/80020
27531 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
27532 * builtins.def (aligned_alloc): Use it.
27535 * Makefile.in (GTFILES): Add calls.c.
27536 * calls.c: Include "gt-calls.h".
27538 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
27540 PR rtl-optimization/79728
27541 * regs.h (struct target_regs): New field
27542 x_contains_allocatable_regs_of_mode.
27543 (contains_allocatable_regs_of_mode): New macro.
27544 * reginfo.c (init_reg_sets_1): Initialize it, and change
27545 contains_reg_of_mode so it includes global regs as well.
27546 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
27547 rather than contains_regs_of_mode.
27549 2017-03-14 Martin Liska <mliska@suse.cz>
27551 * doc/invoke.texi: Document options that can't be combined with
27552 -fcheck-pointer-bounds.
27554 2017-03-14 Martin Liska <mliska@suse.cz>
27556 PR middle-end/79831
27557 * doc/invoke.texi (-Wchkp): Document the option.
27559 2017-03-14 Martin Liska <mliska@suse.cz>
27561 * Makefile.in: Install gcov-dump.
27563 2017-03-14 Martin Liska <mliska@suse.cz>
27565 * multiple_target.c (expand_target_clones): Bail out for
27566 an invalid attribute.
27568 2017-03-14 Richard Biener <rguenther@suse.de>
27570 * alias.c (struct alias_set_entry): Pack properly.
27571 * cfgloop.h (struct loop): Likewise.
27572 * cse.c (struct set): Likewise.
27573 * ipa-utils.c (struct searchc_env): Likewise.
27574 * loop-invariant.c (struct invariant): Likewise.
27575 * lra-remat.c (struct cand): Likewise.
27576 * recog.c (struct change_t): Likewise.
27577 * rtl.h (struct address_info): Likewise.
27578 * symbol-summary.h (function_summary): Likewise.
27579 * tree-loop-distribution.c (struct partition): Likewise.
27580 * tree-object-size.c (struct object_size_info): Likewise.
27581 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
27582 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
27583 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
27584 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
27585 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
27586 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
27587 (struct _stmt_vec_info): Likewise.
27589 2017-03-14 Martin Liska <mliska@suse.cz>
27592 * multiple_target.c (create_dispatcher_calls): Check that
27593 a target can create a function dispatcher.
27595 2017-03-14 Martin Liska <mliska@suse.cz>
27598 * multiple_target.c (expand_target_clones): Drop local.local
27599 flag for default implementation.
27601 2017-03-14 Richard Biener <rguenther@suse.de>
27603 PR tree-optimization/80030
27604 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
27606 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
27608 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
27609 gcc_fallthrough() instead of __attribute__((fallthrough));
27611 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
27613 * doc/gcc.texi: Remove "up" link to (DIR).
27614 * doc/gccint.texi: Ditto.
27616 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
27618 * doc/install.texi (Specific) <avr>: Remove reference to
27621 2017-03-13 Jeff Law <law@redhat.com>
27623 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
27624 attribute rather than comments.
27626 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
27627 match_scratch operand is highest.
27629 2017-03-13 Martin Liska <mliska@suse.cz>
27631 PR middle-end/78339
27632 * ipa-pure-const.c (warn_function_noreturn): If the declarations
27633 is a CHKP clone, use original declaration.
27635 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27637 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
27638 (arc_conditional_register_usage): Use a different allocation order
27639 when optimizing for size.
27640 * common/config/arc/arc-common.c (arc_option_optimization_table):
27641 Section anchors default on when optimizing for size.
27643 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27645 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
27647 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27649 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
27650 * config/arc/arc.md (cpu_facility): Add cd variant.
27651 (*movqi_insn): Add code density variant.
27652 (*movhi_insn): Likewise.
27653 (*movqi_insn): Likewise.
27654 (*addsi3_mixed): Likewise.
27655 (subsi3_insn): Likewise.
27657 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27659 * config/arc/arc.md (movsi_cond_exec): Update constraint.
27661 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27663 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
27664 expressions with MINUS and UNARY ops.
27666 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27669 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
27671 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
27672 between vec_select and vector argument.
27673 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
27674 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
27675 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
27676 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
27677 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
27678 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
27680 2017-03-13 Richard Biener <rguenther@suse.de>
27683 * params.def (vect-max-peeling-for-alignment): Fix typo.
27685 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
27687 * doc/install.texi (Specific) <mips-*-*>: Remove description of
27688 issue that only occurred with binutils below 2.18.
27690 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
27692 * doc/install.texi (Specific) <cris-axis-elf>: No longer
27693 refer to binutils 2.11/2.12 minimum.
27695 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
27697 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
27698 ftp.kernel.org and simplify binutils requirement.
27700 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
27702 * doc/invoke.texi (Warning Options): Fix spelling of link-time
27704 (Optimize Options): Ditto. Also remove redundancy.
27706 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27708 PR translation/79848
27709 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
27711 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
27712 to G_ to avoid double translation.
27714 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27716 PR translation/79923
27717 * auto-profile.c (get_combined_location): Convert leading
27718 character of diagnostics to lower case and remove trailing period.
27719 (read_profile): Likewise for various diagnostics.
27720 * config/arm/arm.c (arm_option_override): Remove trailing period
27721 from various diagnostics.
27722 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
27723 (msp430_expand_delay_cycles): Likewise.
27725 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27728 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
27729 full command-line argument, rather than just "str".
27730 (aarch64_validate_march): Likewise.
27731 (aarch64_validate_mtune): Likewise.
27733 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
27735 PR rtl-optimization/78911
27736 * lra-assigns.c (must_not_spill_p): New function.
27737 (spill_for): Use it.
27739 2017-03-10 Jakub Jelinek <jakub@redhat.com>
27741 PR tree-optimization/79981
27742 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
27743 ATOMIC_COMPARE_EXCHANGE ifn result.
27744 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
27745 IFN_ATOMIC_COMPARE_EXCHANGE.
27747 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27750 * opts.c (parse_sanitizer_options): Add missing question mark to
27751 "did you mean" message.
27753 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27755 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
27757 (VMULEUH_UNS): Likewise.
27758 (VMULOUB_UNS): Likewise.
27759 (VMULOUH_UNS): Likewise.
27760 * config/rs6000/rs6000.c (builtin_function_type): Remove
27761 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
27763 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27766 * read-rtl-function.c (function_reader::read_rtx_operand): Update
27767 x with result of extra_parsing_for_operand_code_0.
27768 (function_reader::extra_parsing_for_operand_code_0): Convert
27769 return type from void to rtx, returning x. When reading
27770 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
27771 larger size containing struct block_symbol.
27773 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
27775 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
27776 -mfloat128-hardware without -m64.
27778 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
27781 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
27782 entries to the case statement that marks unsigned arguments to
27783 overloaded functions.
27785 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
27787 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
27788 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
27790 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
27793 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
27794 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
27796 2017-03-10 Martin Liska <mliska@suse.cz>
27800 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
27801 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
27804 2017-03-10 Olivier Hainque <hainque@adacore.com>
27806 * tree-switch-conversion (array_value_type): Start by resetting
27807 candidate type to it's main variant.
27809 2017-03-10 Jakub Jelinek <jakub@redhat.com>
27811 PR rtl-optimization/79909
27812 * combine.c (try_combine): Use simplify_replace_rtx on individual
27813 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
27814 of the whole CALL_INSN_FUNCTION_USAGE.
27816 PR tree-optimization/79972
27817 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
27818 get_range_info on SSA_NAMEs. Formatting fixes.
27820 2017-03-10 Richard Biener <rguenther@suse.de>
27821 Jakub Jelinek <jakub@redhat.com>
27823 PR tree-optimization/77975
27824 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
27825 edge to be constant.
27826 (get_val_for): For constant x return it. Formatting fix.
27827 (loop_niter_by_eval): Avoid pointless looping if the next iteration
27828 would use the same bases as the current one.
27830 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27832 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
27833 instead of vec_select for V1TImode.
27834 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
27836 (VSX_LE_128): Add V1TI to this mode iterator.
27837 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
27838 (*vsx_le_perm_store_<mode>): Likewise.
27839 (pre-reload splitter for VSX stores): Likewise.
27840 (post-reload splitter for VSX stores): Likewise.
27841 (*vsx_xxpermdi2_le_<mode>): Likewise.
27842 (*vsx_lxvd2x2_le_<mode>): Likewise.
27843 (*vsx_stxvd2x2_le_<mode>): Likewise.
27845 2017-03-09 Michael Eager <eager@eagercon.com>
27847 Correct failures with --enable-checking=yes,rtl.
27849 * config/microblaze/microblaze.c (microblaze_expand_shift):
27850 Replace GET_CODE test with CONST_INT_P and INTVAL test with
27851 test for const0_rtx.
27852 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
27853 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
27855 2017-03-09 Richard Biener <rguenther@suse.de>
27857 PR tree-optimization/79977
27858 * graphite-scop-detection.c (scop_detection::merge_sese):
27859 Handle the case of extra exits to blocks dominating the entry.
27861 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
27863 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
27866 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
27868 PR rtl-optimization/79949
27869 * lra-constraints.c (process_alt_operands): Check memory when
27870 trying to predict a cycle. Print about the overall increase.
27872 2017-03-09 Richard Biener <rguenther@suse.de>
27874 PR middle-end/79971
27875 * gimple-expr.c (useless_type_conversion_p): Preserve
27876 TYPE_SATURATING for fixed-point types.
27878 2017-03-09 Richard Biener <rguenther@suse.de>
27881 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
27882 alignment of BLKmode params.
27884 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27887 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
27888 (VALL_NO_V2Q): Likewise.
27890 * config/aarch64/aarch64-simd.md
27891 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
27893 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
27894 VALL_NO_V2Q mode iterator.
27895 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
27897 2017-03-09 Martin Liska <mliska@suse.cz>
27899 PR tree-optimization/79631
27900 * tree-chkp-opt.c (chkp_is_constant_addr): Call
27901 tree_int_cst_sign_bit just for INTEGER constants.
27903 2017-03-09 Martin Liska <mliska@suse.cz>
27907 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
27910 2017-03-09 Marek Polacek <polacek@redhat.com>
27913 * tree.c (inchash::add_expr): Handle TREE_VEC.
27915 2017-03-09 Martin Liska <mliska@suse.cz>
27918 (chkp_narrow_size_and_offset): New function.
27919 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
27920 (void chkp_parse_bit_field_ref): New function.
27921 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
27922 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
27924 2017-03-09 Martin Liska <mliska@suse.cz>
27927 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
27928 (chkp_find_bounds_1): Remove gcc_unreachable.
27930 2017-03-09 Jakub Jelinek <jakub@redhat.com>
27933 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
27934 BUILT_IN_SYNC*, determine the access type from the size suffix and
27935 always build a MEM_REF with that type. Handle forgotten
27936 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
27939 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
27940 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
27941 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
27942 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
27943 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
27944 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
27945 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
27946 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
27947 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
27948 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
27949 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
27950 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
27951 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
27952 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
27953 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
27954 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
27955 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
27956 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
27957 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
27958 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
27959 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
27960 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
27961 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
27962 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
27963 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
27964 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
27965 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
27966 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
27967 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
27968 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
27969 definitions outside of __OPTIMIZE__ guarded section.
27972 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
27973 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
27974 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
27975 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
27978 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27980 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
27981 ("vfenez<mode>"): Add missing constraints.
27983 2017-03-08 Martin Sebor <msebor@redhat.com>
27986 * config/nds32/nds32.c (nds32_option_override):
27987 Fix misspelled diagnostic.
27989 2017-03-08 Jakub Jelinek <jakub@redhat.com>
27992 * gimplify.c (gimplify_omp_for): Replace index var in outer
27993 taskloop statement with an artificial variable and add
27994 OMP_CLAUSE_PRIVATE clause for it.
27996 2017-03-08 Richard Biener <rguenther@suse.de>
27998 PR tree-optimization/79955
27999 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
28000 for accesses that are completely outside of the variable.
28002 2017-03-08 Andrew Haley <aph@redhat.com>
28004 PR tree-optimization/79943
28005 * tree-ssa-loop-split.c (compute_new_first_bound): When
28006 calculating the new upper bound, (END-BEG) should be added, not
28009 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28011 * config/avr/avr.md (setmemhi): Make sure match_dup
28012 operand number comes before match_scratch.
28014 2017-03-08 Richard Biener <rguenther@suse.de>
28016 PR tree-optimization/79920
28017 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
28018 with ncopies == 1 to ...
28019 (vect_transform_slp_perm_load): ... here. Properly compute
28020 all element loads by iterating VF times over the group. Do
28021 not handle ncopies (computed in a broken way) in
28022 vect_create_mask_and_perm.
28024 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28027 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
28028 is a uniform vector, use uniform_vector_p return value instead of
28029 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
28031 2017-03-07 Marek Polacek <polacek@redhat.com>
28033 PR middle-end/79809
28034 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
28035 (alloca_call_type): Likewise.
28037 2017-03-07 Martin Liska <mliska@suse.cz>
28039 * gcov.c (process_args): Put comment to correct location.
28041 2017-03-07 Martin Liska <mliska@suse.cz>
28043 PR middle-end/68270
28044 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
28045 Use array_at_struct_end_p instead of DECL_CHAIN (field).
28046 (chkp_narrow_bounds_for_field): Likewise.
28047 (chkp_parse_array_and_component_ref): Pass one more argument to
28050 2017-03-07 Richard Biener <rguenther@suse.de>
28052 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
28055 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
28057 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
28058 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
28060 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28063 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
28064 to end of description.
28065 (PARAM_MAX_STORES_TO_MERGE): Likewise.
28067 2017-03-07 Jakub Jelinek <jakub@redhat.com>
28069 PR rtl-optimization/79901
28070 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
28072 (*avx512f_<code><mode>3<mask_name>): ... this.
28073 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
28074 iterator instead of VI8_AVX2_AVX512BW.
28076 PR rtl-optimization/79901
28077 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
28078 min/max expander, expand it using expand_vec_cond_expr.
28081 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
28084 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28087 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
28088 to void * for PCH reasons.
28089 * dwarf2out.c (output_loc_operands, output_die): Cast
28090 v.val_vec.array to unsigned char *.
28092 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
28095 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
28098 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
28100 PR rtl-optimization/79571
28101 * lra-constraints.c (process_alt_operands): Calculate static
28102 reject and subtract it from overall when only addresses will be
28105 2017-03-06 Julia Koval <julia.koval@intel.com>
28108 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
28109 incoming stack boundary to 128 for 64-bit targets.
28111 2017-03-06 Richard Biener <rguenther@suse.de>
28113 PR tree-optimization/79894
28114 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
28115 to NULL after folding it.
28117 2017-03-06 Richard Biener <rguenther@suse.de>
28119 PR tree-optimization/79824
28120 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
28121 check disabling peeling for gaps.
28123 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
28125 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28126 attributes): Document gettimeofday.
28128 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28130 * config/s390/s390.c (s390_option_override_internal): Set
28131 PARAM_MIN_VECT_LOOP_BOUND
28133 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28135 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
28136 * config/s390/s390.md: Likewise.
28138 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28141 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
28142 (<avx2_avx512>_perm<mode>): Rename to ...
28143 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
28145 (<avx512>_perm<mode>_mask): Rename to ...
28146 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
28148 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
28149 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
28150 instead of VI8F_256_512.
28151 (avx512f_perm<mode>): New define_expand.
28152 (avx512f_perm<mode>_mask): Likewise.
28153 (avx512f_perm<mode>_1<mask_name>): New define_insn.
28154 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
28156 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28158 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
28159 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
28161 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
28163 2017-03-06 Martin Liska <mliska@suse.cz>
28166 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
28167 when having a SSA NAME w/o VAR_DECL assigned to it.
28169 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28171 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
28172 msa_dpsub_<su>_d): Fix MODE for vec_select.
28174 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28176 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
28178 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
28180 2017-03-06 Richard Biener <rguenther@suse.de>
28182 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
28183 * plugin.c (register_plugin_info): Likewise.
28184 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
28186 2017-03-05 Jakub Jelinek <jakub@redhat.com>
28188 * config/i386/sse.md (sse_storehps, sse_storelps,
28189 avx_<castmode><avxsizesuffix>_<castmode>,
28190 avx512f_<castmode><avxsizesuffix>_<castmode>,
28191 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
28192 in condition that at least one operand is not a MEM.
28194 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28196 PR middle-end/79805
28197 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
28198 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
28200 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
28201 gimple_call_nothrow_p flag based on whether original builtin can throw.
28202 If it can, emit following stmts on the fallthrough edge.
28203 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
28204 don't create new bb if inserting just debug stmts on the edge, try to
28205 insert them on the fallthru bb or just reset debug stmts.
28207 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
28210 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
28211 restore recog_data (including the operand rtxes inside it) around
28212 the call to get_insn_template.
28214 2017-03-03 Martin Sebor <msebor@redhat.com>
28216 PR tree-optimization/79699
28217 * context.c (context::~context): Free MPFR caches to avoid
28218 a memory leak on program exit.
28220 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28222 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
28223 Use wide_int::ulow () instead of .elt (0).
28225 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28227 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
28228 (*pushxf): Limit oF constraint to 32bit targets and add oC
28229 constraint for 64bit targets.
28230 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
28231 (*pushdf): Change rmF constraint to rmC.
28233 2017-03-03 Martin Liska <mliska@suse.cz>
28235 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
28236 Remove unused variable.
28238 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28241 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
28242 is a memory operand, increase num_memory.
28243 (ix86_expand_args_builtin): Likewise.
28245 2017-03-03 Jan Hubicka <jh@suse.cz>
28248 * ipa-devirt.c (maybe_record_node): Properly handle
28249 __cxa_pure_virtual visibility.
28251 2017-03-03 Martin Liska <mliska@suse.cz>
28253 PR tree-optimization/79803
28254 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
28256 (pass_loop_prefetch::execute): Disabled optimization if an
28257 assumption about L1 cache size is not met.
28259 2017-03-03 Martin Liska <mliska@suse.cz>
28261 PR rtl-optimization/79574
28262 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
28263 (hash_scan_set): Likewise.
28264 (dump_hash_table): Likewise.
28265 (hoist_code): Likewise.
28267 2017-03-03 Richard Biener <rguenther@suse.de>
28269 * fixed-value.c (fixed_from_string): Restore use of elt (1)
28270 in place of uhigh ().
28271 (fixed_convert_from_real): Likewise.
28273 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28276 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
28278 2017-03-03 Richard Biener <rguenther@suse.de>
28280 PR middle-end/79818
28281 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
28282 TYPE_OVERFLOW_UNDEFINED check.
28284 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28286 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
28288 (vector_ae_<mode>_p): Likewise.
28289 (vector_nez_<mode>_p): Likewise.
28290 (vector_ne_v2di_p): Likewise.
28291 (vector_ae_v2di_p): Likewise.
28292 (vector_ne_<mode>_p): Likewise.
28293 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
28295 (vsx_tsqrt<mode>2_fe): Likewise.
28297 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
28300 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
28302 2017-03-02 Jakub Jelinek <jakub@redhat.com>
28304 PR rtl-optimization/79780
28305 * cprop.c (one_cprop_pass): When second and further conditional trap
28306 in a single basic block is turned into an unconditional trap, turn it
28307 into a deleted note to avoid RTL verification failures.
28309 2017-03-02 Richard Biener <rguenther@suse.de>
28311 * fold-const.c (const_binop): Use ulow () instead of elt (0).
28313 2017-03-02 Richard Biener <rguenther@suse.de>
28315 PR tree-optimization/79345
28317 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
28318 param and abort the walk, returning -1 if it is hit.
28319 (walk_aliased_vdefs): Take a limit param and pass it on.
28320 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
28321 defaulting to 0 and return a signed int.
28322 * tree-ssa-uninit.c (struct check_defs_data): New struct.
28323 (check_defs): New helper.
28324 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
28325 about uninitialized memory.
28326 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
28327 bogus uninitialized warning.
28328 (fixed_convert_from_real): Likewise.
28330 2017-03-02 Bin Cheng <bin.cheng@arm.com>
28332 PR tree-optimization/66768
28333 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
28334 iv_use if base object can't be determined.
28336 2017-03-02 Jakub Jelinek <jakub@redhat.com>
28338 PR tree-optimization/79345
28339 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
28340 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
28341 (get_pattern_stats): Initialize it.
28342 * genemit.c (gen_expand): Verify match_scratch numbers come after
28343 match_operand/match_dup numbers.
28344 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
28345 match_scratch numbers.
28346 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
28348 * config/s390/s390.md (trunctdsd2): Likewise.
28350 2017-03-02 Richard Biener <rguenther@suse.de>
28352 * wide-int.h (wide_int_storage::operator=): Implement in terms
28355 2017-03-02 Richard Biener <rguenther@suse.de>
28357 PR tree-optimization/79777
28358 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
28359 the to insert expression to sth existing.
28361 2017-03-01 Martin Sebor <msebor@redhat.com>
28363 PR middle-end/79692
28364 * gimple-ssa-sprintf.c
28365 (directive::known_width_and_precision): New function.
28366 (format_integer): Use it.
28367 (get_mpfr_format_length): Consider the full range of precision
28368 when computing %g output with the # flag. Set the likely byte
28369 count to 3 rather than 1 when precision is indeterminate.
28370 (format_floating): Correct the lower bound of precision.
28372 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28374 * doc/invoke.texi: Document default code model for 64-bit Linux.
28376 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28379 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
28380 udiv rather than div since input pattern is unsigned.
28382 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
28384 * config/i386/i386.c (print_reg): Warn for values of
28385 unsupported size in integer register.
