1 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
3 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
6 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8 * tree-vrp.c (vrp_int_const_binop): Return true on success and
9 return the value by pointer.
10 (extract_range_from_multiplicative_op_1): Update accordingly.
11 Return as soon as an operation fails.
13 2017-11-05 Tom de Vries <tom@codesourcery.com>
16 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
17 (DEF_SANITIZER_BUILTIN): ... here.
18 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
19 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
21 2017-11-05 Tom de Vries <tom@codesourcery.com>
24 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
26 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
27 ASM_OUTPUT_BEFORE_CASE_LABEL call.
28 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
30 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
31 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
32 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
34 2017-11-05 Tom de Vries <tom@codesourcery.com>
37 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
40 2017-11-04 Michael Clark <michaeljclark@mac.com>
42 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
43 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
45 (*addsi3_extended): Likewise.
46 (*addsi3_extended2): Likewise.
47 (<optab>si3): Likewise.
48 (<optab>di3): Likewise.
49 (<optab><mode>3): Likewise.
50 (<*optabe>si3_internal): Likewise.
51 (zero_extendqi<SUPERQI:mode>2): Likewise.
52 (*add<mode>hi3): Likewise.
53 (*xor<mode>hi3): Likewise.
54 (<optab>di3): Likewise.
55 (*<optab>si3_extend): Likewise.
56 (*sge<u>_<X:mode><GPR:mode>): Likewise.
57 (*slt<u>_<X:mode><GPR:mode>): Likewise.
58 (*sle<u>_<X:mode><GPR:mode>): Likewise.
60 2017-11-04 Andrew Waterman <andrew@sifive.com>
62 * config/riscv/riscv.c (riscv_option_override): Conditionally set
63 TARGET_STRICT_ALIGN based upon -mtune argument.
65 2017-11-04 Andrew Waterman <andrew@sifive.com>
67 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
69 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
71 config/i386/i386.c (choose_basereg): Use optional scratch
72 register and add assertion.
73 (x86_emit_outlined_ms2sysv_save): Use scratch register when
74 needed, and don't allocate stack.
75 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
76 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
77 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
79 2017-11-03 Jeff Law <law@redhat.com>
81 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
82 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
83 to probe at the start of a noreturn function.
85 2017-11-03 Jakub Jelinek <jakub@redhat.com>
87 PR tree-optimization/78821
88 * gimple-ssa-store-merging.c: Update the file comment.
89 (MAX_STORE_ALIAS_CHECKS): Define.
90 (struct store_operand_info): New type.
91 (store_operand_info::store_operand_info): New constructor.
92 (struct store_immediate_info): Add rhs_code and ops data members.
93 (store_immediate_info::store_immediate_info): Add rhscode, op0r
94 and op1r arguments to the ctor, initialize corresponding data members.
95 (struct merged_store_group): Add load_align_base and load_align
97 (merged_store_group::merged_store_group): Initialize them.
98 (merged_store_group::do_merge): Update them.
99 (merged_store_group::apply_stores): Pick the constant for
100 encode_tree_to_bitpos from one of the two operands, or skip
101 encode_tree_to_bitpos if neither operand is a constant.
102 (class pass_store_merging): Add process_store method decl. Remove
103 bool argument from terminate_all_aliasing_chains method decl.
104 (pass_store_merging::terminate_all_aliasing_chains): Remove
105 var_offset_p argument and corresponding handling.
106 (stmts_may_clobber_ref_p): New function.
107 (compatible_load_p): New function.
108 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
109 if there is overlap and rhs_code is not INTEGER_CST. For
110 non-overlapping stores terminate group if rhs is not mergeable.
111 (get_alias_type_for_stmts): Change first argument from
112 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
113 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
114 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
116 (get_location_for_stmts): Change first argument from
117 auto_vec<gimple *> & to vec<gimple *> &.
118 (struct split_store): Remove orig_stmts data member, add orig_stores.
119 (split_store::split_store): Create orig_stores rather than orig_stmts.
120 (find_constituent_stmts): Renamed to ...
121 (find_constituent_stores): ... this. Change second argument from
122 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
123 to info structures rather than the statements.
124 (split_group): Rename ALLOW_UNALIGNED argument to
125 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
126 it. Adjust find_constituent_stores caller.
127 (imm_store_chain_info::output_merged_store): Handle rhs_code other
128 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
129 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
130 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
131 (mem_valid_for_store_merging): New function.
132 (handled_load): New function.
133 (pass_store_merging::process_store): New method.
134 (pass_store_merging::execute): Use process_store method. Adjust
135 terminate_all_aliasing_chains caller.
137 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
139 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
140 Return true for more constants, symbols and label references.
141 (aarch64_valid_floating_const): Remove unused function.
143 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
146 * config/aarch64/aarch64.c (aarch64_layout_frame):
147 Undo forcing of LR at bottom of frame.
149 2017-11-03 Jeff Law <law@redhat.com>
152 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
153 for int_registers_saved.
155 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
156 extracted from tree-ssa-dom.c.
157 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
158 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
159 (record_equivalences_from_incoming_edge): Add additional argument
160 to single_pred_edge_ignoring_loop_edges call.
161 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
162 (uncprop_dom_walker::before_dom_children): Add additional argument
163 to single_pred_edge_ignoring_loop_edges call.
164 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
165 single_pred_edge_ignoring_loop_edges rather than open coding.
166 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
168 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
170 * match.pd (-(-A)): Rewrite.
172 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
174 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
175 (rs6000_emit_int_cmove): New declaration.
176 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
177 (rs6000_emit_sISEL): Delete.
178 (rs6000_emit_int_cmove): Make non-static.
179 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
180 instead of rs6000_emit_sISEL.
182 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
184 * asan.c (create_cond_insert_point): Maintain profile.
185 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
187 * basic-block.h (struct basic_block_def): Remove frequency.
188 (EDGE_FREQUENCY): Use to_frequency
189 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
191 (find_traces): Update to use to_frequency.
192 (find_traces_1_round): Likewise; use only IPA counts.
193 (bb_to_key): Likewise.
194 (connect_traces): Use IPA counts only.
195 (copy_bb_p): Update to use to_frequency.
196 (fix_up_crossing_landing_pad): Likewise.
197 (sanitize_hot_paths): Likewise.
198 * bt-load.c (basic_block_freq): Likewise.
199 * cfg.c (init_flow): Set count_max to uninitialized.
200 (check_bb_profile): Remove frequencies; check counts.
201 (dump_bb_info): Do not dump frequencies.
202 (update_bb_profile_for_threading): Update counts only.
203 (scale_bbs_frequencies_int): Likewise.
204 (MAX_SAFE_MULTIPLIER): Remove.
205 (scale_bbs_frequencies_gcov_type): Update counts only.
206 (scale_bbs_frequencies_profile_count): Update counts only.
207 (scale_bbs_frequencies): Update counts only.
208 * cfg.h (struct control_flow_graph): Add count-max.
209 (update_bb_profile_for_threading): Update prototype.
210 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
211 (find_many_sub_basic_blocks): Likewise.
212 * cfgcleanup.c (try_forward_edges): Likewise.
213 (try_crossjump_to_edge): Likewise.
214 * cfgexpand.c (expand_gimple_cond): Likewise.
215 (expand_gimple_tailcall): Likewise.
216 (construct_init_block): Likewise.
217 (construct_exit_block): Likewise.
218 * cfghooks.c (verify_flow_info): Check consistency of counts.
219 (dump_bb_for_graph): Do not dump frequencies.
220 (split_block_1): Do not update frequencies.
221 (split_edge): Do not update frequencies.
222 (make_forwarder_block): Do not update frequencies.
223 (duplicate_block): Do not update frequencies.
224 (account_profile_record): Do not update frequencies.
225 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
226 for global heuristics.
227 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
228 (expected_loop_iterations_unbounded): Use counts only.
229 * cfgloopmanip.c (scale_loop_profile): Simplify.
230 (create_empty_loop_on_edge): Simplify
232 (duplicate_loop_to_header_edge): Simplify
233 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
234 (update_br_prob_note): Take care of removing note when profile
236 (relink_block_chain): Do not dump frequency.
237 (rtl_account_profile_record): Use to_frequency.
238 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
239 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
240 (cgraph_update_edges_for_call_stmt_node): Likewise.
241 (cgraph_edge::verify_count_and_frequency): Update.
242 (cgraph_node::verify_node): Temporarily disable frequency verification.
243 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
245 (cgraph_edge::rebuild_edges): Convert to ipa counts.
246 * cgraphunit.c (init_lowered_empty_function): Do not initialize
248 (cgraph_node::expand_thunk): Update profile.
249 * except.c (dw2_build_landing_pads): Do not update frequency.
250 * final.c (compute_alignments): Use to_frequency.
251 (dump_basic_block_info): Do not dump frequency.
252 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
253 (dump_gimple_bb_header): Do not dump frequency.
254 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
256 * gimple-streamer-in.c (input_bb): Do not stream frequency.
257 * gimple-streamer-out.c (output_bb): Do not stream frequency.
258 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
259 (init_before_recovery): Do not update frequency.
260 (sched_create_recovery_edges): Do not update frequency.
261 * hsa-gen.c (convert_switch_statements): Do not update frequency.
262 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
263 (ipa_cp_c_finalize): Set max_count to uninitialized.
264 * ipa-fnsummary.c (get_minimal_bb): Use counts.
265 (param_change_prob): Use counts.
266 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
268 * ipa-split.c (consider_split): Use to_frequency.
269 (split_function): Use to_frequency.
270 * ira-build.c (loop_compare_func): Likewise.
271 (mark_loops_for_removal): Likewise.
272 (mark_all_loops_for_removal): Likewise.
273 * loop-doloop.c (doloop_modify): Do not update frequency.
274 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
276 * lto-streamer-in.c (input_function): Update count_max.
277 * omp-expand.c (expand_omp_taskreg): Update count_max.
278 * omp-simd-clone.c (simd_clone_adjust): Update profile.
279 * predict.c (maybe_hot_frequency_p): Use to_frequency.
280 (maybe_hot_count_p): Use ipa counts only.
281 (maybe_hot_bb_p): Simplify.
282 (maybe_hot_edge_p): Simplify.
283 (probably_never_executed): Do not take frequency argument.
284 (probably_never_executed_bb_p): Do not pass frequency.
285 (probably_never_executed_edge_p): Likewise.
286 (combine_predictions_for_bb): Check that profile is nonzero.
287 (propagate_freq): Do not set frequency.
288 (drop_profile): Simplify.
289 (counts_to_freqs): Simplify.
290 (expensive_function_p): Use to_frequency.
291 (propagate_unlikely_bbs_forward): Simplify.
292 (determine_unlikely_bbs): Simplify.
293 (estimate_bb_frequencies): Add hack to silence graphite issues.
294 (compute_function_frequency): Use ipa counts.
295 (pass_profile::execute): Update.
296 (rebuild_frequencies): Use counts only.
297 (force_edge_cold): Use counts only.
298 * profile-count.c (profile_count::dump): Dump new count types.
299 (profile_count::differs_from_p): Check compatiblity.
300 (profile_count::to_frequency): New function.
301 (profile_count::to_cgraph_frequency): New function.
302 * profile-count.h (struct function): Declare.
303 (enum profile_quality): Add profile_guessed_local and
304 profile_guessed_global0.
305 (class profile_proability): Decrease number of bits to 29;
306 update from_reg_br_prob_note and to_reg_br_prob_note.
307 (class profile_count: Update comment; decrease number of bits
308 to 61. Check compatibility.
309 (profile_count::compatible_p): New private member function.
310 (profile_count::ipa_p): New member function.
311 (profile_count::operator<): Handle global zero correctly.
312 (profile_count::operator>): Handle global zero correctly.
313 (profile_count::operator<=): Handle global zero correctly.
314 (profile_count::operator>=): Handle global zero correctly.
315 (profile_count::nonzero_p): New member function.
316 (profile_count::force_nonzero): New member function.
317 (profile_count::max): New member function.
318 (profile_count::apply_scale): Handle IPA scalling.
319 (profile_count::guessed_local): New member function.
320 (profile_count::global0): New member function.
321 (profile_count::ipa): New member function.
322 (profile_count::to_frequency): Declare.
323 (profile_count::to_cgraph_frequency): Declare.
324 * profile.c (OVERLAP_BASE): Delete.
325 (compute_frequency_overlap): Delete.
326 (compute_branch_probabilities): Do not use compute_frequency_overlap.
327 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
328 * sched-ebb.c (rank): Use counts only.
329 * shrink-wrap.c (handle_simple_exit): Use counts only.
330 (try_shrink_wrapping): Use counts only.
331 (place_prologue_for_one_component): Use counts only.
332 * tracer.c (find_best_predecessor): Use to_frequency.
333 (find_trace): Use to_frequency.
334 (tail_duplicate): Use to_frequency.
335 * trans-mem.c (expand_transaction): Do not update frequency.
336 * tree-call-cdce.c: Do not update frequency.
337 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
338 (gimple_merge_blocks): Likewise.
339 (gimple_split_edge): Likewise.
340 (gimple_duplicate_sese_region): Likewise.
341 (gimple_duplicate_sese_tail): Likewise.
342 (move_sese_region_to_fn): Likewise.
343 (gimple_account_profile_record): Likewise.
344 (insert_cond_bb): Likewise.
345 * tree-complex.c (expand_complex_div_wide): Likewise.
346 * tree-eh.c (lower_resx): Update profile.
347 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
349 (initialize_cfun): Do not initialize frequencies
350 (freqs_to_counts): Delete.
351 (copy_cfg_body): Ignore count parameter.
353 (expand_call_inline): Update count_max.
354 (optimize_inline_calls): Update count_max.
355 (tree_function_versioning): Update count_max.
356 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
357 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
359 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
360 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
361 (try_peel_loop): Likewise.
362 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
364 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
365 (tree_transform_and_unroll_loop): Do not use frequencies
366 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
367 Use reliable prediction only.
368 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
369 * tree-ssa-sink.c (select_best_block): Use to_frequency.
370 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
372 * tree-ssa-threadupdate.c (create_block_for_threading): Do
374 (any_remaining_duplicated_blocks): Likewise.
375 (update_profile): Likewise.
376 (estimated_freqs_path): Delete.
377 (freqs_to_counts_path): Delete.
378 (clear_counts_path): Delete.
379 (ssa_fix_duplicate_block_edges): Likewise.
380 (duplicate_thread_path): Likewise.
381 * tree-switch-conversion.c (gen_inbound_check): Use counts.
382 * tree-tailcall.c (decrease_profile): Do not update frequency.
383 (eliminate_tail_call): Likewise.
384 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
385 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
386 (optimize_mask_stores): Likewise.
387 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
388 * ubsan.c (ubsan_expand_null_ifn): Update profile.
389 (ubsan_expand_ptr_ifn): Update profile.
390 * value-prof.c (gimple_ic): Simplify.
391 * value-prof.h (gimple_ic): Update prototype.
392 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
393 * ipa-inline.c (compute_uninlined_call_time): Be sure that
395 (want_inline_self_recursive_call_p): Likewise.
396 (resolve_noninline_speculation): Only cummulate defined counts.
397 (inline_small_functions): Use nonzero_p.
398 (ipa_inline): Do not access freed node.
400 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
402 * config/aarch64/aarch64.c (aarch64_override_options_internal):
403 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
405 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
407 * config/riscv/riscv.c (riscv_legitimize_move): Handle
408 non-legitimate address.
410 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
412 * config/rs6000/rs6000.md (*lt0_disi): Delete.
413 (*lt0_<mode>di, *lt0_<mode>si): New.
415 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
417 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
420 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
421 Jim Wilson <jim.wilson@linaro.org>
423 * config/aarch64/aarch64-cores.def (saphira): New CPU.
424 * config/aarch64/aarch64-tune.md: Regenerated.
425 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
426 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
428 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
430 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
431 (arc_expand_prologue): Restore blink for millicode.
432 * config/arc/linux.h (LINK_EH_SPEC): Defined.
434 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
437 * config/i386/i386.c (ix86_vector_duplicate_value): Use
438 gen_vec_duplicate after forcing the scalar into a register.
440 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
442 * combine (try_combine): Print the insns input to try_combine to the
445 2017-11-02 Steve Ellcey <sellcey@cavium.com>
448 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
449 Remove second argument from aarch64_process_target_attr call.
450 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
452 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
454 (aarch64_handle_attr_arch): Remove second argument.
455 (aarch64_handle_attr_cpu): Ditto.
456 (aarch64_handle_attr_tune): Ditto.
457 (aarch64_handle_attr_isa_flags): Ditto.
458 (aarch64_process_one_target_attr): Ditto.
459 (aarch64_process_target_attr): Ditto.
460 (aarch64_option_valid_attribute_p): Remove second argument.
461 on aarch64_process_target_attr call.
463 2017-11-02 David Malcolm <dmalcolm@redhat.com>
465 * diagnostic.c: Include "selftest-diagnostic.h".
466 (selftest::assert_location_text): New function.
467 (selftest::test_diagnostic_get_location_text): New function.
468 (selftest::diagnostic_c_tests): Call it.
470 2017-11-02 David Malcolm <dmalcolm@redhat.com>
472 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
473 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
474 (class selftest::test_diagnostic_context): Move to...
475 * selftest-diagnostic.c: New file.
476 * selftest-diagnostic.h: New file.
478 2017-11-02 James Bowman <james.bowman@ftdichip.com>
480 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
481 offset range for FT32B.
482 * config/ft32/ft32.h: option "mcompress" enables relaxation.
483 * config/ft32/ft32.md: Add TARGET_NOPM.
484 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
485 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
487 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
489 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
491 2017-11-02 Jeff Law <law@redhat.com>
493 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
494 virtual keyword on FINAL OVERRIDE members.
496 * tree-ssa-propagate.h (ssa_propagation_engine): Group
497 virtuals together. Add virtual destructor.
498 (substitute_and_fold_engine): Similarly.
500 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
502 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
504 2017-11-02 Richard Biener <rguenther@suse.de>
506 PR tree-optimization/82795
507 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
509 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
511 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
513 (gcc_GAS_CHECK_FEATURE): Remove.
514 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
515 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
516 * configure: Regenerate.
518 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
520 * config/arc/arc.c (hwloop_optimize): Account for empty
523 2017-11-02 Richard Biener <rguenther@suse.de>
526 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
527 Truncate ARRAY_REF index and element size.
529 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
531 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
533 2017-11-01 Jeff Law <law@redhat.com>
535 * tree-ssa-ccp.c (ccp_folder): New class derived from
536 substitute_and_fold_engine.
537 (ccp_folder::get_value): New member function.
538 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
539 (ccp_fold_stmt): Remove prototype.
540 (ccp_finalize): Call substitute_and_fold from the ccp_class.
541 * tree-ssa-copy.c (copy_folder): New class derived from
542 substitute_and_fold_engine.
543 (copy_folder::get_value): Renamed from get_value.
544 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
545 * tree-vrp.c (vrp_folder): New class derived from
546 substitute_and_fold_engine.
547 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
548 (vrp_folder::get_value): New member function.
549 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
550 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
551 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
552 provide a class interface to folder/substitute routines.
553 (ssa_prop_fold_stmt_fn): Remove typedef.
554 (ssa_prop_get_value_fn): Likewise.
555 (subsitute_and_fold): Remove prototype.
556 (replace_uses_in): Likewise.
557 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
558 Renamed from replace_uses_in. Call the virtual member function
559 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
560 (substitute_and_fold_dom_walker): Remove initialization of
561 data member entries for calbacks. Add substitute_and_fold_engine
562 member and initialize it.
563 (substitute_and_fold_dom_walker::before_dom_children0: Use the
564 member functions for get_value, replace_phi_args_in c
565 replace_uses_in, and fold_stmt calls.
566 (substitute_and_fold_engine::substitute_and_fold): Renamed from
567 substitute_and_fold. Remove assert. Update ctor call.
569 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
570 (ssa_prop_visit_phi_fn): Likewise.
571 (class ssa_propagation_engine): New class to provide an interface
573 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
575 (ssa_prop_visit_phi): Likewise.
576 (ssa_propagation_engine::simulate_stmt): Moved into class.
577 Call visit_phi/visit_stmt from the class rather than via
578 file scoped static variables.
579 (ssa_propagation_engine::simulate_block): Moved into class.
580 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
581 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
582 set file scoped statics for the visit_stmt/visit_phi callbacks.
583 * tree-complex.c (complex_propagate): New class derived from
584 ssa_propagation_engine.
585 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
586 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
587 (tree_lower_complex): Call ssa_propagate via the complex_propagate
589 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
590 ssa_propagation_engine.
591 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
592 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
593 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
594 * tree-ssa-copy.c (copy_prop): New class derived from
595 ssa_propagation_engine.
596 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
597 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
598 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
599 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
600 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
601 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
602 (execute_vrp): Call ssa_propagate from the vrp_prop class.
604 2017-11-01 Jakub Jelinek <jakub@redhat.com>
606 PR rtl-optimization/82778
607 PR rtl-optimization/82597
608 * compare-elim.c (struct comparison): Add in_a_setter field.
609 (find_comparison_dom_walker::before_dom_children): Remove killed
610 bitmap and df_simulate_find_defs call, instead walk the defs.
611 Compute last_setter and initialize in_a_setter. Merge definitions
612 with first initialization for a few variables.
613 (try_validate_parallel): Use insn_invalid_p instead of
614 recog_memoized. Return insn rather than just the pattern.
615 (try_merge_compare): Fix up comment. Don't uselessly test if
616 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
618 (execute_compare_elim_after_reload): Remove df_chain_add_problem
621 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
622 Alan Hayward <alan.hayward@arm.com>
623 David Sherwood <david.sherwood@arm.com>
625 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
626 aarch64_hard_regno_nregs to get the number of registers
629 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
630 Alan Hayward <alan.hayward@arm.com>
631 David Sherwood <david.sherwood@arm.com>
633 * config/aarch64/constraints.md (Upl): Rename to...
635 * config/aarch64/aarch64.md
636 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
639 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
640 Alan Hayward <alan.hayward@arm.com>
641 David Sherwood <david.sherwood@arm.com>
643 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
644 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
647 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
648 Alan Hayward <alan.hayward@arm.com>
649 David Sherwood <david.sherwood@arm.com>
651 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
652 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
653 (aarch64_evpc_dup): Generate rtl direcly, rather than using
655 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
657 * config/aarch64/iterators.md: Add a comment above the permute
658 unspecs to say that they are generated directly by
659 aarch64_expand_vec_perm_const.
660 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
662 2017-11-01 Nathan Sidwell <nathan@acm.org>
664 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
666 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
668 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
669 and medany code models, and describe what they do.
671 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
673 Revert accidental duplicate:
675 * combine.c (can_change_dest_mode): Reject changes in
676 REGMODE_NATURAL_SIZE.
678 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
680 PR rtl-optimization/64682
681 PR rtl-optimization/69567
682 PR rtl-optimization/69737
683 PR rtl-optimization/82683
684 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
685 register mentioned in the note, drop the note, unless it came from I3,
686 in which case it should go to I3 again.
688 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
690 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
691 and return false if not.
692 (clear_bytes_written_by, live_bytes_read): Update accordingly.
694 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
696 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
697 range is known to be empty.
699 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
700 Alan Hayward <alan.hayward@arm.com>
701 David Sherwood <david.sherwood@arm.com>
703 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
704 and CONST_VECTOR_NUNITS instead of computing the number of units from
705 the byte sizes of the vector and element.
706 (simplify_binary_operation_1): Likewise.
707 (simplify_const_binary_operation): Likewise.
708 (simplify_ternary_operation): Likewise.
710 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
711 Alan Hayward <alan.hayward@arm.com>
712 David Sherwood <david.sherwood@arm.com>
714 * var-tracking.c (INT_MEM_OFFSET): Replace with...
715 (int_mem_offset): ...this new function.
716 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
717 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
718 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
721 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
722 Alan Hayward <alan.hayward@arm.com>
723 David Sherwood <david.sherwood@arm.com>
725 * lower-subreg.c (interesting_mode_p): New function.
726 (compute_costs, find_decomposable_subregs, decompose_register)
727 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
728 (resolve_clobber, dump_choices): Use it.
730 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
731 Alan Hayward <alan.hayward@arm.com>
732 David Sherwood <david.sherwood@arm.com>
734 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
736 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
737 Alan Hayward <alan.hayward@arm.com>
738 David Sherwood <david.sherwood@arm.com>
740 * alias.c (find_base_value, find_base_term): Only process integer
741 truncations. Check the precision rather than the size.
743 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
744 Alan Hayward <alan.hayward@arm.com>
745 David Sherwood <david.sherwood@arm.com>
747 * machmode.h (is_narrower_int_mode): New function
748 * optabs.c (expand_float, expand_fix): Use it.
749 * dwarf2out.c (rotate_loc_descriptor): Likewise.
751 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
752 Alan Hayward <alan.hayward@arm.com>
753 David Sherwood <david.sherwood@arm.com>
755 * rtl.h (narrower_subreg_mode): New function.
756 * ira-color.c (update_costs_from_allocno): Use it.
758 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
759 Alan Hayward <alan.hayward@arm.com>
760 David Sherwood <david.sherwood@arm.com>
762 * optabs-query.h (convert_optab_p): New function, split out from...
763 (convert_optab_handler): ...here.
764 (widening_optab_handler): Delete.
765 (find_widening_optab_handler): Remove permit_non_widening parameter.
766 (find_widening_optab_handler_and_mode): Likewise. Provide an
767 override that operates on mode class wrappers.
768 * optabs-query.c (widening_optab_handler): Delete.
769 (find_widening_optab_handler_and_mode): Remove permit_non_widening
770 parameter. Assert that the two modes are the same class and that
771 the "from" mode is narrower than the "to" mode. Use
772 convert_optab_handler instead of widening_optab_handler.
773 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
774 instead of widening_optab_handler.
775 * expr.c (expand_expr_real_2): Update calls to
776 find_widening_optab_handler.
777 * optabs.c (expand_widen_pattern_expr): Likewise.
778 (expand_binop_directly): Take the insn_code as a parameter.
779 (expand_binop): Only call find_widening_optab_handler for
780 conversion optabs; use optab_handler otherwise. Update calls
781 to find_widening_optab_handler and expand_binop_directly.
782 Use convert_optab_handler instead of widening_optab_handler.
783 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
784 find_widening_optab_handler and use scalar_mode rather than
786 (convert_plusminus_to_widen): Likewise.
788 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
789 Alan Hayward <alan.hayward@arm.com>
790 David Sherwood <david.sherwood@arm.com>
792 * machmode.h (fixed_size_mode): New class.
793 * rtl.h (get_pool_mode): Return fixed_size_mode.
794 * gengtype.c (main): Add fixed_size_mode.
795 * target.def (get_raw_result_mode): Return a fixed_size_mode.
796 (get_raw_arg_mode): Likewise.
797 * doc/tm.texi: Regenerate.
798 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
799 * targhooks.c (default_get_reg_raw_mode): Likewise.
800 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
801 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
802 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
803 (msp430_get_raw_result_mode): Likewise.
804 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
805 * dbxout.c (dbxout_parms): Require fixed-size modes.
806 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
807 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
808 * omp-low.c (lower_oacc_reductions): Likewise.
809 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
810 (simplify_subreg): Update accordingly.
811 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
812 (force_const_mem): Update accordingly. Return NULL_RTX for modes
813 that aren't fixed-size.
814 (get_pool_mode): Return a fixed_size_mode.
815 (output_constant_pool_2): Take a fixed_size_mode.
817 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
818 Alan Hayward <alan.hayward@arm.com>
819 David Sherwood <david.sherwood@arm.com>
821 * doc/rtl.texi (vec_series): Document.
822 (const): Say that the operand can be a vec_series.
823 * rtl.def (VEC_SERIES): New rtx code.
824 * rtl.h (const_vec_series_p_1): Declare.
825 (const_vec_series_p): New function.
826 * emit-rtl.h (gen_const_vec_series): Declare.
827 (gen_vec_series): Likewise.
828 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
829 (gen_vec_series): Likewise.
830 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
831 * simplify-rtx.c (simplify_unary_operation): Handle negations
833 (simplify_binary_operation_series): New function.
834 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
835 (test_vector_ops_series): New function.
836 (test_vector_ops): Call it.
837 * config/powerpcspe/altivec.md (altivec_lvsl): Use
838 gen_const_vec_series.
839 (altivec_lvsr): Likewise.
840 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
842 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
843 Alan Hayward <alan.hayward@arm.com>
844 David Sherwood <david.sherwood@arm.com>
846 * doc/rtl.texi (const): Update description of address constants.
847 Say that vector constants are allowed too.
848 * common.md (E, F): Use CONSTANT_P instead of checking for
850 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
851 checking for CONST_VECTOR.
852 * expmed.c (make_tree): Use build_vector_from_val for a CONST
854 * expr.c (expand_expr_real_2): Check for vector modes instead
855 of checking for CONST_VECTOR.
856 * rtl.h (const_vec_p): New function.
857 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
858 (unwrap_const_vec_duplicate): Handle them here too.
860 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
861 David Malcolm <dmalcolm@redhat.com>
862 Alan Hayward <alan.hayward@arm.com>
863 David Sherwood <david.sherwood@arm.com>
865 * rtl.h (vec_duplicate_p): New function.
866 * selftest-rtl.c (assert_rtx_eq_at): New function.
867 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
868 (assert_rtx_eq_at): Declare.
869 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
870 * selftest-run-tests.c (selftest::run_tests): Call it.
871 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
872 (simplify_unary_operation_1): Recursively handle vector duplicates.
873 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
875 (simplify_subreg): Handle subregs of vector duplicates.
876 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
877 (selftest::simplify_rtx_c_tests): New functions.
879 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
880 Alan Hayward <alan.hayward@arm.com>
881 David Sherwood <david.sherwood@arm.com>
883 * emit-rtl.h (gen_const_vec_duplicate): Declare.
884 (gen_vec_duplicate): Likewise.
885 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
887 (gen_const_vector): ...here.
888 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
889 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
890 whose elements are all equal.
891 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
892 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
893 (simplify_relational_operation): Likewise.
894 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
896 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
897 (aarch64_expand_vector_init): Likewise.
898 * config/arm/arm.c (neon_vdup_constant): Likewise.