28387 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
28390 * config/rs6000/predicates.md (current_file_function_operand): Do
28391 not allow self calls to be local if the function is replaceable.
28393 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
28396 * config/rs6000/altivec.h (vec_ctz and others): Change the
28397 preprocessor macro that controls conditional compilation from
28398 _ARCH_PWR9 to __POWER9_VECTOR__.
28399 (vec_all_ne): Change parameterization of __altivec_scalar_pred
28400 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
28401 control (instead of _ARCH_PWR9 control) so that template
28402 definition uses power9-specific function.
28403 (vec_any_eq): Likewise.
28404 (vec_all_ne): Change macro definition to use a power9-specific
28405 expansion under #ifdef __POWER9_VECTOR__ control (instead of
28406 _ARCH_PWR9 control).
28407 (vec_any_eq) Likewise.
28408 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
28409 expansion for CMPNEF to remove support for xvcmpnesp instruction.
28410 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
28411 support for xvcmpnedp instruction.
28412 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
28413 macro expansion so that Power9 implementation of vec_all_ne does
28414 not use the AltiVec predicate framework.
28415 (VCMPNEH_P): Likewise.
28416 (VCMPNEW_P): Likewise.
28417 (VCMPNED_P): Likewise.
28418 (VCMPNEFP_P): Likewise.
28419 (VCMPNEDP_P): Likewise.
28420 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
28421 implementation of vec_any_eq to not use AltiVec predicate
28423 (VCMPAEH_P): Likewise.
28424 (VCMPAEW_P): Likewise.
28425 (VCMPAED_P): Likewise.
28426 (VCMPAEFP_P): Likewise.
28427 (VCMPAEDP_P): Likewise.
28428 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
28429 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
28430 not use the AltiVec predicate framework.
28431 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
28432 of vec_any_eq to not use AltiVec predicate framework.
28433 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
28434 support for predefined __POWER9_VECTOR__ macro to indicate that
28435 Power9 instruction selection is enabled.
28436 (altivec_overloaded_builtins): Remove extraneous
28437 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
28438 function argument types RS6000_BTI_bool_V16QI and
28439 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
28440 entry for overloaded function argument types RS6000_BTI_bool_V4SI
28441 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
28442 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
28443 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
28444 Power9 for implementations of vec_cmpne. Change the signature for
28445 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
28446 (representing vec_all_ne) to remove the previously described first
28447 argument of type RS6000_BTI_INTSI, as this was an artifact of
28448 reliance on the AltiVec predicate framework, which is no longer
28449 used in the implementation of these functions. Add
28450 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
28451 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
28452 since, unlike the AltiVec predicate framework implementation, we
28453 do not share function descriptors between vec_alle and vec_anyeq.
28454 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
28455 set of modes that receive special treatment even when
28456 TARGET_P9_VECTOR is true. The special treatment emits code that
28457 does not depend on Power9 instructions.
28458 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
28459 define_expand to not rely on AltiVec predicate framework.
28460 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
28462 (vector_ne_v2di_p): Change this define_expand to not rely on
28463 AltiVec predicate framework.
28464 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
28466 (vector_ne_<mode>_p): Change this define_expand to not rely on
28467 AltiVec predicate framework.
28468 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
28470 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
28471 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
28472 define_insn pattern.
28473 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
28474 define_insn pattern because the xvcmpne<VSs>. instruction is not
28476 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
28477 instruction is not supported.
28479 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28481 * config/nvptx/nvptx.c: Include intl.h.
28483 2017-03-01 Martin Jambor <mjambor@suse.cz>
28486 * ipa-prop.h (ipa_bits): Removed field known.
28487 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
28488 to pointers. Adjusted their comments to warn about their sharing.
28489 (ipcp_transformation_summary): Change bits to a vector of pointers.
28490 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
28491 (ipa_get_ipa_bits_for_value): Declare.
28492 * tree-vrp.h (value_range): Mark as GTY((for_user)).
28493 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
28494 (ipa_bits_hash_table): Likewise.
28495 (ipa_vr_ggc_hash_traits): Likewise.
28496 (ipa_vr_hash_table): Likewise.
28497 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
28498 being pointers and vr_known being removed.
28499 (ipa_set_jf_unknown): Likewise.
28500 (ipa_get_ipa_bits_for_value): New function.
28501 (ipa_set_jfunc_bits): Likewise.
28502 (ipa_get_value_range): New overloaded functions.
28503 (ipa_set_jfunc_vr): Likewise.
28504 (ipa_compute_jump_functions_for_edge): Use the above functions to
28505 construct bits and vr parts of jump functions.
28506 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
28507 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
28509 (ipcp_grow_transformations_if_necessary): Also allocate
28510 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
28512 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
28513 them. Fix too long lines.
28514 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
28515 vr_known being removed.
28516 (ipa_read_jump_function): Use new setter functions to construct bits
28517 and vr parts of jump functions or set them to NULL.
28518 (write_ipcp_transformation_info): Adjust for bits being pointers.
28519 (read_ipcp_transformation_info): Likewise.
28520 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
28522 Include gt-ipa-prop.h.
28523 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
28525 (ipcp_store_bits_results): Likewise.
28526 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
28527 Do not write to existing jump functions but use a temporary instead.
28529 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28532 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
28533 attempt to use its first operand as BIT_FIELD_REF base.
28535 2017-03-01 Richard Biener <rguenther@suse.de>
28537 PR middle-end/79721
28538 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
28539 interpolating formula in wrapping arithmetic.
28540 (chrec_apply): Convert chrec_evaluate return value to wanted type.
28542 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28544 PR tree-optimization/79734
28545 * tree-vect-generic.c (expand_vector_condition): Optimize
28546 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
28547 Handle VEC_COND_EXPR where comparison has different inner width from
28548 type's inner width.
28550 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
28552 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
28553 markup, and similar issues. Remove @opindex entries for things
28554 that aren't options. Add missing -mmpy-option entries.
28556 2017-02-28 Jakub Jelinek <jakub@redhat.com>
28558 PR tree-optimization/79737
28559 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
28560 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
28561 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
28562 instead of byte_size. Formatting fix.
28563 (shift_bytes_in_array_right): Formatting fix.
28565 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
28568 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
28569 condition on optimize for the leaf function test.
28571 2017-02-28 Martin Liska <mliska@suse.cz>
28574 * read-rtl-function.c (function_reader::handle_unknown_directive):
28575 Bail out when one uses -flto.
28577 2017-02-28 Martin Liska <mliska@suse.cz>
28579 * common.opt: Replace space with tabular for options of <number>
28581 * config/i386/i386.opt: Show <number> value for
28582 -mlarge-data-threshold.
28583 * opts.c (print_filtered_help): Do not display number in hexadecimal
28586 2017-02-28 Martin Liska <mliska@suse.cz>
28588 * common.opt: Fix --help=option -Q for options which are of
28591 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
28593 * config/i386/i386.c (print_reg): Error out for values
28594 of 8-bit size in invalid integer register.
28596 2017-02-28 Martin Sebor <msebor@redhat.com>
28598 PR tree-optimization/79691
28599 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
28601 2017-02-28 Jakub Jelinek <jakub@redhat.com>
28604 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
28605 gcc_unreachable with output_operand_lossage.
28607 2017-02-28 Richard Biener <rguenther@suse.de>
28609 PR tree-optimization/79740
28610 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
28612 (visit_nary_op): Insert the nary into the hashtable if we
28613 pattern-matched sth.
28614 * tree-ssa-pre.c (eliminate_insert): Robustify.
28616 2017-02-28 Richard Biener <rguenther@suse.de>
28618 PR middle-end/79731
28619 * fold-const.c (decode_field_reference): Reject out-of-bound
28622 2017-02-28 Jakub Jelinek <jakub@redhat.com>
28624 * config/i386/i386.c: Include intl.h.
28625 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
28626 instead of just cond ? "..." : "...".
28627 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
28628 * coverage.c (read_counts_file): Likewise.
28629 * omp-offload.c: Include intl.h.
28630 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
28631 of just cond ? "..." : "...".
28632 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
28633 of just cond ? "..." : "...".
28635 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
28638 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
28640 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
28641 'tune for' CPU name.
28642 * config/arm/arm-cpu-data.h: Regenerated.
28644 2017-02-28 Richard Biener <rguenther@suse.de>
28646 PR tree-optimization/79732
28647 * tree-inline.c (expand_call_inline): Do not shadow var.
28649 2017-02-28 Richard Biener <rguenther@suse.de>
28651 PR tree-optimization/79723
28652 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
28653 address-space properly.
28655 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
28657 * doc/optinfo.texi (Optimization groups): Fix option used for
28659 * doc/invoke.texi (-fopt-info): Document "omp".
28660 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
28661 (OPTGROUP_ALL): Add OPTGROUP_OMP.
28662 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
28663 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
28664 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
28666 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
28668 * dumpfile.c (optgroup_options): Instead of "openmp", associate
28669 OPTGROUP_OMP with "omp".
28671 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
28674 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
28675 for arithmetic shift of unsigned V2DI.
28677 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
28679 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
28680 arc/linux.h headers.
28681 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
28682 (LINK_SPEC): Likewise.
28683 (ARC_TLS_EXTRA_START_SPEC): Likewise.
28684 (EXTRA_SPECS): Likewise.
28685 (STARTFILE_SPEC): Likewise.
28686 (ENDFILE_SPEC): Likewise.
28687 (LIB_SPEC): Likewise.
28688 (TARGET_SDATA_DEFAULT): Likewise.
28689 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
28690 (MULTILIB_DEFAULTS): Likewise.
28691 (DWARF2_UNWIND_INFO): Likewise.
28692 * config/arc/big.h: New file.
28693 * config/arc/elf.h: Likewise.
28694 * config/arc/linux.h: Likewise.
28695 * config/arc/t-uClibc: Remove.
28697 2017-02-27 Bin Cheng <bin.cheng@arm.com>
28699 PR tree-optimization/77536
28700 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
28701 (tree_transform_and_unroll_loop): Use above function to compute the
28702 estimated niter of unrolled loop and use it when scaling profile.
28703 Also use count info rather than frequency if it's non-zero.
28704 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
28705 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
28706 (vect_transform_loop): Call above function.
28708 2017-02-27 Richard Biener <rguenther@suse.de>
28710 PR tree-optimization/45397
28711 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
28712 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
28713 (visit_nary_op): Add pattern matching for CSEing sign-changed
28714 or truncated operations with wider ones.
28716 2017-02-27 Richard Biener <rguenther@suse.de>
28718 PR tree-optimization/79690
28719 * tree-vect-stmts.c (vectorizable_store): Use vector type
28720 built from the DR with address-space.
28722 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
28724 * doc/invoke.texi (Optimize Options): Refine the description
28725 of asan-use-after-return.
28727 2017-02-25 Alan Modra <amodra@gmail.com>
28729 PR rtl-optimization/79584
28730 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
28731 base, not ad->base_term, the reg within base. Remove assertion
28732 that ad->base == ad->base_term. Replace gen_int_mode using
28733 bogus mode with const0_rtx.
28735 2017-02-25 Jakub Jelinek <jakub@redhat.com>
28737 PR middle-end/79396
28738 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
28739 FMA_EXPR like tcc_binary or tcc_unary.
28741 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
28744 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
28746 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
28747 (output_loc_operands): Handle DW_OP_call_ref and
28748 DW_OP_GNU_variable_value.
28749 (struct variable_value_struct): New type.
28750 (struct variable_value_hasher): Likewise.
28751 (variable_value_hash): New variable.
28752 (string_types): Remove.
28753 (copy_loc_descr): New function.
28754 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
28755 (prepend_loc_descr_to_each): New function.
28756 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
28757 instead of add_loc_descr_to_each if the first argument is single
28758 location list and the second has multiple.
28759 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
28760 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
28761 when looking for variable value which doesn't have other location info.
28762 (loc_list_from_tree): Formatting fix.
28763 (gen_array_type_die): Simplify DW_AT_string_length handling.
28764 (adjust_string_types): Remove.
28765 (gen_subprogram_die): Don't call adjust_string_types nor test/set
28766 string_types. Call resolve_variable_values.
28767 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
28768 (resolve_addr_in_expr): Likewise. Add A argument.
28769 (copy_deref_exprloc): Remove deref argument. Adjust for the
28770 original expression being DW_OP_GNU_variable_value with optionally
28771 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
28772 optionally after it.
28773 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
28774 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
28775 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
28776 (variable_value_hasher::hash, variable_value_hasher::equal): New
28778 (resolve_variable_value_in_expr, resolve_variable_value,
28779 resolve_variable_values, note_variable_value_in_expr,
28780 note_variable_value): New functions.
28781 (dwarf2out_early_finish): Call note_variable_value on all toplevel
28784 2017-02-24 Jakub Jelinek <jakub@redhat.com>
28787 * opts.h (handle_generated_option): Add GENERATED_P argument.
28788 * opts-common.c (handle_option): Adjust function comment.
28789 (handle_generated_option): Add GENERATED_P argument, pass it to
28791 (control_warning_option): Pass false to handle_generated_option
28793 * opts.c (maybe_default_option): Pass true to handle_generated_option
28795 * optc-gen.awk: Likewise.
28797 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28799 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
28800 a REG, look at the REG it is a SUBREG of.
28801 (splitter for cmpeqsi_t): Ditto.
28803 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28805 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
28806 the special USEs with the pattern of the insn, not the insn itself.
28808 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
28811 * doc/invoke.texi: Document -mload-store-pairs.
28813 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28814 Sandra Loosemore <sandra@codesourcery.com>
28816 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
28817 argument isn't a CONST_INT.
28818 (nios2_alternate_compare_const): Assert op is a CONST_INT.
28819 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
28820 (nios2_validate_compare): Bypass alternate compare logic if *op2
28821 is not a CONST_INT.
28822 (ldstwm_operation_p): Return false if first_base is not a REG or
28823 if first_offset is not a CONST_INT.
28825 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28827 * config/cris/cris.md: Use correct operand in a define_peephole2.
28829 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28831 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
28833 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28835 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
28836 this_insn if it is an INSN or JUMP_INSN.
28837 (force_offsettable): Look at base, not at addr.
28838 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
28839 on things that aren't necessarily CONST_INTs.
28841 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
28843 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
28844 -mfpmath=sse is the default also for x86-32 targets with SSE2
28845 instruction set when @option{-ffast-math} is enabled
28847 2017-02-24 Jeff Law <law@redhat.com>
28849 PR rtl-optimizatoin/79286
28850 * ira.c (update_equiv_regs): Drop may_trap_p exception to
28853 2017-02-24 Richard Biener <rguenther@suse.de>
28855 PR tree-optimization/79389
28856 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
28859 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
28861 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
28862 function comment to reflect reality.
28863 (loop_exits_before_overflow): Fix typo in function description.
28865 2017-02-24 Richard Biener <rguenther@suse.de>
28867 PR tree-optimization/79389
28868 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
28869 properly that a threading opportunity exists. Detect conditional
28870 copy/constant propagation opportunities.
28872 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
28874 * config/visium/visium.md (type): Add trap.
28875 (b): New mode attribute.
28876 (*btst): Rename into...
28877 (*btst<mode>): ...this and adjust.
28878 (*cbranchsi4_btst_insn): Rename into...
28879 (*cbranch<mode>4_btst_insn): ...this and adjust.
28880 (trap): New define_insn.
28882 2017-02-23 Jakub Jelinek <jakub@redhat.com>
28884 PR tree-optimization/79389
28885 * ifcvt.c (struct noce_if_info): Add rev_cond field.
28886 (noce_reversed_cond_code): New function.
28887 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
28888 reversed_comparison_code. Formatting fix.
28889 (noce_try_store_flag): Test rev_cond != NULL in addition to
28890 reversed_comparison_code.
28891 (noce_try_store_flag_constants): Likewise.
28892 (noce_try_store_flag_mask): Likewise.
28893 (noce_try_addcc): Use rev_cond if non-NULL instead of
28894 reversed_comparison_code.
28895 (noce_try_cmove_arith): Likewise. Formatting fixes.
28896 (noce_try_minmax, noce_try_abs): Clear rev_cond.
28897 (noce_find_if_block): Initialize rev_cond.
28898 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
28899 instead of false as last argument never attempt to reverse it
28902 2017-02-23 Bin Cheng <bin.cheng@arm.com>
28904 PR tree-optimization/79663
28905 * tree-predcom.c (combine_chains): Process refs in reverse order
28906 only for ZERO length chains, and add explaining comment.
28908 2017-02-23 Jeff Law <law@redhat.com>
28910 PR tree-optimization/79578
28911 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
28912 in call to operand_equal_p.
28914 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
28917 * config/i386/cpuid.h: Fix another undefined behavior.
28919 2017-02-23 Richard Biener <rguenther@suse.de>
28921 PR tree-optimization/79683
28922 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
28923 vector types for data-refs.
28925 2017-02-23 Martin Liska <mliska@suse.cz>
28927 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
28929 2017-02-23 Jakub Jelinek <jakub@redhat.com>
28931 PR middle-end/79665
28932 * internal-fn.c (get_range_pos_neg): Moved to ...
28933 * tree.c (get_range_pos_neg): ... here. No longer static.
28934 * tree.h (get_range_pos_neg): New prototype.
28935 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
28936 are known to be in between 0 and signed maximum inclusive, try to
28937 expand both unsigned and signed divmod and use the cheaper one from
28940 2017-02-22 Jeff Law <law@redhat.com>
28942 PR tree-optimization/79578
28943 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
28944 to compare base operands.
28946 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
28949 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
28950 gpc_reg_operand instead of fpr_reg_operand.
28952 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
28954 * config/mips/mips.c (mips_return_in_memory): Force FP
28955 vector types to be returned in memory for o32 ABI.
28957 2017-02-22 Jakub Jelinek <jakub@redhat.com>
28959 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
28960 instead of DW_TAG_member for static data member declarations and don't
28961 set no_linkage_name for static inline data members.
28962 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
28965 2017-02-22 Martin Liska <mliska@suse.cz>
28967 * doc/invoke.texi: Replace inequality signs with square brackets
28970 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
28973 * lra-constraints.c (simplify_operand_subreg): Handle
28974 WORD_REGISTER_OPERATIONS targets.
28976 2017-02-22 Jakub Jelinek <jakub@redhat.com>
28979 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
28980 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
28981 elimination by swapping fld*.
28983 2017-02-22 Richard Biener <rguenther@suse.de>
28985 PR tree-optimization/79673
28986 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
28987 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
28988 irrelevant address-space qualifiers and avoiding a
28989 ADDR_SPACE_CONVERT_EXPR from fold_convert.
28991 2017-02-22 Richard Biener <rguenther@suse.de>
28993 PR tree-optimization/79666
28994 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
28995 to not symbolically negate if that may introduce undefined
28998 2017-02-22 Martin Liska <mliska@suse.cz>
29001 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
29002 * data-streamer-out.c (streamer_write_gcov_count_stream):
29004 * value-prof.c (stream_out_histogram_value): Make assert more
29005 precise based on type of counter.
29007 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
29010 * config/i386/i386.md (standard_x87sse_constant_load splitter):
29011 Use nonimmediate_operand instead of memory_operand for operand 1.
29012 (float-extend standard_x87sse_constant_load splitter): Ditto.
29014 2017-02-21 Jeff Law <law@redhat.com>
29016 PR tree-optimization/79621
29017 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
29018 blocks with edges to themselves.
29020 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29023 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
29024 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
29025 Use gimple_call_builtin_p.
29028 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
29029 on temporarily removed DEBUG_INSNs.
29031 PR tree-optimization/79649
29032 * tree-loop-distribution.c (classify_partition): Give up on
29033 non-generic address space loads/stores.
29035 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
29037 * doc/loop.texi (Loop manipulation): Remove nonexistent
29038 tree_ssa_loop_version from the documentation.
29039 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
29041 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29044 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
29045 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
29046 * config/rs6000/rs6000.c: Include except.h.
29047 (rs6000_expand_split_stack_prologue): Call
29048 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
29050 2017-02-21 Martin Jambor <mjambor@suse.cz>
29053 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
29054 have been analyzed.
29056 2017-02-21 Martin Jambor <mjambor@suse.cz>
29058 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
29059 for backward compatibility only.
29060 * doc/invoke.texi (Option Summary): Remove all references to
29061 -fipa-cp-alignment.
29063 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
29067 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29069 * lra-constraints.c (curr_insn_transform): Handle
29070 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29072 2017-02-21 Martin Liska <mliska@suse.cz>
29074 * config/i386/i386.opt: Replace -masm-dialect with -masm.
29076 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
29078 PR translation/79638
29079 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
29081 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
29084 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
29085 (arm_function_ok_for_sibcall): Return false for an indirect call by
29086 descriptor if all the argument registers are used.
29087 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
29088 alignment of the function.
29090 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29092 PR tree-optimization/61441
29093 * simplify-rtx.c (simplify_const_unary_operation): For
29094 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
29095 the sNaN unmodified.
29097 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
29099 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29100 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29101 instead of SYSTEM_HEADER_DIR.
29103 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
29104 Martin Liška <mliska@suse.cz>
29106 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
29107 Fix typos and grammar, use active voice, and clarify.
29109 2017-02-20 Marek Polacek <polacek@redhat.com>
29111 PR middle-end/79537
29112 * gimplify.c (gimplify_expr): Handle unused *&&L;.
29115 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
29117 2017-02-20 Jakub Jelinek <jakub@redhat.com>
29120 * config/i386/i386.c (ix86_expand_builtin): Handle
29121 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
29122 ix86_builtins_isa[fcode].isa as a requirement of those
29123 flags and any other flag in the bitmask.
29124 (ix86_init_mmx_sse_builtins): Use 0 instead of
29125 ~OPTION_MASK_ISA_64BIT as mask.
29126 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
29127 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
29128 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
29129 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
29131 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29134 * lra-constraints.c (split_reg): Check requested split mode
29135 is supported by the register.
29137 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29139 * lra-constraints.c (simplify_operand_subreg): Remove early
29142 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29145 * lra-constraints.c (curr_insn_transform): Tighten condition
29146 for converting SUBREG reloads from OP_OUT to OP_INOUT.
29148 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29151 * lra-constraints.c (curr_insn_transform): Handle
29152 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29154 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
29157 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
29159 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
29161 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
29164 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
29167 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29169 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
29172 2017-02-18 Jakub Jelinek <jakub@redhat.com>
29174 * final.c (last_columnnum, override_columnnum): New variables.
29175 (final_start_function): Set last_columnnum, pass it to begin_prologue
29176 hook and pass 0 to dwarf2out_begin_prologue.
29177 (final_scan_insn): Update override_columnnum. Pass last_columnnum
29178 to source_line debug hook.
29179 (notice_source_line): Compute last_columnnum and for debug_column_info
29180 return true on column changes.
29181 * debug.h (struct gcc_debug_hooks): Add column argument to
29182 source_line and begin_prologue hooks.
29183 (debug_nothing_int_charstar_int_bool): Remove prototype.
29184 (debug_nothing_int_int_charstar,
29185 debug_nothing_int_int_charstar_int_bool): New prototypes.
29186 (dwarf2out_begin_prologue): Add column argument.
29187 * debug.c (do_nothing_debug_hooks): Adjust source_line and
29188 begin_prologue hooks.
29189 (debug_nothing_int_charstar_int_bool): Remove.
29190 (debug_nothing_int_int_charstar,
29191 debug_nothing_int_int_charstar_int_bool): New functions.
29192 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
29193 through to dwarf2out_source_line.
29194 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
29195 (dwarf2out_source_line): Add column argument, emit it if requested.
29196 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
29198 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29199 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29200 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
29201 through to dwarf2out_begin_prologue.
29202 (vmsdbgout_source_line): Add column argument, pass it through to
29203 dwarf2out_source_line.
29204 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
29205 dbxout_source_line caller.
29206 (dbxout_source_line): Add column argument.
29208 * common.opt (gno-column-info, gcolumn-info): New options.
29209 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
29210 (check_die): Also test for multiple DW_AT_decl_column attributes.
29211 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
29212 DW_AT_decl_column if requested.
29213 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
29215 (gen_variable_die): Likewise.
29216 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
29217 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
29220 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
29221 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
29222 (ix86_handle_option): Handle OPT_m3dnowa.
29223 * doc/invoke.texi (-m3dnowa): Document.
29224 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
29225 -m3dnowa instead of -m3dnow -march=athlon.
29228 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
29229 instead of gcc_assert for K, r and R code checks. Formatting fixes.
29231 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29234 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
29235 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
29236 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
29237 generator for vsx_xxpermdi_<mode>_be.
29238 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
29239 force big-endian semantics.
29240 (vsx_xxpermdi_<mode>_be): New define_expand with same
29241 implementation as previous version of vsx_xxpermdi_<mode>.
29243 2017-02-17 Jakub Jelinek <jakub@redhat.com>
29245 PR tree-optimization/79327
29246 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
29247 variable, its initialization and use.
29249 2017-02-17 Julia Koval <julia.koval@intel.com>
29251 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
29252 (OPTION_MASK_ISA_PKU_UNSET): New.
29253 (ix86_handle_option): Handle -mrdpid.
29254 * config/i386/cpuid.h (bit_RDPID): New.
29255 * config/i386/driver-i386.c (host_detect_local_cpu):
29256 Detect RDPID feature.
29257 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
29258 * config/i386/i386-c.c (ix86_target_macros_internal):
29260 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
29261 (ix86_valid_target_attribute_inner_p): Add "rdpid".
29262 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
29263 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
29264 * config/i386/i386.md (define_insn "rdpid"): New.
29265 * config/i386/i386.opt Add -mrdpid.
29266 * config/i386/immintrin.h (_rdpid_u32): New.
29268 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
29270 PR rtl-optimization/79541
29271 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
29272 instead of transforming it into USE.
29274 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
29276 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
29277 If HONOR_SNANS (SFmode) force the input to a register.
29278 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
29279 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
29280 an frsp or similar insn.
29282 2017-02-17 Martin Liska <mliska@suse.cz>
29284 PR rtl-optimization/79577
29285 * params.def (selsched-max-sched-times): Increase minimum to 1.
29287 2017-02-17 Martin Liska <mliska@suse.cz>
29289 PR rtl-optimization/79574
29290 * gcse.c (want_to_gcse_p): Prevent integer overflow.
29292 2017-02-17 Martin Liska <mliska@suse.cz>
29294 PR tree-optimization/79529
29295 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
29296 ssa_defined_default_def_p to handle cases which are implicitly
29298 * tree-ssa.c (ssa_defined_default_def_p): New function.
29299 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
29300 which are implicitly defined.
29301 * tree-ssa.h (ssa_defined_default_def_p): Declare.
29303 2017-02-17 Richard Biener <rguenther@suse.de>
29305 PR middle-end/79576
29306 * params.def (max-ssa-name-query-depth): Limit to 10.
29308 2017-02-17 Richard Biener <rguenther@suse.de>
29310 PR tree-optimization/79552
29311 * tree-ssa-structalias.c (visit_loadstore): Properly verify
29314 2017-02-17 Richard Biener <rguenther@suse.de>
29317 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
29319 2017-02-17 Marek Polacek <polacek@redhat.com>
29321 PR middle-end/79536
29322 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
29323 (fold_negate_expr): New wrapper.
29325 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
29327 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
29328 Correct terminology and de-emphasize pre-standard behavior.
29330 2017-02-16 Alan Modra <amodra@gmail.com>
29332 PR rtl-optimization/79286
29333 * ira.c (def_dominates_uses): New function.
29334 (update_equiv_regs): Don't create an equivalence for insns that
29335 may trap where the register def does not dominate the use.
29337 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
29339 PR rtl-optimization/78127
29340 * lra.c (lra): Call lra_eliminate before finish the loop after
29343 2017-02-16 Richard Biener <rguenther@suse.de>
29345 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
29347 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
29348 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
29349 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
29350 (isl_val_int_from_wi): New function.
29351 (extract_affine_gmp): Rename to ...
29352 (extract_affine_wi): ... this, take a widest_int.
29353 (extract_affine_int): Just wrap extract_affine_wi.
29354 (add_param_constraints): Use isl_val_int_from_wi.
29355 (add_loop_constraints): Likewise, and extract_affine_wi.
29357 2017-02-15 Jeff Law <law@redhat.com>
29359 PR middle-end/79521
29360 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
29361 ira_init_register_move_cost_if_necessary.
29363 2017-02-15 Martin Sebor <msebor@redhat.com>
29365 PR middle-end/32003
29366 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
29367 removed in a prior commit.
29369 2017-02-15 Bin Cheng <bin.cheng@arm.com>
29371 PR tree-optimization/79347
29372 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
29373 counters during peeling.
29375 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
29377 * Makefile.in (site.exp): Remove "set ISLVER".
29379 2017-02-15 Jakub Jelinek <jakub@redhat.com>
29382 * real.c (real_from_integer): Call real_convert even for decimal.
29384 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
29387 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
29389 2017-02-14 Andrew Pinski <apinski@cavium.com>
29391 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
29392 cores and change the partno/implementer to be correct.
29393 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
29394 the 'B" as the implementer.
29395 * config/aarch64/aarch64-tune.md: Regenerate.
29397 2017-02-14 Carl Love <cel@us.ibm.com>
29399 * config/rs6000/rs6000.c: Add case statement entry to make the
29400 xvcvuxdsp built-in argument unsigned.
29401 * config/rs6000/vsx.md: Fix the source and return operand types so they
29402 match the instruction definitions from the ISA document. Fix typo
29403 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
29406 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
29409 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
29410 member early_clobber_alts.
29411 * lra-lives.c (reg_early_clobber_p): New.
29412 (process_bb_lives): Use it.
29413 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
29414 (debug_operand_data): Initialize early_clobber_alts.
29415 (setup_operand_alternative): Set up early_clobber_alts.
29416 (collect_non_operand_hard_regs): Ditto. Pass early clobber
29417 alternatives to new_insn_reg.
29418 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
29420 (lra_update_insn_regno_info): Pass the new arg.
29422 2017-02-14 Jakub Jelinek <jakub@redhat.com>
29424 PR middle-end/79505
29425 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
29426 (new_oacc_loop_raw): Don't clear already cleared fields.
29429 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
29430 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
29431 _mm512_prefetch_i64gather_ps): New inline functions and macros.
29433 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
29436 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
29438 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
29441 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
29442 the extra instruction to the right place to store 128-bit constant
29445 2017-02-14 Martin Sebor <msebor@redhat.com>
29447 PR middle-end/79448
29448 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
29449 warning for strings of unknown length.
29451 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
29453 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
29455 2017-02-14 Jeff Law <law@redhat.com>
29458 * ira-costs.c (scan_one_insn): Initialize register move costs
29459 for pseudos seen in USE/CLOBBER insns.
29461 PR tree-optimization/79095
29462 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
29463 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
29464 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
29465 if the operands are known to be not equal, then the resulting range
29467 (intersect_ranges): If the new range is ~[0,0] and the old range is
29468 wide, then prefer ~[0,0].
29469 * tree-vrp.c (overflow_comparison_p_1): New function.
29470 (overflow_comparison_p): New function.
29471 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
29472 if NAME is used in an overflow test.
29473 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
29474 overflow check that can be expressed as an equality test, then adjust
29475 ops to be that equality test.
29477 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29479 * config/s390/s390-builtin-types.def: Remove flags argument.
29480 * config/s390/s390.c (s390_init_builtins): Likewise.
29482 2017-02-14 Martin Liska <mliska@suse.cz>
29484 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
29485 vector. Fix trailing white spaces.
29487 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
29489 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
29492 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29494 PR rtl-optimization/68664
29495 * config/arm/arm.c (arm_sched_can_speculate_insn):
29496 New function. Declare prototype.
29497 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
29499 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29501 PR rtl-optimization/68664
29502 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
29504 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
29506 2017-02-14 Amit Pawar <amit.pawar@amd.com>
29508 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
29509 max skip bytes for function, loop and jump.
29511 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29513 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
29514 ABS_EXPR for gimple dump.
29516 2017-02-14 Jakub Jelinek <jakub@redhat.com>
29519 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
29521 PR tree-optimization/79408
29522 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
29523 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
29524 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
29525 also if rhs1 is INTEGER_CST.
29527 2017-02-14 Richard Biener <rguenther@suse.de>
29529 PR middle-end/79432
29530 * tree-into-ssa.c (insert_phi_nodes): When the function can
29531 have abnormal edges rewrite SSA names with broken use-def
29532 dominance out of SSA and register them for PHI insertion.
29534 2017-02-13 Martin Sebor <msebor@redhat.com>
29536 PR middle-end/79496
29537 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
29538 clearing info.nowrite flag when snprintf size argument is a range.
29540 2017-02-13 Jakub Jelinek <jakub@redhat.com>
29542 * cprop.c (cprop_jump): Add missing space in string literal.
29543 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
29544 (get_constraint_for_component_ref): Likewise.
29545 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
29546 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
29547 * lra-constraints.c (process_alt_operands): Likewise.
29548 * ipa-inline.c (inline_small_functions): Likewise.
29549 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
29550 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
29551 * trans-mem.c (diagnose_tm_1_op): Likewise.
29552 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
29553 (grid_parallel_clauses_gridifiable): Likewise.
29555 * config/nvptx/mkoffload.c (process): Add space in between
29558 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
29559 "MOD4_SSE_REGS" and "ALL_REGS".
29561 * spellcheck.c (test_data): Add , in between "foo" and "food".
29563 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29566 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
29567 boundary crossing check and subsequent code generation agree.
29569 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29571 * config/aarch64/aarch64.c (has_memory_op): Delete.
29572 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
29575 2017-02-13 Jakub Jelinek <jakub@redhat.com>
29577 PR rtl-optimization/79388
29578 PR rtl-optimization/79450
29579 * combine.c (distribute_notes): When removing TEM_INSN for which
29580 corresponding dest has last value recorded, invalidate that last
29583 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29585 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
29586 of explicit '@'. Add missing assembly comment marker on branch costs
29589 2017-02-13 Nathan Sidwell <nathan@acm.org>
29591 * gengtype-lex.l (<in_struct>): Add '/'.
29593 2017-02-13 Martin Liska <mliska@suse.cz>
29596 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
29598 2017-02-13 Richard Biener <rguenther@suse.de>
29600 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
29602 * configure: Re-generate.
29603 * config.in: Likewise.
29604 * graphite-dependences.c: Simplify as if
29605 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
29606 * graphite-isl-ast-to-gimple.c: Likewise.
29607 * graphite-optimize-isl.c: Likewise.
29608 * graphite-poly.c: Likewise.
29609 * graphite-sese-to-poly.c: Likewise.
29610 * graphite.h: Likewise.
29611 * toplev.c: Include isl/version.h and use isl_version () for
29612 printing the ISL version.
29613 * doc/install.texi: Update ISL requirement.
29615 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
29617 * doc/standards.texi (Standards): Update reference to
29620 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
29622 * doc/extend.texi (Named Address Spaces): sourceware.org now
29624 * doc/install.texi (Binaries): Ditto.
29627 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
29629 * doc/cpp.texi: Replace "stringify"/"stringification" with C
29630 standard terminology "stringize"/"stringizing" throughout.
29631 * doc/cppinternals.texi: Likewise.
29633 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
29635 * doc/extend.texi: Fix some spelling mistakes and typos.
29636 * doc/invoke.texi: Likewise.
29638 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
29641 * params.def (inline-min-speedup) Change from 10 to 8.
29643 2017-02-11 Jakub Jelinek <jakub@redhat.com>
29645 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
29648 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
29651 * ipa-inline-analysis.c (get_minimal_bb): New function.
29652 (record_modified): Use it.
29653 (remap_edge_change_prob): Handle also ancestor functions.
29655 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
29657 * doc/contrib.texi (Contributors): Remove broken link into
29658 the Mauve CVS repository.
29660 2017-02-11 Jakub Jelinek <jakub@redhat.com>
29662 PR middle-end/79454
29663 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
29664 result computation whenever lhs doesn't have vector mode, not
29665 just when it has BLKmode.
29667 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
29669 * doc/makefile.texi (profiledbootstrap): Refer to the
29670 installation instructions only in textual form.
29672 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29675 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
29677 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
29679 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
29680 (Specific): Update mingw-w64 reference.
29682 (Specific): Remove broken link to Renesas RX processor.
29684 2017-02-10 Richard Biener <rguenther@suse.de>
29686 * toplev.c (process_options): Do not mention obsolete graphite
29687 options when printing sorry message about missing graphite support.
29688 Mention -floop-nest-optimize.
29690 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
29692 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
29693 (vtst_p16): Likewise.
29694 (vtstq_p8): Likewise.
29695 (vtstq_p16): Likewise.
29697 (vtstq_p64): Likewise.
29698 * config/arm/arm_neon.h (vgetq_lane_p64): New.
29699 (vset_lane_p64): New.
29700 (vsetq_lane_p64): New.
29702 2017-02-10 Jakub Jelinek <jakub@redhat.com>
29704 PR tree-optimization/79411
29705 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
29706 stmt operands are SSA_NAMEs used in abnormal phis.
29707 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
29710 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
29713 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
29716 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
29718 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
29720 2017-02-09 Jakub Jelinek <jakub@redhat.com>
29722 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
29726 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
29727 not arbitrary TREE_CONSTANT.
29730 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
29731 "omp declare target link" attribute unless is_global_var.
29732 * omp-offload.c (find_link_var_op): Likewise.
29734 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
29735 Chung-Lin Tang <cltang@codesourcery.com>
29737 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
29739 (gimplify_adjust_omp_clauses): Don't delete TILE.
29740 (gimplify_omp_for): Deal with TILE.
29741 * internal-fn.c (expand_GOACC_TILE): New function.
29742 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
29744 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
29745 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
29747 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
29748 avoid DIV for outermost collapse var.
29749 (expand_oacc_for): Insert tile element loop as needed. Adjust.
29750 Remove out of date comments, fix whitespace.
29751 * omp-general.c (omp_extract_for_data): Deal with tiling.
29752 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
29753 adjust OLF_DIM_BASE value.
29754 (struct omp_for_data): Add tiling field.
29755 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
29756 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
29757 for auto loops. Remove default auto determining, moved to
29758 oacc_loop_fixed_partitions.
29759 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
29760 stmts, add e_mask field.
29761 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
29762 (oacc_thread_numbers): Use oacc_dim_call.
29763 (oacc_xform_tile): New.
29764 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
29765 (finish_oacc_loop): Adjust for ifns vector.
29766 (oacc_loop_discover_walk): Append loop abstraction sites to list,
29767 add case for GOACC_TILE fns.
29768 (oacc_loop_xform_loop): Delete.
29769 (oacc_loop_process): Iterate over call list directly, and add
29770 handling for GOACC_TILE fns.
29771 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
29773 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
29774 vector partitioning to outer loops. Assign 2 partitions to loops
29775 when available. Add TILE handling.
29776 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
29777 (execite_oacc_device_lower): Process GOACC_TILE fns,
29778 ignore unknown specs.
29779 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
29780 * tree.c (omp_clause_num_ops): Adjust TILE ops.
29781 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
29783 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
29785 * configure.ac (ACX_BUGURL): Update.
29786 * configure: Regenerate.
29788 2017-02-09 Richard Biener <rguenther@suse.de>
29790 PR tree-optimization/69823
29791 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
29792 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
29794 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
29796 * config/arc/arc-c.def: Add __NPS400__ definition.
29797 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
29798 (TARGET_NPS400): Define.
29800 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
29802 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
29804 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
29805 pointer, arch_info.
29806 (arc_cpu_types): Fill the arch_info field with a pointer into the
29807 arc_arch_types table.
29808 (arc_selected_cpu): Declare.
29809 * config/arc/arc.c (arc_selected_cpu): Make global.
29810 (arc_selected_arch): Delete.
29811 (arc_base_cpu): Delete.
29812 (arc_override_options): Remove references to deleted variables,
29813 update access to arch information.
29814 (ARC_OPT): Update access to arch information.
29815 (ARC_OPTX): Likewise.
29816 * config/arc/arc.h (arc_base_cpu): Remove declaration.
29817 (TARGET_ARC600): Update access to arch information.
29818 (TARGET_ARC601): Likewise.
29819 (TARGET_ARC700): Likewise.
29820 (TARGET_EM): Likewise.
29821 (TARGET_HS): Likewise.
29822 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
29825 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
29828 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
29829 condition/operands for integer GE/LE/GEU/LEU operations.
29831 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
29833 PR translation/79397
29834 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
29837 2017-02-08 Martin Jambor <mjambor@suse.cz>
29840 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
29841 whether allocation happened.
29842 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
29843 nothing was allocated.
29845 2017-02-08 Jakub Jelinek <jakub@redhat.com>
29847 PR tree-optimization/79408
29848 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
29849 constant, but SSA_NAME with a known integer range, use the minimum
29850 of that range instead of op1 to determine if modulo can be replaced
29851 with its first operand.
29853 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29855 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
29857 2017-02-08 Richard Biener <rguenther@suse.de>
29859 PR tree-optimization/71824
29860 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
29861 Check all loops contained in the merged region.
29863 2017-02-07 Andrew Pinski <apinski@cavium.com>
29865 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
29867 2017-02-07 Andrew Pinski <apinski@cavium.com>
29869 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
29870 (thunderxt88): Likewise.
29871 (thunderxt81): Disable LSE and change v8.1 to v8.
29872 (thunderxt83): Likewise.
29874 2017-02-07 Jakub Jelinek <jakub@redhat.com>
29875 Richard Biener <rguenther@suse.de>
29877 PR middle-end/79399
29878 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
29879 type from int to size_t.
29880 * ira-costs.c (struct_costs_size): Change type from int to size_t.
29882 2017-02-07 Jakub Jelinek <jakub@redhat.com>
29884 PR rtl-optimization/79386
29885 * cprop.c (bypass_conditional_jumps): Initialize
29886 bypass_last_basic_block already before splitting bbs after
29887 unconditional traps...
29888 (bypass_conditional_jumps): ... rather than here.
29891 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
29892 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
29893 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
29894 fix -masm=intel patterns.
29896 2017-02-07 Richard Biener <rguenther@suse.de>
29898 PR tree-optimization/79256
29899 PR middle-end/79278
29900 * builtins.c (get_object_alignment_2): Use min_align_of_type
29901 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
29902 and ADJUST_FIELD_ALIGN.
29904 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
29906 * doc/tm.texi: Regenerate.
29907 * stor-layout.c (layout_decl): Adjust.
29908 (update_alignment_for_field): Likewise.
29909 (place_field): Likewise.
29910 (min_align_of_type): Likewise.
29911 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
29912 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
29913 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
29914 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
29915 * config/frv/frv.c (frv_adjust_field_align): Likewise.
29916 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
29917 * config/i386/i386.c (x86_field_alignment): Likewise.