899 (neon_expand_vector_init): Likewise.
900 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
901 (arm_block_set_unaligned_vect): Likewise.
902 (arm_block_set_aligned_vect): Likewise.
903 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
904 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
905 (expand_vec_perm_even_odd_pack): Likewise.
906 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
907 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
908 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
909 gen_const_vec_duplicate.
910 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
911 * config/mips/mips.c (mips_gen_const_int_vector): Use
912 gen_const_vec_duplicate.
913 (mips_expand_vector_init): Use CONST0_RTX.
914 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
915 (define_split): Use gen_const_vec_duplicate.
916 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
917 (define_split): Use gen_const_vec_duplicate.
918 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
919 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
920 * config/spu/spu.c (spu_const): Likewise.
922 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
923 Alan Hayward <alan.hayward@arm.com>
924 David Sherwood <david.sherwood@arm.com>
926 * combine.c (can_change_dest_mode): Reject changes in
927 REGMODE_NATURAL_SIZE.
929 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
931 * configure.ac (--enable-libssp): New.
932 (gcc_cv_libc_provides_ssp): Check for explicit setting before
933 trying to determine target-specific default. Adjust indentation.
934 * configure: Regenerated.
935 * doc/install.texi (Configuration): Expand --disable-libssp
938 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
940 config/i386/i386.c (ix86_expand_epilogue): Correct stack
943 2017-10-31 Martin Jambor <mjambor@suse.cz>
946 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
948 2017-10-31 David Malcolm <dmalcolm@redhat.com>
950 * auto-profile.c (autofdo_source_profile::read): Use
951 UNKNOWN_LOCATION rather than 0.
952 * diagnostic-core.h (warning_at_rich_loc): Rename to...
953 (warning_at): ...this overload.
954 (warning_at_rich_loc_n): Rename to...
955 (warning_n): ...this overload.
956 (error_at_rich_loc): Rename to...
957 (error_at): ...this overload.
958 (pedwarn_at_rich_loc): Rename to...
959 (pedwarn): ...this overload.
960 (permerror_at_rich_loc): Rename to...
961 (permerror): ...this overload.
962 (inform_at_rich_loc): Rename to...
963 (inform): ...this overload.
964 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
965 (diagnostic_n_impl_richloc): Rename to...
966 (diagnostic_n_impl): ...this rich_location *-based decl.
967 (inform_at_rich_loc): Rename to...
968 (inform): ...this, and add an assertion.
969 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
970 (warning_at_rich_loc): Rename to...
971 (warning_at): ...this, and add an assertion.
972 (warning_at_rich_loc_n): Rename to...
973 (warning_n): ...this, and add an assertion.
974 (warning_n): Update location_t-based implementation for removal of
975 location_t-based diagnostic_n_impl.
976 (pedwarn_at_rich_loc): Rename to...
977 (pedwarn): ...this, and add an assertion.
978 (permerror_at_rich_loc): Rename to...
979 (permerror): ...this, and add an assertion.
980 (error_n): Update for removal of location_t-based diagnostic_n_impl.
981 (error_at_rich_loc): Rename to...
982 (error_at): ...this, and add an assertion.
983 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
984 (driver::do_spec_on_infiles): Likewise.
985 * substring-locations.c (format_warning_va): Update for renaming
986 of inform_at_rich_loc.
988 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
990 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
991 _Float<N>X built-in functions so that the variant without the
992 "__builtin_" prefix is only enabled for the GNU C and Objective C
993 languages when they are in non-strict ANSI/ISO mode.
994 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
995 * target.def (floatn_builtin_p): Add a target hook to control
996 whether _Float<N> and _Float<N>X built-in functions without the
997 "__builtin_" prefix are enabled, and return true for C and
998 Objective C in the default hook. Include langhooks.h in
1000 * targhooks.h (default_floatn_builtin_p): Likewise.
1001 * targhooks.c (default_floatn_builtin_p): Likewise.
1002 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
1003 floatn_builtin_p target hook.
1004 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
1006 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
1007 Eric Botcazou <ebotcazou@adacore.com>
1009 PR rtl-optimization/81803
1010 * lra-constraints.c (curr_insn_transform): Also reload the whole
1011 register for a strict subreg no wider than a word if this is for
1012 a WORD_REGISTER_OPERATIONS target.
1014 2017-10-31 Jason Merrill <jason@redhat.com>
1016 * gdbinit.in: Skip over inlines from timevar.h.
1018 2017-10-31 Martin Liska <mliska@suse.cz>
1020 * doc/gcov.texi: Document new option.
1021 * gcov.c (print_usage): Likewise print it.
1022 (process_args): Support the argument.
1023 (format_count): New function.
1024 (format_gcov): Use the function.
1026 2017-10-31 Martin Liska <mliska@suse.cz>
1028 * gcov.c (struct name_map): do not use typedef.
1029 Define operator== and operator<.
1030 (name_search): Remove.
1031 (name_sort): Remove.
1032 (main): Do not allocate names.
1033 (process_file): Add vertical space.
1034 (generate_results): Use std::find.
1035 (release_structures): Do not release memory.
1036 (find_source): Use std::find.
1038 2017-10-31 Martin Liska <mliska@suse.cz>
1040 * gcov.c (struct line_info): Remove it's typedef.
1041 (line_info::line_info): Add proper ctor.
1042 (line_info::has_block): Do not use a typedef.
1043 (struct source_info): Do not use typedef.
1044 (circuit): Likewise.
1045 (get_cycles_count): Likewise.
1046 (output_intermediate_file): Iterate via vector iterator.
1047 (add_line_counts): Use std::vector methods.
1048 (accumulate_line_counts): Likewise.
1049 (output_lines): Likewise.
1051 2017-10-31 Martin Liska <mliska@suse.cz>
1053 * gcov.c (struct source_info): Remove typedef.
1054 (source_info::source_info): Add proper ctor.
1055 (accumulate_line_counts): Use struct, not it's typedef.
1056 (output_gcov_file): Likewise.
1057 (output_lines): Likewise.
1058 (main): Do not allocate an array.
1059 (output_intermediate_file): Use size of vector container.
1060 (process_file): Resize the vector.
1061 (generate_results): Do not preallocate, use newly added vector
1063 (release_structures): Do not release sources.
1064 (find_source): Use vector methods.
1065 (add_line_counts): Do not use typedef.
1067 2017-10-31 Martin Liska <mliska@suse.cz>
1069 * doc/gcov.texi: Document that.
1070 * gcov.c (add_line_counts): Mark lines with a non-executed
1072 (output_line_beginning): Handle such lines.
1073 (output_lines): Pass new argument.
1074 (output_intermediate_file): Print it in intermediate format.
1076 2017-10-31 Martin Liska <mliska@suse.cz>
1078 * color-macros.h: New file.
1079 * diagnostic-color.c: Factor out color related to macros to
1081 * doc/gcov.texi: Document -k option.
1082 * gcov.c (INCLUDE_STRING): Include string.h.
1083 (print_usage): Add -k option.
1084 (process_args): Parse it.
1085 (pad_count_string): New function.
1086 (output_line_beginning): Likewise.
1087 (DEFAULT_LINE_START): New macro.
1088 (output_lines): Support color output.
1090 2017-10-31 Martin Liska <mliska@suse.cz>
1092 PR gcov-profile/82633
1093 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
1094 their interaction with GCOV infrastructure.
1095 * configure.ac: Add -fkeep-{inline,static}-functions to
1097 * configure: Regenerate.
1099 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
1102 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
1104 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
1107 * config/rs6000/rs6000.md (allocate_stack): Force update interval
1108 into a register if it does not fit into an immediate offset field.
1110 2017-10-31 Olivier Hainque <hainque@adacore.com>
1112 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
1114 2017-10-31 Julia Koval <julia.koval@intel.com>
1116 * config.gcc: Add gfniintrin.h.
1117 * config/i386/gfniintrin.h: New.
1118 * config/i386/i386-builtin-types.def
1119 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1120 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1121 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1122 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1123 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1124 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
1125 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
1126 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
1127 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
1128 V64QI_FTYPE_V64QI_V64QI_INT): New types.
1129 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1130 * config/i386/immintrin.h: Include gfniintrin.h.
1131 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
1133 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
1135 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
1137 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1139 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
1140 (arm_ashldi3_1bit): Remove pattern.
1141 (ashrdi3): Remove shift by 1 expansion.
1142 (arm_ashrdi3_1bit): Remove pattern.
1143 (lshrdi3): Remove shift by 1 expansion.
1144 (arm_lshrdi3_1bit): Remove pattern.
1145 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
1146 cost of ashldi3 by 1.
1147 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
1148 (<shift>di3_neon): Likewise.
1150 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
1152 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
1153 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
1154 and (*aarch64_simd_mov<VQ:mode>).
1155 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
1156 pattern alternative.
1157 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
1158 attributes to match pattern alternative.
1160 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
1162 * config.gcc (powerpc*-*-*): Add emmintrin.h.
1163 * config/rs6000/emmintrin.h: New file.
1164 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
1166 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1168 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
1169 * (movdi_vfp_cortexa8): Remove pattern.
1171 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1173 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
1175 (Specific, alpha*-dec-osf5.1): Remove.
1176 (Specific, mips-sgi-irix5): Remove.
1177 (Specific, mips-sgi-irix6): Remove.
1179 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1182 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
1183 arguments to clear_bit_region_be.
1185 2017-10-30 Jim Wilson <wilson@tuliptree.org>
1187 * gimplify.c: Include memmodel.h.
1189 2017-10-30 Martin Jambor <mjambor@suse.cz>
1191 * omp-grid.c (grid_attempt_target_gridification): Also insert a
1192 condition whether loop should be executed at all.
1194 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
1196 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1197 gimple folding of vec_madd() intrinsics.
1198 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
1199 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
1200 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
1202 2017-10-30 Richard Biener <rguenther@suse.de>
1204 PR tree-optimization/82762
1206 2017-10-23 Richard Biener <rguenther@suse.de>
1208 PR tree-optimization/82129
1210 2017-08-01 Richard Biener <rguenther@suse.de>
1212 PR tree-optimization/81181
1213 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
1214 (compute_antic): ... end of iteration here.
1216 2017-10-30 Joseph Myers <joseph@codesourcery.com>
1218 * doc/invoke.texi (C Dialect Options): Document -std=c17,
1219 -std=iso9899:2017 and -std=gnu17.
1220 * doc/standards.texi (C Language): Document C17 support.
1221 * doc/cpp.texi (Overview): Mention -std=c17.
1222 (Standard Predefined Macros): Document C11 and C17 values of
1223 __STDC_VERSION__. Do not refer to C99 support as incomplete.
1224 * doc/extend.texi (Inline): Do not list individual options for
1225 standards newer than C99.
1226 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
1228 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
1231 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
1233 * asan.c (asan_finish_file): Align asan globals array by shadow
1236 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1239 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
1240 (struct store_immediate_info): Add bitregion_start and bitregion_end
1242 (store_immediate_info::store_immediate_info): Add brs and bre
1243 arguments and initialize bitregion_{start,end} from those.
1244 (struct merged_store_group): Add bitregion_start, bitregion_end,
1245 align_base and mask fields. Drop unnecessary struct keyword from
1246 struct store_immediate_info. Add do_merge method.
1247 (clear_bit_region_be): Use memset instead of loop storing zeros.
1248 (merged_store_group::do_merge): New method.
1249 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
1250 stores as long as the surrounding bitregions have no gaps.
1251 (merged_store_group::merge_overlapping): Use do_merge.
1252 (merged_store_group::apply_stores): Test that bitregion_{start,end}
1253 is byte aligned, rather than requiring that start and width are
1254 byte aligned. Drop unnecessary struct keyword from
1255 struct store_immediate_info. Allocate and populate also mask array.
1256 Make start of the arrays relative to bitregion_start rather than
1257 start and size them according to bitregion_{end,start} difference.
1258 (struct imm_store_chain_info): Drop unnecessary struct keyword from
1259 struct store_immediate_info.
1260 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
1261 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
1262 struct keyword from struct store_immediate_info.
1263 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
1264 between stores as long as the surrounding bitregions have no gaps.
1266 (struct split_store): Add orig non-static data member.
1267 (split_store::split_store): Initialize orig to false.
1268 (find_constituent_stmts): Return store_immediate_info *, non-NULL
1269 if there is exactly a single original stmt. Change stmts argument
1270 to pointer from reference, if NULL, don't push anything to it. Add
1271 first argument, use it to optimize skipping over orig stmts that
1272 are known to be before bitpos already. Simplify.
1273 (split_group): Return unsigned int count how many stores are or
1274 would be needed rather than a bool. Add allow_unaligned argument.
1275 Change split_stores argument from reference to pointer, if NULL,
1276 only do a dry run computing how many stores would be produced.
1277 Rewritten algorithm to use both alignment and misalign if
1278 !allow_unaligned and handle bitfield stores with gaps.
1279 (imm_store_chain_info::output_merged_store): Set start_byte_pos
1280 from bitregion_start instead of start. Compute allow_unaligned
1281 here, if true, do 2 split_group dry runs to compute which one
1282 produces fewer stores and prefer aligned if equal. Punt if
1283 new count is bigger or equal than original before emitting any
1284 statements, rather than during that. Remove no longer needed
1285 new_ssa_names tracking. Replace num_stmts with
1286 split_stores.length (). Use 32-bit stack allocated entries
1287 in split_stores auto_vec. Try to reuse original store lhs/rhs1
1288 if possible. Handle bitfields with gaps.
1289 (pass_store_merging::execute): Ignore bitsize == 0 stores.
1290 Compute bitregion_{start,end} for the stores and construct
1291 store_immediate_info with that. Formatting fixes.
1293 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
1296 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
1297 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
1299 2017-10-29 Jim Wilson <wilson@tuliptree.org>
1301 * gimplify.c: Include tm_p.h.
1303 * common.opt (gcoff): Re-add as ignored option.
1304 (gcoff1, gcoff2, gcoff3): Likewise.
1306 * Makefile.in (OBJS): Delete sdbout.o.
1307 (GTFILES): Delete $(srcdir)/sdbout.c.
1308 * debug.h: Delete sdb_debug_hooks.
1309 * final.c: Delete sdbout.h include.
1310 (final_scan_insn): Delete SDB_DEBUG check.
1311 (rest_of_clean_state): Likewise.
1312 * output.h: Delete sdb_begin_function_line.
1315 * toplev.c: Delete sdbout.h include.
1316 (process_options): Delete SDB_DEBUG check.
1317 * tree-core.h (tree_type_common): Delete pointer field of
1319 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
1320 TYPE_SYMTAB_POINTER.
1321 * tree.h (TYPE_SYMTAB_POINTER): Delete.
1322 (TYPE_SYMTAB_IS_POINTER): Delete.
1323 (TYPE_SYMTAB_IS_DIE): Renumber.
1324 * xcoffout.c: Refer to former sdbout.c file.
1325 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
1327 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
1328 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
1329 (Debugging Options): Delete -gcoff.
1330 (-gstabs): Delete SDB reference.
1332 (-gcoff@var{level}): Delete.
1333 * doc/passes.texi (Debugging information output): Delete SDB and
1334 sdbout.c references.
1335 * doc/tm.texi: Regenerate.
1336 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
1337 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
1339 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
1340 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
1341 (SDB_DEBUGGING_INFO): Delete.
1342 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
1343 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
1344 * target.def (output_source_filename): Delete COFF reference.
1346 * common.opt (gcoff): Delete.
1347 (gxcoff+): Update Negative chain.
1348 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
1350 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
1351 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
1352 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
1354 (expand_function_start): Change sdb reference to past tense.
1355 (expand_function_end): Change sdb reference to past tense.
1356 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
1357 * opts.c (debug_type_names): Delete coff entry.
1358 (common_handle_option): Delete OPT_gcoff case.
1359 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
1361 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
1362 * config/cris/cris.h: Delete SDB reference in comment.
1363 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
1364 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
1365 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
1366 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
1368 (ix86_expand_prologue): Likewise.
1369 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
1370 * config/ia64/ia64.h: Likewise.
1371 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
1372 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
1374 * config/mmix/mmix.h: Likewise.
1375 * config/nds32/nds32.c: Likewise.
1376 * config/stormy/storym16.h: Likewise.
1377 * config/visium/visium.h: Likewise.
1378 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
1380 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
1382 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
1383 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
1384 FRAME_POINTER_REGNUM point at high end of local var area.
1386 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
1388 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
1389 Move comment around. Do not reset best_edge for a copiable
1390 destination if the copy would cause a partition change.
1391 (better_edge_p): Remove redundant check.
1393 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
1395 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
1397 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
1399 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
1400 for math functions that have _Float<N> and _Float<N>X variants.
1401 (mathfn_built_in_2): Add support for math functions that have
1402 _Float<N> and _Float<N>X variants.
1403 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
1404 (expand_builtin_mathfn_ternary): Add support for fma with
1405 _Float<N> and _Float<N>X variants.
1406 (expand_builtin): Likewise.
1407 (fold_builtin_3): Likewise.
1408 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
1409 create math function _Float<N> and _Float<N>X variants as external
1411 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
1412 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
1413 the __builtin_ prefix and if not strict ansi, without the prefix.
1414 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
1415 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
1416 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
1417 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
1418 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
1419 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
1420 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
1421 function signatures for fma _Float<N> and _Float<N>X variants.
1422 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
1423 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
1424 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
1425 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
1426 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
1427 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
1428 * gencfn-macros.c (print_case_cfn): Add support for math functions
1429 that have _Float<N> and _Float<N>X variants.
1430 (print_define_operator_list): Likewise.
1431 (fltfn_suffixes): Likewise.
1433 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
1434 for math functions that have _Float<N> and _Float<N>X variants.
1435 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
1436 and _Float<N>X variants.
1437 (COPYSIGN): Likewise.
1440 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
1441 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
1443 (integer_valued_read_call_p): Likewise.
1444 * fold-const-call.c (fold_const_call_ss): Likewise.
1445 (fold_const_call_sss): Add support for copysign, fmin, and fmax
1446 _Float<N> and _Float<N>X variants.
1447 (fold_const_call_ssss): Add support for fma _Float<N> and
1448 _Float<N>X variants.
1449 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
1450 support for copysign and fma _Float<N> and _Float<N>X variants.
1451 (backprop::process_builtin_call_use): Likewise.
1452 * tree-call-cdce.c (can_test_argument_range); Add support for
1453 sqrt _Float<N> and _Float<N>X variants.
1454 (edom_only_function): Likewise.
1455 (get_no_error_domain): Likewise.
1456 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
1457 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
1458 copysign _Float<N> and _Float<N>X variants.
1459 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
1460 handled by machine independent code.
1461 (FMAF128): Likewise.
1462 * doc/cpp.texi (Common Predefined Macros): Document defining
1463 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
1464 fma _Float<N> and _Float<N>X variants.
1466 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
1469 * config/i386/i386-modes.def (CCFPU): Remove definition.
1470 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
1471 (ix86_cc_modes_compatible): Ditto.
1472 (ix86_expand_carry_flag_compare): Ditto.
1473 (ix86_expand_int_movcc): Ditto.
1474 (ix86_expand_int_addcc): Ditto.
1475 (ix86_reverse_condition): Ditto.
1476 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
1477 Return true/false for unordered/ordered fp comparisons.
1478 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
1479 (ix86_prepare_fp_compare_args): Update for rename.
1480 (ix86_expand_fp_compare): Update for rename. Generate unordered
1481 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
1482 (ix86_expand_sse_compare_and_jump): Ditto.
1483 * config/i386/predicates.md (fcmov_comparison_operator):
1484 Remove CCFPU mode handling.
1485 (ix86_comparison_operator): Ditto.
1486 (ix86_carry_flag_operator): Ditto.
1487 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
1488 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
1489 (*cmpu<mode>_cc_i387): Ditto.
1490 (FPCMP): Remove mode iterator.
1491 (unord): Remove mode attribute.
1492 (unord_subst): New define_subst transformation
1493 (unord): New define_subst attribute.
1495 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
1496 (*cmpi<unord>xf_i387): Ditto.
1497 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
1498 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
1499 using unord_subst transformation.
1500 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
1501 (round_saeonly): Also handle CCFP mode.
1502 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
1503 Remove UNSPEC_SAHF unspec handling.
1505 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
1507 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
1509 2017-10-27 Jeff Law <law@redhat.com>
1511 * gimple-ssa-sprintf.c: Include domwalk.h.
1512 (class sprintf_dom_walker): New class, derived from dom_walker.
1513 (sprintf_dom_walker::before_dom_children): New function.
1514 (struct call_info): Moved into sprintf_dom_walker class
1515 (compute_formath_length, handle_gimple_call): Likewise.
1516 (sprintf_length::execute): Call the dominator walker rather
1517 than walking the statements.
1519 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
1520 gimple statement locations.
1521 (check_array_bounds): Corresponding changes. Get the statement's
1522 location directly from wi->stmt.
1524 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
1527 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
1529 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
1531 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
1532 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
1534 2017-10-27 Jakub Jelinek <jakub@redhat.com>
1537 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
1538 * config/i386/i386.c (maybe_get_pool_constant): Removed.
1539 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
1540 maybe_get_pool_constant.
1541 * config/i386/predicates.md (zero_extended_scalar_load_operand):
1544 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1546 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
1547 2.26 caveat. Update gas and gld versions.
1548 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
1551 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1553 * cgraph.h (set_malloc_flag): Declare.
1554 * cgraph.c (set_malloc_flag_1): New function.
1555 (set_malloc_flag): Likewise.
1556 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
1557 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
1559 (read_ipa_call_summary): Add support for reading is_return_callee.
1560 (write_ipa_call_summary): Stream is_return_callee.
1561 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
1562 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
1563 ipa-prop.h, ipa-fnsummary.h.
1564 (pure_const_names): Change to static.
1565 (malloc_state_e): Define.
1566 (malloc_state_names): Define.
1567 (funct_state_d): Add field malloc_state.
1568 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
1569 (check_retval_uses): New function.
1570 (malloc_candidate_p): Likewise.
1571 (analyze_function): Add support for malloc attribute.
1572 (pure_const_write_summary): Stream malloc_state.
1573 (pure_const_read_summary): Add support for reading malloc_state.
1574 (dump_malloc_lattice): New function.
1575 (propagate_malloc): New function.
1576 (warn_function_malloc): New function.
1577 (ipa_pure_const::execute): Call propagate_malloc and
1578 ipa_free_fn_summary.
1579 (pass_local_pure_const::execute): Add support for malloc attribute.
1580 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
1581 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
1583 2017-10-27 Martin Liska <mliska@suse.cz>
1585 PR gcov-profile/82457
1586 * doc/invoke.texi: Document that one needs a non-strict ISO mode
1587 for fork-like functions to be properly instrumented.
1589 2017-10-27 Richard Biener <rguenther@suse.de>
1592 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
1593 info when we redirected EH.
1595 2017-10-26 Michael Collison <michael.collison@arm.com>
1597 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
1599 (<optab>_trunchf<GPI:mode>2: New pattern.
1600 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
1601 * config/aarch64/iterators.md (wv): New mode attribute.
1602 (vf, VF): New mode attributes.
1603 (vgp, VGP): New mode attributes.
1604 (s): Update attribute with SImode and DImode prefixes.
1606 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
1608 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
1609 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
1610 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
1611 (nios2_option_overide): Initialize it. Don't allow R0-relative
1612 addressing with PIC.
1613 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
1614 (nios2_symbolic_constant_p): Likewise.
1615 (nios2_legitimate_address_p): Likewise.
1616 (nios2_r0rel_section_name_p): New.
1617 (nios2_symbol_ref_in_r0rel_data_p): New.
1618 (nios2_emit_move_sequence): Handle r0rel_constant_p.
1619 (r0rel_constant_p): New.
1620 (nios2_print_operand_address): Handle r0rel_constant_p.
1621 (nios2_cdx_narrow_form_p): Likewise.
1622 * config/nios2/nios2.opt (mr0rel-sec=): New option.
1623 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
1624 (Nios II Options): Document -mr0rel-sec.
1626 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
1628 * config/nios2/nios2.c: Include xregex.h.
1629 (nios2_gprel_sec_regex): New.
1630 (nios2_option_overide): Initialize it. Don't allow GP-relative
1631 addressing with PIC.
1632 (nios2_small_section_name_p): Check for regex match.
1633 * config/nios2/nios2.opt (mgprel-sec=): New option.
1634 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
1635 (Nios II Options): Document -mgprel-sec.
1637 2017-10-26 Jim Wilson <wilson@tuliptree.org>
1639 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
1641 2017-10-26 Tom de Vries <tom@codesourcery.com>
1643 PR tree-optimization/82707
1644 * gimple.c (gimple_copy): Fix unsharing of
1645 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
1647 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
1649 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
1650 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
1651 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
1652 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
1653 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
1654 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
1655 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
1656 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
1657 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
1658 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
1659 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
1660 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
1661 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
1662 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
1663 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
1664 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
1665 _mm512_mask_cmpunord_ps_mask): New intrinsics.
1667 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
1669 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
1671 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
1672 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
1673 warning to rs6000.c. Remove the Undocumented flag, since it has
1675 (-mabi=ibmlongdouble): Likewise.
1676 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
1677 already set, set the default format for long double.
1678 (rs6000_debug_reg_global): Print whether long double is IBM or
1680 (rs6000_option_override_internal): Rework setting long double
1681 format. Only warn if the user is changing the long double default
1682 and they did not use -Wno-psabi.
1683 * doc/invoke.texi (PowerPC options): Update the documentation for
1684 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
1686 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
1687 Alan Hayward <alan.hayward@arm.com>
1688 David Sherwood <david.sherwood@arm.com>
1690 * rtl.h (wider_subreg_mode): New function.
1691 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
1692 rather than an unsigned int *.
1693 * ira-color.c (regno_max_ref_width): Replace with...
1694 (regno_max_ref_mode): ...this new variable.
1695 (coalesced_pseudo_reg_slot_compare): Update accordingly.
1696 Use wider_subreg_mode.
1697 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
1698 rather than an unsigned int *.
1699 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
1700 (process_alt_operands): Likewise.
1701 (invariant_p): Likewise.
1702 * lra-spills.c (assign_mem_slot): Likewise.
1703 (add_pseudo_to_slot): Likewise.
1704 * lra.c (collect_non_operand_hard_regs): Likewise.
1705 (add_regs_to_insn_regno_info): Likewise.
1706 * reload1.c (regno_max_ref_width): Replace with...
1707 (regno_max_ref_mode): ...this new variable.
1708 (reload): Update accordingly. Update call to
1709 ira_sort_regnos_for_alter_reg.
1710 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
1711 (init_eliminable_invariants): Update to use regno_max_ref_mode.
1712 (scan_paradoxical_subregs): Likewise.
1714 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
1716 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
1717 (aarch64_frame): Add emit_frame_chain boolean.
1718 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1719 Move eh_return case to aarch64_layout_frame.
1720 (aarch64_layout_frame): Initialize emit_frame_chain.
1721 (aarch64_expand_prologue): Use emit_frame_chain.
1723 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
1725 * config/aarch64/aarch64.c (aarch64_layout_frame):
1726 Ensure LR is always stored at the bottom of the callee-saves.
1727 Remove rarely used frame layout which saves callee-saves at top of
1728 frame, so the store of LR can be used as a valid probe in all cases.
1730 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
1732 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
1733 Improve unaligned TImode/TFmode base/offset split.
1735 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
1736 Alan Hayward <alan.hayward@arm.com>
1737 David Sherwood <david.sherwood@arm.com>
1739 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
1740 * combine.c (find_single_use_1): Likewise.
1741 (expand_field_assignment): Likewise.
1742 (move_deaths): Likewise.
1743 * lra-constraints.c (simplify_operand_subreg): Likewise.
1744 (curr_insn_transform): Likewise.
1745 * lra.c (collect_non_operand_hard_regs): Likewise.
1746 (add_regs_to_insn_regno_info): Likewise.
1747 * rtlanal.c (reg_referenced_p): Likewise.
1748 (covers_regno_no_parallel_p): Likewise.
1750 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
1752 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
1753 Don't print any bits outside the precision of the value.
1754 * wide-int.cc (test_printing): Add some new tests.
1756 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1758 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
1759 supports -xbrace_comment option.
1760 * configure: Regenerate.
1761 * config.in: Regenerate.
1762 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
1763 (ASM_CPU_SPEC): Use it.
1765 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
1767 * target.def (static_rtx_alignment): New hook.
1768 * targhooks.h (default_static_rtx_alignment): Declare.
1769 * targhooks.c (default_static_rtx_alignment): New function.
1770 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
1771 * doc/tm.texi: Regenerate.
1772 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
1773 instead of targetm.constant_alignment. Remove call to
1775 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
1776 (cris_preferred_mininum_alignment): New function, split out from...
1777 (cris_constant_alignment): ...here.
1778 (cris_static_rtx_alignment): New function.
1779 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
1781 (ix86_constant_alignment): ...here.
1782 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
1783 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
1784 (mmix_static_rtx_alignment): New function.
1785 * config/spu/spu.c (spu_static_rtx_alignment): New function.
1786 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
1788 2017-10-26 Tamar Christina <tamar.christina@arm.com>
1791 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
1792 Add flag_trapping_math and flag_fp_int_builtin_inexact.
1794 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
1796 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
1797 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
1798 mark as a sign-extending load.
1799 (local_pic_load_u): Define.
1801 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
1804 * fold-const.c (operand_equal_for_comparison_p): Also return true
1805 if ARG0 is a simple variant of ARG1 with narrower precision.
1806 (fold_ternary_loc): Always pass unstripped operands to the predicate.
1808 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
1810 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
1812 * i386.h (processor_costs): Add gather_static, gather_per_elt,
1813 scatter_static, scatter_per_elt.
1814 * x86-tune-costs.h: Add new cost entries.
1816 2017-10-25 Richard Biener <rguenther@suse.de>
1818 * tree-ssa-sccvn.h (vn_eliminate): Declare.
1819 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
1820 class pass_fre): Move to ...
1821 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
1822 class pass_fre): ... here and adjust for statistics.
1824 2017-10-25 Jakub Jelinek <jakub@redhat.com>
1827 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
1828 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
1829 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
1830 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
1833 2017-10-25 Richard Biener <rguenther@suse.de>
1835 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
1836 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
1837 eliminate_push_avail, eliminate_insert): Move inside...