29918 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
29919 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
29920 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
29921 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
29922 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
29923 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
29927 2017-01-30 Richard Biener <rguenther@suse.de>
29929 PR tree-optimization/79256
29930 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
29931 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
29934 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
29936 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
29937 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
29938 builtins to SImode and emit a zero-extend, if necessary.
29940 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
29942 * docs/invoke.texi (RISC-V Options): Alphabetize.
29944 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
29946 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
29949 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
29951 * config/riscv/riscv.c: New file.
29952 * common/config/riscv/riscv-common.c: Likewise.
29953 * config.gcc: Likewise.
29954 * config/riscv/constraints.md: Likewise.
29955 * config/riscv/elf.h: Likewise.
29956 * config/riscv/generic.md: Likewise.
29957 * config/riscv/linux.h: Likewise.
29958 * config/riscv/multilib-generator: Likewise.
29959 * config/riscv/peephole.md: Likewise.
29960 * config/riscv/pic.md: Likewise.
29961 * config/riscv/predicates.md: Likewise.
29962 * config/riscv/riscv-builtins.c: Likewise.
29963 * config/riscv/riscv-c.c: Likewise.
29964 * config/riscv/riscv-ftypes.def: Likewise.
29965 * config/riscv/riscv-modes.def: Likewise.
29966 * config/riscv/riscv-opts.h: Likewise.
29967 * config/riscv/riscv-protos.h: Likewise.
29968 * config/riscv/riscv.h: Likewise.
29969 * config/riscv/riscv.md: Likewise.
29970 * config/riscv/riscv.opt: Likewise.
29971 * config/riscv/sync.md: Likewise.
29972 * config/riscv/t-elf-multilib: Likewise.
29973 * config/riscv/t-linux: Likewise.
29974 * config/riscv/t-linux-multilib: Likewise.
29975 * config/riscv/t-riscv: Likewise.
29976 * configure.ac: Likewise.
29977 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
29978 Waterman as RISC-V maintainers.
29979 * doc/install.texi: Add RISC-V entries.
29980 * doc/invoke.texi: Add RISC-V options section.
29981 * doc/md.texi: Add RISC-V constraints section.
29982 * configure: Regenerated.
29984 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
29987 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
29988 false values to be constant vectors with all 0 or all 1 bits set.
29989 (vcondu<mode><mode>): Likewise.
29990 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
29992 (fpmask_comparison_operator): Update comment.
29993 (vecint_comparison_operator): New predicate.
29994 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
29995 vector conditionals when the true and false values are constant
29996 vectors with all 0 bits or all 1 bits set.
29998 2017-02-06 Martin Sebor <msebor@redhat.com>
30000 PR tree-optimization/79376
30001 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
30003 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
30005 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
30006 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
30007 to simplify split condition.
30009 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30011 * omp-expand.c (oxpand_omp_atomic_fetch_op,
30012 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
30015 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
30017 PR rtl-optimization/68664
30018 * target.def (can_speculate_insn): New hook.
30019 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
30020 * doc/tm.texi: Regenerate.
30021 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
30022 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
30023 (rs6000_sched_can_speculate_insn): New function.
30025 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30027 PR tree-optimization/79284
30028 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
30029 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
30030 vectorizable_mask_load_store, vectorizable_operation,
30031 vect_is_simple_cond, get_same_sized_vectype): Use it instead
30032 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
30033 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
30034 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
30035 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
30036 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
30037 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
30038 is_gimple_assign (stmt). Replace another such test with
30039 is_gimple_assign (stmt).
30041 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30044 * config/avr/avr.c (rtl-iter.h): Include it.
30045 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
30046 (avr_legitimate_combined_insn): ...and implementation.
30048 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
30050 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
30051 * config/s390/s390.c (s390_const_operand_ok)
30052 (s390_canonicalize_comparison, s390_extract_part)
30053 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
30054 (s390_contiguous_bitmask_p, s390_rtx_costs)
30055 (legitimize_pic_address): Likewise.
30056 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
30057 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
30058 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
30059 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
30060 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
30062 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30064 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
30065 REGNO($0) == REGNO($1).
30067 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30069 * config/s390/linux.h(SIZE_TYPE): Add comment.
30071 2017-02-06 Julian Brown <julian@codesourcery.com>
30072 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30073 Virendra Pathak <virendra.pathak@broadcom.com>
30075 * config/aarch64/aarch64-cores.def: Change the scheduler
30077 * config/aarch64/aarch64.md: Include thunderx2t99.md.
30078 * config/aarch64/thunderx2t99.md: New file.
30080 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30082 * doc/standards.texi (Go Language): Update link to language
30085 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30087 * tree-eh.c (lower_resx): Sanitize profile.
30088 (cleanup_empty_eh_move_lp): Likewise.
30090 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30093 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
30095 * cfgloopmanip.h (loop_version): Update prototype.
30096 * modulo-sched.c (sms_schedule): Update call of loop_version.
30097 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
30098 * tree-parloops.c (gen_parallel_loop): Likewise.
30099 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
30100 * tree-ssa-loop-split.c (split_loop): Likewise.
30101 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
30102 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
30104 2017-02-05 Martin Liska <mliska@suse.cz>
30107 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
30109 (print_operand_address): Initialize a struct to zero.
30111 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30113 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
30114 garbage collector only in textual form.
30116 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30118 * doc/extend.texi (x86 specific memory model extensions for
30119 transactional memory): Simplify a phrase.
30121 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
30124 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
30125 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
30126 (atomic_storedi_1): Likewise.
30128 2017-02-04 Jakub Jelinek <jakub@redhat.com>
30130 PR tree-optimization/79338
30131 * tree-parloops.c (gather_scalar_reductions): Don't call
30132 vect_analyze_loop_form for loop->inner before destroying loop's
30135 2017-02-03 Martin Sebor <msebor@redhat.com>
30137 PR tree-optimization/79327
30138 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
30139 when precision has resulted in leading zeros.
30140 (format_integer): Adjust the likely counter to assume an unknown
30141 argument that may be zero is non-zero.
30143 2017-02-03 Jason Merrill <jason@redhat.com>
30146 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
30147 avoid copying non-taken branch.
30149 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30151 PR tree-optimization/79340
30152 * tree-vect-loop.c (vectorizable_reduction): Release
30153 vec_defs elements after safe_splicing them into other vectors.
30156 PR tree-optimization/79327
30157 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
30158 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
30160 (format_integer): Use wide_int_to_tree instead of build_int_cst
30161 + to_?hwi. If argmin is NULL, just set argmin and argmax to
30162 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
30163 of shortest and longest sequence.
30165 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
30167 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
30168 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
30170 2017-02-03 Walter Lee <walt@tilera.com>
30173 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
30174 after initial stackframe link reg save.
30175 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
30177 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30180 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
30181 wu for stxssp alternative.
30183 2017-02-03 Martin Sebor <msebor@redhat.com>
30185 PR tree-optimization/79352
30186 * gimple-fold.c (get_range_strlen): Add argument.
30187 (get_range_strlen): Change return type to bool.
30188 (get_maxval_strlen): Pass in a dummy argument.
30189 * gimple-fold.h (get_range_strlen): Change return type to bool.
30190 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
30191 * tree.h (array_at_struct_end_p): Add argument.
30192 * tree.c (array_at_struct_end_p): Handle it.
30194 2017-02-03 Martin Liska <mliska@suse.cz>
30197 * multiple_target.c (create_dispatcher_calls): Redirect edge
30198 from a caller of a dispatcher.
30199 (expand_target_clones): Make the clones local.
30200 (ipa_target_clone): Do both target clones and resolvers.
30201 (ipa_dispatcher_calls): Remove the pass.
30202 (pass_dispatcher_calls::gate): Likewise.
30203 (make_pass_dispatcher_calls): Likewise.
30204 * passes.def (pass_target_clone): Put as very first IPA early
30207 2017-02-03 Martin Liska <mliska@suse.cz>
30209 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
30210 in case of a function with ifunc attribute.
30212 2017-02-03 Martin Liska <mliska@suse.cz>
30214 * cgraph.c (cgraph_node::dump): Dump function version info.
30215 * symtab.c (symtab_node::dump_base): Add missing new line.
30217 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30219 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
30220 (ifcombine_ifandif): Use it.
30222 2017-02-03 Martin Liska <mliska@suse.cz>
30224 * doc/invoke.texi: Document default value for
30225 use-after-scope-direct-emission-threshold.
30227 2017-02-03 Martin Liska <mliska@suse.cz>
30229 PR tree-optimization/79339
30230 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
30231 (format_floating): Likewise.
30233 2017-02-03 Martin Liska <mliska@suse.cz>
30236 * ipa-prop.c (ipa_node_params_t::insert): Remove current
30238 (ipa_node_params_t::remove): Likewise.
30239 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
30240 initialization from removed ipa_node_params_t::insert.
30241 (ipa_node_params::~ipa_node_params): Move from removed
30242 ipa_node_params_t::release.
30243 * symbol-summary.h (symbol_summary::m_released): New member.
30244 Do not release a summary twice. Do not allow to call finalizer
30245 for types of a summary that live in GGC memory.
30247 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30249 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
30252 2017-02-02 Martin Sebor <msebor@redhat.com>
30254 PR middle-end/79275
30255 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
30256 (format_string): Tighten up the range of output for non-constant
30257 strings and correct the expected range for wide non-constant strings.
30259 2017-02-02 Martin Sebor <msebor@redhat.com>
30261 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
30263 PR middle-end/32003
30264 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
30266 (-fdump-tree-@var): Add to index and document how to come up
30267 with pass-specific option and dump file names.
30268 (-fdump-passes): Clarify where to look for output.
30270 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30272 PR middle-end/77445
30273 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
30274 statistics of the analyzed path; allow threading for speed when
30275 any of BBs along the path are optimized for speed.
30277 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
30279 PR middle-end/78468
30280 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
30281 settings of the virtual registers.
30284 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
30286 * explow.c (get_dynamic_stack_size): Take known alignment of stack
30287 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
30290 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30292 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
30293 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
30295 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30297 * config/s390/s390.md: Add missing comments with the expanded
30299 * config/s390/vector.md: Likewise.
30300 * config/s390/vx-builtins.md: Likewise.
30302 2017-02-02 Jakub Jelinek <jakub@redhat.com>
30305 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
30306 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
30307 conditions on a single line.
30309 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30311 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
30312 __S390_VX__ to __VX__.
30314 2017-02-01 Andrew Pinski <apinski@cavium.com>
30316 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
30317 stmt_info to record_stmt_cost.
30318 (vect_get_known_peeling_cost): Pass stmt_info if known to
30320 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
30321 cpu_vector_cost field into
30322 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
30323 field into vec_int_stmt_cost and vec_fp_stmt_cost.
30324 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
30325 splitting of scalar_stmt_cost and vec_stmt_cost.
30326 (thunderx_vector_cost): Likewise.
30327 (cortexa57_vector_cost): LIkewise.
30328 (exynosm1_vector_cost): Likewise.
30329 (xgene1_vector_cost): Likewise.
30330 (thunderx2t99_vector_cost): Improve after the splitting of the two
30332 (aarch64_builtin_vectorization_cost): Update for the splitting of
30333 scalar_stmt_cost and vec_stmt_cost.
30335 2017-02-01 Torvald Riegel <triegel@redhat.com>
30336 Richard Henderson <rth@redhat.com>
30338 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
30339 conditional on existance of a fast atomic load.
30340 * optabs-query.c (can_atomic_load_p): New function.
30341 * optabs-query.h (can_atomic_load_p): Declare it.
30342 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
30343 no fast atomic load is available for the particular size of access.
30344 (expand_atomic_compare_and_swap): Likewise.
30345 (expand_atomic_load): Likewise.
30346 (expand_atomic_store): Likewise.
30347 (expand_atomic_fetch_op): Likewise.
30348 * testsuite/lib/target-supports.exp
30349 (check_effective_target_sync_int_128): Remove x86 because it provides
30350 no fast atomic load.
30351 (check_effective_target_sync_int_128_runtime): Likewise.
30353 2017-02-01 Richard Biener <rguenther@suse.de>
30355 * graphite.c: Include tree-vectorizer.h for find_loop_location.
30356 (graphite_transform_loops): Provide opt-info for optimized nests.
30357 * tree-parloop.c (parallelize_loops): Provide opt-info for
30358 parallelized loops.
30360 2017-02-01 Richard Biener <rguenther@suse.de>
30362 PR middle-end/79315
30363 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
30364 was not set before.
30366 2017-02-01 Richard Biener <rguenther@suse.de>
30368 PR tree-optimization/71824
30369 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
30370 Verify the loops are valid in the merged SESE region.
30371 (scop_detection::can_represent_loop_1): Check analyzing the
30372 evolution of the number of iterations in the region succeeds.
30374 2017-01-31 Ian Lance Taylor <iant@golang.org>
30376 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
30377 REG_ARGS_SIZE note to 32-bit push insns and call insn.
30379 2017-01-31 David Malcolm <dmalcolm@redhat.com>
30381 PR preprocessor/79210
30382 * input.c (get_substring_ranges_for_loc): Replace line_width
30383 assertion with error-handling.
30385 2017-01-31 Richard Biener <rguenther@suse.de>
30387 PR tree-optimization/77318
30388 * graphite-sese-to-poly.c (extract_affine): Fix assert.
30389 (create_pw_aff_from_tree): Take loop parameter.
30390 (add_condition_to_pbb): Pass loop of the condition to
30391 create_pw_aff_from_tree.
30393 2017-01-31 Jakub Jelinek <jakub@redhat.com>
30395 * config/s390/s390.c (s390_asan_shadow_offset): New function.
30396 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
30398 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
30402 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
30404 (convert_int_to_float128): Likewise.
30405 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
30406 (convert_int_to_float128): Likewise.
30407 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
30408 (UNSPEC_IEEE128_CONVERT): Likewise.
30409 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
30410 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
30411 Use local variables for IBM extended format.
30412 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
30413 (fix_trunc<mode>si2_fprs): Likewise.
30414 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
30415 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
30416 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
30417 to know that we can now have integers of all sizes in vector
30419 (fix<uns>_<mode>di2_hw): Likewise.
30420 (float<uns>_<mode>si2_hw): Likewise.
30421 (fix_<mode>si2_hw): Likewise.
30422 (fixuns_<mode>si2_hw): Likewise.
30423 (float<uns>_<mode>di2_hw): Likewise.
30424 (float_<mode>di2_hw): Likewise.
30425 (float_<mode>si2_hw): Likewise.
30426 (floatuns_<mode>di2_hw): Likewise.
30427 (floatuns_<mode>si2_hw): Likewise.
30428 (xscvqp<su>wz_<mode>): Delete, no longer used.
30429 (xscvqp<su>dz_<mode>): Likewise.
30430 (xscv<su>dqp_<mode>): Likewise.
30431 (ieee128_mfvsrd_64bit): Likewise.
30432 (ieee128_mfvsrd_32bit): Likewise.
30433 (ieee128_mfvsrwz): Likewise.
30434 (ieee128_mtvsrw): Likewise.
30435 (ieee128_mtvsrd_64bit): Likewise.
30436 (ieee128_mtvsrd_32bit): Likewise.
30438 2017-01-31 Martin Liska <mliska@suse.cz>
30441 * ipa-prop.c (ipa_free_all_node_params): Call release method
30442 instead of ~sumbol_summary to not to trigger double times
30445 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
30447 PR tree-optimization/71691
30448 * bitmap.h (class auto_bitmap): New.
30449 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
30450 is_maybe_undefined instead of ssa_undefined_value_p.
30452 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30454 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
30455 __S390_ARCH_LEVEL__ to __ARCH__.
30457 2017-01-31 Jakub Jelinek <jakub@redhat.com>
30459 PR tree-optimization/79267
30460 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
30461 if should_remove_lhs_p is true.
30463 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
30466 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
30467 (add_alignment_attribute): New.
30468 (base_type_die): Add alignment attribute.
30469 (subrange_type_die): Likewise.
30470 (modified_type_die): Likewise.
30471 (gen_array_type_die): Likewise.
30472 (gen_descr_array_type_die: Likewise.
30473 (gen_enumeration_type_die): Likewise.
30474 (gen_subprogram_die): Likewise.
30475 (gen_variable_die): Likewise.
30476 (gen_field_die): Likewise.
30477 (gen_ptr_to_mbr_type_die): Likewise.
30478 (gen_struct_or_union_type_die): Likewise.
30479 (gen_subroutine_type_die): Likewise.
30480 (gen_typedef_die): Likewise.
30481 (base_type_cmp): Compare alignment attribute.
30483 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30486 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
30487 (setb_unsigned) New pattern for setb with CCUNS.
30488 * config/rs6000/rs6000.c (expand_block_compare): Use a different
30489 subfc./subfe sequence to avoid overflow problems. Generate a
30490 shorter sequence with cmpld/setb for power9.
30491 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
30492 for generating subfc. instruction.
30493 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
30494 now uses this instruction.
30496 2017-01-30 Ian Lance Taylor <iant@google.com>
30499 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
30500 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
30502 2017-01-30 Martin Sebor <msebor@redhat.com>
30504 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
30505 Move constant to the right of a relational operator.
30506 (get_mpfr_format_length, format_character, format_string): Ditto.
30507 (should_warn_p, maybe_warn): Same.
30509 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
30511 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
30514 * asan.c (get_translation_unit_decl): Remove function.
30515 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
30517 2017-01-30 Martin Liska <mliska@suse.cz>
30519 PR gcov-profile/79259
30520 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
30521 -fprofile-generate.
30523 2017-01-30 Martin Liska <mliska@suse.cz>
30526 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
30527 Initialize variables with NULL value.
30529 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
30532 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
30534 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
30536 2017-01-30 Richard Biener <rguenther@suse.de>
30538 PR tree-optimization/79276
30539 * tree-vrp.c (process_assert_insertions): Properly adjust common
30540 when removing a duplicate.
30542 2017-01-30 Richard Biener <rguenther@suse.de>
30544 PR tree-optimization/79256
30545 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
30546 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
30548 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
30550 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
30553 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
30554 ("*r<noxa>sbg_<mode>_sll_bitmask")
30555 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
30556 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
30557 Use contiguous_bitmask_nowrap_operand.
30559 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30562 * config/rs6000/altivec.h (vec_xl): Revise #define.
30563 (vec_xst): Likewise.
30565 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
30567 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
30569 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
30571 PR rtl-optimization/79194
30572 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
30573 traps before call to bypass_conditional_jumps.
30575 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
30577 PR tree-optimization/71374
30578 * lra-constraints.c (check_conflict_input_operands): New.
30579 (match_reload): Use it.
30581 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
30584 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
30585 account to calculate conflict_set.
30587 2017-01-27 Bin Cheng <bin.cheng@arm.com>
30589 PR rtl-optimization/78559
30590 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
30591 other_insn in combine.
30593 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
30595 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
30596 uint16_type_node for BT_UINT16.
30598 2017-01-27 David Malcolm <dmalcolm@redhat.com>
30600 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
30601 "RTL Tests" to menu.
30602 (GIMPLE Tests): New node.
30603 (RTL Tests): New node.
30605 2017-01-27 Richard Biener <rguenther@suse.de>
30607 PR tree-optimization/79245
30608 * tree-loop-distribution.c (distribute_loop): Apply cost
30609 modeling also to detected patterns.
30611 2017-01-27 Richard Biener <rguenther@suse.de>
30613 PR tree-optimization/71433
30614 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
30615 (compare_assert_loc): New function.
30616 (process_assert_insertions): Sort and optimize assert locations
30617 to remove duplicates and push down identical assertions on
30618 edges to their destination block.
30620 2017-01-27 Richard Biener <rguenther@suse.de>
30622 PR tree-optimization/79244
30623 * tree-vrp.c (remove_range_assertions): Forcefully propagate
30624 out SSA names even if abnormal.
30626 2017-01-27 Jakub Jelinek <jakub@redhat.com>
30628 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
30629 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
30630 instead of MPFR_RNDN.
30632 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
30635 * arm.c (arm_option_override): Don't call build_target_option_node
30636 until after doing all option overrides.
30637 (arm_valid_target_attribute_tree): Likewise.
30639 2017-01-27 Martin Liska <mliska@suse.cz>
30641 * doc/invoke.texi (-fprofile-arcs): Document profiling support
30642 for {cd}tors and C++ {cd}tors.
30644 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
30646 * config/s390/s390.md ("*setmem_long_and")
30647 ("*setmem_long_and_31z"): Use zero_extend instead of and.
30649 2017-01-26 Martin Sebor <msebor@redhat.com>
30651 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
30654 2017-01-26 Martin Sebor <msebor@redhat.com>
30656 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
30657 HAVE_DFmode before using XFmode or DFmode.
30658 (parse_directive): Avoid using the z length modifier to avoid
30659 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
30661 PR middle-end/78703
30662 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
30663 to accept adjustment as an array.
30664 (get_int_range): New function.
30665 (struct directive): Make width and prec arrays.
30666 (directive::set_width, directive::set_precision): Call get_int_range.
30667 (format_integer, format_floating): Handle width and precision ranges.
30668 (format_string, parse_directive): Same.
30670 2017-01-26 Jakub Jelinek <jakub@redhat.com>
30673 * dwarf2out.c (generate_skeleton_bottom_up): For children with
30674 comdat_type_p set, just clone them, but keep the children in the
30678 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
30679 which have direct callers with -fvar-tracking-assignments enabled
30681 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
30682 inside of type units.
30684 2017-01-26 Martin Sebor <msebor@redhat.com>
30686 PR middle-end/78703
30687 * gimple-ssa-sprintf.c (struct result_range): Add likely and
30689 (struct format_result): Replace number_chars, number_chars_min,
30690 and number_chars_max with a single member of struct result_range.
30692 (format_result::operator+=): Adjust.
30693 (struct fmtresult): Remove bounded. Handle likely and unlikely
30695 (fmtresult::adjust_for_width_or_precision): New function.
30696 (fmtresult:type_max_digits): New function.
30697 (bytes_remaining): Handle likely and unlikely counters.
30698 (min_bytes_remaining): Remove.
30699 (format_percent): Simplify.
30700 (format_integer, format_floating): Set likely and unlikely counters.
30701 (get_string_length, format_character, format_string): Same.
30702 (format_plain, should_warn_p): New function.
30703 (maybe_warn): Call should_warn_p. Update diagnostic messages
30704 and handle those for all directives, including plain strings.
30705 (format_directive): Handle likely and unlikely counters.
30706 Remove unnecessary quoting from diagnostics. Add an informational
30708 (add_bytes): Remove.
30709 (pass_sprintf_length::compute_format_length): Simplify.
30710 (try_substitute_return_value): Handle likely and unlikely counters.
30712 2017-01-26 Carl Love <cel@us.ibm.com>
30714 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
30715 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
30717 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
30720 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
30721 endianess for subregs into account.
30722 * lra-constraints.c (lra_constraints): Do risky transformations
30723 always on the first iteration.
30724 * lra-lives.c (check_pseudos_live_through_calls): Add arg
30725 last_call_used_reg_set.
30726 (process_bb_lives): Define and use last_call_used_reg_set.
30727 * lra.c (lra): Always continue after lra_constraints on the first
30730 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
30732 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
30734 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
30736 2017-01-26 Jakub Jelinek <jakub@redhat.com>
30738 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
30739 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
30740 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
30741 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
30742 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
30743 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
30744 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
30745 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
30746 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
30748 2017-01-26 Marek Polacek <polacek@redhat.com>
30751 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
30752 for the third operand.
30754 2017-01-26 Jakub Jelinek <jakub@redhat.com>
30756 PR middle-end/79236
30757 * omp-low.c (struct omp_context): Add simt_stmt field.
30758 (scan_omp_for): Return omp_context *.
30759 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
30760 context to the _simt_ SIMD stmt.
30761 (lower_omp_for): For combined SIMD with sibling _simt_
30762 SIMD, make sure to use the same decls in _looptemp_
30763 clauses as in the sibling.
30765 2017-01-26 David Sherwood <david.sherwood@arm.com>
30767 PR middle-end/79212
30768 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
30771 2017-01-26 Jakub Jelinek <jakub@redhat.com>
30774 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
30775 emit fld b; fld a; if possible.
30777 * brig-builtins.def: Update copyright years.
30778 * config/arm/arm_acle_builtins.def: Update copyright years.
30780 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
30783 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
30784 constraint instead of o for the stxsd instruction.
30786 2017-01-25 Carl Love <cel@us.ibm.com>
30788 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
30789 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
30791 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
30793 * doc/invoke.texi (C++ Dialect Options): Fix typo.
30795 2017-01-25 Richard Biener <rguenther@suse.de>
30797 PR tree-optimization/69264
30798 * target.def (vector_alignment_reachable): Improve documentation.
30799 * doc/tm.texi: Regenerate.
30800 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
30802 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
30803 earlier changes with respect to TYPE_USER_ALIGN.
30804 (vector_alignment_reachable_p): Likewise. Improve dumping.
30806 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30809 * config/arm/arm.md (xordi3): Force constant operand into a register
30812 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30814 * doc/invoke.texi (-fstore-merging): Correct default optimization
30815 levels at which it is enabled.
30816 (-O): Move -fstore-merging from list to...
30819 2017-01-25 Richard Biener <rguenther@suse.de>
30822 * omp-expand.c: Include debug.h.
30823 (expand_omp_taskreg): Make sure to generate early debug before
30824 outlining anything from a function.
30825 (expand_omp_target): Likewise.
30826 (grid_expand_target_grid_body): Likewise.
30828 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
30831 * asan.c (get_translation_unit_decl): New function.
30832 (asan_add_global): Extract modules file name from globals
30833 TRANSLATION_UNIT_DECL name.
30835 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
30838 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
30839 for long calls with APCS frame and VFP.
30841 2017-01-24 David Malcolm <dmalcolm@redhat.com>
30843 * cfg.c (original_copy_tables_initialized_p): New function.
30844 * cfg.h (original_copy_tables_initialized_p): New decl.
30845 * cfgrtl.c (relink_block_chain): Guard the call to
30846 free_original_copy_tables with a call to
30847 original_copy_tables_initialized_p.
30848 * cgraph.h (symtab_node::native_rtl_p): New decl.
30849 * cgraphunit.c (symtab_node::native_rtl_p): New function.
30850 (symtab_node::needed_p): Don't assert for early assembly output
30851 for __RTL functions.
30852 (cgraph_node::finalize_function): Set "force_output" for __RTL
30854 (cgraph_node::analyze): Bail out early for __RTL functions.
30855 (analyze_functions): Update assertion to support __RTL functions.
30856 (cgraph_node::expand): Bail out early for __RTL functions.
30857 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
30859 * function.h (struct function): Update comment for field
30861 * gimple-expr.c: Include "tree-pass.h".
30862 (gimple_has_body_p): Return false for __RTL functions.
30863 * Makefile.in (OBJS): Add run-rtl-passes.o.
30864 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
30866 (gcc::pass_manager::get_clean_slate): New accessor.
30867 * passes.c: Include "insn-addr.h".
30868 (should_skip_pass_p): Add logging. Update logic for running
30869 "expand" to be compatible with both __GIMPLE and __RTL. Guard
30870 property-provider override so it is only done for gimple passes.
30872 (skip_pass): New function.
30873 (execute_one_pass): Call skip_pass when skipping passes.
30874 * read-md.c (md_reader::read_char): Support filtering
30875 the input to a subset of line numbers.
30876 (md_reader::md_reader): Initialize fields
30877 m_first_line and m_last_line.
30878 (md_reader::read_file_fragment): New function.
30879 * read-md.h (md_reader::read_file_fragment): New decl.
30880 (md_reader::m_first_line): New field.
30881 (md_reader::m_last_line): New field.
30882 * read-rtl-function.c (function_reader::create_function): Only
30883 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
30884 curr_properties. Set DECL_INITIAL to a dummy block.
30885 (read_rtl_function_body_from_file_range): New function.
30886 * read-rtl-function.h (read_rtl_function_body_from_file_range):
30888 * run-rtl-passes.c: New file.
30889 * run-rtl-passes.h: New file.
30891 2017-01-24 Jeff Law <law@redhat.com>
30893 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
30896 2017-01-24 Bin Cheng <bin.cheng@arm.com>
30898 PR tree-optimization/79159
30899 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
30900 (record_nonwrapping_iv): Improve boundary using above function if no
30901 value range information.
30903 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
30904 Martin Jambor <mjambor@suse.cz>
30906 * brig-builtins.def: New file.
30907 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
30908 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
30909 (DEF_HSAIL_SAT_BUILTIN): Likewise.
30910 (DEF_HSAIL_INTR_BUILTIN): Likewise.
30911 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
30912 * builtin-types.def (BT_INT8): New.
30913 (BT_INT16): Likewise.
30914 (BT_UINT8): Likewise.
30915 (BT_UINT16): Likewise.
30916 (BT_FN_ULONG): Likewise.
30917 (BT_FN_UINT_INT): Likewise.
30918 (BT_FN_UINT_ULONG): Likewise.
30919 (BT_FN_UINT_LONG): Likewise.
30920 (BT_FN_UINT_PTR): Likewise.
30921 (BT_FN_ULONG_PTR): Likewise.
30922 (BT_FN_INT8_FLOAT): Likewise.
30923 (BT_FN_INT16_FLOAT): Likewise.
30924 (BT_FN_UINT32_FLOAT): Likewise.
30925 (BT_FN_UINT16_FLOAT): Likewise.
30926 (BT_FN_UINT8_FLOAT): Likewise.
30927 (BT_FN_UINT64_FLOAT): Likewise.
30928 (BT_FN_UINT16_UINT32): Likewise.
30929 (BT_FN_UINT32_UINT16): Likewise.
30930 (BT_FN_UINT16_UINT16_UINT16): Likewise.
30931 (BT_FN_INT_PTR_INT): Likewise.
30932 (BT_FN_UINT_PTR_UINT): Likewise.
30933 (BT_FN_LONG_PTR_LONG): Likewise.
30934 (BT_FN_ULONG_PTR_ULONG): Likewise.
30935 (BT_FN_VOID_UINT64_UINT64): Likewise.
30936 (BT_FN_UINT8_UINT8_UINT8): Likewise.
30937 (BT_FN_INT8_INT8_INT8): Likewise.
30938 (BT_FN_INT16_INT16_INT16): Likewise.
30939 (BT_FN_INT_INT_INT): Likewise.
30940 (BT_FN_UINT_FLOAT_UINT): Likewise.
30941 (BT_FN_FLOAT_UINT_UINT): Likewise.
30942 (BT_FN_ULONG_UINT_UINT): Likewise.
30943 (BT_FN_ULONG_UINT_PTR): Likewise.
30944 (BT_FN_ULONG_ULONG_ULONG): Likewise.
30945 (BT_FN_UINT_UINT_UINT): Likewise.
30946 (BT_FN_VOID_UINT_PTR): Likewise.
30947 (BT_FN_UINT_UINT_PTR: Likewise.
30948 (BT_FN_UINT32_UINT64_PTR): Likewise.
30949 (BT_FN_INT_INT_UINT_UINT): Likewise.
30950 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
30951 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
30952 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
30953 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
30954 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
30955 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
30956 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
30957 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
30958 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
30959 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
30960 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
30961 * doc/frontends.texi: List BRIG FE.
30962 * doc/install.texi (Testing): Add BRIG tesring requirements.
30963 * doc/invoke.texi (Overall Options): Mention BRIG.
30964 * doc/standards.texi (Standards): Doucment BRIG HSA version.
30966 2017-01-24 Richard Biener <rguenther@suse.de>
30968 PR translation/79208
30969 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
30971 2017-01-24 Martin Jambor <mjambor@suse.cz>
30974 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
30975 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
30976 and known_contexts.
30978 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
30980 PR middle-end/79123
30981 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
30982 casts from signed to unsigned really don't have a range.
30984 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
30986 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
30987 GMP_RNDx for compatiblity.
30989 2017-01-24 Martin Liska <mliska@suse.cz>
30992 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
30993 that would prevent us to call alloca with -1 as argument.
30995 2017-01-24 Jakub Jelinek <jakub@redhat.com>
30997 * dwarf2out.c (output_compilation_unit_header, output_file_names):
30998 Avoid -Wformat-security warning.
31000 2017-01-23 Andrew Pinski <apinski@cavium.com>
31002 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
31005 2017-01-23 Martin Sebor <msebor@redhat.com>
31007 PR middle-end/78703
31008 * gimple-ssa-sprintf.c (warn_level): New global.
31009 (format_integer): Use it here and throughout the rest of the file.
31010 Use the same switch to compute sign as base.
31011 (maybe_warn): New function.
31012 (format_directive): Factor out warnings into maybe_warn.
31013 Add debugging output. Use warn_level.
31014 (add_bytes): Use warn_level.
31015 (pass_sprintf_length::compute_format_length): Add debugging output.
31016 (try_substitute_return_value): Same.
31017 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
31019 PR middle-end/78703
31020 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
31021 (struct fmtresult, format_integer, format_floating): Adjust.
31022 (fmtresult::fmtresult): Set max correctly in two argument ctor.
31023 (get_string_length, format_string,format_directive): Same.
31024 (pass_sprintf_length::compute_format_length): Same.
31025 (try_substitute_return_value): Simplify slightly.
31027 PR middle-end/78703
31028 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
31029 (fmtresult::operator+=): Outlined.
31030 (struct fmtresult): Add ctors.
31031 (struct conversion_spec): Rename...
31032 (struct directive): ...to this. Add and remove data members.
31033 (directive::set_width, directive::set_precision): New functions.
31034 (format_percent): Use fmtresult ctor.
31035 (get_width_and_precision): Remove.
31036 (format_integer): Make naming changes. Avoid computing width and
31038 (format_floating): Same. Adjust indentation.
31039 (format_character, format_none): New functions.
31040 (format_string): Moved character handling to format_character.
31041 (format_directive): Remove arguments, change return type.
31042 (parse_directive): New function.
31043 (pass_sprintf_length::compute_format_length): Move directive
31044 parsing to parse_directive.
31046 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31048 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
31049 (assign_assembler_name_if_needed): ... this.
31050 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
31051 (assign_assembler_name_if_needed): ... this.
31052 (free_lang_data_in_cgraph): Adjust callers.
31053 * cgraphunit.c (cgraph_node::analyze): Likewise.
31054 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
31057 2017-01-23 Richard Biener <rguenther@suse.de>
31059 PR tree-optimization/79088
31060 PR tree-optimization/79188
31061 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
31062 resetting loop bounds after last path deletion. Reset loop
31063 bounds of the target loop, make code match the comments.
31064 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
31065 Make sure loops need no fixups.
31067 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
31069 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
31070 exponent support with double type for first argument.
31071 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
31072 type returned by __builtin_vec_extract_sig,
31073 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
31074 functions from "vector int" to "vector unsigned int" or from
31075 "vector long long int" to "vector unsigned long long int".
31076 Changed type returned by __builtin_vec_extract_exp,
31077 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
31078 functions from "vector int" to "vector unsigned int" or from
31079 "vector long long int" to "vector unsigned long long int".
31080 Changed return type of __builtin_vec_test_data_class,
31081 __builtin_vec_test_data_class_sp, and
31082 __builtin_vec_test_data_class_dp from "vector int" to
31083 "vector bool int" or from "vector long long int" to "vector bool
31084 long long int" and changed second argument type from "unsigned
31085 int" to "int". Added new overloaded function forms "vector float
31086 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
31087 "vector float __builtin_vec_insert_exp_sp (vector float, vector
31088 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
31089 double, vector unsigned long long int)" and "vector double
31090 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
31091 long int)". Changed return type of
31092 __builtin_scalar_test_data_class and
31093 __builtin_scalar_test_data_class_sp and
31094 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
31095 int" and changed second argument from "unsigned int" to "int".
31096 Changed type returned by __builtin_scalar_test_neg,
31097 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
31098 from "int" to "bool int". Added new overloaded function form
31099 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
31100 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
31101 exponent double-precision with floating point first argument.
31102 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
31103 documentation of scalar_test_data_class, scalar_test_neg,
31104 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
31105 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
31106 vec_test_data_class built-in functions to reflect refinements in
31107 their type signatures.
31109 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
31111 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
31113 (aarch64_elf_asm_destructor): Likewise.
31115 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
31117 PR rtl-optimization/78634
31118 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
31119 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
31120 * ifcvt.c (noce_try_cmove): Add missing cost check.
31122 PR rtl-optimization/71724
31123 * combine.c (if_then_else_cond): Look for situations where it is
31124 beneficial to undo the work of one of the recursive calls.
31126 2017-01-23 Bin Cheng <bin.cheng@arm.com>
31128 PR tree-optimization/70754
31129 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
31130 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
31131 combined stmt before it if not NULL.
31132 (combine_chains): Process refs reversely and compute dominance point
31135 2017-01-23 Martin Liska <mliska@suse.cz>
31137 PR tree-optimization/79196
31138 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
31139 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
31141 (strlen_optimize_stmt): Call the renamed function.
31143 2017-01-23 Michael Matz <matz@suse.de>
31145 PR tree-optimization/78384
31146 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
31148 2017-01-23 Richard Biener <rguenther@suse.de>
31150 PR tree-optimization/79186
31151 * tree-vrp.c (register_new_assert_for): Make sure we've seen
31152 both incoming edges before moving an assert.
31154 2017-01-23 Martin Jambor <mjambor@suse.cz>
31156 * ipa-prop.c (load_from_param_1): Removed.
31157 (load_from_unmodified_param): Bits from load_from_param_1 put back
31159 (load_from_param): Removed.
31160 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
31161 with stmt. Reverted back to use of load_from_unmodified_param.
31163 2017-01-23 Martin Jambor <mjambor@suse.cz>
31166 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
31167 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
31168 field a pointer to garbage collected vector, mark lattices and
31169 ipcp_orig_node with GTY((skip)).
31170 (ipa_get_param_count): Adjust to descriptors being a pointer.
31171 (ipa_get_param): Likewise.
31172 (ipa_get_type): Likewise.
31173 (ipa_get_param_move_cost): Likewise.
31174 (ipa_set_param_used): Likewise.
31175 (ipa_get_controlled_uses): Likewise.
31176 (ipa_set_controlled_uses): Likewise.
31177 (ipa_is_param_used): Likewise.
31178 (ipa_node_params_t): Move into garbage collector. New methods insert
31180 (ipa_node_params_sum): Annotate wth GTY(()).
31181 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
31183 (ipa_load_from_parm_agg): Adjust declaration.
31184 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
31185 * ipa-profile.c (ipa_profile): Likewise.
31186 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
31187 (ipa_populate_param_decls): Make descriptors parameter garbage
31189 (ipa_dump_param): Adjust to descriptors being a pointer.
31190 (ipa_alloc_node_params): Likewise.
31191 (ipa_initialize_node_params): Likewise.
31192 (load_from_param_1): Make descriptors parameter garbage collected.
31193 (load_from_unmodified_param): Likewise.
31194 (load_from_param): Likewise.
31195 (ipa_load_from_parm_agg): Likewise.
31196 (ipa_node_params::~ipa_node_params): Removed.
31197 (ipa_free_all_node_params): Remove call to delete operator.
31198 (ipa_node_params_t::insert): New.
31199 (ipa_node_params_t::remove): Likewise.
31200 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
31201 copy known_csts and known_contexts vectors.
31202 (ipa_read_node_info): Adjust to descriptors being a pointer.
31203 (ipcp_modif_dom_walker): Make m_descriptors field garbage
31205 (ipcp_transform_function): Make descriptors variable garbage
31208 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
31210 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
31211 * config/i386/avx512dqintrin.h: Ditto.
31212 * config/i386/avx512fintrin.h: Ditto.
31213 * config/i386/i386.c: Handle new builtins.
31214 * config/i386/i386-builtin.def: Add new builtins.
31215 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
31216 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
31218 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31219 Martin Liska <mliska@suse.cz>
31221 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
31222 * asan.c (asan_expand_poison_ifn): Support stores and use
31223 appropriate ASAN report function.
31224 * internal-fn.c (expand_ASAN_POISON_USE): New function.
31225 * internal-fn.def (ASAN_POISON_USE): Declare.
31226 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
31227 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
31228 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
31229 ASAN_POISON calls w/o LHS.
31230 * tree-ssa.c (execute_update_addresses_taken): Create clobber
31231 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
31232 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
31233 * gimplify.c (asan_poison_variables): Add attribute
31234 use_after_scope_memory to variables that really needs to live
31236 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
31237 having the attribute.
31239 2017-01-23 Martin Liska <mliska@suse.cz>
31241 * asan.c (create_asan_shadow_var): New function.
31242 (asan_expand_poison_ifn): Likewise.
31243 * asan.h (asan_expand_poison_ifn): New declaration.
31244 * internal-fn.c (expand_ASAN_POISON): Likewise.
31245 * internal-fn.def (ASAN_POISON): New builtin.
31246 * sanopt.c (pass_sanopt::execute): Expand
31247 asan_expand_poison_ifn.
31248 * tree-inline.c (copy_decl_for_dup_finish): Make function
31250 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
31251 * tree-ssa.c (is_asan_mark_p): New function.
31252 (execute_update_addresses_taken): Rewrite local variables
31253 (identified just by use-after-scope as addressable) into SSA.
31255 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31257 * doc/install.texi (Specific): opensource.apple.com uses https
31258 now. Remove trailing slash.
31260 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31262 * README.Portability: Remove note on an Irix compatibility issue.
31264 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
31266 * gcov.c (INCLUDE_ALGORITHM): Define.
31267 (INCLUDE_VECTOR): Define.
31268 No longer include <vector> and <algorithm> directly.
31270 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
31272 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
31274 * doc/invoke.texi (Code Gen Options): Ditto.
31276 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
31279 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
31281 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
31283 rtl-optimization/79125
31284 * cprop.c (local_cprop_pass): Handle cases where we make an
31285 unconditional trap.
31287 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
31291 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
31292 read from, for big endian.
31294 2017-01-20 Jiong Wang <jiong.wang@arm.com>
31296 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
31297 register pauth builtins for LP64 only.
31299 2017-01-20 Marek Polacek <polacek@redhat.com>
31302 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
31305 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
31307 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
31309 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
31310 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
31311 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
31313 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31316 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
31317 in big-endian mode when they are not a single duplicated value.
31319 2017-01-20 Richard Biener <rguenther@suse.de>
31321 * BASE-VER: Bump to 7.0.1.
31323 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
31325 * omp-low.c (omplow_simd_context): New struct. Use it...
31326 (lower_rec_simd_input_clauses): ...here and...
31327 (lower_rec_input_clauses): ...here to hold common data. Adjust all
31328 references to idx, lane, max_vf, is_simt.
31330 2017-01-20 Graham Markall <graham.markall@embecosm.com>
31332 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
31335 2017-01-20 Martin Jambor <mjambor@suse.cz>
31337 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
31338 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
31340 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
31341 (GTFILES): Rename hsa.c to hsa-common.c.
31342 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
31343 * hsa-dump.c: Likewise.
31344 * hsa-gen.c: Likewise.
31345 * hsa-regalloc.c: Likewise.