1838 (class eliminate_dom_walker): ... this class in preparation
1840 (fini_eliminate): Remove by merging with ...
1841 (eliminate): ... this function. Adjust for class changes.
1842 (pass_pre::execute): Remove fini_eliminate call.
1843 (pass_fre::execute): Likewise.
1845 2017-10-24 Jakub Jelinek <jakub@redhat.com>
1848 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
1849 (VPERMI2, VPERMI2I): New mode iterators.
1850 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
1851 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
1853 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
1854 mode iterator. Remove 3 old define_insn patterns.
1855 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
1856 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
1857 VPERMI2 mode iterator, remove the other two expanders.
1858 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
1859 to use VPERMI2 mode iterator, add another alternative for vpermi2*
1860 instructions, remove the other two patterns.
1861 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
1862 mode iterator, remove the other two patterns.
1863 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
1864 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
1865 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
1866 and adjust argument order accordingly.
1867 (ix86_expand_vec_perm): Adjust caller.
1868 (expand_vec_perm_1): Likewise.
1869 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
1870 (expand_vec_perm_vpermt2_vpshub2): ... this.
1871 (ix86_expand_vec_perm_const_1): Adjust caller.
1872 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
1875 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
1876 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
1877 (vec_shl_<mode>): Remove unused expander.
1878 (avx512bw_<shift_insn><mode>3): New define_insn.
1879 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
1880 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
1882 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
1885 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
1888 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
1890 PR rtl-optimization/82396
1891 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
1892 (autopref_multipass_init): Simplify initialization.
1893 (autopref_rank_data): Simplify sort order.
1894 * gcc/sched-int.h (autopref_multipass_data_): Remove
1895 multi_mem_insn_p, min_offset and max_offset.
1897 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
1900 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1901 Check special value of flag_omit_frame_pointer.
1902 (aarch64_can_eliminate): Likewise.
1903 (aarch64_override_options_after_change_1): Simplify handling of
1904 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
1906 2017-10-24 Richard Biener <rguenther@suse.de>
1908 PR tree-optimization/82697
1909 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
1910 zero for conditional load and unconditional store.
1912 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
1914 * doc/install.texi: Document bootstrap-cet.
1916 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
1919 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
1920 ENDBR instruction at function entrance if function is only
1923 2017-10-24 Jakub Jelinek <jakub@redhat.com>
1926 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
1927 patterns to better describe from which operation the CF is computed.
1928 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
1929 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
1930 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
1931 is 0, use _0 suffixed expanders instead of emitting a comparison
1934 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
1936 * config/i386/i386.md(*movsf_internal, *movdf_internal):
1937 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
1939 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
1942 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
1943 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
1944 * loop-iv.c (iv_get_reaching_def): Likewise.
1945 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
1946 variable is promoted and the partition contains undefined values.
1948 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
1950 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
1952 (nios2_address_cost): Define.
1953 (nios2_legitimize_address): Recognize (exp + constant) directly.
1954 (TARGET_ADDRESS_COST): Define.
1956 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
1958 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
1959 (nios2_symbolic_memory_operand_p): Declare.
1960 (nios2_split_large_constant): Declare.
1961 (nios2_split_symbolic_memory_operand): Declare.
1962 * config/nios2/nios2.c: Adjust includes.
1963 (nios2_symbolic_constant_allowed): New.
1964 (nios2_symbolic_constant_p): New.
1965 (nios2_plus_symbolic_constant_p): New.
1966 (nios2_valid_addr_expr_p): Recognize addresses involving
1968 (nios2_legitimate_address_p): Likewise, also LO_SUM.
1969 (nios2_symbolic_memory_operand_p): New.
1970 (nios2_large_constant_p): New.
1971 (nios2_split_large_constant): New.
1972 (nios2_split_plus_large_constant): New.
1973 (nios2_split_symbolic_memory_operand): New.
1974 (nios2_legitimize_address): Code refactoring. Handle addresses
1975 involving symbolic constants.
1976 (nios2_emit_move_sequence): Likewise.
1977 (nios2_print_operand): Improve error output.
1978 (nios2_print_operand_address): Handle LO_SUM.
1979 (nios2_cdx_narrow_form_p): Likewise.
1980 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
1981 operands involving symbolic constants.
1982 (movhi_internal, movsi_internal): Likewise.
1983 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
1984 (extendhisi2, extendqi<mode>2): Likewise.
1986 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
1988 * tree-pass.h (PROP_rtl_split_insns): Define.
1989 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
1991 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
1993 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
1995 2017-10-23 Jakub Jelinek <jakub@redhat.com>
1998 * target.def (const_not_ok_for_debug_p): Default to
1999 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
2000 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
2001 * targhooks.c (default_const_not_ok_for_debug_p): New function.
2002 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
2003 which targetm.const_not_ok_for_debug_p returned true.
2004 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
2006 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
2008 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
2009 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
2010 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
2012 (ix86_const_not_ok_for_debug_p): New function.
2013 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
2014 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
2016 2017-10-23 David Malcolm <dmalcolm@redhat.com>
2019 * system.h: Conditionally include "unique-ptr.h" if
2020 INCLUDE_UNIQUE_PTR is defined.
2021 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
2022 of defining INCLUDE_UNIQUE_PTR before including "system.h".
2024 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
2026 * config/rl78/rl78.md: New define_expand "subdi3".
2028 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
2031 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
2032 DF_REF_INSN if DF_REF_INSN_INFO is false.
2034 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
2036 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
2037 xmm_move instead of sse_move.
2038 (sse_store_index): New function.
2039 (ix86_register_move_cost): Be more sensible about mismatch stall;
2040 model AVX moves correctly; make difference between sse->integer and
2042 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
2043 moves; make difference between SSE and AVX.
2044 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
2045 and zmm_move. Increase size of sse load and store tables;
2046 add unaligned load and store tables; add ssemmx_to_integer.
2047 * x86-tune-costs.h: Update all entries according to real
2048 move latencies from Agner Fog's manual and chip documentation.
2050 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2053 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
2054 * config/i386/constraints.md (Wf): New constraint.
2055 * config/i386/i386.md (UNSPEC_SBB): New unspec.
2056 (cmp<dwi>_doubleword): Removed.
2057 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
2058 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
2059 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
2060 expand with cmp<dwi>_doubleword. For LTU and GEU use
2061 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
2063 * common.opt (gcolumn-info): Enable by default.
2064 * doc/invoke.texi (gcolumn-info): Document new default.
2066 2017-10-23 Richard Biener <rguenther@suse.de>
2068 PR tree-optimization/82672
2069 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
2070 Fold the stmt if we propagated into it.
2072 2017-10-23 Richard Biener <rguenther@suse.de>
2074 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
2075 (bitmap_remove_expr_from_set): ... this. All callers call this
2076 for non-constant values.
2077 (bitmap_set_subtract): Rename to...
2078 (bitmap_set_subtract_expressions): ... this. Adjust and
2080 (bitmap_set_contains_value): Remove superfluous check.
2081 (bitmap_set_replace_value): Inline into single caller ...
2082 (bitmap_value_replace_in_set): ... here and simplify.
2083 (dependent_clean): Merge into ...
2084 (clean): ... this using an overload. Adjust.
2085 (prune_clobbered_mems): Adjust.
2086 (compute_antic_aux): Likewise.
2087 (compute_partial_antic_aux): Likewise.
2089 2017-10-23 Richard Biener <rguenther@suse.de>
2091 PR tree-optimization/82129
2093 2017-08-01 Richard Biener <rguenther@suse.de>
2095 PR tree-optimization/81181
2096 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2097 (compute_antic): ... end of iteration here.
2099 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2101 * target.def (starting_frame_offset): New hook.
2102 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
2103 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
2104 * doc/tm.texi.in: Regenerate.
2105 * hooks.h (hook_hwi_void_0): Declare.
2106 * hooks.c (hook_hwi_void_0): New function.
2107 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
2108 STARTING_FRAME_OFFSET.
2109 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
2110 * reload1.c (reload): Likewise.
2111 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
2112 instead of STARTING_FRAME_OFFSET.
2113 * function.c (try_fit_stack_local): Likewise.
2114 (assign_stack_local_1): Likewise
2115 (instantiate_virtual_regs): Likewise.
2116 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
2117 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
2118 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
2119 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
2120 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
2121 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
2122 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
2123 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
2124 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
2125 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
2126 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
2127 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
2128 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
2129 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
2130 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
2131 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
2132 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
2133 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
2134 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
2135 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
2136 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
2137 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
2138 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
2139 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
2140 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
2141 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
2142 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
2143 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
2144 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
2145 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
2146 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
2147 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
2148 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
2149 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
2150 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
2151 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
2152 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
2153 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
2154 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
2155 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
2156 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
2157 * config/avr/avr.c (avr_starting_frame_offset): Make static and
2158 return a HOST_WIDE_INT.
2159 (avr_builtin_setjmp_frame_value): Use it instead of
2160 STARTING_FRAME_OFFSET.
2161 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2162 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
2163 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
2165 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2166 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
2167 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
2168 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2169 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
2170 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
2171 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2172 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
2173 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
2174 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2175 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
2176 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
2178 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2179 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
2180 * config/mips/mips.c (mips_compute_frame_info): Refer to
2181 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2182 (mips_starting_frame_offset): New function.
2183 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2184 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
2185 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
2186 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
2187 and return a HOST_WIDE_INT.
2188 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2189 (mmix_initial_elimination_offset): Refer to
2190 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2191 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
2192 * config/pa/pa.c (pa_starting_frame_offset): New function.
2193 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
2194 (pa_expand_prologue): Likewise.
2195 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2196 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
2197 !FRAME_GROWS_DOWNWARD handling to...
2198 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2199 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
2200 !FRAME_GROWS_DOWNWARD handling to...
2201 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2202 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
2203 !FRAME_GROWS_DOWNWARD handling to...
2204 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2205 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
2207 (rs6000_starting_frame_offset): New function.
2208 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
2209 !FRAME_GROWS_DOWNWARD handling to...
2210 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2211 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
2212 !FRAME_GROWS_DOWNWARD handling to...
2213 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2214 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
2215 !FRAME_GROWS_DOWNWARD handling to...
2216 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2217 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
2218 (rs6000_starting_frame_offset): New function.
2219 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
2220 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
2221 * config/vax/vax.c (vax_starting_frame_offset): New function.
2222 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
2223 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2224 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
2225 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
2226 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2227 * system.h (STARTING_FRAME_OFFSET): Poison.
2229 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2231 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
2232 SCALAR_TYPE_MODE instead of TYPE_MODE.
2234 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2235 Alan Hayward <alan.hayward@arm.com>
2236 David Sherwood <david.sherwood@arm.com>
2238 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
2240 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2241 Alan Hayward <alan.hayward@arm.com>
2242 David Sherwood <david.sherwood@arm.com>
2244 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
2246 2017-10-23 Richard Biener <rguenther@suse.de>
2248 PR tree-optimization/82129
2249 * tree-ssa-pre.c (bitmap_set_and): Remove.
2250 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
2251 canonicalizing expressions in the set to those with lowest
2252 ID rather than taking that from the first edge.
2254 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2256 * combine.c (rtx_equal_for_field_assignment_p): Use
2257 byte_lowpart_offset.
2259 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2260 Alan Hayward <alan.hayward@arm.com>
2261 David Sherwood <david.sherwood@arm.com>
2263 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
2264 to a SUBREG_PROMOTED_VAR.
2266 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2267 Alan Hayward <alan.hayward@arm.com>
2268 David Sherwood <david.sherwood@arm.com>
2270 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
2271 (expand_debug_source_expr): Likewise.
2272 * combine.c (combine_simplify_rtx): Likewise.
2273 * cse.c (fold_rtx): Likewise.
2274 * optabs.c (expand_float): Likewise.
2275 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2276 (simplify_binary_operation_1): Likewise.
2278 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2279 Alan Hayward <alan.hayward@arm.com>
2280 David Sherwood <david.sherwood@arm.com>
2282 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
2283 (record_promoted_value): Likewise.
2284 * expr.c (expand_expr_real_2): Likewise.
2285 * ree.c (update_reg_equal_equiv_notes): Likewise.
2286 (combine_set_extension): Likewise.
2287 * rtlanal.c (low_bitmask_len): Likewise.
2288 * simplify-rtx.c (neg_const_int): Likewise.
2289 (simplify_binary_operation_1): Likewise.
2291 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2292 Alan Hayward <alan.hayward@arm.com>
2293 David Sherwood <david.sherwood@arm.com>
2295 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
2296 * regcprop.c (maybe_mode_change): Likewise.
2297 * reload1.c (alter_reg): Likewise.
2299 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2301 * inchash.h (inchash::hash::add_wide_int): New function.
2302 * lto-streamer-out.c (hash_tree): Use it.
2304 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2306 * inchash.h (inchash::hash::add_wide_int): Rename to...
2307 (inchash::hash::add_hwi): ...this.
2308 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
2309 (polymorphic_call_target_hasher::hash): Likewise.
2310 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
2311 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
2312 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
2313 * lto-streamer-out.c (hash_tree): Likewise.
2314 * optc-save-gen.awk: Likewise.
2315 * tree.c (add_expr): Likewise.
2317 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
2320 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
2321 for ordered inequality comparisons even with TARGET_IEEE_FP.
2323 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
2326 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
2327 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
2328 Expand with cmp<dwi>_doubleword.
2330 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2332 * extend.texi: Add x86 specific to 'nocf_check' attribute.
2333 List CET intrinsics.
2334 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
2335 specific to -fcf-protection option.
2337 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2339 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
2340 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
2341 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
2342 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
2343 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
2344 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
2345 (extra_objs): Add cet.o for Linux/x86 targets.
2346 (tmake_file): Add i386/t-cet for Linux/x86 targets.
2347 * config/i386/cet.c: New file.
2348 * config/i386/cetintrin.h: Likewise.
2349 * config/i386/t-cet: Likewise.
2350 * config/i386/cpuid.h (bit_SHSTK): New.
2351 (bit_IBT): Likewise.
2352 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
2353 pass IBT and SHSTK bits.
2354 * config/i386/i386-builtin-types.def
2355 (VOID_FTYPE_UNSIGNED_PVOID): New.
2356 (VOID_FTYPE_UINT64_PVOID): Likewise.
2357 * config/i386/i386-builtin.def: Add CET intrinsics.
2358 * config/i386/i386-c.c (ix86_target_macros_internal): Add
2359 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
2360 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
2361 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
2363 * config/i386/i386.c (rest_of_insert_endbranch): New.
2364 (pass_data_insert_endbranch): Likewise.
2365 (pass_insert_endbranch): Likewise.
2366 (make_pass_insert_endbranch): Likewise.
2367 (ix86_notrack_prefixed_insn_p): Likewise.
2368 (ix86_target_string): Add -mibt, -mshstk flags.
2369 (ix86_option_override_internal): Add flag_cf_protection
2371 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
2372 (ix86_print_operand): Add 'notrack' prefix output.
2373 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
2374 (ix86_expand_builtin): Expand CET intrinsics.
2375 (x86_output_mi_thunk): Add 'endbranch' instruction.
2376 * config/i386/i386.h (TARGET_IBT): New.
2377 (TARGET_IBT_P): Likewise.
2378 (TARGET_SHSTK): Likewise.
2379 (TARGET_SHSTK_P): Likewise.
2380 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
2381 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
2382 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
2383 (builtin_setjmp_setup): New pattern.
2384 (builtin_longjmp): Likewise.
2385 (rdssp<mode>): Likewise.
2386 (incssp<mode>): Likewise.
2387 (saveprevssp): Likewise.
2388 (rstorssp): Likewise.
2389 (wrss<mode>): Likewise.
2390 (wruss<mode>): Likewise.
2391 (setssbsy): Likewise.
2392 (clrssbsy): Likewise.
2393 (nop_endbr): Likewise.
2394 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
2396 * config/i386/immintrin.h: Include <cetintrin.h>.
2397 * config/i386/linux-common.h
2398 (file_end_indicate_exec_stack_and_cet): New prototype.
2399 (TARGET_ASM_FILE_END): New.
2401 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
2403 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
2404 latencies instead of having separate table; make difference between
2405 integer and float costs.
2406 * i386.h (processor_costs): Remove scalar_stmt_cost,
2407 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
2408 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
2410 * x86-tune-costs.h: Remove entries which has been removed in
2411 procesor_costs from all tables; make cond_taken_branch_cost
2412 and cond_not_taken_branch_cost COST_N_INSNS based.
2414 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
2416 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
2418 2017-10-20 Jakub Jelinek <jakub@redhat.com>
2420 * config/i386/i386.md (isa): Remove fma_avx512f.
2421 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
2422 <avx512>_fmadd_<mode>_mask3<round_name>,
2423 <avx512>_fmsub_<mode>_mask<round_name>,
2424 <avx512>_fmsub_<mode>_mask3<round_name>,
2425 <avx512>_fnmadd_<mode>_mask<round_name>,
2426 <avx512>_fnmadd_<mode>_mask3<round_name>,
2427 <avx512>_fnmsub_<mode>_mask<round_name>,
2428 <avx512>_fnmsub_<mode>_mask3<round_name>,
2429 <avx512>_fmaddsub_<mode>_mask<round_name>,
2430 <avx512>_fmaddsub_<mode>_mask3<round_name>,
2431 <avx512>_fmsubadd_<mode>_mask<round_name>,
2432 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
2433 (*vec_widen_umult_even_v16si<mask_name>,
2434 *vec_widen_smult_even_v16si<mask_name>): Likewise.
2435 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
2437 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2439 * extend.texi: Add 'nocf_check' documentation.
2440 * gimple.texi: Add second parameter to
2441 gimple_build_call_from_tree.
2442 * invoke.texi: Add -fcf-protection documentation.
2443 * rtl.texi: Add REG_CALL_NOTRACK documenation.
2445 2017-10-20 Richard Biener <rguenther@suse.de>
2447 PR tree-optimization/82473
2448 * tree-vect-loop.c (vectorizable_reduction): Properly get at
2449 the largest input type.
2451 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2453 * c-attribs.c (handle_nocf_check_attribute): New function.
2454 (c_common_attribute_table): Add 'nocf_check' handling.
2455 * gimple-parser.c: Add second argument NULL to
2456 gimple_build_call_from_tree.
2457 * attrib.c (comp_type_attributes): Check nocf_check attribute.
2458 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
2460 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
2461 * common.opt: Add fcf-protection flag.
2462 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
2463 * flag-types.h: Add enum cf_protection_level.
2464 * gimple.c (gimple_build_call_from_tree): Add second parameter.
2465 Add 'nocf_check' attribute propagation to gimple call.
2466 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
2467 (gimple_build_call_from_tree): Update prototype.
2468 (gimple_call_nocf_check_p): New function.
2469 (gimple_call_set_nocf_check): Likewise.
2470 * gimplify.c: Add second argument to gimple_build_call_from_tree.
2471 * ipa-icf.c: Add nocf_check attribute in statement hash.
2472 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
2473 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
2474 * toplev.c (process_options): Add flag_cf_protection handling.
2476 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
2478 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
2480 2017-10-20 Richard Biener <rguenther@suse.de>
2482 PR tree-optimization/82603
2483 * tree-if-conv.c (predicate_mem_writes): Make sure to only
2484 remove false predicated stores.
2486 2017-10-20 Richard Biener <rguenther@suse.de>
2488 * graphite-isl-ast-to-gimple.c
2489 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
2490 Remove return value and simplify, dump copied stmt after lhs
2492 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
2493 Reduce dump verbosity.
2494 (gsi_insert_earliest): Likewise.
2495 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
2496 * graphite.c (print_global_statistics): Adjust dumping.
2497 (print_graphite_scop_statistics): Likewise.
2498 (print_graphite_statistics): Do not dump loops here.
2499 (graphite_transform_loops): But here.
2501 2017-10-20 Nicolas Roche <roche@adacore.com>
2503 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
2504 * configure: Regenerate.
2506 2017-10-20 Jakub Jelinek <jakub@redhat.com>
2509 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
2510 functions when optimizing replace GIMPLE_RETURN stmts with
2511 calls to __builtin_unreachable ().
2514 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
2515 for -fsanitize=thread link of executables.
2516 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
2517 link of executables.
2520 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
2522 (<shift_insn><mode>3<mask_name>): Change the last of the 3
2523 define_insns for logical vector shifts to use VI248_AVX512BW
2524 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
2525 condition, useless isa and prefix attributes. Change the first
2526 2 of these define_insns to ...
2527 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
2528 define_insn for avx512vl.
2529 (<shift_insn><mode>3): ... and this, new define_insn without
2530 masking for non-avx512vl.
2533 * config/i386/sse.md (*andnot<mode>3,
2534 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
2535 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
2536 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
2537 not applied use empty suffix even for TARGET_AVX512VL.
2538 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
2539 is applied, supply evex,evex or evex,evex,evex instead of just
2542 2017-10-20 Julia Koval <julia.koval@intel.com>
2544 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
2545 (OPTION_MASK_ISA_GFNI_UNSET): New.
2546 (ix86_handle_option): Handle OPT_mgfni.
2547 * config/i386/cpuid.h (bit_GFNI): New.
2548 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
2549 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
2550 * config/i386/i386.c (ix86_target_string): Add -mgfni.
2551 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
2552 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
2553 * config/i386/i386.opt: Add mgfni.
2555 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
2557 * config/msp430/msp430.c (msp430_option_override): Disable
2558 -fdelete-null-pointer-checks.
2559 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
2561 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
2563 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
2564 of x87 and SSE instructions.
2566 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
2568 * asan.c (create_cond_insert_point): Do not update edge count.
2569 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
2570 (afdo_propagate_circuit): Likewise.
2571 (afdo_calculate_branch_prob): Likewise.
2572 (afdo_annotate_cfg): Likewise.
2573 * basic-block.h (struct edge_def): Remove count.
2574 (edge_def::count): New accessor.
2575 * bb-reorder.c (rotate_loop): Update.
2576 (find_traces_1_round): Update.
2577 (connect_traces): Update.
2578 (sanitize_hot_paths): Update.
2579 * cfg.c (unchecked_make_edge): Update.
2580 (make_single_succ_edge): Update.
2581 (check_bb_profile): Update.
2582 (dump_edge_info): Update.
2583 (update_bb_profile_for_threading): Update.
2584 (scale_bbs_frequencies_int): Update.
2585 (scale_bbs_frequencies_gcov_type): Update.
2586 (scale_bbs_frequencies_profile_count): Update.
2587 (scale_bbs_frequencies): Update.
2588 * cfganal.c (connect_infinite_loops_to_exit): Update.
2589 * cfgbuild.c (compute_outgoing_frequencies): Update.
2590 (find_many_sub_basic_blocks): Update.
2591 * cfgcleanup.c (try_forward_edges): Update.
2592 (try_crossjump_to_edge): Update
2593 * cfgexpand.c (expand_gimple_cond): Update
2594 (expand_gimple_tailcall): Update
2595 (construct_exit_block): Update
2596 * cfghooks.c (verify_flow_info): Update
2597 (redirect_edge_succ_nodup): Update
2598 (split_edge): Update
2599 (make_forwarder_block): Update
2600 (duplicate_block): Update
2601 (account_profile_record): Update
2602 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
2603 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
2604 * cfgloopmanip.c (scale_loop_profile): Update.
2606 (lv_adjust_loop_entry_edge): Update.
2607 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
2608 (force_nonfallthru_and_redirect): Update.
2609 (purge_dead_edges): Update.
2610 (rtl_flow_call_edges_add): Update.
2611 * cgraphunit.c (init_lowered_empty_function): Update.
2612 (cgraph_node::expand_thunk): Update.
2613 * gimple-pretty-print.c (dump_probability): Update.
2614 (dump_edge_probability): Update.
2615 * gimple-ssa-isolate-paths.c (isolate_path): Update.
2616 * haifa-sched.c (sched_create_recovery_edges): Update.
2617 * hsa-gen.c (convert_switch_statements): Update.
2618 * ifcvt.c (dead_or_predicable): Update.
2619 * ipa-inline-transform.c (inline_transform): Update.
2620 * ipa-split.c (split_function): Update.
2621 * ipa-utils.c (ipa_merge_profiles): Update.
2622 * loop-doloop.c (add_test): Update.
2623 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
2624 * lto-streamer-in.c (input_cfg): Update.
2625 (input_function): Update.
2626 * lto-streamer-out.c (output_cfg): Update.
2627 * modulo-sched.c (sms_schedule): Update.
2628 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
2629 * predict.c (maybe_hot_edge_p): Update.
2630 (unlikely_executed_edge_p): Update.
2631 (probably_never_executed_edge_p): Update.
2632 (dump_prediction): Update.
2633 (drop_profile): Update.
2634 (propagate_unlikely_bbs_forward): Update.
2635 (determine_unlikely_bbs): Update.
2636 (force_edge_cold): Update.
2637 * profile.c (compute_branch_probabilities): Update.
2638 * reg-stack.c (better_edge): Update.
2639 * shrink-wrap.c (handle_simple_exit): Update.
2640 * tracer.c (better_p): Update.
2641 * trans-mem.c (expand_transaction): Update.
2642 (split_bb_make_tm_edge): Update.
2643 * tree-call-cdce.c: Update.
2644 * tree-cfg.c (gimple_find_sub_bbs): Update.
2645 (gimple_split_edge): Update.
2646 (gimple_duplicate_sese_region): Update.
2647 (gimple_duplicate_sese_tail): Update.
2648 (gimple_flow_call_edges_add): Update.
2649 (insert_cond_bb): Update.
2650 (execute_fixup_cfg): Update.
2651 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
2652 * tree-complex.c (expand_complex_div_wide): Update.
2653 * tree-eh.c (lower_resx): Update.
2654 (unsplit_eh): Update.
2655 (cleanup_empty_eh_move_lp): Update.
2656 * tree-inline.c (copy_edges_for_bb): Update.
2657 (freqs_to_counts): Update.
2658 (copy_cfg_body): Update.
2659 * tree-ssa-dce.c (remove_dead_stmt): Update.
2660 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
2661 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
2662 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
2663 (unloop_loops): Update.
2664 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
2665 * tree-ssa-loop-split.c (connect_loops): Update.
2666 (split_loop): Update.
2667 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
2668 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
2669 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
2670 * tree-ssa-reassoc.c (branch_fixup): Update.
2671 * tree-ssa-tail-merge.c (replace_block_by): Update.
2672 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
2673 (compute_path_counts): Update.
2674 (update_profile): Update.
2675 (recompute_probabilities): Update.
2676 (update_joiner_offpath_counts): Update.
2677 (estimated_freqs_path): Update.
2678 (freqs_to_counts_path): Update.
2679 (clear_counts_path): Update.
2680 (ssa_fix_duplicate_block_edges): Update.
2681 (duplicate_thread_path): Update.
2682 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
2683 (case_bit_test_cmp): Update.
2684 (collect_switch_conv_info): Update.
2685 (gen_inbound_check): Update.
2686 (do_jump_if_equal): Update.
2687 (emit_cmp_and_jump_insns): Update.
2688 * tree-tailcall.c (decrease_profile): Update.
2689 (eliminate_tail_call): Update.
2690 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
2691 (vect_do_peeling): Update.
2692 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
2693 * ubsan.c (ubsan_expand_null_ifn): Update.
2694 (ubsan_expand_ptr_ifn): Update.
2695 * value-prof.c (gimple_divmod_fixed_value): Update.
2696 (gimple_mod_pow2): Update.
2697 (gimple_mod_subtract): Update.
2698 (gimple_ic): Update.
2699 (gimple_stringop_fixed_value): Update.
2701 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
2704 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
2706 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
2708 PR rtl-optimization/82395
2709 * ira-color.c (allocno_priority_compare_func): Fix comparison step
2710 based on non_spilled_static_chain_regno_p.
2712 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
2714 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
2715 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
2716 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
2718 2017-10-19 Martin Sebor <msebor@redhat.com>
2720 PR tree-optimization/82596
2721 * tree.c (array_at_struct_end_p): Handle STRING_CST.
2723 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
2725 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
2726 (get_mem_refs_of_builtin_call): Likewise.
2727 * builtins.c (expand_builtin_apply): Adjust call to
2728 allocate_dynamic_stack_space.
2729 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
2730 the third argument to allocate_dynamic_stack_space, otherwise -1.
2731 (expand_builtin): Deal with all alloca variants.
2732 (is_inexpensive_builtin): Likewise.
2733 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
2734 * calls.c (special_function_p): Deal with all alloca variants.
2735 (initialize_argument_information): Adjust call to
2736 allocate_dynamic_stack_space.
2737 (expand_call): Likewise.
2738 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
2739 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
2740 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
2741 use it for the stack usage computation.
2742 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
2743 * function.c (gimplify_parameters): Call build_alloca_call_expr.
2744 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
2745 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
2746 (in_loop_p): Remove first argument and useless check.
2747 (pass_walloca::execute): Remove useless test and adjust call to above.
2748 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
2749 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
2750 (gimplify_call_expr): Deal with all alloca variants.
2751 * hsa-gen.c (gen_hsa_alloca): Likewise.
2752 (gen_hsa_insns_for_call): Likewise.
2753 * ipa-pure-const.c (special_builtin_state): Likewise.
2754 * tree-chkp.c (chkp_build_returned_bound): Likewise.
2755 * tree-object-size.c (alloc_object_size): Likewise.
2756 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
2757 (call_may_clobber_ref_p_1): Likewise.
2758 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
2759 (ccp_fold_stmt): Likewise.
2760 (optimize_stack_restore): Likewise.
2761 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
2762 (mark_all_reaching_defs_necessary_1): Likewise.
2763 (propagate_necessity): Likewise.
2764 (eliminate_unnecessary_stmts): Likewise.
2765 * tree.c (build_common_builtin_nodes): Build
2766 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
2767 (build_alloca_call_expr): New function.
2768 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
2769 (CASE_BUILT_IN_ALLOCA): Likewise.
2770 (build_alloca_call_expr): Declare.
2771 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
2773 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
2776 * dwarf2out.c (new_die_raw): New static inline function.
2777 (new_die): Use it to create the DIE.
2778 (add_AT_external_die_ref): Likewise.
2779 (clone_die): Likewise.
2780 (clone_as_declaration): Likewise.
2781 (dwarf2out_vms_debug_main_pointer): Likewise.
2782 (base_type_die): Likewise. Remove early return for corner cases.
2783 Do not call add_pubtype on the DIE here.
2784 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
2785 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
2786 typedefs for base types with DW_AT_endianity. Make sure a DIE with
2787 native order exists for base types, attach the DIE manually and call
2788 add_pubtype on it. Do not equate a reverse order DIE to the type.
2790 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
2792 * config/arm/arm.c (align_ok_ldrd_strd): New function.
2793 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
2795 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
2797 2017-10-19 Jakub Jelinek <jakub@redhat.com>
2799 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
2800 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
2801 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
2802 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
2803 * opts.c (sanitizer_opts): Add builtin.
2804 * ubsan.c (instrument_builtin): New function.
2805 (pass_ubsan::execute): Call it.
2806 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
2807 * doc/invoke.texi: Document -fsanitize=builtin.
2809 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
2810 builtins, store max (log2 (align), 0) into uchar field instead of
2811 align into uptr field.
2812 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
2813 store uchar 0 field instead of uptr 0 field.
2814 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
2815 instead of passing one address of struct with 2 locations pass
2816 two addresses of structs with 1 location each.
2817 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
2818 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
2819 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
2820 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
2821 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
2822 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
2823 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
2824 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
2826 2017-10-19 Martin Liska <mliska@suse.cz>
2829 * file-find.c (remove_prefix): Remove.
2830 * file-find.h (remove_prefix): Likewise.
2831 * gcc-ar.c: Remove smartness of lookup.
2833 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
2835 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
2836 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
2837 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
2839 2017-10-19 Jakub Jelinek <jakub@redhat.com>
2842 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
2843 (setcc + and to xor + setcc): New peephole2.
2845 2017-10-19 Tom de Vries <tom@codesourcery.com>
2847 * doc/sourcebuild.texi (Test Directives, Variants of
2848 dg-require-support): Add dg-require-stack-size.
2850 2017-10-19 Martin Liska <mliska@suse.cz>
2853 * gimplify.c (gimplify_decl_expr): Do not instrument variables
2854 that have a large alignment.
2855 (gimplify_target_expr): Likewise.
2857 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
2859 PR rtl-optimization/82602
2860 * ira.c (rtx_moveable_p): Return false for volatile asm.
2862 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
2865 * config/i386/i386-modes.def (CCGZ): New CC mode.
2866 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
2867 * config/i386/predicates.md (ix86_comparison_operator):
2869 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
2870 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
2871 with double-word subtraction.
2872 (put_condition_code): Handle CCGZmode.
2874 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
2876 * wide-int.cc (debug (const wide_int &)): New.
2877 (debug (const wide_int *)): New.
2878 (debug (const widest_int &)): New.
2879 (debug (const widest_int *)): New.
2881 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
2884 * lra-constraints.c (curr_insn_transform): Use non-input operand
2885 instead of output one for matched reload.
2887 2017-10-18 Bin Cheng <bin.cheng@arm.com>
2889 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
2890 (tree-ssa-loop-ivopts.h): New header file.
2891 (struct builtin_info): New fields.
2892 (classify_builtin_1): Compute and record base and offset parts for
2893 memset builtin partition by calling strip_offset.
2894 (offset_cmp, fuse_memset_builtins): New functions.
2895 (finalize_partitions): Fuse adjacent memset partitions by calling
2897 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
2898 Expose the interface.
2899 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
2901 2017-10-18 Bin Cheng <bin.cheng@arm.com>
2903 PR tree-optimization/82574
2904 * tree-loop-distribution.c (find_single_drs): New parameter. Check
2905 that data reference must be executed exactly once per iteration
2906 against the outermost loop in nest.
2907 (classify_partition): Update call to above function.
2909 2017-10-18 Richard Biener <rguenther@suse.de>
2911 PR tree-optimization/82591
2912 * graphite.c (graphite_transform_loops): Move code gen message
2914 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
2915 Here. Handle scop_to_isl_ast failing.
2916 (scop_to_isl_ast): Limit the number of ISL operations.
2918 2017-10-18 Richard Biener <rguenther@suse.de>
2920 * graphite-isl-ast-to-gimple.c
2921 (translate_isl_ast_to_gimple::set_rename): Simplify.
2922 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
2923 (graphite_copy_stmts_from_block): ... here.
2924 (copy_bb_and_scalar_dependences): Simplify.
2925 (add_parameters_to_ivs_params): Canonicalize.
2926 (generate_entry_out_of_ssa_copies): Simplify.
2927 * graphite-sese-to-poly.c (extract_affine_name): Simplify
2928 by passing in ISL dimension.
2929 (parameter_index_in_region_1): Rename to ...
2930 (parameter_index_in_region): ... this.
2931 (extract_affine): Adjust assert, pass down parameter index.
2932 (add_param_constraints): Use range-info when available.
2933 (build_scop_context): Adjust.
2934 * sese.c (new_sese_info): Adjust.
2935 (free_sese_info): Likewise.
2936 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
2937 Remove unused typedefs.
2938 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
2940 2017-10-18 Martin Liska <mliska@suse.cz>
2942 * combine.c (simplify_compare_const): Add gcc_fallthrough.
2944 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
2946 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
2947 (s390_sched_init): Do not reset s390_sched_state if we entered the
2948 current basic block via a fallthru edge and all others are unlikely.
2950 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
2952 * config/s390/s390.c (NUM_SIDES): New variable.
2953 (LONGRUNNING_THRESHOLD): New variable.
2954 (LATENCY_FACTOR): New variable.
2955 (s390_sched_score): Decrease score for long-running instructions on
2957 (s390_sched_variable_issue): Perform bookkeeping for long-running
2960 2017-10-18 Richard Biener <rguenther@suse.de>
2962 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
2963 Simplify with removal of the parameter rename map.
2964 (set_rename): Likewise.
2965 (should_copy_to_new_region): Likewise.
2966 (graphite_copy_stmts_from_block): Likewise.
2967 (copy_bb_and_scalar_dependences): Remove initialization of
2968 unused copied_bb_map.
2970 (copy_internal_parameters): Likewise.
2971 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
2972 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
2973 Use INTEGRAL_TYPE_P.
2974 (parameter_index_in_region_1): Rename to ...
2975 (assign_parameter_index_in_region): ... this. Assert we have
2976 a parameter we handle.
2977 (scan_tree_for_params): Adjust.
2978 * sese.h (parameter_rename_map_t): Remove.
2979 (struct sese_info_t): Remove unused parameter_rename_map and
2980 copied_bb_map members.
2981 * sese.c (new_sese_info): Adjust.
2982 (free_sese_info): Likewise.
2984 2017-10-18 Martin Liska <mliska@suse.cz>
2987 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
2988 on an abnormal edge.
2990 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
2992 * doc/invoke.texi (ffunction-sections and fdata-sections):
2995 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
2997 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
2998 the use statement can throw internally.
3000 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3002 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
3003 any RTX present on the RHS of a SET.
3004 * compare-elim.c (try_eliminate_compare): Restore comment.
3006 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3008 * langhooks.h (struct lang_hooks): Document that tree_size langhook
3009 may be also called on tcc_type nodes.
3010 * langhooks.c (lhd_tree_size): Likewise.
3012 2017-10-17 David Malcolm <dmalcolm@redhat.com>
3014 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
3015 format_warning_at_substring.
3016 (maybe_warn): Convert source_range * param to a location_t. Pass
3017 UNKNOWN_LOCATION rather than NULL to fmtwarn.
3018 (format_directive): Remove code to extract source_ranges and
3019 source_range * in favor of just a location_t.
3020 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
3022 * substring-locations.c (format_warning_va): Convert
3023 source_range * param to a location_t.
3024 (format_warning_at_substring): Likewise.
3025 * substring-locations.h (format_warning_va): Likewise.
3026 (format_warning_at_substring): Likewise.
3028 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
3030 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
3032 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
3033 and scatter/gather ops.
3035 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
3036 vec_gather_load and vec_scatter_store.
3037 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
3038 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
3039 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
3040 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
3041 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
3042 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
3044 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
3046 * reg-stack.c (compare_for_stack_reg): Add bool argument.
3047 Detect FTST instruction and handle its register pops. Only pop
3048 second operand if can_pop_second_op is true.
3049 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
3050 set can_pop_second_op to false in the compare_for_stack_reg call.
3052 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
3053 output_fp_compare for stack register operands.
3054 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
3055 instructions here. Do not emit stack register pops here. Assert
3056 that FCOMPP pops next to top stack register. Rewrite function.
3058 2017-10-17 Nathan Sidwell <nathan@acm.org>
3061 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
3062 use DECL_ASSEMBLER_NAME_RAW.
3065 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
3068 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
3069 Wilco Dijkstra <wilco.dijkstra@arm.com>
3071 * builtins.c (expand_builtin_update_setjmp_buf): Add a
3072 converstion to Pmode from the buf_addr.
3074 2017-10-17 Richard Biener <rguenther@suse.de>
3076 * graphite-dependences.c (scop_get_reads_and_writes): Change
3077 output parameters to references.
3079 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
3081 PR 71026/tree-optimization
3082 * fold-const.c (distribute_real_division): Removed.
3083 (fold_binary_loc): Remove calls to distribute_real_divison.
3085 2017-10-17 Richard Biener <rguenther@suse.de>
3087 * graphite-scop-detection.c
3088 (scop_detection::stmt_has_simple_data_refs_p): Always use
3089 the full nest as region.
3090 (try_generate_gimple_bb): Likewise.
3091 * sese.c (scalar_evolution_in_region): Simplify now that
3092 SCEV can handle instantiation in regions.
3093 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
3094 in the non-loop part of a function if requested.
3096 2017-10-17 Richard Biener <rguenther@suse.de>
3098 PR tree-optimization/82563
3099 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
3101 (graphite_regenerate_ast_isl): Call it.
3102 * graphite-scop-detection.c (build_scops): Remove entry edge split.
3104 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3106 PR tree-optimization/82549
3107 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
3108 Formatting fixes. Instead of calling make_bit_field_ref with negative
3111 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
3113 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
3114 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
3115 _mm_maskz_reduce_ss): New.
3116 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
3117 __builtin_ia32_reducess_mask): Ditto..
3118 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
3119 * config/i386/sse.md (reduces<mode>): Renamed to ...
3120 (reduces<mode><mask_scalar_name>): ... this.
3121 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
3123 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
3124 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
3126 2017-10-16 David Malcolm <dmalcolm@redhat.com>
3128 * Makefile.in (OBJS): Add unique-ptr-tests.o.
3129 * selftest-run-tests.c (selftest::run_tests): Call
3130 selftest::unique_ptr_tests_cc_tests.
3131 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
3132 * unique-ptr-tests.cc: New file.
3134 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
3137 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3139 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3140 (make_hard_regno_born, make_hard_regno_dead): Update
3141 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
3143 2017-10-16 Jeff Law <law@redhat.com>
3145 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
3147 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
3149 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
3151 2017-10-16 Olivier Hainque <hainque@adacore.com>
3153 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
3154 with_cpu if we were configured for an e500v2 target cpu name.
3156 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
3158 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
3159 * doc/invoke.texi: Document +nodsp as a valid extension for
3162 2017-10-16 Martin Liska <mliska@suse.cz>
3164 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
3165 (test_set_range): Likewise.
3166 (test_range_functions): Rename to ...
3167 (test_bit_in_range): ... this.
3168 (sbitmap_c_tests): Add new test.
3170 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3172 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
3174 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
3175 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
3177 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3179 * config/aarch64/aarch64-builtins.c
3180 (aarch64_types_quadopu_lane_qualifiers): New.
3181 (TYPES_QUADOPU_LANE): New.
3182 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
3183 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
3184 (aarch64_<sur>dot_laneq<vsi2qi>): New.
3185 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
3186 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
3187 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
3188 (Vdottype, DOTPROD): New.
3189 (sur): Add SDOT and UDOT.
3191 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3193 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
3194 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
3195 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
3197 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
3198 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
3199 Enable TARGET_DOTPROD.
3200 (cortex-a75.cortex-a55): Likewise.
3201 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
3203 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3205 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
3206 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
3208 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
3210 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
3211 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
3212 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
3213 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
3214 * config/arm/types.md (neon_dot, neon_dot_q): New.
3215 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
3217 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3219 * config/arm/arm.h (TARGET_DOTPROD): New.
3220 * config/arm/arm.c (arm_arch_dotprod): New.
3221 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
3222 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
3223 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
3224 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
3225 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
3226 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
3227 * doc/invoke.texi (armv8.2-a): Document dotprod
3229 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3231 * i386.c (ix86_vec_cost): New function.
3232 (ix86_rtx_costs): Handle vector operations better.
3233 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
3234 * x86-tune-costs.h: Add new costs to all tables.
3236 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3238 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
3240 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
3241 divsd, sqrtss and sqrtsd
3242 * x86-tune-costs.h: Add new entries to all costs.
3243 (znver1_cost): Fix to match real instruction latencies.
3245 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3246 Michael Collison <michael.collison@arm.com>
3248 * compare-elim.c: Include emit-rtl.h.
3249 (can_merge_compare_into_arith): New function.
3250 (try_validate_parallel): Likewise.
3251 (try_merge_compare): Likewise.
3252 (try_eliminate_compare): Call the above when no previous clobber
3254 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
3257 2017-10-14 Jakub Jelinek <jakub@redhat.com>
3261 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
3262 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
3266 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
3267 to be any operand_equal_p operands. For & (B - 1) require
3268 B to be power of 2. Recognize
3269 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
3271 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3274 * optabs.c (expand_memory_blockage): Fix call of
3275 targetm.have_memory_blockage.
3277 2017-10-14 Jakub Jelinek <jakub@redhat.com>
3280 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
3281 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
3282 objects to extra_objs instead of overwriting it.
3284 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3286 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
3287 Use any_fp_register_operand as operand[3] predicate. Simplify
3288 equality test for operands[2] and operands[4] memory location.
3289 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
3290 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
3291 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
3292 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
3293 any_fp_register_operand as operand[1] predicate. Simplify
3294 equality test for operands[0] and operands[3] memory location.
3295 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
3296 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
3297 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
3299 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3301 * target-insns.def: Add memory_blockage.
3302 * optabs.c (expand_memory_blockage): New function.
3303 (expand_asm_memory_barrier): Rename ...
3304 (expand_asm_memory_blockage): ... to this.
3305 (expand_mem_thread_fence): Call expand_memory_blockage
3306 instead of expand_asm_memory_barrier.
3307 (expand_mem_singnal_fence): Ditto.
3308 (expand_atomic_load): Ditto.
3309 (expand_atomic_store): Ditto.
3310 * doc/md.texi (Standard Pattern Names For Generation):
3311 Document memory_blockage instruction pattern.
3313 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
3315 * config/rl78/rl78.c (rl78_emit_libcall): New function.
3316 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
3317 * config/rl78/rl78.md: New define_expand "adddi3".
3319 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
3321 * cfghooks.c (verify_flow_info): Disable check that all probabilities
3324 2017-10-13 Jeff Law <law@redhat.com>
3326 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
3328 2017-10-13 Jakub Jelinek <jakub@redhat.com>
3331 * internal-fn.c (expand_mul_overflow): If both operands have
3332 the same highpart of -1 or 0 and the topmost bit of lowpart
3333 is different, overflow is if res <= 0 rather than res < 0.
3335 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
3337 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
3338 TARGET_P9_VECTOR code for unaligned_load case.
3340 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
3342 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
3344 2017-10-13 Nathan Sidwell <nathan@acm.org>
3346 * tree-core.h (tree_contains_struct): Make bool.
3347 * tree.c (tree_contains_struct): Likewise.
3348 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
3349 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
3350 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
3351 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
3352 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
3354 2017-10-13 Richard Biener <rguenther@suse.de>
3356 * graphite-isl-ast-to-gimple.c
3357 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
3358 parameters and dominance check.
3359 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
3360 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
3361 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
3362 Do not update SSA form here or do intermediate IL verification.
3363 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
3364 (graphite_initialize): Remove check on the number of loops in
3365 the function and inline into graphite_transform_loops.
3366 (graphite_finalize): Inline into graphite_transform_loops.
3367 (graphite_transform_loops): Perform SSA update and IL verification
3369 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
3371 2017-10-13 Richard Biener <rguenther@suse.de>
3373 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
3374 graphite_expression_type_precision): Avoid global constructor
3376 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
3377 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
3378 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
3379 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
3380 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
3381 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
3383 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
3386 * config/i386/i386.h (ix86_red_zone_size): New.
3387 * config/i386/i386.md (push peephole2s): Replace
3388 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
3390 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3391 Alan Hayward <alan.hayward@arm.com>
3392 David Sherwood <david.sherwood@arm.com>
3394 * combine.c (can_change_dest_mode): Reject changes in
3395 REGMODE_NATURAL_SIZE.
3397 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3398 Alan Hayward <alan.hayward@arm.com>
3399 David Sherwood <david.sherwood@arm.com>
3401 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
3402 (expand_debug_source_expr): Likewise.
3403 * combine.c (combine_simplify_rtx): Likewise.
3404 * cse.c (fold_rtx): Likewise.
3405 * fwprop.c (canonicalize_address): Likewise.
3406 * targhooks.c (default_shift_truncation_mask): Likewise.
3408 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3409 Alan Hayward <alan.hayward@arm.com>
3410 David Sherwood <david.sherwood@arm.com>
3412 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
3413 (widened_mode): Likewise.
3414 (expand_unop): Likewise.
3415 * ree.c (transform_ifelse): Likewise.
3416 (merge_def_and_ext): Likewise.
3417 (combine_reaching_defs): Likewise.
3418 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3420 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3421 Alan Hayward <alan.hayward@arm.com>
3422 David Sherwood <david.sherwood@arm.com>
3424 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
3425 * combine.c (gen_lowpart_for_combine): Likewise.
3426 * dwarf2out.c (rtl_for_decl_location): Likewise.
3427 * final.c (alter_subreg): Likewise.
3428 * rtlhooks.c (gen_lowpart_general): Likewise.
3429 (gen_lowpart_if_possible): Likewise.
3431 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3432 Alan Hayward <alan.hayward@arm.com>
3433 David Sherwood <david.sherwood@arm.com>
3435 * calls.c (expand_call): Use subreg_lowpart_offset.
3436 * cse.c (cse_insn): Likewise.
3437 * regcprop.c (copy_value): Likewise.
3438 (copyprop_hardreg_forward_1): Likewise.
3440 2017-10-13 Jakub Jelinek <jakub@redhat.com>
3443 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
3444 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
3445 =Q constraints to +Q and into insn condition add check
3446 that operands[0] and operands[1] are equal.
3447 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
3448 =Q constraints to +Q and into insn condition add check
3449 that operands[0] is equal to either operands[1] or operands[2].
3452 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
3453 instead of handling MINUS_EXPR twice (once for each argument),
3454 canonicalize operand order and handle just once, use rtype where
3455 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
3458 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
3459 any values of __C while still being pattern recognizable as a simple
3462 2017-10-13 Richard Biener <rguenther@suse.de>
3464 PR tree-optimization/82451
3466 2017-10-02 Richard Biener <rguenther@suse.de>
3468 PR tree-optimization/82355
3469 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
3470 a mapping for the enclosing loop but avoid generating one for
3472 (copy_bb_and_scalar_dependences): Remove premature codegen
3473 error on PHIs in blocks duplicated into multiple places.
3474 * graphite-scop-detection.c
3475 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
3476 in the region use it as loop and nest to analyze the DR in.
3477 (try_generate_gimple_bb): Likewise.
3478 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
3479 (add_loop_constraints): For blocks in a loop not in the region
3480 create a dimension with a single iteration.
3481 * sese.h (gbb_loop_at_index): Remove assert.
3483 * cfgloop.c (loop_preheader_edge): For the loop tree root
3484 return the single successor of the entry block.
3485 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3486 Reset the SCEV hashtable and niters.
3487 * graphite-scop-detection.c
3488 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
3489 assert that we only have POLYNOMIAL_CHREC that vary in loops
3490 contained in the region.
3491 (scop_detection::graphite_can_represent_expr): Adjust.
3492 (scop_detection::stmt_has_simple_data_refs_p): For loops
3493 not in the region set loop to NULL. The nest is now the
3494 entry edge to the region.
3495 (try_generate_gimple_bb): Likewise.
3496 * sese.c (scalar_evolution_in_region): Adjust for
3497 instantiate_scev change.
3498 * tree-data-ref.h (graphite_find_data_references_in_stmt):
3499 Make nest parameter the edge into the region.
3500 (create_data_ref): Likewise.
3501 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
3502 entry edge into a region and adjust instantiate_scev calls.
3503 (create_data_ref): Likewise.
3504 (graphite_find_data_references_in_stmt): Likewise.
3505 (find_data_references_in_stmt): Pass the loop preheader edge
3506 from the nest argument.
3507 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
3508 parameter the edge into the region.
3509 (instantiate_parameters): Use the loop preheader edge as entry.
3510 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
3512 (get_instantiated_value_entry): Make instantiate_below parameter
3513 the edge into the region.
3514 (instantiate_scev_name): Likewise. Adjust dominance checks,
3515 when we cannot use loop-based instantiation instantiate by
3516 walking use-def chains.
3517 (instantiate_scev_poly): Adjust.
3518 (instantiate_scev_binary): Likewise.
3519 (instantiate_scev_convert): Likewise.
3520 (instantiate_scev_not): Likewise.
3521 (instantiate_array_ref): Remove.
3522 (instantiate_scev_3): Likewise.
3523 (instantiate_scev_2): Likewise.
3524 (instantiate_scev_1): Likewise.
3525 (instantiate_scev_r): Do not blindly handle N-operand trees.
3526 Do not instantiate array-refs. Handle all constants and invariants.
3527 (instantiate_scev): Make instantiate_below parameter
3528 the edge into the region.
3529 (resolve_mixers): Use the loop preheader edge for the region
3530 parameter to instantiate_scev_r.
3531 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
3533 2017-10-13 Richard Biener <rguenther@suse.de>
3535 PR tree-optimization/82525
3536 * graphite-isl-ast-to-gimple.c
3537 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
3539 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
3540 Fail code generation when we cannot represent the isl integer.
3541 (binary_op_to_tree): Elide modulo operations that are no-ops
3542 in the type we code generate. Remove now superfluous code
3545 2017-10-13 Richard Biener <rguenther@suse.de>
3547 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
3548 (scop_detection::harmful_loop_in_region): Remove premature
3549 IV type restriction.
3550 (scop_detection::graphite_can_represent_scev): We can handle
3551 pointer IVs just fine.
3553 2017-10-13 Alan Modra <amodra@gmail.com>
3555 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
3556 "Clobbers and Scratch Registers". Add paragraph on
3557 alternative to clobbers for scratch registers and OpenBLAS
3560 2017-10-13 Alan Modra <amodra@gmail.com>
3562 * doc/extend.texi (Clobbers): Correct vax example. Delete old
3563 example of a memory input for a string of known length. Move
3564 commentary out of table. Add a number of new examples
3565 covering array memory inputs.
3567 2017-10-12 Martin Liska <mliska@suse.cz>
3569 PR tree-optimization/82493
3570 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
3571 (test_range_functions): New function.
3572 (sbitmap_c_tests): Likewise.
3573 * selftest-run-tests.c (selftest::run_tests): Run new tests.
3574 * selftest.h (sbitmap_c_tests): New function.
3576 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
3578 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
3580 * config/rs6000/amo.h: Fix spacing issue.
3582 2017-10-12 Jakub Jelinek <jakub@redhat.com>
3585 * config/i386/i386.md (*ashl<mode>3_mask_1,
3586 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
3587 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
3590 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
3592 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
3593 (profile_probability): Set max_probability
3594 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
3596 * profile-count.c (profile_probability::differs_from_p): Do not
3597 rely on max_probaiblity == 10000
3599 2017-10-12 Jeff Law <law@redhat.com>
3601 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
3604 2017-10-12 Martin Sebor <msebor@redhat.com>
3608 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
3609 (handle_alias_pairs): Call it.
3610 * common.opt (-Wattribute-alias): New option.
3611 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
3612 * doc/invoke.texi (-Wattribute-alias): Document.
3614 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
3617 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
3619 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3621 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3622 (make_hard_regno_born, make_hard_regno_dead): Update
3623 bb_killed_pseudos and bb_gen_pseudos.
3625 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
3627 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
3629 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
3631 * config/alpha/alpha.c (alpha_split_conditional_move):
3632 Use std::swap instead of manually swapping.
3633 (alpha_stdarg_optimize_hook): Ditto.
3634 (alpha_canonicalize_comparison): Ditto.
3636 2017-10-12 Bin Cheng <bin.cheng@arm.com>
3638 * tree-loop-distribution.c (struct builtin_info): New struct.
3639 (struct partition): Refactor fields into struct builtin_info.
3640 (partition_free): Free struct builtin_info.
3641 (build_size_arg_loc, build_addr_arg_loc): Delete.
3642 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
3643 information from struct builtin_info.
3644 (find_single_drs): New function refactored from classify_partition.
3645 Also moved builtin validity checks to this function.
3646 (compute_access_range, alloc_builtin): New functions.
3647 (classify_builtin_st, classify_builtin_ldst): New functions.
3648 (classify_partition): Refactor code into functions find_single_drs,
3649 classify_builtin_st and classify_builtin_ldst.
3650 (distribute_loop): Don't do runtime alias check when distributing
3652 (find_seed_stmts_for_distribution): New function.
3653 (pass_loop_distribution::execute): Refactor code finding seed
3654 stmts into above function. Support distribution for the innermost
3655 two-level loop nest. Adjust dump information.
3657 2017-10-12 Bin Cheng <bin.cheng@arm.com>
3659 * tree-loop-distribution.c: Adjust the general comment.
3660 (NUM_PARTITION_THRESHOLD): New macro.
3661 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
3662 (classify_partition): Skip builtin pattern of loop nest's inner loop.
3663 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
3664 in call to build_partition_graph.
3665 (finalize_partitions): New parameter. Make loop distribution more
3666 conservative by fusing more partitions.
3667 (distribute_loop): Don't do runtime alias check in case of loop nest
3669 (find_seed_stmts_for_distribution): New function.
3670 (prepare_perfect_loop_nest): New function.
3671 (pass_loop_distribution::execute): Refactor code finding seed stmts
3672 and loop nest into above functions. Support loop nest distribution.
3673 Adjust dump information accordingly.
3675 2017-10-12 Bin Cheng <bin.cheng@arm.com>
3677 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
3678 and set PTYPE_SEQUENTIAL for merged partition.
3680 2017-10-12 Richard Biener <rguenther@suse.de>
3682 PR tree-optimization/69728
3684 2017-09-19 Richard Biener <rguenther@suse.de>
3686 PR tree-optimization/69728
3687 * graphite-sese-to-poly.c (schedule_error): New global.
3688 (add_loop_schedule): Handle empty domain by failing the
3690 (build_original_schedule): Handle schedule_error.
3692 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
3693 domain by returning an unchanged schedule.
3695 2017-10-12 Jakub Jelinek <jakub@redhat.com>
3697 * genrecog.c (validate_pattern): For VEC_SELECT verify that
3698 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
3700 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
3702 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
3705 2017-10-12 Jakub Jelinek <jakub@redhat.com>
3708 * expr.c (store_field): Don't optimize away bitsize == 0 store
3709 from CALL_EXPR with addressable return type.
3711 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
3713 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
3714 * config/rs6000/rs6000.md (sel): Delete mode attribute.
3715 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
3716 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
3717 TARGET_ISEL instead of TARGET_ISEL<sel>.
3719 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
3721 * config/rs6000/rs6000.c
3722 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
3724 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
3726 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
3728 (reg_or_cint_operand): Fix comment.
3729 (reg_or_zero_operand): New predicate.
3730 * config/rs6000/rs6000-protos.h (output_isel): Delete.
3731 * config/rs6000/rs6000.c (output_isel): Delete.
3732 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
3733 instead of reg_or_cint_operand. Output instruction directly (not via
3735 (isel_unsigned_<mode>): Ditto.
3736 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
3737 gpc_reg_operand. Add an instruction alternative for this. Output
3738 instruction directly.
3739 (*isel_reversed_unsigned_<mode>): Ditto.
3741 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
3743 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
3744 (TARGET_CANONICALIZE_COMPARISON): Define.
3746 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
3749 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
3750 Check whether the dest is REG before adding REG_EQUIV note.
3752 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
3755 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3757 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3758 (make_hard_regno_born, make_hard_regno_dead): Update
3759 bb_killed_pseudos and bb_gen_pseudos.
3761 2017-10-11 Nathan Sidwell <nathan@acm.org>
3763 * incpath.h (enum incpath_kind): Name enum, prefix values.
3764 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
3765 * incpath.c (heads, tails): Use INC_MAX.
3766 (add_env_var_paths, add_standard_paths): Use incpath_kind.
3767 (merge_include_chains, split_quote_chain,
3768 register_include_chains): Update incpath_kind names.
3769 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
3770 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
3772 (add_framework_path, darwin_register_objc_includes): Likewise.
3773 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
3775 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
3777 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
3778 Do not use float_operator operator predicate.
3779 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
3780 * config/i386/predicates.md (float_operator): Remove predicate.
3782 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
3784 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
3785 (*jccxf_i387): Ditto.
3786 (*jcc<mode>_i387): Ditto.
3787 (*jccu<mode>_i387): Ditto.
3788 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
3789 (*jcc_*_i387 splitters): Remove.
3790 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
3791 * config/i386/i386.c (ix86_split_fp_branch): Remove.
3792 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
3795 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
3797 * profile-count.h (slow_safe_scale_64bit): New function.
3798 (safe_scale_64bit): New inline.
3799 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
3800 * profile-count.c: Include wide-int.h
3801 (slow_safe_scale_64bit): New.
3803 2017-10-11 Nathan Sidwell <nathan@acm.org>
3805 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
3806 HAS_DECL_ASSEMBLER_NAME_P.
3807 * gimple-expr.c (gimple_decl_printable_name: Check
3808 HAS_DECL_ASSEMBLER_NAME_P too.
3809 * ipa-utils.h (type_in_anonymous_namespace_p): Check
3810 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
3811 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
3812 * passes.c (rest_of_decl_compilation): Check
3813 HAS_DECL_ASSEMBLER_NAME_P too.
3814 * recog.c (verify_changes): Likewise.