31346 * ipa-hsa.c: Likewise.
31347 * omp-expand.c: Likewise.
31348 * omp-low.c: Likewise.
31349 * toplev.c: Likewise.
31351 2017-01-20 Marek Polacek <polacek@redhat.com>
31354 * doc/invoke.texi: Document -Wduplicated-branches.
31355 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
31356 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
31357 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
31358 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
31359 return 0 only when not OEP_LEXICOGRAPHIC.
31360 (fold_build_cleanup_point_expr): Use the expression
31361 location when building CLEANUP_POINT_EXPR.
31362 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
31363 * tree.c (add_expr): Handle error_mark_node.
31365 2017-01-20 Martin Liska <mliska@suse.cz>
31368 * tree-profile.c (init_ic_make_global_vars): Do not call
31370 (gimple_init_gcov_profiler): Likewise.
31372 2017-01-20 Martin Liska <mliska@suse.cz>
31375 * cgraph.h (maybe_create_reference): Remove argument and
31377 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
31379 * ipa-cp.c (create_specialized_node): Likewise.
31380 * symtab.c (symtab_node::maybe_create_reference): Handle
31381 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
31383 2017-01-20 Martin Liska <mliska@suse.cz>
31385 * read-rtl-function.c (function_reader::create_function): Use
31386 build_decl instread of build_decl_stat.
31388 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
31390 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
31391 * config/i386/avx512dqintrin.h: Ditto.
31392 * config/i386/avx512fintrin.h: Ditto.
31393 * config/i386/i386-builtin-types.def: Add new types.
31394 * config/i386/i386.c: Handle new types.
31395 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
31396 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
31397 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
31398 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
31399 (__builtin_ia32_kshiftridi): New.
31400 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
31402 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
31406 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
31407 define to rs6000_init_stack_protect_guard.
31408 (rs6000_init_stack_protect_guard): New function.
31410 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
31411 Yunqiang Su <yunqiang.su@imgtec.com>
31413 * config.gcc (supported_defaults): Add madd4.
31414 (with_madd4): Add validation.
31415 (all_defaults): Add madd4.
31416 * config/mips/mips.opt (mmadd4): New option.
31417 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
31419 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
31421 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
31422 (ISA_HAS_FUSED_MADD4): Likewise.
31423 * doc/invoke.texi (-mmadd4): Document the new option.
31424 * doc/install.texi (--with-madd4): Document the new option.
31426 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31428 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
31429 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
31430 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
31431 (aarch64_init_pauth_hint_builtins): New.
31432 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
31433 (aarch64_expand_builtin): Expand new builtins.
31435 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31437 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
31438 * combine-stack-adj.c (no_unhandled_cfa): Handle
31439 REG_CFA_TOGGLE_RA_MANGLE.
31440 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
31441 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
31442 info for return address signing.
31443 (aarch64_expand_epilogue): Likewise.
31445 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31447 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
31448 * config/aarch64/aarch64-protos.h
31449 (aarch64_return_address_signing_enabled): New declaration.
31450 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
31452 (aarch64_expand_prologue): Sign return address before it's pushed onto
31454 (aarch64_expand_epilogue): Authenticate return address fetched from
31456 (aarch64_override_options): Sanity check for ILP32 and ISA level.
31457 (aarch64_attributes): New function attributes for "sign-return-address".
31458 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
31459 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
31460 ("*do_return"): Generate combined instructions according to key index.
31461 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
31462 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
31464 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
31465 * config/aarch64/aarch64.opt (msign-return-address=): New.
31466 * doc/extend.texi (AArch64 Function Attributes): Documents
31467 "sign-return-address=".
31468 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
31470 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
31472 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
31473 overall option summary.
31475 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31477 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
31478 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
31479 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
31480 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
31482 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
31484 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
31485 -mpower9-minmax by default for -mcpu=power9.
31486 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
31487 128-bit floating point.
31489 2017-01-20 Alan Modra <amodra@gmail.com>
31491 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
31492 optimizing for size.
31494 2017-01-20 Alan Modra <amodra@gmail.com>
31497 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
31498 for strcmp and strncmp from corresponding builtin decl.
31500 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
31502 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
31503 instead of i386/rtems-64.h.
31504 * config/i386/rtems-64.h: Remove.
31506 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
31510 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
31512 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
31514 2017-01-19 Tamar Christina <tamar.christina@arm.com>
31516 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
31517 Change int to HOST_WIDE_INT.
31518 * config/aarch64/aarch64-protos.h
31519 (aarch64_simd_gen_const_vector_dup): Likewise.
31520 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
31522 2017-01-19 David Malcolm <dmalcolm@redhat.com>
31524 * langhooks-def.h (lhd_type_for_size): New decl.
31525 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
31526 * langhooks.c (lhd_type_for_size): New function, taken from
31529 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
31531 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
31532 define_bypass for CR latency.
31533 (power9-cracked-alu): Update bypass latency and remove power9-branch.
31534 (power9-alu2): Add define_bypass for CR latency.
31536 (power9-mul): Update insn latency.
31537 (power9-mul-compare): Update insn latency, bypass latency and remove
31540 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31542 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
31544 * config/aarch64/aarch64.md
31545 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
31546 aarch64_nopcrelative_literal_loads.
31547 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
31549 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
31551 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
31552 TARGET_LOONGSON_3A.
31553 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
31555 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
31558 * config.gcc (supported_defaults): Add lxc1-sxc1.
31559 (with_lxc1_sxc1): Add validation.
31560 (all_defaults): Add lxc1-sxc1.
31561 * config/mips/mips.opt (mlxc1-sxc1): New option.
31562 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
31564 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
31565 __mips_no_lxc1_sxc1.
31566 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
31567 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
31568 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
31570 2017-01-19 Richard Biener <rguenther@suse.de>
31572 PR tree-optimization/72488
31573 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
31574 sure to restore SSA info.
31575 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
31577 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
31579 PR rtl-optimization/79121
31580 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
31581 of the inner type when shifting an extended value.
31583 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
31586 * symtab.c (symtab_node::equal_address_to): Fix comparing of
31587 interposable aliases.
31589 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
31592 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
31593 Use the evmergelohi instruction.
31594 (mov_si<mode>_e500_subreg4_2_le): Likewise.
31595 (mov_sitf_e500_subreg8_2_be): Likewise.
31596 (mov_sitf_e500_subreg12_2_le): Likewise.
31597 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
31598 (mov_si<mode>_e500_subreg4_2_be): Likewise.
31599 (mov_sitf_e500_subreg8_2_le): Likewise.
31600 (mov_sitf_e500_subreg12_2_be): Likewise.
31602 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31604 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
31605 attribute from vecsimple to vecperm.
31606 (altivec_vbpermq2): Likewise.
31608 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31611 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
31613 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31614 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
31615 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
31616 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
31617 case where N arg is SIZE_MAX.
31618 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
31619 (cmpstrsi): Add pattern.
31621 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
31623 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
31624 __builtin_vec_revb builtins.
31625 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
31626 built-in functions to support generation of the ISA 3.0 XXBR<x>
31627 vector byte reverse instructions.
31628 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
31629 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
31630 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
31631 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
31632 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
31633 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
31634 (P9V_BUILTIN_VEC_REVB): Likewise.
31635 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
31636 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
31637 (p9_xxbrq_v16qi): Likewise.
31638 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
31639 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
31640 (p9_xxbrh_v8hi): Likewise.
31641 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
31642 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
31643 vec_revb built-in functions.
31645 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
31647 PR rtl-optimization/78952
31648 * config/i386/i386.md (any_extract): New code iterator.
31649 (*insvqi_2): Use any_extract for source operand.
31650 (*insvqi_3): Use any_shiftrt for source operand.
31652 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
31654 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
31656 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
31658 2017-01-18 Matthias Klose <doko@ubuntu.com>
31660 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
31662 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31664 * config/rs6000/altivec.h (vec_bperm): Change #define.
31665 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
31666 (altivec_vbpermq2): New define_insn.
31667 (altivec_vbpermd): Likewise.
31668 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
31669 function interface.
31670 (VBPERMD): Likewise.
31671 (VBPERM): New polymorphic function interface.
31672 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
31673 Add entries for P9V_BUILTIN_VEC_VBPERM.
31674 * doc/extend.texi: Add interfaces for vec_bperm.
31676 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31678 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
31679 first letter of error messages.
31680 (s390_resolve_overloaded_builtin): Likewise.
31681 * config/s390/s390.c (s390_expand_builtin): Likewise.
31682 (s390_invalid_arg_for_unprototyped_fn): Likewise.
31683 (s390_valid_target_attribute_inner_p): Likewise.
31684 * config/s390/s390.md ("tabort"): Likewise.
31686 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
31688 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
31689 (ISA_AVOID_DIV_HILO): New macro.
31690 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
31691 (ISA_HAS_DDIV): Likewise.
31693 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
31695 * doc/invoke.texi (fabi-version): Correct number of occurrences.
31697 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
31699 * doc/invoke.texi (fabi-version): Spelling fix.
31701 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
31704 * doc/invoke.texi (fabi-version): Mention mangling fix for
31707 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
31710 * doc/invoke.texi (fabi-version): Document discriminator mangling.
31712 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
31715 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
31716 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
31718 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
31720 (stack_protect_test): Ditto.
31721 * config/rs6000/rs6000.opt (mstack-protector-guard=,
31722 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
31724 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
31725 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
31726 -mstack-protector-guard-offset=.
31727 (RS/6000 and PowerPC Options): Ditto.
31729 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
31731 * config/i386/i386.h (MASK_CLASS_P): New define.
31732 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
31733 there are no registers from different register sets also when
31734 mask registers are used. Update function comment.
31735 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
31736 to (*k/*r) and (*k/*km) alternatives.
31738 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
31740 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
31741 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
31742 (EH_RETURN_HANDLER_RTX): New define.
31743 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
31744 Force frame pointer in EH return functions.
31745 (aarch64_expand_epilogue): Add barrier for eh_return.
31746 (aarch64_final_eh_return_addr): Remove.
31747 (aarch64_eh_return_handler_rtx): New function.
31748 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
31750 (aarch64_eh_return_handler_rtx): New prototype.
31752 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31754 * config/rs6000/altivec.h (vec_rlmi): New #define.
31755 (vec_vrlnm): Likewise.
31756 (vec_rlnm): Likewise.
31757 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
31758 (UNSPEC_VRLNM): Likewise.
31759 (VIlong): New mode iterator.
31760 (altivec_vrl<VI_char>mi): New define_insn.
31761 (altivec_vrl<VI_char>nm): Likewise.
31762 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
31764 (VRLDNM): Likewise.
31765 (RLNM): New polymorphic function entry.
31766 (VRLWMI): New monomorphic function entry.
31767 (VRLDMI): Likewise.
31768 (RLMI): New polymorphic function entry.
31769 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
31770 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
31771 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
31774 2017-01-17 Jakub Jelinek <jakub@redhat.com>
31777 * dwarf2out.c (field_byte_offset): Restore the
31778 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
31779 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
31780 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
31783 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
31786 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
31788 2017-01-17 Jakub Jelinek <jakub@redhat.com>
31791 * dwarf2out.c (add_data_member_location_attribute): For constant
31792 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
31793 instead of DW_AT_data_member_location, DW_AT_bit_offset and
31794 DW_AT_byte_size attributes.
31796 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
31798 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
31799 after forcing to constant memory when the code model is medium.
31801 2017-01-17 Julia Koval <julia.koval@intel.com>
31804 * config/i386/avx512fintrin.h
31805 (_mm512_i32gather_ps): Change __addr type to void const*.
31806 (_mm512_mask_i32gather_ps): Ditto.
31807 (_mm512_i32gather_pd): Ditto.
31808 (_mm512_mask_i32gather_pd): Ditto.
31809 (_mm512_i64gather_ps): Ditto.
31810 (_mm512_mask_i64gather_ps): Ditto.
31811 (_mm512_i64gather_pd): Ditto.
31812 (_mm512_mask_i64gather_pd): Ditto.
31813 (_mm512_i32gather_epi32): Ditto.
31814 (_mm512_mask_i32gather_epi32): Ditto.
31815 (_mm512_i32gather_epi64): Ditto.
31816 (_mm512_mask_i32gather_epi64): Ditto.
31817 (_mm512_i64gather_epi32): Ditto.
31818 (_mm512_mask_i64gather_epi32): Ditto.
31819 (_mm512_i64gather_epi64): Ditto.
31820 (_mm512_mask_i64gather_epi64): Ditto.
31821 (_mm512_i32scatter_ps): Change __addr type to void*.
31822 (_mm512_mask_i32scatter_ps): Ditto.
31823 (_mm512_i32scatter_pd): Ditto.
31824 (_mm512_mask_i32scatter_pd): Ditto.
31825 (_mm512_i64scatter_ps): Ditto.
31826 (_mm512_mask_i64scatter_ps): Ditto.
31827 (_mm512_i64scatter_pd): Ditto.
31828 (_mm512_mask_i64scatter_pd): Ditto.
31829 (_mm512_i32scatter_epi32): Ditto.
31830 (_mm512_mask_i32scatter_epi32): Ditto.
31831 (_mm512_i32scatter_epi64): Ditto.
31832 (_mm512_mask_i32scatter_epi64): Ditto.
31833 (_mm512_i64scatter_epi32): Ditto.
31834 (_mm512_mask_i64scatter_epi32): Ditto.
31835 (_mm512_i64scatter_epi64): Ditto.
31836 (_mm512_mask_i64scatter_epi64): Ditto.
31837 * config/i386/avx512pfintrin.h
31838 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
31839 (_mm512_mask_prefetch_i32gather_ps): Ditto.
31840 (_mm512_mask_prefetch_i64gather_pd): Ditto.
31841 (_mm512_mask_prefetch_i64gather_ps): Ditto.
31842 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
31843 (_mm512_prefetch_i32scatter_ps): Ditto.
31844 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
31845 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
31846 (_mm512_prefetch_i64scatter_pd): Ditto.
31847 (_mm512_prefetch_i64scatter_ps): Ditto.
31848 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
31849 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
31850 * config/i386/avx512vlintrin.h
31851 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
31852 (_mm_mmask_i32gather_ps): Ditto.
31853 (_mm256_mmask_i32gather_pd): Ditto.
31854 (_mm_mmask_i32gather_pd): Ditto.
31855 (_mm256_mmask_i64gather_ps): Ditto.
31856 (_mm_mmask_i64gather_ps): Ditto.
31857 (_mm256_mmask_i64gather_pd): Ditto.
31858 (_mm_mmask_i64gather_pd): Ditto.
31859 (_mm256_mmask_i32gather_epi32): Ditto.
31860 (_mm_mmask_i32gather_epi32): Ditto.
31861 (_mm256_mmask_i32gather_epi64): Ditto.
31862 (_mm_mmask_i32gather_epi64): Ditto.
31863 (_mm256_mmask_i64gather_epi32): Ditto.
31864 (_mm_mmask_i64gather_epi32): Ditto.
31865 (_mm256_mmask_i64gather_epi64): Ditto.
31866 (_mm_mmask_i64gather_epi64): Ditto.
31867 (_mm256_i32scatter_ps): Change __addr type to void*.
31868 (_mm256_mask_i32scatter_ps): Ditto.
31869 (_mm_i32scatter_ps): Ditto.
31870 (_mm_mask_i32scatter_ps): Ditto.
31871 (_mm256_i32scatter_pd): Ditto.
31872 (_mm256_mask_i32scatter_pd): Ditto.
31873 (_mm_i32scatter_pd): Ditto.
31874 (_mm_mask_i32scatter_pd): Ditto.
31875 (_mm256_i64scatter_ps): Ditto.
31876 (_mm256_mask_i64scatter_ps): Ditto.
31877 (_mm_i64scatter_ps): Ditto.
31878 (_mm_mask_i64scatter_ps): Ditto.
31879 (_mm256_i64scatter_pd): Ditto.
31880 (_mm256_mask_i64scatter_pd): Ditto.
31881 (_mm_i64scatter_pd): Ditto.
31882 (_mm_mask_i64scatter_pd): Ditto.
31883 (_mm256_i32scatter_epi32): Ditto.
31884 (_mm256_mask_i32scatter_epi32): Ditto.
31885 (_mm_i32scatter_epi32): Ditto.
31886 (_mm_mask_i32scatter_epi32): Ditto.
31887 (_mm256_i32scatter_epi64): Ditto.
31888 (_mm256_mask_i32scatter_epi64): Ditto.
31889 (_mm_i32scatter_epi64): Ditto.
31890 (_mm_mask_i32scatter_epi64): Ditto.
31891 (_mm256_i64scatter_epi32): Ditto.
31892 (_mm256_mask_i64scatter_epi32): Ditto.
31893 (_mm_i64scatter_epi32): Ditto.
31894 (_mm_mask_i64scatter_epi32): Ditto.
31895 (_mm256_i64scatter_epi64): Ditto.
31896 (_mm256_mask_i64scatter_epi64): Ditto.
31897 (_mm_i64scatter_epi64): Ditto.
31898 (_mm_mask_i64scatter_epi64): Ditto.
31899 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
31900 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
31901 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
31902 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
31903 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
31904 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
31905 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
31906 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
31907 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
31908 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
31909 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
31910 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
31911 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
31912 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
31913 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
31914 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
31915 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
31916 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
31917 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
31918 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
31919 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
31920 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
31921 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
31922 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
31923 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
31924 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
31925 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
31926 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
31927 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
31928 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
31929 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
31930 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
31931 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
31932 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
31933 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
31934 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
31935 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
31936 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
31937 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
31938 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
31939 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
31940 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
31941 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
31942 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
31943 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
31944 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
31945 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
31946 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
31947 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
31948 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
31949 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
31950 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
31951 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
31952 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
31953 definitions accordingly.
31955 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
31956 Kuan-Lin Chen <kuanlinchentw@gmail.com>
31959 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
31962 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
31965 * ira-conflicts.c (ira_build_conflicts): Update total conflict
31966 hard regs for inner regno.
31968 2017-01-17 Martin Liska <mliska@suse.cz>
31971 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
31972 assumption and add comment.
31974 2017-01-17 Nathan Sidwell <nathan@acm.org>
31976 * ipa-visibility.c (localize_node): New function, broken out of ...
31977 (function_and_variable_visibility): ... here. Call it.
31979 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
31981 PR middle-end/77445
31982 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
31983 correctly set frequency of oudgoing edge.
31984 (duplicate_thread_path): Fix profile updating.
31986 2017-01-17 Jakub Jelinek <jakub@redhat.com>
31989 * configure.ac: Add GCC_BASE_VER.
31990 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
31991 version from BASE-VER file.
31992 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
31993 (gcc.o): Depend on $(BASEVER).
31994 * common.opt (dumpfullversion): New option.
31995 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
31996 * doc/invoke.texi: Document -dumpfullversion.
31997 * doc/install.texi: Document --with-gcc-major-version-only.
31998 * configure: Regenerated.
32000 2017-01-17 Richard Biener <rguenther@suse.de>
32002 PR tree-optimization/71433
32003 * tree-vrp.c (register_new_assert_for): Merge same asserts
32004 on all incoming edges.
32005 (process_assert_insertions_for): Handle insertions at the
32008 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
32010 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
32011 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
32013 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
32016 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
32019 2017-01-17 Alan Modra <amodra@gmail.com>
32022 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
32023 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
32024 symbolic stack limit when pic.
32026 2017-01-16 Martin Sebor <msebor@redhat.com>
32028 PR tree-optimization/78608
32029 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
32031 2017-01-16 Jeff Law <law@redhat.com>
32034 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
32035 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
32036 for several include directories that may be relative to sysroot.
32037 * config/i386/x-mingw32 (gplus_includedir): Define.
32038 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
32039 (native_system_includedir): Likewise.
32040 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
32041 override if TARGET_SYSTEM_ROOT is defined.
32042 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
32044 PR tree-optimization/79090
32045 PR tree-optimization/33562
32046 PR tree-optimization/61912
32047 PR tree-optimization/77485
32048 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
32049 and computed trims into the dump file.
32051 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32053 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
32055 2017-01-16 Jakub Jelinek <jakub@redhat.com>
32058 * gimplify.c (gimplify_init_constructor): If want_value and
32059 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
32063 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
32064 sequence. Formatting fixes.
32065 (doloop_optimize): Formatting fixes.
32068 * gcc.c (debug_level_greater_than_spec_func): New function.
32069 (static_spec_functions): Add debug-level-gt spec function.
32070 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
32072 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32073 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32074 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
32075 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
32076 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
32077 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
32079 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
32081 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
32082 QImode fixups to general and mask registers only.
32084 2017-01-16 Carl Love <cel@us.ibm.com>
32086 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
32087 for built-in functions
32088 vector signed char vec_nabs (vector signed char)
32089 vector signed short vec_nabs (vector signed short)
32090 vector signed int vec_nabs (vector signed int)
32091 vector signed long long vec_nabs (vector signed long long)
32092 vector float vec_nabs (vector float)
32093 vector double vec_nabs (vector double)
32094 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
32096 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
32097 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
32098 * doc/extend.texi: Update the documentation file for the new built-in
32101 2017-01-16 Martin Sebor <msebor@redhat.com>
32103 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
32106 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32108 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
32109 UNSPEC_VSX__XXSPLTD to require special splat handling.
32111 2017-01-16 David Malcolm <dmalcolm@redhat.com>
32114 * system.h: Poison strndup.
32116 2017-01-16 Alan Modra <amodra@gmail.com>
32119 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
32122 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
32124 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
32126 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
32128 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
32129 call recog here. Assert that INSN_CODE (insn) is non-negative.