3815 * tree-pretty-print.c (dump_decl_name): Likewise.
3816 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
3818 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
3819 (SET_DECL_ASSEMBLER_NAME): Use it.
3820 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
3821 (COPY_DECL_ASSEMBLER_NAME): Likewise.
3822 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
3824 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
3826 * config.gcc (i386, x86_64): Add extra objects.
3827 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
3828 (ix86_min_insn_size): Declare.
3829 (ix86_issue_rate): Declare.
3830 (ix86_adjust_cost): Declare.
3831 (ia32_multipass_dfa_lookahead): Declare.
3832 (ix86_macro_fusion_p): Declare.
3833 (ix86_macro_fusion_pair_p): Declare.
3834 (ix86_bd_has_dispatch): Declare.
3835 (ix86_bd_do_dispatch): Declare.
3836 (ix86_core2i7_init_hooks): Declare.
3837 (ix86_atom_sched_reorder): Declare.
3838 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
3839 (COSTS_N_BYTES): Move to x86-tune-costs.h.
3840 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
3841 (rip_relative_addr_p): Rename to ...
3842 (ix86_rip_relative_addr_p): ... this one; export.
3843 (memory_address_length): Update.
3844 (ix86_issue_rate): Move to x86-tune-sched.c.
3845 (ix86_flags_dependent): Move to x86-tune-sched.c.
3846 (ix86_agi_dependent): Move to x86-tune-sched.c.
3847 (exact_dependency_1): Move to x86-tune-sched.c.
3848 (exact_store_load_dependency): Move to x86-tune-sched.c.
3849 (ix86_adjust_cost): Move to x86-tune-sched.c.
3850 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
3851 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
3852 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
3853 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
3854 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
3855 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
3856 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
3857 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
3858 (min_insn_size): Rename to ...
3859 (ix86_min_insn_size): ... this one; export.
3860 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
3861 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
3862 (core2i7_first_cycle_multipass_backtrack): Move to
3863 x86-tune-sched-core.c.
3864 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
3865 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
3866 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
3867 (ix86_avoid_jump_mispredicts): Update.
3868 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
3869 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
3870 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
3871 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
3872 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
3873 (MAX_INSN): Move to ix86-tune-sched-bd.c.
3874 (MAX_IMM): Move to ix86-tune-sched-bd.c.
3875 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
3876 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
3877 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
3878 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
3879 (MAX_STORE): Move to ix86-tune-sched-bd.c.
3880 (BIG): Move to ix86-tune-sched-bd.c.
3881 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
3882 (enum insn_path): Move to ix86-tune-sched-bd.c.
3883 (get_mem_group): Move to ix86-tune-sched-bd.c.
3884 (is_cmp): Move to ix86-tune-sched-bd.c.
3885 (dispatch_violation): Move to ix86-tune-sched-bd.c.
3886 (is_branch): Move to ix86-tune-sched-bd.c.
3887 (is_prefetch): Move to ix86-tune-sched-bd.c.
3888 (init_window): Move to ix86-tune-sched-bd.c.
3889 (allocate_window): Move to ix86-tune-sched-bd.c.
3890 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
3891 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
3892 (process_end_window): Move to ix86-tune-sched-bd.c.
3893 (allocate_next_window): Move to ix86-tune-sched-bd.c.
3894 (find_constant): Move to ix86-tune-sched-bd.c.
3895 (get_num_immediates): Move to ix86-tune-sched-bd.c.
3896 (has_immediate): Move to ix86-tune-sched-bd.c.
3897 (get_insn_path): Move to ix86-tune-sched-bd.c.
3898 (get_insn_group): Move to ix86-tune-sched-bd.c.
3899 (count_num_restricted): Move to ix86-tune-sched-bd.c.
3900 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
3901 (add_insn_window): Move to ix86-tune-sched-bd.c.
3902 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
3903 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
3904 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
3905 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
3906 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
3907 (do_dispatch): Move to ix86-tune-sched-bd.c.
3908 (has_dispatch): Move to ix86-tune-sched-bd.c.
3909 * config/i386/t-i386: Add new object files.
3910 * config/i386/x86-tune-costs.h: New file.
3911 * config/i386/x86-tune-sched-atom.c: New file.
3912 * config/i386/x86-tune-sched-bd.c: New file.
3913 * config/i386/x86-tune-sched-core.c: New file.
3914 * config/i386/x86-tune-sched.c: New file.
3916 2017-10-11 Liu Hao <lh_mouse@126.com>
3918 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
3919 the generic version below instead.
3920 (should_colorize): Recognize Windows consoles as terminals
3922 * pretty-print.c [__MINGW32__] (write_all): New function.
3923 [__MINGW32__] (find_esc_head): Likewise.
3924 [__MINGW32__] (find_esc_terminator): Likewise.
3925 [__MINGW32__] (eat_esc_sequence): Likewise.
3926 [__MINGW32__] (mingw_ansi_fputs): New function that handles
3928 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
3931 2017-10-11 Richard Biener <rguenther@suse.de>
3933 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
3934 Properly call analyze_scalar_evolution with the loop of the stmt.
3936 2017-10-11 Richard Biener <rguenther@suse.de>
3938 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
3939 * tree-core.h (tree_base): Add chrec_var union member.
3940 * tree.h (CHREC_VAR): Remove.
3941 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
3942 * tree-chrec.h (build_polynomial_chrec): Adjust.
3943 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
3944 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
3946 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
3948 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
3949 * match.pd: ... here.
3950 ((T) X == (T) Y): Relax condition.
3952 2017-10-11 Bin Cheng <bin.cheng@arm.com>
3954 PR tree-optimization/82472
3955 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
3957 (break_alias_scc_partitions): Update postorder number.
3959 2017-10-11 Martin Liska <mliska@suse.cz>
3962 * opts.c (parse_no_sanitize_attribute): Do not use error_value
3964 * opts.h (parse_no_sanitize_attribute): Remove last argument.
3966 2017-10-11 Martin Liska <mliska@suse.cz>
3968 * print-rtl.c (print_insn): Move declaration of idbuf
3969 to same scope as name.
3971 2017-10-11 Martin Liska <mliska@suse.cz>
3976 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
3979 2017-10-11 Martin Liska <mliska@suse.cz>
3982 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
3985 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
3987 * config.gcc (default_gnu_indirect_function): Default to yes for
3988 arm*-*-linux* with glibc.
3990 2017-10-11 Richard Biener <rguenther@suse.de>
3992 * tree-scalar-evolution.c (get_scalar_evolution): Handle
3993 default-defs and types we do not want to analyze.
3994 (interpret_loop_phi): Replace unreachable code with an assert.
3995 (compute_scalar_evolution_in_loop): Remove and inline ...
3996 (analyze_scalar_evolution_1): ... here, replacing condition with
3997 what makes the intent clearer. Remove handling of cases
3998 get_scalar_evolution now handles.
4000 2017-10-10 Jim Wilson <wilson@tuliptree.org>
4002 PR rtl-optimization/81434
4003 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
4004 comment for main loop. In sched_group_found if, also add checks for
4005 pass and min_cost_group.
4007 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
4009 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
4010 (rs6000_insn_cost): New function.
4011 * config/rs6000/rs6000.md (cost): New attribute.
4013 2017-10-10 Jakub Jelinek <jakub@redhat.com>
4014 H.J. Lu <hongjiu.lu@intel.com>
4018 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
4019 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
4020 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
4021 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
4022 to OPTION_MASK_ISA_AVX512VL - builtins that have both
4023 OPTION_MASK_ISA_MMX and some other bit set require both
4024 mmx and the ISAs without the mmx bit.
4025 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
4026 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
4027 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
4028 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
4029 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
4030 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
4031 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
4032 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
4033 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
4034 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
4035 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
4036 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
4037 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
4038 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
4039 Add OPTION_MASK_ISA_MMX.
4041 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
4043 * config.gcc (armv7*-*-freebsd*): New target.
4044 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
4046 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
4048 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
4049 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
4052 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
4054 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
4055 a template parameter.
4056 (WIDE_INT_REF_FOR): Update accordingly.
4057 * tree.h (wi::int_traits <const_tree>): Delete.
4058 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
4059 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
4060 (wi::tree_to_wide_ref): New typedef.
4061 (wi::to_wide): New function.
4062 * calls.c (get_size_range): Use wi::to_wide when operating on
4064 * cgraph.c (cgraph_node::create_thunk): Likewise.
4065 * config/i386/i386.c (ix86_data_alignment): Likewise.
4066 (ix86_local_alignment): Likewise.
4067 * dbxout.c (stabstr_O): Likewise.
4068 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
4069 * expr.c (const_vector_from_tree): Likewise.
4070 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
4071 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
4072 (fold_negate_expr_1, int_const_binop_1, const_binop)
4073 (fold_convert_const_int_from_real, optimize_bit_field_compare)
4074 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
4075 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
4076 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
4077 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
4078 (fold_not_const, round_up_loc): Likewise.
4079 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
4080 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
4081 (alloca_call_type): Likewise.
4082 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
4083 * godump.c (go_output_typedef): Likewise.
4084 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
4085 * internal-fn.c (get_min_precision): Likewise.
4086 * ipa-cp.c (ipcp_store_vr_results): Likewise.
4087 * ipa-polymorphic-call.c
4088 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
4089 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
4090 (ipa_modify_call_arguments): Likewise.
4091 * match.pd: Likewise.
4092 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
4093 * print-tree.c (print_node_brief, print_node): Likewise.
4094 * stmt.c (expand_case): Likewise.
4095 * stor-layout.c (layout_type): Likewise.
4096 * tree-affine.c (tree_to_aff_combination): Likewise.
4097 * tree-cfg.c (group_case_labels_stmt): Likewise.
4098 * tree-data-ref.c (dr_analyze_indices): Likewise.
4099 (prune_runtime_alias_test_list): Likewise.
4100 * tree-dump.c (dequeue_and_dump): Likewise.
4101 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
4102 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
4103 * tree-pretty-print.c (dump_generic_node): Likewise.
4104 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
4105 (simple_iv_with_niters): Likewise.
4106 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
4107 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
4108 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
4109 * tree-ssa-loop-niter.c (split_to_var_and_offset)
4110 (refine_value_range_using_guard, number_of_iterations_ne_max)
4111 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
4112 (get_cst_init_from_scev, record_nonwrapping_iv)
4113 (scev_var_range_cant_overflow): Likewise.
4114 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
4115 * tree-ssa-pre.c (compute_avail): Likewise.
4116 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
4117 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
4118 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
4119 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
4120 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
4121 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
4122 (dump_case_nodes, try_switch_expansion): Likewise.
4123 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
4124 (vect_do_peeling): Likewise.
4125 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
4126 * tree-vect-stmts.c (vectorizable_load): Likewise.
4127 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
4128 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
4129 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
4130 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
4131 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
4132 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
4133 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
4134 (evrp_dom_walker::before_dom_children): Likewise.
4135 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
4136 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
4137 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
4138 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
4139 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
4140 (get_range_pos_neg): Likewise.
4141 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
4142 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
4143 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
4144 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
4145 * config/avr/avr.c (avr_fold_builtin): Likewise.
4146 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
4147 * config/msp430/msp430.c (msp430_attr): Likewise.
4148 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
4149 * config/powerpcspe/powerpcspe-c.c
4150 (altivec_resolve_overloaded_builtin): Likewise.
4151 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
4152 (rs6000_expand_ternop_builtin): Likewise.
4153 * config/rs6000/rs6000-c.c
4154 (altivec_resolve_overloaded_builtin): Likewise.
4155 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
4156 (rs6000_expand_ternop_builtin): Likewise.
4157 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
4159 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4161 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
4162 when copying loop nest with only one inner loop.
4164 2017-10-10 Richard Biener <rguenther@suse.de>
4166 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
4167 blocks if SCEV is active.
4168 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
4170 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
4171 (scev_initialize): Assert we are not yet initialized.
4173 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4175 * tree-loop-distribution.c (generate_loops_for_partition): Remove
4176 inner loop's exit stmt by making it always exit the loop, otherwise
4177 we would generate an infinite empty loop.
4179 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4181 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
4182 renaming variables in new preheader if it's deleted.
4184 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4186 * tree-loop-distribution.c (struct partition): Remove unused field
4187 loops of the structure.
4188 (partition_alloc, partition_free): Ditto.
4189 (build_rdg_partition_for_vertex): Ditto.
4191 2017-10-09 Jeff Law <law@redhat.com>
4193 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
4194 return type to match prototype and documentation.
4196 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4198 * config/rs6000/rs6000.c (processor_costs): Move to ...
4199 * config/rs6000/rs6000.h: ... here.
4200 (rs6000_cost): Declare.
4202 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
4204 * except.c (setjmp_fn): New global variable.
4205 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
4206 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
4207 if DONT_USE_BUILTIN_SETJMP is defined.
4209 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4211 * target.def (insn_cost): New hook.
4212 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
4213 * doc/tm.texi: Regenerate.
4214 * rtlanal.c (insn_cost): Use the new hook.
4216 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4218 * combine.c (combine_validate_cost): Compute the new insn_cost,
4219 not just pattern_cost.
4220 (try_combine): Adjust comment.
4222 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4224 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
4226 * combine.c (uid_insn_cost): Adjust comment.
4227 (combine_validate_cost): Adjust comment. Use pattern_cost instead
4229 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
4230 * dse.c (find_shift_sequence): Ditto.
4231 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
4232 (bb_valid_for_noce_process_p): Use pattern_cost.
4233 * rtl.h (insn_rtx_cost): Delete.
4234 (pattern_cost): New prototype.
4235 (insn_cost): New prototype.
4236 * rtlanal.c (insn_rtx_cost): Rename to...
4237 (pattern_cost): ... this.
4240 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
4242 * config/i386/i386.md (*jcc_2): Remove insn pattern.
4243 (*jcc<mode>_0_r_i387): Ditto.
4244 (*jccxf_r_i387): Ditto.
4245 (*jcc<mode>_r_i387): Ditto.
4246 (*jccu<mode>_r_i387): Ditto.
4247 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
4248 (*jcc): Rename from *jcc_1.
4250 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4252 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
4253 deferred rescans after the lvx/stvx recombination pre-pass.
4255 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
4257 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
4258 memory operation instruction support.
4259 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
4260 (rs6000-ibm-aix[789]*): Likewise.
4261 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
4262 Document new functions.
4264 2017-10-09 Richard Biener <rguenther@suse.de>
4266 PR tree-optimization/82397
4267 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
4268 equality only for semantically equal trees.
4270 2017-10-09 Richard Biener <rguenther@suse.de>
4272 PR tree-optimization/82449
4273 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
4274 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
4275 allow constant addresses.
4276 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
4279 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4281 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
4284 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4287 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
4290 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4293 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
4295 2017-10-09 Jakub Jelinek <jakub@redhat.com>
4298 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
4299 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
4301 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
4303 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
4304 (WI_BINARY_PREDICATE_RESULT): Likewise.
4305 (wi::binary_traits::operator_result): New type.
4306 (wi::binary_traits::predicate_result): Likewise.
4307 (generic_wide_int::operator~, unary generic_wide_int::operator-)
4308 (generic_wide_int::operator==, generic_wide_int::operator!=)
4309 (generic_wide_int::operator&, generic_wide_int::and_not)
4310 (generic_wide_int::operator|, generic_wide_int::or_not)
4311 (generic_wide_int::operator^, generic_wide_int::operator+
4312 (binary generic_wide_int::operator-, generic_wide_int::operator*):
4314 (operator~, unary operator-, operator==, operator!=, operator&)
4315 (operator|, operator^, operator+, binary operator-, operator*): New
4317 * expr.c (get_inner_reference): Use wi::bit_and_not.
4318 * fold-const.c (fold_binary_loc): Likewise.
4319 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
4320 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
4321 (bit_value_binop): Likewise.
4322 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
4323 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
4324 (extract_range_from_binary_expr_1): Likewise.
4325 (masked_increment): Likewise.
4326 (simplify_bit_ops_using_ranges): Likewise.
4328 2017-10-09 Martin Jambor <mjambor@suse.cz>
4331 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
4332 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
4333 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
4335 (hsa_fixup_mov_insn_type): New function.
4336 (hsa_op_with_type::get_in_type): Use it.
4337 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
4338 immediates in an assert.
4339 (hsa_op_with_type::extend_int_to_32bit): New method.
4340 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
4341 types. Convert to dest type if necessary.
4342 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
4343 (reg_for_gimple_ssa): Pass false as min32int to
4344 hsa_type_for_scalar_tree_type.
4345 (gen_hsa_addr): Fixup type when creating addresable temporary.
4346 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
4347 (gen_hsa_unary_operation): Extend operands and convert to dest type if
4348 necessary. Call hsa_fixup_mov_insn_type.
4349 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
4350 extend operands and convert to dest type if necessary.
4351 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
4352 to dest type if necessary.
4353 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
4354 if conversion nt necessary and size matches.
4355 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
4356 to dest type if necessary.
4357 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
4358 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
4360 (gen_hsa_clrsb): Likewise.
4361 (gen_hsa_ffs): Likewise.
4362 (gen_hsa_divmod): Extend operands and convert to dest type if
4364 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
4366 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
4368 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
4369 Remove empty default arguments. Use a brace block as output
4371 (conditional return): Ditto.
4373 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
4374 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
4376 (group_ending_nop): Ditto.
4377 (doloop_end): Ditto.
4378 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
4379 (splitters for those): Ditto.
4381 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
4383 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
4384 a conditional jump (and the compare for it) so that pc_rtx is the
4386 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
4387 for the deleted and renamed ctr<mode>_internal[234] patterns.
4388 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
4389 Delete second conditional return pattern.
4390 (ctr<mode>_internal2): Delete this second bdnz pattern.
4391 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
4392 (ctr<mode>_internal4): Delete this second bdz pattern.
4394 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
4396 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
4397 (always_initialized_rtx_for_ssa_name_p): New predicate.
4398 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
4399 (finish_out_of_ssa): Free new field of SA.
4400 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
4401 * tree-ssa-coalesce.c: Include tree-ssa.h.
4402 (get_parm_default_def_partitions): Remove extern keyword.
4403 (get_undefined_value_partitions): New function.
4404 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
4405 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
4407 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
4409 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
4411 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
4413 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
4415 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
4416 for targets that preffer 128bit.
4418 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
4420 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
4422 2017-10-08 Olivier Hainque <hainque@adacore.com>
4424 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
4425 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
4427 (thumb_set_return_address): Likewise.
4429 2017-10-08 Olivier Hainque <hainque@adacore.com>
4431 * common/config/arm/arm-common.c (arm_except_unwind_info):
4432 Handle DWARF2_UNWIND_INFO.
4434 2017-10-07 Michael Collison <michael.collison@arm.com>
4436 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
4439 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
4441 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
4442 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
4443 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
4444 defined, force the creation of a new block for a dispatch label.
4446 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
4448 * invoke.texi (Wsuggest-attribute=cold): Document.
4449 * common.opt (Wsuggest-attribute=cold): New
4450 * ipa-pure-const.c (warn_function_cold): New function.
4451 * predict.c (compute_function_frequency): Use it.
4452 * predict.h (warn_function_cold): Declare.
4454 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
4456 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
4459 2017-10-06 Martin Liska <mliska@suse.cz>
4461 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
4462 keyword for member functions.
4463 (struct sanopt_tree_couple): New struct.
4464 (struct sanopt_tree_couple_hash): New function.
4465 (struct sanopt_ctx): Add new hash_map.
4466 (has_dominating_ubsan_ptr_check): New function.
4467 (record_ubsan_ptr_check_stmt): Likewise.
4468 (maybe_optimize_ubsan_ptr_ifn): Likewise.
4469 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
4470 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
4472 2017-10-06 Sudakshina Das <sudi.das@arm.com>
4475 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
4476 aarch64_simd_valid_immediate on CONST_VECTORs.
4477 (aarch64_reg_or_bic_imm): Likewise.
4479 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
4481 PR rtl-optimization/82396
4482 * haifa-sched.c (ready_sort_real): Disable qsort checking.
4484 2017-10-06 Sebastian Pop <sebpop@gmail.com>
4486 * graphite-dependences.c (scop_get_reads): Move code to...
4487 (scop_get_must_writes): Move code to...
4488 (scop_get_may_writes): Move code to...
4489 (scop_get_reads_and_writes): ... here.
4490 (scop_get_dependences): Call scop_get_reads_and_writes.
4492 2017-10-06 Jakub Jelinek <jakub@redhat.com>
4494 PR tree-optimization/82434
4495 * fold-const.h (can_native_encode_type_p,
4496 can_native_encode_string_p): Remove.
4497 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
4498 don't encode anything, just return what would be otherwise returned.
4499 (native_encode_fixed, native_encode_complex, native_encode_vector):
4501 (native_encode_string): Likewise. Inline by hand
4502 can_native_encode_string_p.
4503 (can_native_encode_type_p): Remove.
4504 (can_native_encode_string_p): Remove.
4505 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
4506 STRING_CSTs using can_native_encode_string_p, test all
4507 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
4508 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
4509 argument from native_encode_expr.
4510 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
4511 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
4514 2017-10-06 Richard Biener <rguenther@suse.de>
4516 PR tree-optimization/82397
4517 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
4518 operand_equal_p but rely on data_ref_compare_tree for detecting
4520 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
4521 to match up with dr_group_sort_cmp.
4523 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4526 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
4528 * config/s390/s390-builtin-types.def: Regenerate.
4530 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4533 * config/s390/s390-builtin-types.def: Regenerate.
4534 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
4535 Change flag from B_VXE to B_VX.
4536 (s390_vec_min_dbl): Remove B_VXE flag.
4538 2017-10-06 Richard Biener <rguenther@suse.de>
4540 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
4541 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
4542 translate_isl_ast_to_gimple::is_valid_rename,
4543 translate_isl_ast_to_gimple::get_rename,
4544 translate_isl_ast_to_gimple::get_def_bb_for_const,
4545 translate_isl_ast_to_gimple::get_new_name,
4546 translate_isl_ast_to_gimple::collect_all_ssa_names,
4547 translate_isl_ast_to_gimple::copy_loop_phi_args,
4548 translate_isl_ast_to_gimple::collect_all_ssa_names,
4549 translate_isl_ast_to_gimple::copy_loop_phi_args,
4550 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
4551 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
4552 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
4553 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
4554 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
4555 translate_isl_ast_to_gimple::copy_cond_phi_args,
4556 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
4557 translate_isl_ast_to_gimple::edge_for_new_close_phis,
4558 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
4559 translate_isl_ast_to_gimple::rename_uses,
4560 translate_isl_ast_to_gimple::rename_all_uses): Remove.
4561 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
4562 (set_rename_for_each_def): Likewise.
4563 (graphite_copy_stmts_from_block): Handle debug stmt resetting
4564 here. Handle rewriting SCEV analyzable uses here.
4565 (copy_bb_and_scalar_dependences): Generate code for PHI
4567 (graphite_regenerate_ast_isl): Adjust.
4568 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
4569 (add_write, add_read): New functions.
4570 (build_cross_bb_scalars_def): Use it and simplify.
4571 (build_cross_bb_scalars_use): Likewise.
4572 (graphite_find_cross_bb_scalar_vars): Inline into...
4573 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
4574 simulating out-of-SSA. Compute liveout and add dependencies.
4575 (build_scops): Force an empty entry block.
4576 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
4578 (sese_build_liveouts): Declare.
4579 (sese_trivially_empty_bb_p): Likewise.
4580 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
4581 compute liveout and debug_liveout.
4582 (sese_bad_liveouts_use): Remove.
4583 (sese_reset_debug_liveouts_bb): Likewise.
4584 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
4585 (sese_build_liveouts): Build liveout and debug_liveout and store
4587 (new_sese_info): Adjust.
4588 (free_sese_info): Likewise.
4589 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
4590 do not build liveout here.
4591 (move_sese_in_condition): Adjust region entry.
4592 (scev_analyzable_p): Match up with chrec_apply requirements.
4593 (sese_trivially_empty_bb_p): New.
4594 * tree-into-ssa.c (get_reaching_def): Properly support generating
4595 default-defs for incremental rewrite of anonymous names.
4597 2017-10-06 Richard Biener <rguenther@suse.de>
4599 * graphite-sese-to-poly.c (extract_affine): For casts increasing
4600 precision do not perform modulo reduction.
4602 2017-10-06 Richard Biener <rguenther@suse.de>
4604 PR tree-optimization/82436
4605 * tree-vect-slp.c (vect_supported_load_permutation_p): More
4606 conservatively choose the vectorization factor when checking
4607 whether we can perform the required load permutation.
4608 (vect_transform_slp_perm_load): Assert when we may not fail.
4610 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
4612 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
4613 message for incompatible -msdata=* and -mcall-* options.
4615 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
4617 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
4618 rate for post-reload scheduling.
4620 2017-10-05 Tamar Christina <tamar.christina@arm.com>
4622 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
4624 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
4626 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
4627 to improve monte carlo in scimark.
4629 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
4631 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
4632 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
4633 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
4634 pentium4_cost, nocona_cost): Set reassociation width to 1.
4635 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
4636 width to 2 for fp operations and 1 otherwise.
4637 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
4639 (atom_cost): Set reassociation width to 2.
4640 (slm_cost, generic_cost): Set fp reassociation width
4641 to 2 and 1 otherwise.
4642 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
4643 (core_cost): Set fp reassociation width to 4 and vector to 2.
4644 (ix86_reassociation_width): Rewrite using cost table; special case
4645 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
4646 and TARGET_AVX128_OPTIMAL.
4647 * config/i386/i386.h (processor_costs): Add
4648 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
4649 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
4650 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
4651 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
4652 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
4653 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
4655 2017-10-05 Nathan Sidwell <nathan@acm.org>
4657 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
4659 2017-10-05 Tamar Christina <tamar.christina@arm.com>
4661 * config/arm/arm.c (arm_test_fpu_data): New.
4662 (arm_run_selftests): Call arm_test_fpu_data.
4664 2017-10-04 Nathan Sidwell <nathan@acm.org>
4666 * toplev.c (toplev::main): Remove excess parens on pretty_printer
4668 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
4670 2017-10-04 Sudakshina Das <sudi.das@arm.com>
4672 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
4673 check type for aarch64_simd_valid_immediate.
4674 (aarch64_output_simd_mov_immediate): Update prototype.
4675 (aarch64_simd_valid_immediate): Update prototype.
4676 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
4677 support for ORR-immediate.
4678 (and<mode>3): modified pattern to add support for BIC-immediate.
4679 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
4680 now checks for valid immediate for BIC and ORR based on new enum
4682 (aarch64_output_simd_mov_immediate): Function now used to output
4683 BIC/ORR imm as well based on new enum argument.
4684 * config/aarch64/constraints.md (Do): New vector immediate constraint.
4686 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
4687 (aarch64_reg_or_bic_imm): Likewise.
4689 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4691 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
4692 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
4694 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
4698 PR rtl-optimization/82396
4699 * haifa-sched.c (autopref_multipass_init): Simplify
4701 (autopref_rank_data): Simplify sort order.
4702 * sched-int.h (autopref_multipass_data_): Remove
4703 multi_mem_insn_p, min_offset and max_offset.
4705 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4707 * doc/sourcebuild.texi: Document vect_peeling_profitable.
4709 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4711 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
4714 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4716 * doc/sourcebuild.texi: Document vect_long_mult.
4718 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
4720 PR tree-optimization/82413
4721 * fold-const.c (build_range_check): Use widest_int when comparing
4722 the maximum ETYPE value with HIGH.
4724 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
4726 PR rtl-optimization/82396
4727 * haifa-sched.c (autopref_multipass_init): Simplify
4729 (autopref_rank_data): Simplify sort order.
4730 * sched-int.h (autopref_multipass_data_): Remove
4731 multi_mem_insn_p, min_offset and max_offset.
4733 2017-10-04 Jakub Jelinek <jakub@redhat.com>
4735 PR tree-optimization/82381
4736 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
4737 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
4740 PR tree-optimization/82374
4741 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
4742 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
4743 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
4744 current_function_decl to the new decl.
4746 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
4748 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
4749 helper macro for IEEE float128 hardware built-in functions.
4750 (SQRTF128_ODD): Add built-in functions with the round-to-odd
4752 (TRUNCF128_ODD): Likewise.
4753 (ADDF128_ODD): Likewise.
4754 (SUBF128_ODD): Likewise.
4755 (MULF128_ODD): Likewise.
4756 (DIVF128_ODD): Likewise.
4757 (FMAF128_ODD): Likewise.
4758 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
4759 UNSPEC_TRUNC_ROUND_TO_ODD.
4760 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
4761 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
4762 floating point round to odd instructions.
4763 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
4764 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
4765 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
4766 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
4767 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
4768 (trunc<mode>sf2_hw): Change the truncate with round to odd
4769 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
4770 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
4771 to odd hardware instructions.
4772 (sub<mode>3_odd): Likewise.
4773 (mul<mode>3_odd): Likewise.
4774 (div<mode>3_odd): Likewise.
4775 (sqrt<mode>2_odd): Likewise.
4776 (fma<mode>4_odd): Likewise.
4777 (fms<mode>4_odd): Likewise.
4778 (nfma<mode>4_odd): Likewise.
4779 (nfms<mode>4_odd): Likewise.
4780 (trunc<mode>df2_odd): Change the truncate with round to odd
4781 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
4783 * doc/extend.texi (PowerPC built-in functions): Update documentation
4784 for existing IEEE float128-bit built-in functions. Add built-in
4785 functions that generate the IEEE 128-bit floating point round to
4788 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
4790 PR rtl-optimization/77729
4791 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
4792 to (X&(C1&~C2))|C2 transformations.
4794 2017-10-03 Martin Jambor <mjambor@suse.cz>
4796 PR tree-optimization/82363
4797 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
4798 mismatch, mark lacc written regardless of racc.
4800 2017-10-03 Jakub Jelinek <jakub@redhat.com>
4802 PR tree-optimization/82381
4803 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
4804 stmt_to_insert nor wheather SSA_NAMEs are default defs.
4805 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
4806 fallthrough into reassoc_stmt_dominates_stmt_p.
4809 * combine.c (combine_instructions): Don't combine in unreachable
4812 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
4815 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
4816 function to not use the have_cpu variable. Do not set cpu_index,
4817 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
4819 (rs6000_valid_attribute_p): Remove duplicate initializations of
4820 old_optimize and func_optimize.
4821 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
4822 (rs6000_activate_target_options): Make global.
4823 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
4826 2017-10-02 Jakub Jelinek <jakub@redhat.com>
4828 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
4829 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
4830 Set *poffset to 0 and *psize and *pmax_size to -1 if
4831 *poffset + *psize overflows in HOST_WIDE_INT.
4833 PR tree-optimization/82387
4834 PR tree-optimization/82388
4835 PR tree-optimization/82389
4836 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
4837 instead of live_bytes non-NULL.
4839 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
4842 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
4845 2017-10-02 Richard Biener <rguenther@suse.de>
4847 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
4848 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
4849 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
4851 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
4853 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
4854 requested precision matches the type's.
4855 * calls.c (alloc_max_size): Calculate the new candidate size as
4856 a widest_int and use wi::to_widest when comparing it with the
4857 current candidate size.
4858 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
4859 zero rather than integer_zero_node.
4860 * match.pd: Check for a no-op conversion before using wi::add
4861 rather than after. Use tree_to_uhwi when summing small shift
4862 counts into an unsigned int.
4864 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
4865 Alan Hayward <alan.hayward@arm.com>
4866 David Sherwood <david.sherwood@arm.com>
4869 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
4870 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
4871 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
4872 POINTER_AND_FP_REGS.
4874 2017-10-02 Richard Biener <rguenther@suse.de>
4876 PR tree-optimization/82355
4877 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
4878 a mapping for the enclosing loop but avoid generating one for
4880 (copy_bb_and_scalar_dependences): Remove premature codegen
4881 error on PHIs in blocks duplicated into multiple places.
4882 * graphite-scop-detection.c
4883 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
4884 in the region use it as loop and nest to analyze the DR in.
4885 (try_generate_gimple_bb): Likewise.
4886 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
4887 (add_loop_constraints): For blocks in a loop not in the region
4888 create a dimension with a single iteration.
4889 * sese.h (gbb_loop_at_index): Remove assert.
4891 2017-10-01 Kevin Buettner <kevinb@redhat.com>
4893 * omp-expand.c (adjust_context_scope): New function.
4894 (expand_parallel_call): Call adjust_context_scope.
4896 2017-10-01 Jeff Law <law@redhat.com>
4898 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
4899 dom_opt_dom_walker class with direct access to private members.
4900 Add comments. Call test_for_singularity.
4901 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
4902 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
4903 m_dummy_cond anymore.
4904 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
4906 (pass_dominator:execute): Build the dummy_cond here and pass it
4907 to the dom_opt_dom_walker ctor.
4908 (test_for_singularity): New function.
4910 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
4911 Maya Rashish <coypu@sdf.org>
4913 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
4914 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
4915 (alpha*-*-netbsd*) Use nbsd_tm_file.
4916 (arm*-*-netbsdelf*) Likewise.
4917 (i[34567]86-*-netbsdelf*) Likewise.
4918 (x86_64-*-netbsd*) Likewise.
4919 (mips*-*-netbsd*) Likewise.
4920 (powerpc-*-netbsd*) Likewise.
4921 (sh*-*-netbsd*) Likewise.
4922 (sparc-*-netbsdelf*) Likewise.
4923 (sparc64-*-netbsd*) Likewise.
4924 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
4926 (vax-*-netbsdelf*) Likewise.
4927 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
4928 (UINT_FAST8_TYPE) Likewise.
4929 (INT_FAST16_TYPE) Check CHAR_FAST16.
4930 (UINT_FAST16_TYPE) Likewise.
4932 2017-09-30 Jakub Jelinek <jakub@redhat.com>
4935 * config/i386/i386.md
4936 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
4937 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
4938 *divmodsi4_zext_2): New define_insn_and_split.
4939 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
4940 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
4941 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
4942 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
4943 New define_insn_and_split.
4944 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
4945 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
4946 operands[1] having DImode when mode is SImode.
4948 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
4949 always SImode for DIV and MOD in REG_EQUAL notes.
4951 2017-09-29 Yury Gribov <tetra2005@gmail.com>
4954 * match.pd: Fix handling of NaNs in pattern.
4956 2017-09-29 Jeff Law <law@redhat.com>
4958 * sbitmap.c (bitmap_bit_in_range_p): New function.
4959 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
4960 * tree-ssa-dse.c (live_bytes_read): New function.
4961 (dse_classify_store): Ignore reads of dead bytes.
4963 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
4964 typos and whitespace errors.
4965 * config/i386/predicates.md (address_no_seg_operand): Likewise.
4966 * config/s390/s390.c (s390_emit_prologue): Likewise.
4968 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
4971 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
4972 with a symbol for LRA.
4974 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
4976 PR rtl-optimization/82338
4977 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
4979 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
4981 * genmodes.c (calc_wider_mode): Suppress qsort macro.
4982 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
4983 (qsort_chk): Declare.
4984 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
4985 (qsort_chk): New function.
4987 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4989 PR tree-optimization/82337
4990 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
4991 phi definition if the PHI result appears in an abnormal PHI.
4992 (find_basis_for_base_expr): Don't record a basis if the LHS of the
4993 basis appears in an abnormal PHI.
4995 2017-09-29 Richard Biener <rguenther@suse.de>
4997 * graphite-isl-ast-to-gimple.c
4998 (translate_isl_ast_to_gimple::set_codegen_error): New function.
4999 (binary_op_to_tree): Use it.
5000 (get_rename_from_scev): Likewise.
5001 (copy_loop_phi_nodes): Likewise.
5002 (copy_bb_and_scalar_dependences): Likewise.
5003 (translate_pending_phi_nodes): Likewise.
5005 2017-09-29 Jakub Jelinek <jakub@redhat.com>
5008 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
5009 for movabsq $(i32 << shift), r64.
5011 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
5013 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
5014 index when encoding %esp as %rsp to avoid 0x67 prefix.
5016 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
5018 * config/i386/i386.md (*movsf_internal, *movdf_internal):
5019 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
5021 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
5023 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
5024 Extensions with more than 16 double VFP registers.
5025 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
5026 to_clear_mask and all code related to it. Replace the remaining
5027 entry by a sbitmap and adapt code accordingly.
5029 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
5031 * brig-builtins.def: Change pure attributes to const.
5033 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5035 * config.gcc (default_gnu_indirect_function): Default to yes for
5036 sparc*-*-linux* with glibc.
5038 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5040 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
5041 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
5042 when creating .init_array and .fini_array sections with priority
5045 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
5048 * config/aarch64/aarch64.c
5049 (aarch64_builtin_support_vector_misalignment): Always return false
5050 when misalignment is unknown.
5052 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
5054 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
5055 this function to return false if the definition used by the swap
5056 instruction is artificial, or if the memory address from which the
5057 constant value is loaded is not represented by a base address held
5058 in a register or if the base address register is a frame or stack
5059 pointer. Additionally, return false if the base address of the
5060 loaded constant is a SYMBOL_REF but is not considered to be a
5062 (replace_swapped_load_constant): New function.
5063 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
5064 loaded constant vector with a load of a swapped constant vector.
5066 2017-09-27 Carl Love <cel@us.ibm.com>
5068 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
5069 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
5070 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
5073 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
5075 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
5076 first, always call autopref_rank_data otherwise.
5078 2017-09-27 Richard Biener <rguenther@suse.de>
5080 * graphite-scop-detection.c (find_scop_parameters): Move
5081 loop bound handling ...
5082 (gather_bbs::before_dom_children): ... here, avoiding the need
5083 to build scop_info->loop_nest.
5084 (record_loop_in_sese): Remove.
5085 * sese.h (sese_info_t::loop_nest): Remove.
5086 * sese.c (new_sese_info): Do not allocate loop_nest.
5087 (free_sese_info): Do not free loop_nest.
5089 2017-09-27 Jakub Jelinek <jakub@redhat.com>
5092 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
5093 lhs from calls if the lhs has addressable type.
5095 2017-09-27 Richard Biener <rguenther@suse.de>
5097 * graphite.h (scop::max_alias_set): New member.
5098 * graphite-scop-detection.c: Remove references to non-existing
5099 --param in comments.
5100 (build_alias_sets): Record the maximum alias set used for drs.
5101 (build_scops): Support zero as unlimited for
5102 --param graphite-max-arrays-per-scop.
5103 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
5105 (build_poly_sr_1): ... here. Compute alias set based on the
5106 maximum alias set used for drs rather than
5107 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
5109 2017-09-27 Richard Biener <rguenther@suse.de>
5111 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
5112 --param loop-block-tile-size=0 to disable tiling.
5114 2017-09-27 Richard Biener <rguenther@suse.de>
5116 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
5117 (graphite-max-nb-scop-params): Document special value zero.
5118 * domwalk.h (dom_walker::STOP): New symbolical constant.
5119 (dom_walker::dom_walker): Add optional parameter for bb to
5121 (dom_walker::~dom_walker): Declare.
5122 (dom_walker::before_dom_children): Document STOP return value.
5123 (dom_walker::m_user_bb_to_rpo): New member.
5124 (dom_walker::m_bb_to_rpo): Likewise.
5125 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
5126 mapping here if not provided by the user.
5127 (dom_walker::~dom_walker): Free bb to RPO mapping if not
5128 provided by the user.
5129 (dom_walker::STOP): Define.
5130 (dom_walker::walk): Do not compute bb to RPO mapping here.
5131 Support STOP return value from before_dom_children to stop
5133 * graphite-optimize-isl.c (optimize_isl): If the schedule
5134 is the same still generate code if -fgraphite-identity
5135 or -floop-parallelize-all are given.
5136 * graphite-scop-detection.c: Include cfganal.h.
5137 (gather_bbs::gather_bbs): Get and pass through bb to RPO
5139 (gather_bbs::before_dom_children): Return STOP for BBs
5141 (build_scops): Compute bb to RPO mapping and pass it to
5142 the domwalk. Treat --param graphite-max-nb-scop-params=0
5143 as not limiting the number of params.
5144 * graphite.c (graphite_initialize): Remove limit on the
5145 number of basic-blocks in a function.
5146 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
5147 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
5148 default value of 10.
5150 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5152 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
5153 Adjust code to eliminate needing to do the shift right 32-bits
5154 operation after XSCVDPSPN.
5156 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5158 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
5159 ((X / Y) != 0 -> X >= Y): Likewise.
5161 2017-09-26 Carl Love <cel@us.ibm.com>
5163 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
5164 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
5165 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
5166 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
5167 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
5168 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
5169 definitions and overloading.
5170 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
5171 statement for P9V_BUILTIN_XST_LEN_R.
5172 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
5173 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
5174 define_expand and define_insn for the instructions and builtins.
5175 * doc/extend.texi: Update the built-in documentation file for the new
5177 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
5178 define_insn for the instructions
5180 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
5183 * gcc/config/netbsd-protos.h: New file.
5184 * gcc/config/netbsd.c: New file.
5185 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
5186 * gcc/config/t-netbsd: New file.
5187 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
5188 (tmake_file) Add t-netbsd.
5189 (extra_objs) Add netbsd.o.
5191 2017-09-26 Janus Weil <janus@gcc.gnu.org>
5195 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
5197 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5199 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
5200 sign extension from a vector register to a GPR by doing a 32-bit
5201 direct move and then an EXTSW.
5202 (extendsi<mode>2 splitter): Likewise.
5203 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
5204 right or vector extract after doing XSCVDPSPN. Use
5205 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
5207 (movdi_from_sf_zero_ext): Likewise.
5208 (reload_gpr_from_vsxsf): Likewise.
5209 (p8_mfvsrd_4_disf): Delete, no longer used.
5210 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
5211 then needing to move the SFmode to a GPR to use the XSCVDPSP
5212 instruction instead of FRSP and XSCVDPSPN.
5213 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
5214 it is adjacent to the other XSCVSPDP insns.
5215 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
5216 SFmode to be in traditional Altivec registers.
5217 (vsx_xscvdpspn): Eliminate useless alternative constraint.
5218 (vsx_xscvspdpn): Likewise.
5219 (vsx_xscvspdpn_scalar): Likewise.
5221 2017-09-26 Martin Jambor <mjambor@suse.cz>
5223 * tree-sra.c (compare_access_positions): Put integral types first,
5224 stabilize sorting of integral types, remove conditions putting
5225 non-full-precision integers last.
5226 (sort_and_splice_var_accesses): Disable scalarization if a
5227 non-integert would be represented by a non-full-precision integer.
5229 2017-09-26 Joseph Myers <joseph@codesourcery.com>
5231 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
5232 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
5233 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
5234 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
5235 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
5236 conditionals inside the function instead of around it. Call
5237 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
5238 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
5240 2017-09-26 Richard Biener <rguenther@suse.de>
5242 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
5244 (scop_detection::build_scop_breadth): ... this. Removed.
5245 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
5246 (scop_detection::harmful_stmt_in_bb): Likewise.
5247 (scop_detection::graphite_can_represent_stmt): Likewise.
5248 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
5249 (scop_detection::can_represent_loop): Remove recursion, fold in ...
5250 (scop_detection::can_represent_loop_1): ... this. Removed.
5251 (scop_detection::harmful_loop_in_region): Simplify after inlining
5252 the above and remove more quadraticness.
5253 (build_scops): Adjust.
5254 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
5257 2017-09-26 Jakub Jelinek <jakub@redhat.com>
5260 * config/i386/i386.c (ix86_print_operand_address_as): Only test
5261 REGNO (base) == SP_REG if base is a REG.
5264 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
5265 if it is different SSA_NAME.
5266 (optimize_range_tests_cmp_bitwise): New function.
5267 (optimize_range_tests): Call it.
5269 2017-09-26 Richard Biener <rguenther@suse.de>
5271 PR tree-optimization/82321
5272 * graphite.c (canonicalize_loop_closed_ssa): Properly check
5273 for the def being inside the loop.
5275 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5277 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
5279 * config/s390/s390-builtins.def: Fix constraint on op4.
5281 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5283 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
5284 independent expanders.
5285 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
5286 ("vec_ordered", "vec_unordered"): New expanders.
5288 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5290 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
5293 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5295 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
5296 vec_unpacks_lo_v16qi.
5297 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
5299 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5301 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
5302 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
5303 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
5305 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5307 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
5309 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
5311 ("*vec_slb<mode>"): New insn pattern.
5312 ("vec_shr_<mode>"): New expander.
5313 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
5314 and force the shift count operand to V16QImode.
5315 ("vec_srb<mode>"): Set shift count mode to V16QI.
5317 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5319 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
5320 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
5321 ("vec_widen_smult_hi_<mode>"): New expander definitions.
5323 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
5326 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
5328 2017-09-26 Richard Biener <rguenther@suse.de>
5330 PR tree-optimization/82320
5331 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
5334 2017-09-25 Jeff Law <law@redhat.com>
5336 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
5337 prototype for new argument.
5338 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
5339 mostly extracted from rs6000_emit_allocate_stack.
5340 (rs6000_emit_probe_stack_range_stack_clash): New function.
5341 (rs6000_emit_allocate_stack): Call
5342 rs6000_emit_probe_stack_range_stack_clash as needed.
5343 (rs6000_emit_probe_stack_range): Add additional argument
5344 to call to gen_probe_stack_range{si,di}.
5345 (output_probe_stack_range): New.
5346 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
5347 (output_probe_stack_range_stack_clash): New.
5348 (rs6000_emit_prologue): Emit notes into dump file as requested.
5349 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
5350 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
5351 Add additional operand and pass it to output_probe_stack_range.
5353 2017-09-25 Bin Cheng <bin.cheng@arm.com>
5355 PR tree-optimization/82163
5356 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
5357 (checking_verify_loop_closed_ssa): New parameter.
5358 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
5359 (check_loop_closed_ssa_stmt): Delete.
5360 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
5361 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
5362 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
5365 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
5367 * brig-builtins.def: Treat HSAIL barrier builtins as
5368 setjmp/longjump style functions.
5370 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
5372 * target.def (constant_alignment): New hook.
5373 * defaults.h (CONSTANT_ALIGNMENT): Delete.
5374 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
5375 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
5376 * doc/tm.texi: Regenerate.
5377 * targhooks.h (default_constant_alignment): Declare.
5378 (constant_alignment_word_strings): Likewise.
5379 * targhooks.c (default_constant_alignment): New function.
5380 (constant_alignment_word_strings): Likewise.
5381 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
5382 instead of CONSTANT_ALIGNMENT.
5383 * varasm.c (align_variable, get_variable_align, build_constant_desc)
5384 (force_const_mem): Likewise.
5385 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
5386 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
5387 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
5388 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5389 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
5391 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
5392 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5393 constant_alignment_word_strings.
5394 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
5395 (CONSTANT_ALIGNMENT): Likewise.
5396 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5397 (arm_constant_alignment): New function.
5398 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
5399 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5400 constant_alignment_word_strings.
5401 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
5402 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5403 constant_alignment_word_strings.
5404 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
5405 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5406 (cris_constant_alignment): New function.
5407 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
5408 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5409 (epiphany_constant_alignment): New function.
5410 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
5411 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5412 constant_alignment_word_strings.
5413 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
5414 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5415 constant_alignment_word_strings.
5416 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
5417 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5418 constant_alignment_word_strings.
5419 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
5420 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
5421 * config/i386/i386.c (ix86_constant_alignment): Make static.
5422 Use the same interface as the target hook.
5423 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5424 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
5425 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5426 constant_alignment_word_strings.
5427 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
5428 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
5429 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5430 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
5431 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5432 constant_alignment_word_strings.
5433 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
5434 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5435 constant_alignment_word_strings.
5436 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
5437 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5438 constant_alignment_word_strings.
5439 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
5440 * config/microblaze/microblaze.c (microblaze_constant_alignment):
5442 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5443 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
5444 * config/mips/mips.c (mips_constant_alignment): New function.
5445 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5446 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
5447 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
5448 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5449 (mmix_constant_alignment): Make static. Use the same interface
5451 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
5452 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5453 constant_alignment_word_strings.
5454 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
5455 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5456 constant_alignment_word_strings.
5457 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
5458 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5459 constant_alignment_word_strings.
5460 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
5461 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5462 (rs6000_constant_alignment): New function.
5463 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
5464 * config/riscv/riscv.c (riscv_constant_alignment): New function.
5465 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5466 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
5467 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5468 (rs6000_constant_alignment): New function.
5469 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
5470 * config/s390/s390.c (s390_constant_alignment): New function.
5471 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5472 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
5473 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5474 constant_alignment_word_strings.
5475 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
5476 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5477 (sparc_constant_alignment): New function.
5478 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
5479 * config/spu/spu.c (spu_constant_alignment): New function.
5480 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5481 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
5482 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5483 constant_alignment_word_strings.
5484 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
5485 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5486 constant_alignment_word_strings.
5487 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
5488 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5489 constant_alignment_word_strings.
5490 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
5491 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5492 (visium_constant_alignment): New function.
5493 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
5494 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5495 (xtensa_constant_alignment): New function.
5496 * system.h (CONSTANT_ALIGNMENT): Poison.
5498 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
5500 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
5501 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
5502 (rs6000_builtin_valid_without_lhs): New helper function.
5503 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5504 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
5506 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
5508 * target.h (vec_perm_indices): Use unsigned short rather than
5510 (auto_vec_perm_indices): Likewise.
5511 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
5512 Use unsigned int rather than unsigned char.
5513 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
5515 2017-09-25 Richard Biener <rguenther@suse.de>
5517 * cfgloop.h (sort_sibling_loops): Declare.
5518 * cfgloop.c (sort_sibling_loops_cmp): New helper.
5519 (sort_sibling_loops): New function sorting the sibling loop list
5521 * graphite.c (graphite_transform_loops): Sort sibling loops.
5523 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
5525 * target.def (vec_perm_const_ok): Change sel parameter to
5527 * optabs-query.c (can_vec_perm_p): Update accordingly.
5528 * doc/tm.texi: Regenerate.
5529 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
5530 auto_vec_perm_indices and remove separate nelt field.
5531 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
5532 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
5533 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
5534 (aarch64_expand_vec_perm_const): Update accordingly.
5535 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
5536 to vec_perm_indices.
5537 * config/arm/arm.c (expand_vec_perm_d): Change perm to
5538 auto_vec_perm_indices and remove separate nelt field.
5539 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
5540 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
5541 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
5543 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
5544 to vec_perm_indices.
5545 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
5546 sel to vec_perm_indices.
5547 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
5548 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
5549 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
5551 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
5554 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
5557 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
5558 on the FUNCTION_DECL function context if it has a DIE that is a
5561 2017-09-25 Richard Biener <rguenther@suse.de>
5563 PR tree-optimization/82285
5564 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
5567 2017-09-25 Tom de Vries <tom@codesourcery.com>
5571 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
5574 2017-09-25 Richard Biener <rguenther@suse.de>
5576 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
5577 ISL errors other than isl_error_quota happen. Dump if the
5578 schedule is the same.
5579 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
5580 errors instead of aborting inside ISL.
5582 2017-09-25 Iain Sandoe <iain@codesourcery.com>
5585 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
5586 of libgcc_eh for m64.
5587 * config/i386/darwin64.h: Likewise.
5589 2017-09-25 Richard Biener <rguenther@suse.de>
5592 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
5593 attribute for incomplete types nor twice for complete ones.
5595 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
5598 * config/i386/i386.c (ix86_print_operand_address_as): Encode
5599 %esp as %rsp to avoid 0x67 prefix if there is no index or base
5602 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
5605 * config/i386/i386.opt (mprefer-avx256): Use
5606 ix86_target_flags variable.
5607 * config/i386/i386.c (ix86_target_string): Move
5608 -mprefer-avx256 to flag2_opts.
5610 2017-09-22 Jakub Jelinek <jakub@redhat.com>
5613 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
5614 and x != -1 | y != -1 into (x & y) != -1.
5616 2017-09-22 Steve Ellcey <sellcey@cavium.com>
5618 * config.gcc: Add new case statement to set
5619 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
5620 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
5621 s390x-*-linux* case statements. Added aarch64 to the list of
5622 supported architectures.
5624 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
5626 PR tree-optimization/82289
5627 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
5628 STMT_VINFO_RELEVANT_P.
5630 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
5631 Alan Hayward <alan.hayward@arm.com>
5632 David Sherwood <david.sherwood@arm.com>
5634 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
5635 for VR_RANGE only; don't allow VR_ANTI_RANGE.
5636 (extract_range_from_binary_expr_1): Don't call
5637 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
5639 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
5640 Alan Hayward <alan.hayward@arm.com>
5641 David Sherwood <david.sherwood@arm.com>
5643 * target.def (preferred_vector_alignment): New hook.
5644 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
5646 * doc/tm.texi: Regenerate.
5647 * targhooks.h (default_preferred_vector_alignment): Declare.
5648 * targhooks.c (default_preferred_vector_alignment): New function.
5649 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
5651 (DR_TARGET_ALIGNMENT): New macro.
5652 (aligned_access_p): Update commentary.
5653 (vect_known_alignment_in_bytes): New function.
5654 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
5656 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
5657 Calculate the misalignment based on the target alignment rather than
5659 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
5660 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
5661 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
5662 the target alignment, rather than masking the element misalignment
5663 with the number of elements in a vector. Also use the target
5664 alignment when calculating the maximum number of peels.
5665 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
5666 instead of TYPE_ALIGN_UNIT.
5667 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
5668 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
5669 (vect_create_addr_base_for_vector_ref): Update call accordingly.
5670 (vect_create_data_ref_ptr): Likewise.
5671 (vect_setup_realignment): Realign by ANDing with
5672 -DR_TARGET_MISALIGNMENT.
5673 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
5674 the number of peels based on DR_TARGET_ALIGNMENT.
5675 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
5676 with the guaranteed alignment boundary when deciding whether
5678 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
5679 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
5680 (ensure_base_align): Remove stmt_info parameter. Get the
5681 target base alignment from DR_TARGET_ALIGNMENT.
5682 (vectorizable_store): Update call accordingly. Interpret
5683 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
5685 (vectorizable_load): Likewise.
5687 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
5688 Alan Hayward <alan.hayward@arm.com>
5689 David Sherwood <david.sherwood@arm.com>
5691 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
5692 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
5693 (vect_enhance_data_refs_alignment): Likewise.
5695 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
5697 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
5698 error. Only quit immediately if parsing is complete.
5699 (BEGIN): Initialize fatal_err and parse_done.
5700 (begin fpu, end fpu): Check number of arguments.
5701 (begin arch, end arch): Likewise.
5702 (begin cpu, end cpu): Likewise.
5703 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
5704 (optalias): Likewise.
5706 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
5708 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
5709 * config/arm/arm-isa.h: Delete. Move definitions to ...
5710 * arm-cpus.in: ... here. Use new feature and fgroup values.
5711 * config/arm/arm.c (arm_option_override): Use lower case for feature
5713 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
5714 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
5715 * config/arm/parsecpu.awk (END): Add new command 'isa'.
5717 (print_isa_bits_for): New function.
5718 (gen_isa): New function.
5719 (gen_comm_data): Use print_isa_bits_for.
5720 (define feature): New keyword.
5721 (define fgroup): New keyword.
5722 * config/arm/t-arm (TM_H): Remove.
5723 (GTM_H): Add arm-isa.h.
5724 (arm-isa.h): Add rule to generate file.
5725 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
5726 case for feature bit names.
5728 2017-09-22 Richard Biener <rguenther@suse.de>
5730 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
5732 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
5733 print of no dependency loops ...
5734 * graphite.c (graphite_transform_loops): ... here.
5735 (canonicalize_loop_closed_ssa_form): Work from inner to outer
5737 (same_close_phi_node, remove_duplicate_close_phi,
5738 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
5739 (canonicalize_loop_closed_ssa): ... here and simplify.
5740 * graphite-optimize-isl.c: Include tree-vectorizer.h.
5741 (optimize_isl): Use dump_printf_loc to tell when we stopped
5742 optimizing because of an ISL timeout.
5744 2017-09-22 Richard Biener <rguenther@suse.de>
5746 PR tree-optimization/82291
5747 * tree-if-conv.c (predicate_mem_writes): Make sure to
5748 remove writes in blocks predicated with false.
5750 2017-09-22 Richard Biener <rguenther@suse.de>
5752 * sese.c: Include cfganal.h.
5753 (if_region_set_false_region): Remove.
5754 (create_if_region_on_edge): Likewise.
5755 (move_sese_in_condition): Re-implement without destroying
5758 2017-09-22 Richard Biener <rguenther@suse.de>
5760 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
5761 Verify both BBs contain loop PHI nodes before dispatching to
5763 (graphite_regenerate_ast_isl): Do not recompute dominators,
5764 do not verify three times. Restructure for clarity.
5765 * graphite-scop-detection.c (same_close_phi_node,
5766 remove_duplicate_close_phi, make_close_phi_nodes_unique,
5767 defined_in_loop_p, canonicalize_loop_closed_ssa,
5768 canonicalize_loop_closed_ssa_form): Simplify, remove excess
5769 checking and SSA rewrite, move to ...
5770 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
5771 (graphite_initialize): Do not pass in ctx, do not reset the
5772 SCEV cache, compute only dominators.
5773 (graphite_transform_loops): Allocate ISL ctx after
5774 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
5775 Maintain post-dominators only around build_scops.
5776 * sese.c (if_region_set_false_region): Make static. Free
5777 and recompute dominators.
5778 (move_sese_in_condition): Assert we don't get called with
5779 post-dominators computed.
5780 * sese.h (if_region_set_false_region): Remove.
5782 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
5784 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
5785 mode attribute for TARGET_AVX512VL.
5787 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
5789 * config/i386/i386.opt (mprefer-avx256): New option.
5790 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
5792 (ix86_preferred_simd_mode): Return 256-bit AVX modes
5793 for TARGET_PREFER_AVX256.
5794 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
5796 2017-09-21 Jeff Law <law@redhat.com>
5798 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
5799 Fix dump output if the only stack space is for pushed registers.
5801 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5803 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
5806 2017-09-21 Martin Sebor <msebor@redhat.com>
5809 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
5810 code (in C++) or code that triggers warnings.
5812 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
5814 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
5816 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
5818 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
5819 * sched-rgn.c: Ditto.
5820 * sel-sched-ir.c: Ditto.
5822 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
5824 * toplev.h (set_random_seed): Adjust return type.
5825 * toplev.c (init_local_tick): Move eager initialization of random_seed
5826 to get_random_seed. Adjust comment.
5827 (init_random_seed): Inline to get_random_seed, delete.
5828 (get_random_seed): Initialize random_seed lazily.
5829 (set_random_seed): Do not return previous value.
5830 (print_switch_value): Do not call get_random_seed.
5832 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
5834 * cgraph.c (delete_function_version): New, broken out from...
5835 (cgraph_node::delete_function_version): ...here. Rename to
5836 cgraph_node::delete_function_version_by_decl. Update all uses.
5837 (cgraph_node::remove): Call delete_function_version.
5839 2017-09-21 Jakub Jelinek <jakub@redhat.com>
5842 * tree-inline.c (expand_call_inline): Emit clobber stmts for
5843 VAR_DECLs to which addressable non-volatile parameters are mapped
5844 and for id->retvar after the return value assignment. Clear
5845 id->retval and id->retbnd after inlining.
5847 2017-09-21 Richard Biener <rguenther@suse.de>
5849 PR tree-optimization/82276
5850 PR tree-optimization/82244
5851 * tree-vrp.c (build_assert_expr_for): Set
5852 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
5854 (remove_range_assertions): Revert earlier change.
5856 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
5859 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
5861 2017-09-21 Richard Biener <rguenther@suse.de>
5863 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5864 Restore valid IL after code generation errors.
5865 * graphite.c (graphite_transform_loops): Diagnose code
5866 generation issues as MSG_MISSED_OPTIMIZATION and continue
5867 with processing SCOPs.
5869 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5870 Alan Hayward <alan.hayward@arm.com>
5871 David Sherwood <david.sherwood@arm.com>
5873 * calls.c (compute_argument_addresses): Use simplify_gen_binary
5874 rather than choosing between plus_constant and gen_rtx_<CODE>.