32131 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32134 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
32136 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
32137 in the currently scheduled RTL fragment.
32139 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32141 PR rtl-optimization/78751
32142 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
32145 2017-01-14 Jeff Law <law@redhat.com>
32147 PR tree-optimization/79090
32148 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
32149 variable length stores.
32150 (compute_trims): Delete dead assignment to *trim_tail.
32151 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
32154 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
32156 PR rtl-optimization/78626
32157 PR rtl-optimization/78727
32158 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
32159 of a block, and split such blocks after everything else is finished.
32161 2017-01-14 Alan Modra <amodra@gmail.com>
32164 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
32165 target legitimate_combined_insn.
32166 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
32167 (rs6000_legitimate_combined_insn): New function.
32168 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
32170 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
32171 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
32172 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
32174 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
32176 * doc/frontends.texi (G++ and GCC): Remove references to Java.
32178 2017-01-13 Jeff Law <law@redhat.com>
32180 PR tree-optimization/33562
32181 PR tree-optimization/61912
32182 PR tree-optimization/77485
32183 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
32185 (delete_dead_assignment): Likewise.
32186 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
32187 statement to delete_dead_call and delete_dead_assignment.
32189 2017-01-13 David Malcolm <dmalcolm@redhat.com>
32192 * substring-locations.c (format_warning_va): Strengthen case 1 so
32193 that both endpoints of the substring must be within the format
32194 range for just the substring to be printed.
32196 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
32198 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
32199 * config/i386/i386.c (ix86_target_string): Add missing options
32200 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
32201 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
32202 flags_other and ix86_target_other to flags2_other. Display unknown
32204 (ix86_valid_target_attribute_inner_p): Add missing options and
32205 reorder options by implied ISAs, as in ix86_target_string.
32207 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32209 * hash-table.h (hash_table::too_empty_p): New function.
32210 (hash_table::expand): Use it.
32211 (hash_table::traverse): Likewise.
32212 (hash_table::empty_slot): Use sizeof (value_type) instead of
32213 sizeof (PTR) to convert bytes to elements. Shrink the table
32214 if the current size is excessive for the current number of
32217 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32219 * ira-costs.c (record_reg_classes): Break from the inner loop
32220 early once alt_fail is known to be true. Update outer loop
32221 handling accordingly.
32223 2017-01-13 Jeff Law <law@redhat.com>
32225 * tree-ssa-dse.c (decrement_count): New function.
32226 (increment_start_addr, maybe_trim_memstar_call): Likewise.
32227 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
32228 when we know the partially dead statement is a mem* function.
32230 PR tree-optimization/61912
32231 PR tree-optimization/77485
32232 * tree-ssa-dse.c: Include expr.h.
32233 (maybe_trim_constructor_store): New function.
32234 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
32236 PR tree-optimization/33562
32237 PR tree-optimization/61912
32238 PR tree-optimization/77485
32239 * doc/invoke.texi: Document new dse-max-object-size param.
32240 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
32241 * tree-ssa-dse.c: Include params.h.
32242 (dse_store_status): New enum.
32243 (initialize_ao_ref_for_dse): New, partially extracted from
32245 (valid_ao_ref_for_dse, normalize_ref): New.
32246 (setup_live_bytes_from_ref, compute_trims): Likewise.
32247 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
32248 (maybe_trim_partially_dead_store): Likewise.
32249 (maybe_trim_complex_store): Likewise.
32250 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
32251 Track what bytes live from the original store. Return tri-state
32252 for dead, partially dead or live.
32253 (dse_dom_walker): Add constructor, destructor and new private members.
32254 (delete_dead_call, delete_dead_assignment): New extracted from
32256 (dse_optimize_stmt): Make a member of dse_dom_walker.
32257 Use initialize_ao_ref_for_dse.
32259 PR tree-optimization/33562
32260 PR tree-optimization/61912
32261 PR tree-optimization/77485
32262 * sbitmap.h (bitmap_count_bits): Prototype.
32263 (bitmap_clear_range, bitmap_set_range): Likewise.
32264 * sbitmap.c (bitmap_clear_range): New function.
32265 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
32267 2017-01-13 Martin Liska <mliska@suse.cz>
32270 * function.c (set_cfun): Add new argument force.
32271 * function.h (set_cfun): Likewise.
32272 * ipa-inline-transform.c (inline_call): Use the function when
32273 strict alising from is dropped for function we inline to.
32275 2017-01-13 Richard Biener <rguenther@suse.de>
32277 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
32278 for dumping GIMPLE INTEGER_CSTs.
32280 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32282 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
32283 to 201112L since C++17.
32285 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
32288 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
32289 if -fsanitize=kernel-address is present.
32291 2017-01-13 Richard Biener <rguenther@suse.de>
32293 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
32294 as _Literal ( type ) number in case usual suffixes do not
32295 preserve all information.
32297 2017-01-13 Richard Biener <rguenther@suse.de>
32299 PR tree-optimization/77283
32300 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
32301 and ssa-iterators.h.
32302 (is_feasible_trace): Implement a cost model based on joiner
32305 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
32308 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
32309 char or short to __float128/_Float128 directly.
32311 2017-01-12 Martin Sebor <msebor@redhat.com>
32313 to -Wformat-overflow.
32314 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
32315 (min_bytes_remaining): Same.
32316 (get_string_length): Same.
32317 (format_string): Same.
32318 (format_directive): Same.
32320 (pass_sprintf_length::handle_gimple_call): Same.
32322 2017-01-12 Jakub Jelinek <jakub@redhat.com>
32324 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
32325 info.nowrite calls with no lhs that can't throw. Return bool
32326 whether gsi_remove has been called or not.
32327 (pass_sprintf_length::handle_gimple_call): Return bool whether
32328 try_substitute_return_value called gsi_remove. Formatting fix.
32329 (pass_sprintf_length::execute): Don't use gsi_remove if
32330 handle_gimple_call returned true.
32333 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
32334 be removed due to side-effects, don't remove following barrier nor
32335 turn the successor edge into fallthru edge.
32337 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32340 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
32341 element-reversing loads and stores as not swappable.
32343 2017-01-12 Nathan Sidwell <nathan@acm.org>
32344 Nicolai Stange <nicstange@gmail.com>
32346 * combine.c (try_combine): Don't ignore result of overlap checking
32347 loop. Combine overlap & asm check into single loop.
32349 2017-01-12 Richard Biener <rguenther@suse.de>
32351 * tree-pretty-print.c (dump_generic_node): Provide -gimple
32352 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
32354 2017-01-12 Richard Biener <rguenther@suse.de>
32356 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
32357 and TS_TARGET_OPTION directly derive from TS_BASE.
32358 * tree-core.h (tree_optimization_option): Derive from tree_base.
32359 (tree_target_option): Likewise.
32361 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
32363 * config/i386/i386.c (memory_address_length): Increase len
32364 only when rip_relative_addr_p returns false.
32366 2017-01-11 Julia Koval <julia.koval@intel.com>
32368 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
32369 (OPTION_MASK_ISA_SGX_SET): New.
32370 (ix86_handle_option): Handle OPT_msgx.
32371 * config.gcc: Added sgxintrin.h.
32372 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
32373 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
32374 * config/i386/i386.c (ix86_target_string): Add -msgx.
32376 (ix86_option_override_internal): Handle new options.
32377 (ix86_valid_target_attribute_inner_p): Add sgx.
32378 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
32379 * config/i386/i386.opt: Add msgx.
32380 * config/i386/sgxintrin.h: New file.
32381 * config/i386/x86intrin.h: Add sgxintrin.h.
32383 2017-01-11 Jakub Jelinek <jakub@redhat.com>
32386 * fold-const.c (maybe_nonzero_address): Return 1 for function
32388 (tree_single_nonzero_warnv_p): Don't handle function local objects
32392 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
32395 2017-01-11 David Malcolm <dmalcolm@redhat.com>
32398 * opts.c: Include "spellcheck.h"
32399 (struct string_fragment): New struct.
32400 (struct edit_distance_traits<const string_fragment &>): New
32402 (get_closest_sanitizer_option): New function.
32403 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
32405 2017-01-11 Jakub Jelinek <jakub@redhat.com>
32407 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
32409 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
32410 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
32411 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
32412 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
32413 for initial die_offset if dwarf_split_debug_info.
32414 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
32415 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
32417 (output_skeleton_debug_sections): Formatting fix. Use
32418 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
32419 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
32421 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
32423 * config/arm/cortex-a53.md: Add bypasses for
32424 cortex_a53_r2f_cvt.
32425 (cortex_a53_r2f): Only use for transfers.
32426 (cortex_a53_f2r): Likewise.
32427 (cortex_a53_r2f_cvt): Add reservation for conversions.
32428 (cortex_a53_f2r_cvt): Likewise.
32430 2017-01-11 Tamar Christina <tamar.christina@arm.com>
32432 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
32433 to all inlined functions, change static to extern.
32435 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
32438 * config/arm/arm.c (legitimize_pic_address): Handle reference to
32440 (arm_assemble_integer): Likewise.
32442 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
32444 * config.gcc: Use new awk script to check CPU, FPU and architecture
32445 parameters for --with-... options.
32446 * config/arm/parsecpu.awk: New file
32447 * config/arm/arm-cpus.in: New file.
32448 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
32450 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
32452 * config/arm/t-arm: Update dependency rules.
32453 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
32454 of processing .def files.
32455 * config/arm/genopt.sh: Deleted.
32456 * config/arm/gentune.sh: Deleted.
32457 * config/arm/arm-cores.def: Deleted.
32458 * config/arm/arm-arches.def: Deleted.
32459 * config/arm/arm-fpus.def: Deleted.
32460 * config/arm/arm-tune.md: Regenerated.
32461 * config/arm/arm-tables.opt: Regenerated.
32462 * config/arm/arm-cpu.h: New generated file.
32463 * config/arm/arm-cpu-data.h: New generated file.
32464 * config/arm/arm-cpu-cdata.h: New generated file.
32466 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
32469 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
32471 (input_varpool_node): Unpack dynamically_initialized bit.
32473 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
32475 PR rtl-optimization/79032
32476 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
32477 the alignment of the adjusted memory reference against that of MODE,
32478 instead of the alignment of the original memory reference.
32480 2017-01-11 Martin Jambor <mjambor@suse.cz>
32482 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
32484 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
32485 decorated functions.
32487 2017-01-11 Richard Biener <rguenther@suse.de>
32489 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
32490 set range/nonnull info for PHI results. Do not set it on
32491 stmts marked for removal.
32493 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
32495 * expr.c (store_field): In the bitfield case, fetch the return value
32496 from the registers before applying a single big-endian adjustment.
32497 Always do a final load for a BLKmode value not larger than a word.
32499 2017-01-10 David Malcolm <dmalcolm@redhat.com>
32502 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
32503 that we correctly handle column numbers greater than
32504 LINE_MAP_MAX_COLUMN_NUMBER.
32506 2017-01-10 Martin Sebor <msebor@redhat.com>
32508 PR middle-end/78245
32509 * gimple-ssa-sprintf.c (get_destination_size): Call
32510 {init,fini}object_sizes.
32511 * tree-object-size.c (addr_object_size): Adjust.
32512 (pass_through_call): Adjust.
32513 (pass_object_sizes::execute): Adjust.
32514 * tree-object-size.h (fini_object_sizes): Declare.
32516 2017-01-10 Martin Sebor <msebor@redhat.com>
32518 PR tree-optimization/78775
32519 * builtins.c (get_size_range): Move...
32520 * calls.c: ...to here.
32521 (alloc_max_size): Accept zero argument.
32522 (operand_signed_p): Remove.
32523 (maybe_warn_alloc_args_overflow): Call get_size_range.
32524 * calls.h (get_size_range): Declare.
32526 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
32528 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
32529 from TI's devices.csv file as of September 2016.
32530 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
32532 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
32534 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
32535 * doc/invoke.texi: Likewise.
32536 * doc/md.texi: Likewise.
32537 * doc/objc.texi: Likewise.
32539 2017-01-10 Joshua Conner <joshconner@google.com>
32541 * config/arm/fuchsia-elf.h: New file.
32542 * config/fuchsia.h: New file.
32543 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
32544 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
32546 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
32548 2016-01-10 Richard Biener <rguenther@suse.de>
32550 PR tree-optimization/79034
32551 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
32552 Propagate out degenerate PHIs in the joiner.
32554 2017-01-10 Martin Liska <mliska@suse.cz>
32556 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
32557 (sort_congruence_classes_by_decl_uid): Likewise.
32558 (sort_congruence_class_groups_by_decl_uid): Likewise.
32559 (sem_item_optimizer::merge_classes): Sort class, groups in these
32560 classes and members in the groups by DECL_UID of declarations.
32561 This would make merge operations stable.
32563 2017-01-10 Martin Liska <mliska@suse.cz>
32565 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
32566 usage of m_classes_vec.
32567 (sem_item_optimizer::~sem_item_optimizer): Likewise.
32568 (sem_item_optimizer::get_group_by_hash): Likewise.
32569 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
32570 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
32571 (sem_item_optimizer::verify_classes): Likewise.
32572 (sem_item_optimizer::process_cong_reduction): Likewise.
32573 (sem_item_optimizer::dump_cong_classes): Likewise.
32574 (sem_item_optimizer::merge_classes): Likewise.
32575 * ipa-icf.h (congruence_class_hash): Rename from
32576 congruence_class_group_hash. Remove declaration of m_classes_vec.
32578 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
32580 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
32581 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
32582 * config.gcc: Add avx512vpopcntdqintrin.h.
32583 * config/i386/avx512vpopcntdqintrin.h: New.
32584 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
32585 * config/i386/i386-builtin-types.def: Add new types.
32586 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
32587 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
32588 __builtin_ia32_vpopcountq_v8di_mask): New.
32589 * config/i386/i386-c.c (ix86_target_macros_internal): Define
32590 __AVX512VPOPCNTDQ__.
32591 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
32592 (PTA_AVX512VPOPCNTDQ): Define.
32593 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
32594 TARGET_AVX512VPOPCNTDQ_P): Define.
32595 * config/i386/i386.opt: Add mavx512vpopcntdq.
32596 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
32597 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
32599 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32601 PR middle-end/77484
32602 * predict.def (PRED_CALL): Set to 67.
32604 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
32606 * expr.c (store_field): In the bitfield case, if the value comes from
32607 a function call and is of an aggregate type returned in registers, do
32608 not modify the field mode; extract the value in all cases if the mode
32609 is BLKmode and the size is not larger than a word.
32611 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
32614 * config/i386/cpuid.h: Fix undefined behavior.
32616 2017-01-04 Jeff Law <law@redhat.com>
32618 PR tree-optimization/79007
32619 PR tree-optimization/67955
32620 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
32621 conservative for pt.null when flag_non_call_exceptions is on.
32623 2017-01-09 Jakub Jelinek <jakub@redhat.com>
32625 PR translation/79019
32626 PR translation/79020
32627 * params.def (PARAM_INLINE_MIN_SPEEDUP,
32628 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
32629 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
32631 * config/avr/avr.opt (maccumulate-args): Likewise.
32632 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
32633 * common.opt (freport-bug): Likewise.
32634 * cif-code.def (CIF_FINAL_ERROR): Likewise.
32635 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
32636 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
32637 translatable string.
32638 * config/i386/i386.c (function_value_32): Likewise.
32639 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
32640 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
32642 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
32643 * common/config/msp430/msp430-common.c (msp430_handle_option):
32645 * symtab.c (symtab_node::verify_base): Likewise.
32646 * opts.c (set_debug_level): Likewise.
32647 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
32648 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
32649 missing whitespace to translatable strings.
32650 * config/avr/avr.md (bswapsi2): Fix typo in comment.
32651 * config/sh/superh.h: Likewise.
32652 * config/i386/xopintrin.h: Likewise.
32653 * config/i386/znver1.md: Likewise.
32654 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
32655 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
32656 * double-int.h (struct double_int): Likewise.
32657 * double-int.c (div_and_round_double): Likewise.
32658 * wide-int.cc: Likewise.
32659 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
32660 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
32661 * cfgcleanup.c (crossjumps_occured): Renamed to ...
32662 (crossjumps_occurred): ... this.
32663 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
32666 PR tree-optimization/78899
32667 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
32668 returning bool return struct loop *, NULL for failure and the new
32670 (versionable_outer_loop_p): Don't version outer loop if it has
32671 dont_vectorized bit set.
32672 (tree_if_conversion): When versioning outer loop, ensure
32673 tree_if_conversion is performed also on the inner loop of the
32674 non-vectorizable outer loop copy.
32675 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
32676 LOOP_VECTORIZED in inner loop of the scalar outer loop and
32677 prevent vectorization of it.
32678 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
32679 the outer loop vectorization of the non-scalar version is attempted
32680 before vectorization of the inner loop in scalar version. If
32681 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
32682 vectorization of its inner loop.
32683 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
32684 has 2 inner loops, rename also on edges from bb whose single pred
32685 is outer_loop->header. Fix typo in function comment.
32687 2017-01-09 Martin Sebor <msebor@redhat.com>
32690 * asan.c (asan_emit_stack_protection): Increase local buffer size
32691 to avoid snprintf truncation warning.
32693 2017-01-09 Andrew Pinski <apinski@cavium.com>
32695 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
32696 to reference thunderx2t99 for the tuning structure
32697 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
32699 (thunderx2t99_extra_costs): This.
32700 * config/aarch64/aarch64-tune.md: Regenerate.
32701 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
32702 (vulcan_addrcost_table): This.
32703 (vulcan_regmove_cost): Rename to ...
32704 (thunderx2t99_regmove_cost): This.
32705 (vulcan_vector_cost): Rename to ...
32706 (thunderx2t99_vector_cost): this.
32707 (vulcan_branch_cost): Rename to ...
32708 (thunderx2t99_branch_cost): This.
32709 (vulcan_tunings): Rename to ...
32710 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
32711 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
32713 2017-01-09 Martin Jambor <mjambor@suse.cz>
32717 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
32718 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
32719 (propagate_vr_accross_jump_function): Use the above function for all
32720 value range computations for pass-through jump functions and type
32721 converasion from explicit value range values.
32722 (ipcp_propagate_stage): Do not attempt to deduce types of formal
32723 parameters from TYPE_ARG_TYPES.
32724 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
32725 (ipa_write_node_info): Stream type of the actual argument.
32726 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
32728 2017-01-09 Martin Liska <mliska@suse.cz>
32731 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
32732 (lookup_compiler): Do not show error message with have_E.
32734 2017-01-09 Jakub Jelinek <jakub@redhat.com>
32736 PR tree-optimization/78938
32737 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
32738 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
32739 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
32740 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
32743 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32745 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
32748 2017-01-09 Richard Biener <rguenther@suse.de>
32750 PR tree-optimization/78997
32751 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
32752 name condition properly.
32754 2017-01-09 Richard Biener <rguenther@suse.de>
32757 * dwarf2out.c (is_cxx): New overload with context.
32758 (is_naming_typedef_decl): Use it.
32760 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
32762 * invoke.texi (Option Summary): Correct spacing in option lists
32763 and add line breaks to fix over-long lines.
32765 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
32767 PR middle-end/17660
32769 * extend.texi (Common Variable Attributes): Add xref to GCC
32770 Internals manual to explain mode attribute keywords.
32772 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
32775 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
32776 and Preprocessor Options.
32777 (Options for Linking): Document -pthread here....
32778 (RS/6000 and PowerPC Options): ...not here.
32779 (Solaris 2 Options): ...or here.
32780 * doc/cppopts.texi: Document -pthread.
32782 2017-01-08 Martin Sebor <msebor@redhat.com>
32784 PR middle-end/77708
32785 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
32786 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
32787 New member functions.
32788 (format_directive): Used them.
32790 (pass_sprintf_length::handle_gimple_call): Same.
32791 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
32792 to avoid truncation for any argument.
32793 (extract_affine_mul): Same.
32794 * tree.c (get_file_function_name): Same.
32796 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32798 PR middle-end/77484
32799 * predict.def (PRED_INDIR_CALL): Set to 86.
32801 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
32803 PR preprocessor/54124
32804 * doc/cppopts.texi: Reformat -d subtable to list the full name
32805 of the options. Add cross-reference to the docs for the general
32806 compiler -d options.
32807 * doc/invoke.texi (Developer Options): Add cross-reference to the
32808 preprocessor-specific -d option documentation.
32810 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
32812 PR preprocessor/13498
32813 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
32814 redudant material, and reflect new command-line options.
32815 (System Headers): Likewise.
32817 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
32819 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
32820 -isystem, and -idirafter. Copy-edit.
32821 * doc/cppopts.texi: Copy-edit. Remove contradiction about
32822 default for -ftrack-macro-expansion. Delete obsolete and
32823 badly-formatted implementation details about -fdebug-cpp output.
32824 * doc/cppwarnopts.texi: Copy-edit.
32826 2017-01-07 David Malcolm <dmalcolm@redhat.com>
32829 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
32830 that the transition from a max line width >= 1<<10 to narrower
32831 lines works correctly.
32833 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
32835 * doc/options.texi (PerFunction): New.
32836 * opt-functions.awk (switch_flags): Map both Optimization and
32837 PerFunction to CL_OPTIMIZATION.
32838 * opth-gen.awk: Test for PerFunction flag along with
32840 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
32841 it only when the latter is present. Skip those that don't in
32842 the hash function generator.
32843 * common.opt (fvar-tracking): Mark as PerFunction instead of
32845 (fvar-tracking-assignments): Likewise.
32846 (fvar-tracking-assignments-toggle): Likewise.