5875 * expr.c (emit_push_insn): Likewise.
5876 (expand_expr_real_2): Likewise.
5878 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5879 Alan Hayward <alan.hayward@arm.com>
5880 David Sherwood <david.sherwood@arm.com>
5882 * loop-unroll.c (split_iv): Call copy_rtx on the step.
5884 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5885 Alan Hayward <alan.hayward@arm.com>
5886 David Sherwood <david.sherwood@arm.com>
5888 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
5889 calling tree_to_uhwi.
5891 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5892 Alan Hayward <alan.hayward@arm.com>
5893 David Sherwood <david.sherwood@arm.com>
5895 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
5896 INTEGER_CST rather than a negative test for ADDR_EXPR.
5898 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5899 Alan Hayward <alan.hayward@arm.com>
5900 David Sherwood <david.sherwood@arm.com>
5902 * tree-vrp.c (extract_range_from_binary_expr_1): Check
5903 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
5905 2017-09-21 Richard Biener <rguenther@suse.de>
5907 PR tree-optimization/71351
5908 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
5909 graphite_create_new_loop_guard): Remove, fold remaining parts
5911 (translate_isl_ast_node_for): ... here and simplify.
5913 2017-09-21 Jakub Jelinek <jakub@redhat.com>
5916 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
5917 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
5918 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
5919 alternative always use QI mode, for -Os imov (=R,R) alternative
5920 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
5923 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5924 Jeff Law <law@redhat.com>
5926 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
5927 (allocate_stack_space): New function, partially extracted from
5929 (s390_emit_prologue): Track offset to most recent stack probe.
5930 Code to allocate space moved into allocate_stack_space.
5931 Dump actions when no stack is allocated.
5932 (s390_prologue_plus_offset): New function.
5933 (s390_emit_stack_probe): Likewise.
5935 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
5937 * common.opt (Wa, Wl, Wp, g, gz=): Add
5939 (gno-column-info): Remove.
5940 (gcolumn-info): Drop RejectNegative.
5942 (gno-record-gcc-switches): Remove.
5943 (grecord-gcc-switches): Drop RejectNegative.
5944 (gno-split-dwarf): Remove.
5945 (gsplit-dwarf): Drop RejectNegative.
5946 (gno-strict-dwarf): Remove.
5947 (gstrict-dwarf): Drop RejectNegative.
5948 * config/darwin.opt (gfull, gused): Add RejectNegative.
5949 * dwarf2out.c (gen_producer_string): Drop
5950 gno-record-gcc-switches handler.
5951 * optc-gen.awk: Add g to prefixes with negative forms.
5952 * opts-common.c (remapping_prefix_p): New.
5953 (find_opt): Check it.
5954 (generate_canonical_option): Test g prefix.
5955 (option_map): Add -gno- mapping.
5956 (add_misspelling_candidates): Check remapping_prefix_p.
5958 2017-09-20 Jeff Law <law@redhat.com>
5960 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
5961 thinko in stack clash protection support.
5963 * explow.c (compute_stack_clash_protection_loop_data): Use
5964 CONST_INT_P instead of explicit test. Verify object is a
5965 CONST_INT_P before looking at INTVAL.
5966 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
5967 instead of explicit test.
5969 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
5972 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
5973 address instead of to r1 and r11.
5975 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
5977 * config.gcc: Support "knm".
5978 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
5979 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
5981 * config/i386/i386.c (m_KNM): Define.
5982 (processor_target_table): Add "knm".
5984 (ix86_option_override_internal): Add "knm".
5985 (ix86_issue_rate): Add PROCESSOR_KNM.
5986 (ix86_adjust_cost): Ditto.
5987 (ia32_multipass_dfa_lookahead): Ditto.
5988 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
5989 (fold_builtin_cpu): Add M_INTEL_KNM.
5990 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
5991 (processor_type): Add PROCESSOR_KNM.
5992 * config/i386/x86-tune.def: Add m_KNM.
5993 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
5995 2017-09-20 Richard Biener <rguenther@suse.de>
5997 PR tree-optimization/80213
5998 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
5999 are allowed in empty BBs as well.
6000 (canonicalize_loop_closed_ssa): Also look for other complex
6002 (scop_detection::get_sese): Include the loop-closed PHI block
6004 (scop_detection::merge_sese): Remove code adding extra blocks.
6005 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
6006 (build_scops): Assert the final returned scop is invalid.
6008 2017-09-20 Richard Biener <rguenther@suse.de>
6010 PR tree-optimization/82264
6011 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
6013 (vn_phi_lookup): Likewise.
6014 (vn_phi_insert): Likewise.
6016 2017-09-20 Jakub Jelinek <jakub@redhat.com>
6018 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
6019 that fits into uhwi or shwi, add DW_AT_const_value regardless
6020 of early_dwarf without going through RTL, using add_AT_unsigned
6023 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
6024 (DEBUG_LTO_ABBREV_SECTION): Likewise.
6025 (DEBUG_LTO_MACINFO_SECTION): Likewise.
6026 (DEBUG_MACRO_SECTION): Likewise.
6027 (DEBUG_LTO_MACRO_SECTION): Likewise.
6028 (DEBUG_STR_DWO_SECTION): Likewise.
6029 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
6030 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
6031 (DEBUG_LTO_DWO_LINE_SECTION): Define.
6032 (DEBUG_LTO_LINE_STR_SECTION): Define.
6033 (init_sections_and_labels): Initialize debug_line_str_section
6034 variable. Initialize debug_loc_section for -gdwarf-5 to
6035 DEBUG_LOCLISTS_SECTION. Formatting fixes.
6037 2017-09-20 Richard Biener <rguenther@suse.de>
6039 * graphite-sese-to-poly.c (extract_affine): Properly handle
6040 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
6042 2017-09-20 Richard Biener <rguenther@suse.de>
6044 PR tree-optimization/81373
6045 * graphite-scop-detection.c (build_cross_bb_scalars_def):
6046 Force SESE live-out defs to be handled even if they are
6049 2017-09-19 Jeff Law <law@redhat.com>
6051 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
6052 nothing for stack adjustments with REG_STACK_CHECK.
6053 * sched-deps.c (parse_add_or_inc): Reject insns with
6054 REG_STACK_CHECK from dependency breaking.
6055 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
6056 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
6057 * reg-notes.def (STACK_CHECK): New note.
6059 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
6060 (ix86_expand_prologue): Dump stack clash info as needed.
6061 Call ix86_adjust_stack_and_probe_stack_clash as needed.
6063 * function.c (dump_stack_clash_frame_info): New function.
6064 * function.h (dump_stack_clash_frame_info): Prototype.
6065 (enum stack_clash_probes): New enum.
6067 * config/alpha/alpha.c (alpha_expand_prologue): Also check
6068 flag_stack_clash_protection.
6069 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
6070 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
6071 (arm_frame_pointer_required): Likewise.
6072 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
6073 (ia64_expand_prologue): Likewise.
6074 * config/mips/mips.c (mips_expand_prologue): Likewise.
6075 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
6076 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6077 (sparc_flat_expand_prologue): Likewise.
6078 * config/spu/spu.c (spu_expand_prologue): Likewise.
6080 * explow.c: Include "params.h".
6081 (anti_adjust_stack_and_probe_stack_clash): New function.
6082 (get_stack_check_protect): Likewise.
6083 (compute_stack_clash_protection_loop_data): Likewise.
6084 (emit_stack_clash_protection_loop_start): Likewise.
6085 (emit_stack_clash_protection_loop_end): Likewise.
6086 (allocate_dynamic_stack_space): Use get_stack_check_protect.
6087 Use anti_adjust_stack_and_probe_stack_clash.
6088 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
6089 (emit_stack_clash_protection_loop_start): Likewise.
6090 (emit_stack_clash_protection_loop_end): Likewise.
6091 * rtl.h (get_stack_check_protect): Prototype.
6092 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
6093 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
6094 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
6096 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
6098 * doc/tm.texi: Rebuilt.
6099 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
6100 get_stack_check_protect.
6101 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
6102 * config/arm/arm.c (arm_expand_prologue): Likewise.
6103 (arm_frame_pointer_required): Likewise.
6104 * config/i386/i386.c (ix86_expand_prologue): Likewise.
6105 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
6106 * config/mips/mips.c (mips_expand_prologue): Likewise.
6107 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
6108 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
6109 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6110 (sparc_flat_expand_prologue): Likewise.
6112 * common.opt (-fstack-clash-protection): New option.
6113 * flag-types.h (enum stack_check_type): Note difference between
6114 -fstack-check= and -fstack-clash-protection.
6115 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
6116 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
6117 * toplev.c (process_options): Issue warnings/errors for cases
6118 not handled with -fstack-clash-protection.
6119 * doc/invoke.texi (-fstack-clash-protection): Document new option.
6120 (-fstack-check): Note additional problem with -fstack-check=generic.
6121 Note that -fstack-check is primarily for Ada and refer users
6122 to -fstack-clash-protection for stack-clash-protection.
6123 Document new params for stack clash protection.
6125 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
6127 * config/i386/i386.c (ix86_split_long_move): Do not handle
6128 address used for LEA in a special way.
6130 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
6132 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
6134 2017-09-19 Martin Sebor <msebor@redhat.com>
6137 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
6138 of incompatible types.
6140 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
6142 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6143 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
6144 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6145 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
6147 2017-09-19 Richard Biener <rguenther@suse.de>
6149 PR tree-optimization/82244
6150 * tree-vrp.c (remove_range_assertions): Do not propagate
6151 a constant to abnormals but replace the assert with a copy.
6153 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
6155 PR rtl-optimization/57878
6156 PR rtl-optimization/68988
6157 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
6158 avoidance test involving non_reload_pseudos. Move frequency test
6159 below the general fragmentation avoidance test.
6161 2017-09-19 Richard Biener <rguenther@suse.de>
6163 PR tree-optimization/69728
6164 * graphite-sese-to-poly.c (schedule_error): New global.
6165 (add_loop_schedule): Handle empty domain by failing the
6167 (build_original_schedule): Handle schedule_error.
6169 2017-09-19 Richard Biener <rguenther@suse.de>
6171 * graphite-scop-detection.c (scop_detection::can_represent_loop):
6172 Do not iterate to sibling loops but only to siblings of inner
6175 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
6178 * config/m68k/m68k.md (moveq feeding equality comparison): Check
6179 that the registers are different.
6181 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
6183 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
6184 to processor_model and "amdfam17h" to arch_names_table.
6185 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
6187 2017-09-18 Jakub Jelinek <jakub@redhat.com>
6190 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
6192 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6193 Alan Hayward <alan.hayward@arm.com>
6194 David Sherwood <david.sherwood@arm.com>
6196 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
6198 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
6200 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
6201 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
6202 vect_schedule_slp_instance.
6203 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
6204 Update call to vect_slp_analyze_node_operations. Simplify return
6206 (vect_slp_analyze_bb_1): Update call accordingly.
6207 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
6208 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
6209 (vect_schedule_slp): Update call accordingly.
6211 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6212 Alan Hayward <alan.hayward@arm.com>
6213 David Sherwood <david.sherwood@arm.com>
6215 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
6216 with types that aren't in fact scalar.
6218 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6220 * tree-vect-slp.c (vect_record_max_nunits): New function,
6222 (vect_build_slp_tree_1): ...here.
6223 (vect_build_slp_tree_2): Call it for phis too.
6225 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6227 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
6228 to vect_get_vec_def_for_operand when getting the mask operand.
6230 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6231 Alan Hayward <alan.hayward@arm.com>
6232 David Sherwood <david.sherwood@arm.com>
6234 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
6237 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6238 Alan Hayward <alan.hayward@arm.com>
6239 David Sherwood <david.sherwood@arm.com>
6241 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
6242 calculation for vector booleans.
6244 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6245 Alan Hayward <alan.hayward@arm.com>
6246 David Sherwood <david.sherwood@arm.com>
6248 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
6250 (vect_transform_stmt): ...here.
6251 (vect_analyze_stmt): Use it instead of calling
6252 vectorizable_live_operation directly.
6254 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
6256 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
6257 non-SIMT targets in acc vector loops.
6259 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
6261 * configure.ac: Add arc and check if assembler supports gdwarf2.
6262 * configure: Regenerate.
6264 2017-09-18 Richard Biener <rguenther@suse.de>
6266 PR tree-optimization/82220
6267 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
6268 epilogue niters from the min_profitable_iters compute.
6270 2017-09-18 Jakub Jelinek <jakub@redhat.com>
6273 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
6274 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
6275 (ix86_init_pic_reg): Revert 2017-09-01 changes.
6277 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
6280 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
6281 switching to a new text section.
6283 2017-09-18 Richard Biener <rguenther@suse.de>
6285 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
6287 (build_alias_set): Reject aliases with no access function.
6289 2017-09-18 Richard Biener <rguenther@suse.de>
6291 PR tree-optimization/79622
6292 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
6294 (build_cross_bb_scalars_use): Likewise.
6296 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
6298 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
6300 2017-09-18 Alan Modra <amodra@gmail.com>
6303 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
6304 stack_pointer_rtx for count 0. Update comments. Break up
6305 large rtl expression.
6307 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
6309 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
6310 Increase to 20 bytes.
6311 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
6312 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
6313 or avx version of the stub.
6315 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
6318 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
6319 compute the minimum stack alignment. Also update preferred stack
6320 boundary for leaf functions.
6322 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
6324 PR tree-optimization/82228
6325 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
6328 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
6330 * common/config/nds32/nds32-common.c
6331 (nds32_option_optimization_table): Refine formatting.
6332 (nds32_option_optimization_table): Use -fsched-pressure and
6333 -fomit-frame-pointer for specific optimization level.
6335 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
6337 * config/nds32/nds32.c: Refine formatting and comments.
6338 * config/nds32/nds32.h: Likewise.
6339 * config/nds32/nds32.md: Likewise.
6340 * config/nds32/nds32-cost.c: Likewise.
6341 * config/nds32/nds32-isr.c: Likewise.
6342 * config/nds32/nds32-md-auxiliary.c: Likewise.
6343 * config/nds32/nds32-multiple.md: Likewise.
6344 * config/nds32/nds32-predicates.c: Likewise.
6346 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
6347 Jakub Jelinek <jakub@redhat.com>
6349 Add support for -std=c++2a.
6350 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
6352 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
6354 2017-09-15 Steve Ellcey <sellcey@cavium.com>
6357 * doc/extend.texi (Common Function Attributes): Add
6358 references to ARM, AArch64, and S/390 specific attributes.
6359 (Function Specific Option Pragmas): Add AArch64 and S/390
6360 to list of back ends that support the target pragma.
6362 2017-09-15 Nathan Sidwell <nathan@acm.org>
6364 * doc/standards.texi: Fix C++17 description. Update URLs for
6367 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
6369 * common.opt (Wcast-align=strict): New warning option.
6370 * doc/invoke.texi: Document -Wcast-align=strict.
6372 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
6374 * cgraph.h (cgraph_thunk_info): Add comments.
6375 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
6376 assert for VIRTUAL_* arguments stricter.
6378 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
6380 PR tree-optimization/71026
6381 * match.pd: Move RDIV patterns from fold-const.c
6382 * fold-const.c (distribute_real_division): Removed.
6383 (fold_binary_loc): Remove calls to distribute_real_divison.
6385 2017-09-15 Jakub Jelinek <jakub@redhat.com>
6387 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
6388 c++1z and gnu++1z as deprecated. Change other references to
6389 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
6390 Change -Wc++1z-compat to -Wc++17-compat.
6391 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
6392 * dwarf2out.c (highest_c_language): Handle C++17.
6393 (gen_compile_unit_die): Likewise.
6395 2017-09-15 Jakub Jelinek <jakub@redhat.com>
6397 PR rtl-optimization/82192
6398 * combine.c (make_extraction): Don't look through non-paradoxical
6399 SUBREGs or TRUNCATE if pos + len is or might be bigger than
6402 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6403 Alan Hayward <alan.hayward@arm.com>
6404 David Sherwood <david.sherwood@arm.com>
6406 * target.def (function_arg_offset): New hook.
6407 * targhooks.h (default_function_arg_offset): Declare.
6408 * targhooks.c (default_function_arg_offset): New function.
6409 * function.c (locate_and_pad_parm): Use
6410 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
6411 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
6412 (TARGET_FUNCTION_ARG_OFFSET): ...this.
6413 * doc/tm.texi: Regenerate.
6414 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
6415 * config/spu/spu.c (spu_function_arg_offset): New function.
6416 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
6417 * system.h (FUNCTION_ARG_OFFSET): Poison.
6419 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6420 Alan Hayard <alan.hayward@arm.com>
6421 David Sherwood <david.sherwood@arm.com>
6423 * target.def (truly_noop_truncation): New hook.
6424 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
6425 than TRULY_NOOP_TRUNCATION.
6426 * hooks.h (hook_bool_uint_uint_true): Declare.
6427 * hooks.c (hook_bool_uint_uint_true): New function.
6428 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
6429 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
6430 * doc/tm.texi: Regenerate.
6431 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
6432 rather than TRULY_NOOP_TRUNCATION in comments.
6433 (simplify_comparison): Likewise.
6434 (record_truncated_value): Likewise.
6435 * expmed.c (extract_bit_field_1): Likewise.
6436 (extract_split_bit_field): Likewise.
6437 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
6438 instead of TRULY_NOOP_TRUNCATION.
6439 * function.c (assign_parm_setup_block): Likewise.
6440 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
6441 * rtlhooks.c: Include target.h.
6442 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
6443 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
6444 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
6445 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
6446 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
6447 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
6448 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
6449 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
6450 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
6451 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
6452 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
6453 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
6454 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
6455 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
6456 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
6457 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
6458 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
6459 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
6460 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
6461 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
6462 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
6463 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
6464 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
6465 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
6466 * config/mips/mips.c (mips_truly_noop_truncation): New function.
6467 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
6468 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
6469 rather than TRULY_NOOP_TRUNCATION in comments.
6470 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
6471 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
6472 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
6473 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
6474 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
6475 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
6476 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
6477 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
6478 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
6479 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
6480 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
6481 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
6482 rather than TRULY_NOOP_TRUNCATION in comments.
6483 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
6484 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
6485 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
6486 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
6487 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
6488 TRULY_NOOP_TRUNCATION condition.
6489 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
6490 (TRULY_NOOP_TRUNCATION): Delete.
6491 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
6492 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
6493 * config/spu/spu.c (spu_truly_noop_truncation): New function.
6494 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
6495 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
6496 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
6497 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
6498 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
6499 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
6500 rather than TRULY_NOOP_TRUNCATION in comments.
6501 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
6502 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
6503 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
6504 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
6505 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
6506 * system.h (TRULY_NOOP_TRUNCATION): Poison.
6508 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
6511 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
6512 (*cmp_ior): Likewise.
6513 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
6514 (*ior_scc_scc_cmp): Likewise.
6515 (*and_scc_scc): Likewise.
6516 (*and_scc_scc_cmp): Likewise.
6518 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6519 Alan Hayard <alan.hayward@arm.com>
6520 David Sherwood <david.sherwood@arm.com>
6522 * target.def (can_change_mode_class): New hook.
6523 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
6524 (hard_regno_nregs): Likewise.
6525 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
6526 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
6527 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
6528 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
6529 (LOAD_EXTEND_OP): Update accordingly.
6530 * doc/tm.texi: Regenerate.
6531 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
6532 CANNOT_CHANGE_MODE_CLASS.
6533 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
6534 (REG_CAN_CHANGE_MODE_P): ...this new macro.
6535 * combine.c (simplify_set): Update accordingly.
6536 * emit-rtl.c (validate_subreg): Likewise.
6537 * recog.c (general_operand): Likewise.
6538 * regcprop.c (mode_change_ok): Likewise.
6539 * reload1.c (choose_reload_regs): Likewise.
6540 (inherit_piecemeal_p): Likewise.
6541 * rtlanal.c (simplify_subreg_regno): Likewise.
6542 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
6543 instead of CANNOT_CHANGE_MODE_CLASS.
6544 (reload_cse_simplify_operands): Likewise.
6545 * reload.c (push_reload): Use targetm.can_change_mode_class
6546 instead of CANNOT_CHANGE_MODE_CLASS.
6547 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
6548 REG_CANNOT_CHANGE_MODE_P.
6549 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6550 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
6551 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6552 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6553 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6554 (arm_can_change_mode_class): New function.
6555 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
6556 than CANNOT_CHANGE_MODE_CLASS in comments.
6557 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6558 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
6559 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
6560 (ix86_can_change_mode_class): ...this new function, inverting the
6561 sense of the return value.
6562 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6563 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6564 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6565 (ia64_can_change_mode_class): New function.
6566 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6567 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
6568 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
6569 (m32c_can_change_mode_class): ...this new function, inverting the
6570 sense of the return value.
6571 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6572 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6573 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
6574 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
6575 (mips_can_change_mode_class): ...this new function, inverting the
6576 sense of the return value.
6577 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6578 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6579 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6580 (msp430_can_change_mode_class): New function.
6581 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6582 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
6583 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6584 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6585 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6586 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
6587 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6588 (pa_cannot_change_mode_class): Replace with...
6589 (pa_can_change_mode_class): ...this new function, inverting the
6590 sense of the return value.
6591 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
6592 than CANNOT_CHANGE_MODE_CLASS in comments.
6593 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6594 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
6595 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6596 (pdp11_cannot_change_mode_class): Replace with...
6597 (pdp11_can_change_mode_class): ...this new function, inverting the
6598 sense of the return value.
6599 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6600 * config/powerpcspe/powerpcspe-protos.h
6601 (rs6000_cannot_change_mode_class_ptr): Delete.
6602 * config/powerpcspe/powerpcspe.c
6603 (rs6000_cannot_change_mode_class_ptr): Delete.
6604 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6605 (rs6000_option_override_internal): Assign to
6606 targetm.can_change_mode_class instead of
6607 rs6000_cannot_change_mode_class_ptr.
6608 (rs6000_cannot_change_mode_class): Replace with...
6609 (rs6000_can_change_mode_class): ...this new function, inverting the
6610 sense of the return value.
6611 (rs6000_debug_cannot_change_mode_class): Replace with...
6612 (rs6000_debug_can_change_mode_class): ...this new function.
6613 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6614 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
6615 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6616 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6617 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
6619 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
6620 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6621 (rs6000_option_override_internal): Assign to
6622 targetm.can_change_mode_class instead of
6623 rs6000_cannot_change_mode_class_ptr.
6624 (rs6000_cannot_change_mode_class): Replace with...
6625 (rs6000_can_change_mode_class): ...this new function, inverting the
6626 sense of the return value.
6627 (rs6000_debug_cannot_change_mode_class): Replace with...
6628 (rs6000_debug_can_change_mode_class): ...this new function.
6629 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6630 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
6631 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
6632 (s390_can_change_mode_class): ...this new function, inverting the
6633 sense of the return value.
6634 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6635 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6636 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
6637 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6638 (sh_cannot_change_mode_class): Replace with...
6639 (sh_can_change_mode_class): ...this new function, inverting the
6640 sense of the return value.
6641 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6642 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6643 (sparc_can_change_mode_class): New function.
6644 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6645 * config/spu/spu.c (spu_can_change_mode_class): New function.
6646 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6647 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6648 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6649 (visium_can_change_mode_class): New function.
6650 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
6652 2017-09-15 Richard Biener <rguenther@suse.de>
6654 PR tree-optimization/82217
6655 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
6656 but not undefined case.
6658 2017-09-15 Jakub Jelinek <jakub@redhat.com>
6661 * postreload.c (reload_cse_simplify_operands): Skip
6662 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
6664 2017-09-15 Richard Biener <rguenther@suse.de>
6666 PR tree-optimization/68823
6667 * graphite-scop-detection.c (build_alias_set): If we have a
6668 possible dependence check whether we can handle them by just
6669 looking at the DRs DR_ACCESS_FNs.
6670 (build_scops): If build_alias_set fails, fail the SCOP.
6672 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
6674 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
6675 to support float128 built-in functions that require the ISA 3.0
6677 (BU_FLOAT128_3_HW): Likewise.
6678 (SQRTF128): Add support for the IEEE 128-bit square root and fma
6680 (FMAF128): Likewise.
6682 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
6683 support for built-in functions that need the ISA 3.0 IEEE 128-bit
6684 floating point instructions.
6685 (rs6000_invalid_builtin): Likewise.
6686 (rs6000_builtin_mask_names): Likewise.
6687 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
6688 (RS6000_BTM_FLOAT128_HW): Likewise.
6689 (RS6000_BTM_COMMON): Likewise.
6690 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
6692 * doc/extend.texi (RS/6000 built-in functions): Document the
6693 IEEE 128-bit floating point square root and fused multiply-add
6696 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
6698 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
6699 reg (r2) isn't in the set of registers defined in the prologue.
6701 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6702 Alan Hayward <alan.hayward@arm.com>
6703 David Sherwood <david.sherwood@arm.com>
6705 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
6706 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
6707 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
6708 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
6709 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
6711 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
6712 max_vectorization_factor.
6713 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
6715 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6716 Alan Hayward <alan.hayward@arm.com>
6717 David Sherwood <david.sherwood@arm.com>
6719 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
6720 (vect_worthwhile_without_simd_p): Declare.
6721 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
6722 (vectorizable_reduction): Use it.
6723 * tree-vect-stmts.c (vectorizable_shift): Likewise.
6724 (vectorizable_operation): Likewise.
6726 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6727 Alan Hayward <alan.hayward@arm.com>
6728 David Sherwood <david.sherwood@arm.com>
6730 * tree-vectorizer.h (vect_get_num_copies): New function.
6731 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
6732 * tree-vect-loop.c (vectorizable_reduction): Likewise.
6733 (vectorizable_induction): Likewise.
6734 (vectorizable_live_operation): Likewise.
6735 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
6736 (vectorizable_bswap): Likewise.
6737 (vectorizable_call): Likewise.
6738 (vectorizable_conversion): Likewise.
6739 (vectorizable_assignment): Likewise.
6740 (vectorizable_shift): Likewise.
6741 (vectorizable_operation): Likewise.
6742 (vectorizable_store): Likewise.
6743 (vectorizable_load): Likewise.
6744 (vectorizable_condition): Likewise.
6745 (vectorizable_comparison): Likewise.
6746 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
6748 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6749 Alan Hayward <alan.hayward@arm.com>
6750 David Sherwood <david.sherwood@arm.com>
6752 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
6753 of vect_init_vector.
6755 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6756 Alan Hayward <alan.hayward@arm.com>
6757 David Sherwood <david.sherwood@arm.com>
6759 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
6760 an inline wrapper that provides a location.
6761 (gimple_build_vector): Likewise.
6762 * gimple-fold.c (gimple_build_vector_from_val): New function.
6763 (gimple_build_vector): Likewise.
6764 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
6765 functions to build the initial value. Always return a gimple value.
6766 (get_initial_defs_for_reduction): Likewise. Only compute
6768 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
6769 vect_init_vector on the results from get_initial_def(s)_for_reduction.
6770 (vectorizable_induction): Use gimple_build_vector rather than
6773 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6774 Alan Hayward <alan.hayward@arm.com>
6775 David Sherwood <david.sherwood@arm.com>
6777 * target.h (vec_perm_indices): New typedef.
6778 (auto_vec_perm_indices): Likewise.
6779 * optabs-query.h: Include target.h
6780 (can_vec_perm_p): Take a vec_perm_indices *.
6781 * optabs-query.c (can_vec_perm_p): Likewise.
6782 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
6783 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
6784 * tree-vect-generic.c (lower_vec_perm): Likewise.
6785 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
6786 (vect_grouped_load_supported): Likewise.
6787 (vect_shift_permute_load_chain): Likewise.
6788 (vect_permute_store_chain): Use auto_vec_perm_indices.
6789 (vect_permute_load_chain): Likewise.
6790 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
6791 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
6792 Update uses of can_vec_perm_p.
6793 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
6794 mode with a number of elements. Take a vec_perm_indices *.
6795 (vect_create_epilog_for_reduction): Update accordingly.
6796 Use auto_vec_perm_indices.
6797 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
6798 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
6799 (vect_transform_slp_perm_load): Likewise.
6800 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
6801 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
6802 (vect_gen_perm_mask_checked): Likewise.
6803 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
6804 (vect_gen_perm_mask_checked): Likewise.
6805 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
6806 (vectorizable_store): Likewise.
6807 (vectorizable_load): Likewise.
6808 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
6809 (vectorizable_bswap): Likewise.
6811 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6812 Alan Hayward <alan.hayward@arm.com>
6813 David Sherwood <david.sherwood@arm.com>
6815 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
6816 * tree.c (build_vector): Likewise.
6817 (build_vector_from_ctor): Update accordingly.
6818 (build_vector_from_val): Likewise.
6819 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
6820 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
6821 * tree-vect-generic.c (add_rshift): Likewise.
6822 (expand_vector_divmod): Likewise.
6823 (optimize_vector_constructor): Likewise.
6824 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
6825 (vect_transform_slp_perm_load): Likewise.
6826 (vect_schedule_slp_instance): Likewise.
6827 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
6828 (vectorizable_call): Likewise.
6829 (vect_gen_perm_mask_any): Likewise. Add elements in order.
6830 * expmed.c (make_tree): Likewise.
6831 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
6832 a vector passed to build_vector.
6833 (fold_convert_const): Likewise.
6834 (exact_inverse): Likewise.
6835 (fold_ternary_loc): Likewise.
6836 (fold_relational_const): Likewise.
6837 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
6838 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
6839 (const_unop): Likewise. Store the reduction accumulator in a
6840 variable rather than an array.
6841 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
6842 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
6843 the new vector, rather than constructing it after the input arrays.
6844 (native_interpret_vector): Use auto_vec<tree> when building
6845 a vector passed to build_vector. Add elements in order.
6846 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
6847 auto_vec<tree> when building a vector passed to build_vector.
6848 (vect_create_epilog_for_reduction): Likewise.
6849 (vectorizable_induction): Likewise.
6850 (get_initial_def_for_reduction): Likewise. Fix indentation of
6852 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
6854 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
6855 passed to build_vector.
6857 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6858 Alan Hayward <alan.hayward@arm.com>
6859 David Sherwood <david.sherwood@arm.com>
6861 * tree-core.h (tree_base::u): Add an "nelts" field.