32847 (fvar-tracking-uninit): Likewise.
32849 2017-01-07 Jakub Jelinek <jakub@redhat.com>
32851 PR translation/79018
32852 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
32855 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
32858 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
32859 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
32860 TARGET_LONG_JUMP_TABLE_OFFSETS.
32861 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
32862 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
32863 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
32864 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
32865 * config/m68k/m68k.md (tablejump expander): Likewise.
32866 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
32867 TARGET_LONG_JUMP_TABLE_OFFSETS.
32868 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
32869 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
32871 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
32872 David Holsgrove <david.holsgrove@xilinx.com>
32874 * common/config/microblaze/microblaze-common.c
32875 (TARGET_EXCEPT_UNWIND_INFO): Remove.
32876 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
32878 * config/microblaze/microblaze.c (microblaze_must_save_register)
32879 (microblaze_expand_epilogue, microblaze_return_addr): Handle
32881 (microblaze_eh_return): New function.
32882 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
32883 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
32884 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
32885 * config/microblaze/microblaze.md (eh_return): New pattern.
32887 2017-01-06 Jakub Jelinek <jakub@redhat.com>
32889 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
32890 GCC_DIAGNOSTIC_STRINGIFY): Define.
32892 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
32894 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32896 * config/arm/arm.md (<mcrr>): New.
32898 * config/arm/arm.c (arm_arch5te): New.
32899 (arm_option_override): Set arm_arch5te.
32900 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
32902 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
32903 (arm_mcrr_qualifiers): ... this. New.
32904 (MRRC_QUALIFIERS): Define to...
32905 (arm_mrrc_qualifiers): ... this. New.
32906 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
32908 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
32909 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
32910 (MRRCI, mrrc, MRRC): New.
32911 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
32912 VUNSPEC_MRRC2): New.
32914 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32916 * config/arm/arm.md (<mcr>): New.
32918 * config/arm/arm.c (arm_coproc_builtin_available): Add
32919 support for mcr, mrc, mcr2 and mrc2.
32920 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
32921 (arm_mcr_qualifiers): ... this. New.
32922 (MRC_QUALIFIERS): Define to ...
32923 (arm_mrc_qualifiers): ... this. New.
32924 (MCR_QUALIFIERS): Define to ...
32925 (arm_mcr_qualifiers): ... this. New.
32926 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
32928 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
32929 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
32930 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
32931 VUNSPEC_MRC2): New.
32933 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32935 * config/arm/arm.md (*ldc): New.
32939 * config/arm/arm.c (arm_coproc_builtin_available): Add
32940 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
32941 (arm_coproc_ldc_stc_legitimate_address): New.
32942 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
32943 'qualifier_const_pointer'.
32944 (LDC_QUALIFIERS): Define to...
32945 (arm_ldc_qualifiers): ... this. New.
32946 (STC_QUALIFIERS): Define to...
32947 (arm_stc_qualifiers): ... this. New.
32948 * config/arm/arm-protos.h
32949 (arm_coproc_ldc_stc_legitimate_address): New.
32950 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
32951 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
32952 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
32953 stc2, stcl, stc2l): New.
32954 * config/arm/constraints.md (Uz): New.
32955 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
32956 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
32957 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
32958 VUNSPEC_STC2L): New.
32960 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32962 * config/arm/arm.md (<cdp>): New.
32963 * config/arm/arm.c (neon_const_bounds): Rename this ...
32964 (arm_const_bounds): ... this.
32965 (arm_coproc_builtin_available): New.
32966 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
32967 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
32968 (CDP_QUALIFIERS): Define to...
32969 (arm_cdp_qualifiers): ... this. New.
32971 (arm_expand_builtin_args): Add case for 6 arguments.
32972 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
32973 (arm_const_bounds): ... this.
32974 (arm_coproc_builtin_available): New.
32975 * config/arm/arm_acle.h (__arm_cdp): New.
32977 * config/arm/arm_acle_builtins.def (cdp): New.
32979 * config/arm/iterators.md (CDPI,CDP,cdp): New.
32980 * config/arm/neon.md: Rename all 'neon_const_bounds' to
32981 'arm_const_bounds'.
32982 * config/arm/types.md (coproc): New.
32983 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
32984 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
32985 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
32986 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
32988 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32990 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
32991 (UBINOP_QUALIFIERS): New.
32993 (acle_builtin_data): New. Change comment.
32994 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
32995 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
32996 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
32997 arm_acle_builtins.def.
32998 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
32999 (arm_init_acle_builtins): New.
33000 (CRC32_BUILTIN): Remove.
33001 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
33002 crc32cb, crc32ch and crc32cw.
33003 (arm_init_crc32_builtins): Remove.
33004 (arm_init_builtins): Use arm_init_acle_builtins rather
33005 than arm_init_crc32_builtins.
33006 (arm_expand_acle_builtin): New.
33007 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
33008 * config/arm/arm_acle_builtins.def: New.
33010 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33012 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
33013 (arm_builtin_datum): ... this.
33014 (arm_init_neon_builtin): Rename to ...
33015 (arm_init_builtin): ... this. Add a new parameters PREFIX
33016 and USE_SIG_IN_NAME.
33017 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
33018 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
33019 'arm_builtin_datum'.
33020 (arm_init_vfp_builtins): Likewise.
33021 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
33022 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
33023 (arm_expand_neon_args): Rename to ...
33024 (arm_expand_builtin_args): ... this. Rename builtin_arg
33025 enum values and differentiate between ARG_BUILTIN_MEMORY
33026 and ARG_BUILTIN_NEON_MEMORY.
33027 (arm_expand_neon_builtin_1): Rename to ...
33028 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
33029 values, arm_expand_builtin_args and add bool parameter NEON.
33030 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
33031 (arm_expand_vfp_builtin): Likewise.
33032 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
33034 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33036 PR middle-end/77484
33037 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
33038 * predict.c (tree_estimate_probability_bb): Reverse direction of
33039 polymorphic call predictor.
33041 2017-01-06 David Malcolm <dmalcolm@redhat.com>
33043 * passes.c (execute_one_pass): Split out pass-skipping logic into...
33044 (determine_pass_name_match): ...this new function and...
33045 (should_skip_pass_p): ...this new function.
33047 2017-01-06 Nathan Sidwell <nathan@acm.org>
33049 * ipa-visibility.c (function_and_variable_visibility): Reformat
33050 comments and long lines. Remove extrneous if.
33051 * symtab.c (symtab_node::make_decl_local): Fix code format.
33052 (symtab_node::set_section_for_node): Fix comment typo.
33054 2017-01-06 Martin Liska <mliska@suse.cz>
33057 * lra-constraints.c: Rename invariant to lra_invariant.
33058 * predict.c (set_even_probabilities): Initialize e to NULL.
33060 2017-01-05 Martin Sebor <msebor@redhat.com>
33062 PR tree-optimization/78910
33063 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
33064 (format_integer): Correct off-by-one error in the handling
33065 of precision with negative numbers in signed conversions..
33067 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
33069 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
33071 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33073 PR tree-optimization/71016
33074 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
33075 factor_out_conditional_conversion. Formatting fix.
33076 (factor_out_conditional_conversion): Add cond_stmt argument.
33077 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
33078 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
33081 2017-01-05 David Malcolm <dmalcolm@redhat.com>
33083 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
33084 read-rtl-function.o, and selftest-rtl.o.
33085 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
33086 (selftest::aarch64_test_loading_full_dump): New function.
33087 (selftest::aarch64_run_selftests): New function.
33088 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
33089 selftest::aarch64_run_selftests.
33090 * config/i386/i386.c
33091 (selftest::ix86_test_loading_dump_fragment_1): New function.
33092 (selftest::ix86_test_loading_call_insn): New function.
33093 (selftest::ix86_test_loading_full_dump): New function.
33094 (selftest::ix86_test_loading_unspec): New function.
33095 (selftest::ix86_run_selftests): Call the new functions.
33096 * emit-rtl.c (maybe_set_max_label_num): New function.
33097 * emit-rtl.h (maybe_set_max_label_num): New decl.
33098 * function.c (instantiate_decls): Guard call to
33099 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
33100 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
33102 * gensupport.c (gen_reader::gen_reader): Pass "false"
33103 for new "compact" param of rtx_reader.
33104 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
33105 rather than an empty string for NULL strings.
33106 * read-md.c: Potentially include config.h rather than bconfig.h.
33107 Wrap include of errors.h with #ifdef GENERATOR_FILE.
33108 (have_error): New global, copied from errors.c.
33109 (md_reader::read_name): Rename to...
33110 (md_reader::read_name_1): ...this, adding "out_loc" param,
33111 and converting "missing name or number" to returning false, rather
33113 (md_reader::read_name): Reimplement in terms of read_name_1.
33114 (md_reader::read_name_or_nil): New function.
33115 (md_reader::read_string): Handle "(nil)" by returning NULL.
33116 (md_reader::md_reader): Add new param "compact".
33117 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
33118 (md_reader::read_file): New method.
33119 * read-md.h (md_reader::md_reader): Add new param "compact".
33120 (md_reader::read_file): New method.
33121 (md_reader::is_compact): New accessor.
33122 (md_reader::read_name): Convert return type from void to file_location.
33123 (md_reader::read_name_or_nil): New decl.
33124 (md_reader::read_name_1): New decl.
33125 (md_reader::m_compact): New field.
33126 (noop_reader::noop_reader): Pass "false" for new "compact" param
33128 (rtx_reader::rtx_reader): Add new "compact" param.
33129 (rtx_reader::read_rtx_operand): Make virtual and convert return
33130 type from void to rtx.
33131 (rtx_reader::read_until): New decl.
33132 (rtx_reader::handle_any_trailing_information): New virtual function.
33133 (rtx_reader::postprocess): New virtual function.
33134 (rtx_reader::finalize_string): New virtual function.
33135 (rtx_reader::m_in_call_function_usage): New field.
33136 (rtx_reader::m_reuse_rtx_by_id): New field.
33137 * read-rtl-function.c: New file.
33138 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
33139 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
33140 (selftest::verify_three_block_rtl_cfg): New decl.
33141 * read-rtl-function.h: New file.
33142 * read-rtl.c: Potentially include config.h rather than bconfig.h.
33143 For host, include function.h, memmodel.h, and emit-rtl.h.
33144 (one_time_initialization): New function.
33145 (struct compact_insn_name): New struct.
33146 (compact_insn_names): New array.
33147 (find_code): Handle insn codes in compact dumps.
33148 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
33149 (bind_subst_iter_and_attr): Likewise.
33150 (add_condition_to_string): Likewise.
33151 (add_condition_to_rtx): Likewise.
33152 (apply_attribute_uses): Likewise.
33153 (add_current_iterators): Likewise.
33154 (apply_iterators): Likewise.
33155 (initialize_iterators): Guard usage of apply_subst_iterator with
33156 #ifdef GENERATOR_FILE.
33157 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
33158 (md_reader::read_mapping): Likewise.
33159 (add_define_attr_for_define_subst): Likewise.
33160 (add_define_subst_attr): Likewise.
33161 (read_subst_mapping): Likewise.
33162 (check_code_iterator): Likewise.
33163 (rtx_reader::read_rtx): Likewise. Move one-time initialization
33165 (one_time_initialization): New function.
33166 (rtx_reader::read_until): New method.
33167 (read_flags): New function.
33168 (parse_reg_note_name): New function.
33169 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
33170 Handle reuse_rtx ids.
33171 Wrap iterator lookup within #ifdef GENERATOR_FILE.
33172 Add parsing support for RTL dumps, mirroring the special-cases in
33173 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
33174 values, and calling handle_any_trailing_information.
33175 (rtx_reader::read_rtx_operand): Convert return type from void
33176 to rtx, returning return_rtx. Handle case 'e'. Call
33177 finalize_string on XSTR and XTMPL fields.
33178 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
33179 "(nil)" values were omitted. Call the postprocess vfunc on the
33181 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
33182 class ctor. Initialize m_in_call_function_usage. Call
33183 one_time_initialization.
33184 * rtl-tests.c (selftest::test_uncond_jump): Call
33185 set_new_first_and_last_insn.
33186 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
33187 * selftest-rtl.c: New file.
33188 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
33189 (selftest::get_insn_by_uid): New decl.
33190 * selftest-run-tests.c (selftest::run_tests): Call
33191 read_rtl_function_c_tests.
33192 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
33193 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
33196 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
33198 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
33199 operands in a special way. Assert that pos+len <= mode precision.
33201 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33203 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
33204 3 argument Alias with unlimited for the negative form.
33205 (fno-vect-cost-model): Removed.
33207 2017-01-05 Martin Liska <mliska@suse.cz>
33209 * hsa-gen.c (gen_hsa_divmod): New function.
33210 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
33212 2017-01-05 Martin Liska <mliska@suse.cz>
33215 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
33218 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33220 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
33221 small constant length operands.
33223 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33225 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
33226 between loop iterations.
33228 2017-01-05 Martin Liska <mliska@suse.cz>
33231 * gimplify.c (gimplify_decl_expr): Compare to
33232 asan_poisoned_variables instread of checking flags.
33233 (gimplify_target_expr): Likewise.
33234 (gimplify_expr): Likewise.
33235 (gimplify_function_tree): Conditionally initialize
33236 asan_poisoned_variables.
33238 2017-01-04 Jeff Law <law@redhat.com>
33240 PR tree-optimizatin/78812
33241 * rtl.h (contains_mem_rtx_p): Prototype.
33242 * ifcvt.c (containts_mem_rtx_p): Move from here to...
33243 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
33244 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
33245 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
33246 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
33248 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33250 * input.c (assert_char_at_range): Default-initialize actual_range.
33252 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33254 * df-scan.c (df_ref_create_structure): Make regno unsigned,
33255 to match the caller.
33257 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33259 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
33260 insns after final jump in test to emit dummy move.
33262 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33264 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
33265 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
33267 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33269 * multiple_target.c (create_dispatcher_calls): Init e_next.
33270 * tree-ssa-loop-split.c (split_loop): Init border.
33271 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
33274 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
33279 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
33280 (altivec_register_operand): Do not return true if the operand
33281 contains a SUBREG mixing SImode and SFmode.
33282 (vsx_register_operand): Likewise.
33283 (vsx_reg_sfsubreg_ok): New predicate.
33284 (vfloat_operand): Do not return true if the operand contains a
33285 SUBREG mixing SImode and SFmode.
33286 (vint_operand): Likewise.
33287 (vlogical_operand): Likewise.
33288 (gpc_reg_operand): Likewise.
33289 (int_reg_operand): Likewise.
33290 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
33291 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
33292 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
33294 (rs6000_emit_move_si_sf_subreg): New helper function.
33295 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
33296 fixup SUBREGs involving SImode and SFmode.
33297 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
33298 numbers for the new peephole2 optimization.
33299 (peephole2 for SFmode unions): New peephole2 to optimize cases in
33300 the GLIBC math library that do AND/IOR/XOR operations on single
33301 precision floating point.
33302 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
33303 target macros to say whether we need to avoid SUBREGs mixing
33305 (TARGET_ALLOW_SF_SUBREG): Likewise.
33306 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
33307 (UNSPEC_SI_FROM_SF): Likewise.
33308 (iorxor): Change spacing.
33309 (and_ior_xor): New iterator for AND, IOR, and XOR.
33310 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
33311 (movdi_from_sf_zero_ext): Likewise.
33312 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
33313 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
33314 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
33315 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
33316 (fms<mode>4): Likewise.
33317 (fnma<mode>4): Likewise.
33318 (fnms<mode>4): Likewise.
33319 (nfma<mode>4): Likewise.
33320 (nfms<mode>4): Likewise.
33322 2017-01-04 Marek Polacek <polacek@redhat.com>
33325 * doc/invoke.texi: Document -Wpointer-compare.
33327 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33329 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
33332 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
33333 descriptions for -gdwarf-5 and emit them as uleb128 instead of
33336 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
33339 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
33340 documentation of the powerpc_popcntb_ok attribute.
33341 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
33342 code to issue warning messages if a requested CPU configuration is
33343 not supported by the binary (assembler and loader) toolchain.
33344 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
33345 made to define a built-in function that has been disabled.
33346 (paired_init_builtins): Add assertion to prevent ICE if attempt is
33347 made to define a built-in function that has been disabled.
33348 (altivec_init_builtins): Add comment explaining why definition
33349 of the DST built-in functions is not preceded by an assertion
33350 check. Add assertions to prevent ICE if attempts are made to
33351 define an altivec predicate or an abs* built-in function that has
33353 (htm_init_builtins): Add comment explaining why definition of the
33354 htm built-in functions is not preceded by an assertion check.
33356 2017-01-04 Jeff Law <law@redhat.com>
33358 PR tree-optimizatin/67955
33359 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
33360 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
33361 the points-to solution does not include pt_null. Use DECL_PT_UID
33364 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
33366 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
33367 Use gen_int_mode instead of gen_lopwart for const_int operands.
33369 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33371 PR tree-optimization/71563
33372 * match.pd: Simplify X << Y into X if Y is known to be 0 or
33373 out of range value - has low bits known to be zero.
33375 2017-01-04 Alan Modra <amodra@gmail.com>
33377 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
33378 * configure: Regenerate.
33379 * config.in: Regenerate.
33381 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33384 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
33385 a substring of the message, but strcmp with the whole message. Ifdef
33386 ENABLE_NLS, translate the message first using dgettext.
33388 2017-01-03 Jeff Law <law@redhat.com>
33390 PR tree-optimizatin/78856
33391 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
33392 (mark_threaded_blocks): Remove code to truncate thread paths that
33393 cross multiple loop headers. Instead invalidate the cached loop
33394 iteration information and handle case of a thread path walking
33395 into an irreducible region.
33397 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
33400 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
33401 assertions. Add support for doing the signbit if the IEEE 128-bit
33402 floating point value is in a GPR.
33403 * config/rs6000/rs6000.md (Fsignbit): Delete.
33404 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
33405 Update the length attribute if the value is in a GPR.
33406 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
33407 the sign or zero extension instruction, since the value is always 0/1.
33408 (signbit<mode>2_dm2): Delete using <Fsignbit>.
33411 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
33412 extracting SImode to a GPR register so that we can generate a
33413 store, limit the vector to be in a traditional Altivec register
33414 for the vextuwrx instruction.
33416 2017-01-03 Ian Lance Taylor <iant@google.com>
33418 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
33420 2017-01-03 Martin Sebor <msebor@redhat.com>
33422 PR tree-optimization/78696
33423 * gimple-ssa-sprintf.c (format_floating): Correct handling of
33424 precision. Use MPFR for %f for greater fidelity. Correct handling
33426 (pass_sprintf_length::compute_format_length): Set width and precision
33427 specified by asrerisk to void_node for vararg functions.
33428 (try_substitute_return_value): Adjust dump output.
33430 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
33432 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
33434 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
33436 * doc/invoke.texi (SPARC options): Document -mlra as the default.
33437 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
33438 -mlra/-mno-lra was passed to the compiler.
33440 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
33442 PR rtl-optimization/65618
33443 * emit-rtl.c (try_split): Move initialization of "before" and
33444 "after" to just before the call to emit_insn_after_setloc.
33446 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
33448 * doc/md.texi (Standard Names): Remove reference to Java frontend.
33450 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
33452 * dwarf2out.c (gen_enumeration_type_die): When
33453 -gno-strict-dwarf, add a DW_AT_encoding attribute.
33455 2017-01-03 Jakub Jelinek <jakub@redhat.com>
33457 PR tree-optimization/78965
33458 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
33459 Change first argument from const call_info & to call_info &. For %n
33460 set info.nowrite to false.
33462 PR middle-end/78901
33463 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
33464 possibly throwing calls.
33466 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
33467 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
33468 and fns handling, rather than in a separate case SSA_NAME.
33470 2017-01-02 Jeff Law <law@redhat.com>
33472 * config/darwin-driver.c (darwin_driver_init): Const-correctness
33473 fixes for first_period and second_period variables.
33475 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
33478 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
33479 (*insvqi_1): New insn pattern.
33480 (*insvqi_1_mem_rex64): Ditto.
33481 (*insvqi_2): Ditto.
33482 (*insvqi_3): Rename from *insvqi.
33484 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
33486 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
33488 * doc/cfg.texi (Edges): Remove reference to Java.
33489 (Maintaining the CFG): Ditto.
33491 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33493 PR middle-end/77674
33494 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
33495 transparent aliases.
33497 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33499 PR middle-end/77484
33500 * predict.def (PRED_CALL): Update hitrate.
33501 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
33502 * predict.c (tree_estimate_probability_bb): Split CALL predictor
33503 into direct/indirect/polymorphic variants.
33505 2017-01-01 Jakub Jelinek <jakub@redhat.com>
33507 Update copyright years.
33509 * gcc.c (process_command): Update copyright notice dates.
33510 * gcov-dump.c (print_version): Ditto.
33511 * gcov.c (print_version): Ditto.
33512 * gcov-tool.c (print_version): Ditto.
33513 * gengtype.c (create_file): Ditto.
33514 * doc/cpp.texi: Bump @copying's copyright year.
33515 * doc/cppinternals.texi: Ditto.
33516 * doc/gcc.texi: Ditto.
33517 * doc/gccint.texi: Ditto.
33518 * doc/gcov.texi: Ditto.
33519 * doc/install.texi: Ditto.
33520 * doc/invoke.texi: Ditto.
33522 Copyright (C) 2017 Free Software Foundation, Inc.
33524 Copying and distribution of this file, with or without modification,
33525 are permitted in any medium without royalty provided the copyright
33526 notice and this notice are preserved.