6862 (tree_vector): Use VECTOR_CST_NELTS as the length.
6863 * tree.c (tree_size): Likewise.
6864 (make_vector): Initialize VECTOR_CST_NELTS.
6865 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
6866 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
6867 TYPE_VECTOR_SUBPARTS.
6868 * expr.c (const_vector_mask_from_tree): Consistently use "units"
6869 as the number of units, setting it from VECTOR_CST_NELTS.
6870 (const_vector_from_tree): Likewise.
6871 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
6872 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
6873 (fold_negate_expr_1): Likewise.
6874 (fold_convert_const): Likewise.
6875 (const_binop): Likewise. Differentiate the number of output and
6877 (const_unop): Likewise.
6878 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
6879 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
6880 in cases that did the opposite.
6882 2017-09-14 Richard Biener <rguenther@suse.de>
6884 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
6887 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
6889 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
6891 2017-09-14 Jakub Jelinek <jakub@redhat.com>
6894 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
6895 if and where to split a bb, except for splitting before debug insn
6896 sequences followed by non-label real insn. Delete debug insns
6897 in between basic blocks.
6899 * combine.c (make_compound_operation_int): Formatting fixes.
6901 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
6902 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6903 * config/netbsd.h (LINK_EH_SPEC): Likewise.
6904 * config/sol2.h (LINK_EH_SPEC): Likewise.
6905 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6906 * config/s390/linux.h (LINK_SPEC): Likewise.
6907 * config/freebsd.h (LINK_EH_SPEC): Likewise.
6908 * config/openbsd.h (LINK_EH_SPEC): Likewise.
6909 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6910 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
6911 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
6912 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6913 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
6914 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
6916 2017-09-13 Jakub Jelinek <jakub@redhat.com>
6918 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
6920 (ENDFILE_LINUX_SPEC): Likewise.
6921 (LINK_EH_SPEC): Likewise.
6922 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
6923 (LINK_OS_LINUX_SPEC32): Likewise.
6924 (LINK_OS_LINUX_SPEC64): Likewise.
6925 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
6926 (LINK_OS_LINUX_SPEC): Likewise.
6928 2017-09-13 Martin Liska <mliska@suse.cz>
6931 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
6932 CASE_HIGH is NULL_TREE.
6934 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
6935 Alan Hayward <alan.hayward@arm.com>
6936 David Sherwood <david.sherwood@arm.com>
6938 * target.def (secondary_memory_needed): New hook.
6939 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
6940 instead of SECONDARY_MEMORY_NEEDED.
6941 (secondary_memory_needed_mode): Likewise.
6942 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
6943 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
6944 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
6945 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
6946 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
6947 * doc/tm.texi: Regenerate.
6948 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
6949 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
6950 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6951 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
6952 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
6953 * config/i386/i386.c (inline_secondary_memory_needed): Put the
6954 mode argument first and change the reg_class arguments to reg_class_t.
6955 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
6956 Make static. Update the call to inline_secondary_memory_needed.
6957 (ix86_register_move_cost): Update the call to
6958 inline_secondary_memory_needed.
6959 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6960 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
6962 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
6963 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
6965 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
6966 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
6967 * config/mips/mips.c (mips_secondary_memory_needed): Make static
6968 and match hook interface. Add comment from mips.h.
6969 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6970 * config/mmix/mmix.md (truncdfsf2): Refer to
6971 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
6973 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
6974 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
6975 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6976 (pa_secondary_memory_needed): New function.
6977 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
6978 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
6979 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6980 (pdp11_secondary_memory_needed): Make static and match hook interface.
6981 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
6982 * config/powerpcspe/powerpcspe-protos.h
6983 (rs6000_secondary_memory_needed_ptr): Delete.
6984 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
6986 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6987 (rs6000_option_override_internal): Assign to
6988 targetm.secondary_memory_needed rather than
6989 rs6000_secondary_memory_needed_ptr.
6990 (rs6000_secondary_memory_needed): Match hook interface.
6991 (rs6000_debug_secondary_memory_needed): Likewise.
6992 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
6993 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
6994 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
6995 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6996 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
6997 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
6999 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
7000 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7001 (rs6000_option_override_internal): Assign to
7002 targetm.secondary_memory_needed rather than
7003 rs6000_secondary_memory_needed_ptr.
7004 (rs6000_secondary_memory_needed): Match hook interface.
7005 (rs6000_debug_secondary_memory_needed): Likewise.
7006 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
7007 * config/s390/s390.c (s390_secondary_memory_needed): New function.
7008 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7009 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
7010 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7011 (sparc_secondary_memory_needed): New function.
7012 * lra-constraints.c (check_and_process_move): Refer to
7013 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7015 (curr_insn_transform): Likewise.
7016 (process_alt_operands): Use targetm.secondary_memory_needed
7017 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7018 (check_secondary_memory_needed_p): Likewise.
7019 (choose_split_class): Likewise.
7020 * reload.c: Unconditionally include code that was previously
7021 conditional on SECONDARY_MEMORY_NEEDED.
7022 (push_secondary_reload): Use targetm.secondary_memory_needed
7023 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7024 (push_reload): Likewise.
7025 * reload1.c: Unconditionally include code that was previously
7026 conditional on SECONDARY_MEMORY_NEEDED.
7027 (choose_reload_regs): Use targetm.secondary_memory_needed
7028 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7029 (gen_reload): Likewise.
7030 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
7032 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7033 Alan Hayward <alan.hayward@arm.com>
7034 David Sherwood <david.sherwood@arm.com>
7036 * target.def (secondary_memory_needed_mode): New hook:
7037 * targhooks.c (default_secondary_memory_needed_mode): Declare.
7038 * targhooks.h (default_secondary_memory_needed_mode): New function.
7039 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
7040 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
7041 * doc/tm.texi: Regenerate.
7042 * lra-constraints.c (check_and_process_move): Use
7043 targetm.secondary_memory_needed_mode instead of
7044 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
7045 (curr_insn_transform): Likewise.
7046 * reload.c (get_secondary_mem): Likewise.
7047 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7048 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
7050 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7051 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7052 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
7053 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7054 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
7056 * config/powerpcspe/powerpcspe-protos.h
7057 (rs6000_secondary_memory_needed_mode): Delete.
7058 * config/powerpcspe/powerpcspe.c
7059 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7060 (rs6000_secondary_memory_needed_mode): Make static.
7061 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7062 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
7064 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7066 (rs6000_secondary_memory_needed_mode): Make static.
7067 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7068 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
7069 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7070 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7071 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7073 (sparc_secondary_memory_needed_mode): New function.
7074 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
7076 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
7078 * config/aarch64/constraints.md (Umq): New constraint.
7079 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
7081 (mov<mode>): Update condition.
7083 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7085 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
7086 when bitposition is the same.
7088 2017-09-13 Richard Biener <rguenther@suse.de>
7090 * dwarf2out.c (output_die_symbol): Remove.
7091 (output_die): Do not output a DIEs symbol.
7093 2017-09-13 Richard Biener <rguenther@suse.de>
7096 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
7097 default-def to avoid breaking iterator update with the weird
7098 interaction with cgraph_update_edges_for_call_stmt_node.
7100 2017-09-13 Richard Biener <rguenther@suse.de>
7102 * tree-cfg.c (verify_gimple_assign_binary): Add verification
7103 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
7104 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
7105 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
7107 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
7109 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
7110 Disable pc relative literal load irrespective of
7111 TARGET_FIX_ERR_A53_84341 for default.
7113 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
7115 * config/sparc/sparc.c (output_return): Output the source location of
7116 the insn in the delay slot, if any.
7117 (output_sibcall): Likewise.
7119 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
7122 * common.opt (-static-pie): New alias.
7123 (shared): Negate static-pie.
7124 (-no-pie): Update help text.
7126 (static-pie): New option.
7127 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
7128 -static-pie support.
7129 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
7130 (LINK_EH_SPEC): Likewise.
7131 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7132 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7133 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7134 * gcc.c (LINK_COMMAND_SPEC): Likewise.
7135 (init_gcc_specs): Likewise.
7136 (init_spec): Likewise.
7137 (display_help): Update help message for -pie.
7138 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
7141 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
7143 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
7144 (movdi_aarch64): Likewise.
7145 (movti_aarch64): Likewise.
7147 2017-09-12 Simon Wright <simon@pushface.org>
7150 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
7151 calculation of the minor version, always output as 0.
7153 2017-09-12 Jakub Jelinek <jakub@redhat.com>
7156 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
7157 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
7158 on it early, rather than manual conversion late. For
7159 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
7160 instead of performing manual conversion.
7162 2017-09-12 Carl Love <cel@us.ibm.com>
7164 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
7165 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
7167 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7168 VMULOSW): Add definitions.
7169 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7170 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7171 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7172 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7173 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7175 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7177 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
7179 (movti_aarch64): Likewise.
7180 (movdf_aarch64): Likewise.
7181 (movtf_aarch64): Likewise.
7182 (load_pairdi): Likewise.
7183 (store_pairdi): Likewise.
7184 (load_pairdf): Likewise.
7185 (store_pairdf): Likewise.
7186 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
7187 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
7188 (ldr_got_small_<mode>): Likewise.
7189 (ldr_got_small_28k_<mode>): Likewise.
7190 (ldr_got_tiny): Likewise.
7191 * config/aarch64/iterators.md (ldst_sz): New.
7192 (ldpstp_sz): Likewise.
7193 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
7195 (thunderx_load): Split load_8 to load_16.
7196 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
7198 (thunderx2t99_storepair_basic): Split store_8 to store_16.
7199 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
7200 (xgene1_store_pair): Split store_8 to store_16.
7201 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
7202 (falkor_st_0_st_sd): Split store_8 to store_16.
7204 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7206 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
7207 and store1/2/3/4 to store_4/8/12/16.
7208 * config/aarch64/aarch64.md: Update for rename.
7209 * config/arm/arm.md: Likewise.: Likewise.
7210 * config/arm/arm.c: Likewise.
7211 * config/arm/thumb1.md: Likewise.
7212 * config/arm/thumb2.md: Likewise.
7213 * config/arm/vfp.md: Likewise.
7214 * config/arm/arm-generic.md: Likewise.
7215 * config/arm/arm1020e.md: Likewise.
7216 * config/arm/arm1026ejs.md: Likewise.
7217 * config/arm/arm1136jfs.md: Likewise.
7218 * config/arm/arm926ejs.md: Likewise.
7219 * config/arm/cortex-a15.md: Likewise.
7220 * config/arm/cortex-a17.md: Likewise.
7221 * config/arm/cortex-a5.md: Likewise.
7222 * config/arm/cortex-a53.md: Likewise.
7223 * config/arm/cortex-a57.md: Likewise.
7224 * config/arm/cortex-a7.md: Likewise.
7225 * config/arm/cortex-a8.md: Likewise.
7226 * config/arm/cortex-a9.md: Likewise.
7227 * config/arm/cortex-m4.md: Likewise.
7228 * config/arm/cortex-m7.md: Likewise.
7229 * config/arm/cortex-r4.md: Likewise.
7230 * config/arm/exynos-m1.md: Likewise.
7231 * config/arm/fa526.md: Likewise.
7232 * config/arm/fa606te.md: Likewise.
7233 * config/arm/fa626te.md: Likewise.
7234 * config/arm/fa726te.md: Likewise.
7235 * config/arm/fmp626.md: Likewise.
7236 * config/arm/iwmmxt.md: Likewise.
7237 * config/arm/ldmstm.md: Likewise.
7238 * config/arm/marvell-pj4.md: Likewise.
7239 * config/arm/xgene1.md: Likewise.
7240 * config/aarch64/thunderx.md: Likewise.
7241 * config/aarch64/thunderx2t99.md: Likewise.
7242 * config/aarch64/falkor.md: Likewise.
7244 2017-09-12 Martin Liska <mliska@suse.cz>
7246 * attribs.c (private_lookup_attribute): New function.
7247 * attribs.h (private_lookup_attribute): Declared here.
7248 (lookup_attribute): Called from this place.
7250 2017-09-12 Richard Biener <rguenther@suse.de>
7252 PR tree-optimization/82157
7253 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
7254 stmts with side-effects.
7256 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7257 Alan Hayward <alan.hayward@arm.com>
7258 David Sherwood <david.sherwood@arm.com>
7260 * target.def (hard_regno_nregs): New hook.
7261 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
7262 * targhooks.h (default_hard_regno_nregs): Declare.
7263 * targhooks.c (default_hard_regno_nregs): New function.
7264 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
7265 (TARGET_HARD_REGNO_NREGS): ...this hook.
7266 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
7267 (CLASS_MAX_NREGS): Likewise.
7268 * doc/tm.texi: Regenerate.
7269 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
7270 instead of HARD_REGNO_NREGS.
7271 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
7272 HARD_REGNO_NREGS in the comment.
7273 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
7274 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
7275 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
7276 Return an unsigned int.
7277 (TARGET_HARD_REGNO_NREGS): Redefine.
7278 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
7279 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
7280 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
7281 (arc_hard_regno_nregs): New function.
7282 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
7283 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
7284 (arm_hard_regno_nregs): New function.
7285 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
7286 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
7287 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
7288 (TARGET_HARD_REGNO_NREGS): Redefine.
7289 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
7290 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
7291 (HARD_REGNO_NREGS): Delete.
7292 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
7293 (cr16_hard_regno_nregs): New function.
7294 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
7295 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
7296 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
7297 (cris_hard_regno_nregs): New function.
7298 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
7299 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
7300 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
7301 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
7302 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
7303 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
7304 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
7305 (frv_hard_regno_nregs): Make static. Take and return an
7307 (frv_class_max_nregs): Remove outdated copy of documentation.
7308 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
7309 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
7310 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
7311 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
7312 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
7313 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
7314 (TARGET_HARD_REGNO_NREGS): Redefine.
7315 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
7316 (CLASS_MAX_NREGS): Update comment.
7317 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
7318 (ia64_hard_regno_nregs): New function.
7319 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
7320 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
7321 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
7322 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
7323 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
7325 (m32c_hard_regno_nregs): Likewise. Make static.
7326 (TARGET_HARD_REGNO_NREGS): Redefine.
7327 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
7328 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
7329 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
7330 (m68k_hard_regno_nregs): New function.
7331 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
7332 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
7333 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
7334 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
7335 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
7336 Take and return an unsigned int.
7337 (TARGET_HARD_REGNO_NREGS): Redefine.
7338 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
7339 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
7340 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
7341 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
7342 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
7343 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
7344 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
7345 (msp430_hard_regno_nregs): Make static. Take and return an
7347 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
7348 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
7349 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
7350 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
7351 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
7352 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
7353 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
7354 (TARGET_HARD_REGNO_NREGS): Redefine.
7355 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
7356 (PA_HARD_REGNO_NREGS): ...this.
7357 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
7358 (PA_HARD_REGNO_NREGS): ...this.
7359 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
7360 (pa_hard_regno_nregs): New function.
7361 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
7362 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
7363 (pdp11_hard_regno_nregs): New function.
7364 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
7365 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
7366 (rs6000_hard_regno_nregs_hook): New function.
7367 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
7368 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
7369 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
7370 Take and return an unsigned int. Move earlier in file.
7371 (TARGET_HARD_REGNO_NREGS): Redefine.
7372 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
7373 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
7374 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
7375 (rl78_hard_regno_nregs): Make static. Take and return an
7377 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
7378 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
7379 (rs6000_hard_regno_nregs_hook): New function.
7380 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
7381 * config/rx/rx.c (rx_hard_regno_nregs): New function.
7382 (TARGET_HARD_REGNO_NREGS): Redefine.
7383 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
7384 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
7385 instead of HARD_REGNO_NREGS.
7386 (s390_hard_regno_nregs): New function.
7387 (s390_hard_regno_mode_ok): Add comment from s390.h.
7388 (TARGET_HARD_REGNO_NREGS): Redefine.
7389 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
7390 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
7391 (sh_hard_regno_nregs): New function.
7392 (sh_pass_in_reg_p): Use it.
7393 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
7394 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
7395 (sparc_hard_regno_nregs): New function.
7396 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
7397 * config/spu/spu.c (spu_hard_regno_nregs): New function.
7398 (spu_function_arg_advance): Use it, supplying a valid register number.
7399 (TARGET_HARD_REGNO_NREGS): Redefine.
7400 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
7401 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
7402 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
7403 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
7404 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
7405 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
7406 (CLASS_MAX_NREGS): Remove copy of old documentation.
7407 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
7408 (visium_hard_regno_nregs): New function.
7409 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
7410 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
7411 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
7412 xtensa_hard_regno_nregs): New function.
7413 * system.h (HARD_REGNO_NREGS): Poison.
7415 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7417 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
7418 hard_regno_nregs instead of HARD_REGNO_NREGS.
7419 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
7420 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
7421 (c6x_expand_epilogue): Likewise.
7422 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
7423 (frv_read_iacc_argument): Likewise.
7424 * config/sh/sh.c: Include regs.h.
7425 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
7426 (regs_used): Likewise.
7427 (output_stack_adjust): Likewise.
7428 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
7429 * expmed.c: Include regs.h.
7430 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
7431 * ree.c: Include regs.h.
7432 (combine_reaching_defs): Use hard_regno_nregs instead of
7434 (add_removable_extension): Likewise.
7436 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7438 * regs.h (hard_regno_nregs): Turn into a function.
7439 (end_hard_regno): Update accordingly.
7440 * caller-save.c (setup_save_areas): Likewise.
7441 (save_call_clobbered_regs): Likewise.
7442 (replace_reg_with_saved_mem): Likewise.
7443 (insert_restore): Likewise.
7444 (insert_save): Likewise.
7445 * combine.c (can_change_dest_mode): Likewise.
7446 (move_deaths): Likewise.
7447 (distribute_notes): Likewise.
7448 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
7449 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
7450 (rs6000_split_multireg_move): Likewise.
7451 (rs6000_register_move_cost): Likewise.
7452 (rs6000_memory_move_cost): Likewise.
7453 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
7454 (rs6000_split_multireg_move): Likewise.
7455 (rs6000_register_move_cost): Likewise.
7456 (rs6000_memory_move_cost): Likewise.
7457 * cselib.c (cselib_reset_table): Likewise.
7458 (cselib_lookup_1): Likewise.
7459 * emit-rtl.c (set_mode_and_regno): Likewise.
7460 * function.c (aggregate_value_p): Likewise.
7461 * ira-color.c (setup_profitable_hard_regs): Likewise.
7462 (check_hard_reg_p): Likewise.
7463 (calculate_saved_nregs): Likewise.
7464 (assign_hard_reg): Likewise.
7465 (improve_allocation): Likewise.
7466 (calculate_spill_cost): Likewise.
7467 * ira-emit.c (modify_move_list): Likewise.
7468 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
7469 (ira_hard_reg_in_set_p): Likewise.
7470 * ira.c (setup_reg_mode_hard_regset): Likewise.
7471 (clarify_prohibited_class_mode_regs): Likewise.
7472 (check_allocation): Likewise.
7473 * lra-assigns.c (find_hard_regno_for_1): Likewise.
7474 (lra_setup_reg_renumber): Likewise.
7475 (setup_try_hard_regno_pseudos): Likewise.
7476 (spill_for): Likewise.
7477 (assign_hard_regno): Likewise.
7478 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
7479 * lra-constraints.c (in_class_p): Likewise.
7480 (lra_constraint_offset): Likewise.
7481 (simplify_operand_subreg): Likewise.
7482 (lra_constraints): Likewise.
7483 (split_reg): Likewise.
7484 (split_if_necessary): Likewise.
7485 (invariant_p): Likewise.
7486 (inherit_in_ebb): Likewise.
7487 * lra-lives.c (process_bb_lives): Likewise.
7488 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
7489 (get_hard_regs): Likewise.
7490 (do_remat): Likewise.
7491 * lra-spills.c (assign_spill_hard_regs): Likewise.
7492 * mode-switching.c (create_pre_exit): Likewise.
7493 * postreload.c (reload_combine_recognize_pattern): Likewise.
7494 * recog.c (peep2_find_free_register): Likewise.
7495 * regcprop.c (kill_value_regno): Likewise.
7496 (set_value_regno): Likewise.
7497 (copy_value): Likewise.
7498 (maybe_mode_change): Likewise.
7499 (find_oldest_value_reg): Likewise.
7500 (copyprop_hardreg_forward_1): Likewise.
7501 * regrename.c (check_new_reg_p): Likewise.
7502 (regrename_do_replace): Likewise.
7503 * reload.c (push_reload): Likewise.
7504 (combine_reloads): Likewise.
7505 (find_dummy_reload): Likewise.
7506 (operands_match_p): Likewise.
7507 (find_reloads): Likewise.
7508 (find_equiv_reg): Likewise.
7509 (reload_adjust_reg_for_mode): Likewise.
7510 * reload1.c (count_pseudo): Likewise.
7511 (count_spilled_pseudo): Likewise.
7512 (find_reg): Likewise.
7513 (clear_reload_reg_in_use): Likewise.
7514 (free_for_value_p): Likewise.
7515 (allocate_reload_reg): Likewise.
7516 (choose_reload_regs): Likewise.
7517 (reload_adjust_reg_for_temp): Likewise.
7518 (emit_reload_insns): Likewise.
7519 (delete_output_reload): Likewise.
7520 * rtlanal.c (subreg_get_info): Likewise.
7521 * sched-deps.c (sched_analyze_reg): Likewise.
7522 * sel-sched.c (init_regs_for_mode): Likewise.
7523 (mark_unavailable_hard_regs): Likewise.
7524 (choose_best_reg_1): Likewise.
7525 (verify_target_availability): Likewise.
7526 * valtrack.c (dead_debug_insert_temp): Likewise.
7527 * var-tracking.c (track_loc_p): Likewise.
7528 (emit_note_insn_var_location): Likewise.
7529 * varasm.c (make_decl_rtl): Likewise.
7530 * reginfo.c (choose_hard_reg_mode): Likewise.
7531 (init_reg_modes_target): Refer directly to
7532 this_target_regs->x_hard_regno_nregs.
7534 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7536 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
7537 instead of hard_regno_nregs.
7539 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7541 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
7542 end_hard_regno instead of hard_regno_nregs.
7543 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
7544 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
7545 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
7546 * ira-color.c (improve_allocation): Likewise.
7547 * lra-assigns.c (find_hard_regno_for_1): Likewise.
7548 * lra-lives.c (mark_regno_live): Likewise.
7549 (mark_regno_dead): Likewise.
7550 * lra-remat.c (operand_to_remat): Likewise.
7551 * lra.c (collect_non_operand_hard_regs): Likewise.
7552 * postreload.c (reload_combine_note_store): Likewise.
7553 (move2add_valid_value_p): Likewise.
7554 * reload.c (regno_clobbered_p): Likewise.
7556 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7558 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
7560 * config/v850/v850.c (v850_reorg): Likewise.
7561 * reload.c (refers_to_regno_for_reload_p): Likewise.
7562 (find_equiv_reg): Likewise.
7563 * reload1.c (reload_reg_reaches_end_p): Likewise.
7565 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7567 * caller-save.c (add_used_regs): Use REG_NREGS instead of
7569 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
7570 * config/arm/arm.c (output_move_neon): Likewise.
7571 (arm_attr_length_move_neon): Likewise.
7572 (neon_split_vcombine): Likewise.
7573 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
7574 (c6x_mark_reg_written): Likewise.
7575 (c6x_dwarf_register_span): Likewise.
7576 * config/i386/i386.c (ix86_save_reg): Likewise.
7577 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
7578 (rws_access_reg): Likewise.
7579 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
7580 * mode-switching.c (create_pre_exit): Likewise.
7581 * ree.c (combine_reaching_defs): Likewise.
7582 (add_removable_extension): Likewise.
7583 * regcprop.c (find_oldest_value_reg): Likewise.
7584 (copyprop_hardreg_forward_1): Likewise.
7585 * reload.c (reload_inner_reg_of_subreg): Likewise.
7586 (push_reload): Likewise.
7587 (combine_reloads): Likewise.
7588 (find_dummy_reload): Likewise.
7589 (reload_adjust_reg_for_mode): Likewise.
7590 * reload1.c (find_reload_regs): Likewise.
7591 (forget_old_reloads_1): Likewise.
7592 (reload_reg_free_for_value_p): Likewise.
7593 (reload_adjust_reg_for_temp): Likewise.
7594 (emit_reload_insns): Likewise.
7595 (delete_output_reload): Likewise.
7596 * sel-sched.c (choose_best_reg_1): Likewise.
7597 (choose_best_pseudo_reg): Likewise.
7599 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7600 Alan Hayward <alan.hayward@arm.com>
7601 David Sherwood <david.sherwood@arm.com>
7603 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
7604 * target.def (slow_unaligned_access): New hook.
7605 * targhooks.h (default_slow_unaligned_access): Declare.
7606 * targhooks.c (default_slow_unaligned_access): New function.
7607 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
7608 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
7609 * doc/tm.texi: Regenerate.
7610 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
7611 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
7612 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
7614 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
7615 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
7617 (rs6000_slow_unaligned_access): New function.
7618 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
7619 (expand_block_compare): Likewise.
7620 (expand_strn_compare): Likewise.
7621 (rs6000_rtx_costs): Likewise.
7622 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
7623 (riscv_slow_unaligned_access): Likewise.
7624 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
7625 (riscv_slow_unaligned_access_p): ...this and make static.
7626 (riscv_option_override): Update accordingly.
7627 (riscv_slow_unaligned_access): New function.
7628 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
7629 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
7630 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
7631 (rs6000_slow_unaligned_access): New function.
7632 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
7633 (rs6000_rtx_costs): Likewise.
7634 * config/rs6000/rs6000-string.c (expand_block_compare)
7635 (expand_strn_compare): Use targetm.slow_unaligned_access instead
7636 of SLOW_UNALIGNED_ACCESS.
7637 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
7638 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
7639 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
7640 of SLOW_UNALIGNED_ACCESS.
7641 * expmed.c (simple_mem_bitfield_p): Likewise.
7642 * expr.c (alignment_for_piecewise_move): Likewise.
7643 (emit_group_load_1): Likewise.
7644 (emit_group_store): Likewise.
7645 (copy_blkmode_from_reg): Likewise.
7646 (emit_push_insn): Likewise.
7647 (expand_assignment): Likewise.
7648 (store_field): Likewise.
7649 (expand_expr_real_1): Likewise.
7650 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
7651 * lra-constraints.c (simplify_operand_subreg): Likewise.
7652 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
7653 * gimple-ssa-store-merging.c: Likewise in block comment at start
7655 * tree-ssa-strlen.c: Include target.h.
7656 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
7657 of SLOW_UNALIGNED_ACCESS.
7658 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
7660 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7662 PR rtl-optimization/82185
7663 * expmed.c (emit_store_flag_int): Only test tem if it has been
7666 2017-09-12 Richard Biener <rguenther@suse.de>
7669 * match.pd ((FTYPE) N CMP CST): Fix typo.
7671 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
7673 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
7675 (mips_near_type_p): Add 'short_call' attribute as a synonym
7677 * doc/extend.texi (short_call): Document new function attribute.
7679 2017-09-12 Jakub Jelinek <jakub@redhat.com>
7682 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
7683 assertion check that in the condition.
7684 (get_atomic_generic_size): Likewise. Before testing if parameter
7685 has pointer type, if it has array type, call for C++
7686 default_conversion to perform array-to-pointer conversion.
7688 2017-09-12 Richard Biener <rguenther@suse.de>
7690 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
7691 for operations we cannot scalarize.
7693 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
7695 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
7696 vectors heap vectors. Clean up comments.
7697 Make visited_bbs a reference.
7698 (profitable_jump_thread_path): Make GC
7699 vectors heap vectors. Clean up comments.
7701 (convert_and_register_jump_thread_path): Make GC vectors heap
7703 (check_subpath_and_update_thread_path): Same. Clean up comments.
7704 Make visited_bbs a reference.
7705 (handle_phi): Abstract common code to to
7706 register_jump_thread_path_if_profitable.
7707 Rename VAR_BB to DEF_BB.
7709 Make GC vectors heap vectors.
7710 Make visited_bbs a reference.
7711 (handle_assignment): Same.
7712 (register_jump_thread_path_if_profitable): New.
7713 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
7715 Make GC vectors heap vectors. Clean up comments.
7716 Make visited_bbs a reference.
7717 (find_jump_threads_backwards): Make visited_bbs live in the stack.
7718 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
7721 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
7724 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
7725 words of E_DImode object are reachable by xtensa_uimm8x4 access.
7727 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
7729 Revert r251800 and r251799.
7731 2017-09-11 Martin Jambor <mjambor@suse.cz>
7734 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
7735 arguments in advance.
7736 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
7737 use it to find predecessor edges.
7738 (naive_outof_ssa): Collect vector of predecessors.
7740 2017-09-08 Jason Merrill <jason@redhat.com>
7742 PR c++/70029 - ICE with ref-qualifier and -flto
7743 * langhooks.h (struct lang_hooks_for_types): Add
7744 copy_lang_qualifiers.
7745 * attribs.c (build_type_attribute_qual_variant): Use it.
7746 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
7748 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
7749 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
7751 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
7754 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
7755 (*mulsi3_sp64): New instruction.
7756 (mulsi3): New expander.
7758 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
7760 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
7762 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7764 * sancov.c: Include memmodel.h.
7766 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
7769 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
7772 2017-09-07 Carl Love <cel@us.ibm.com>
7774 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
7775 the sldi instruction.
7777 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
7779 * sancov.c: Include tm_p.h.
7781 2017-09-07 Jakub Jelinek <jakub@redhat.com>
7784 * output.h (switch_to_other_text_partition): New declaration.
7785 * varasm.c (switch_to_other_text_partition): New function.
7786 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
7787 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
7788 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
7789 to the other text partition before emitting LCL label and switch back
7790 after emitting the word after it.
7792 2017-09-07 Richard Biener <rguenther@suse.de>
7794 * passes.def (pass_split_crit_edges): Remove instance before PRE.
7795 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
7796 critical edges here, after loop init.
7797 (pass_data_pre): Remove PROP_no_crit_edges flags.
7798 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
7799 for valueization of call args to avoid leaking VN_TOP.
7800 (visit_use): Assert we do not visit default defs.
7801 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
7802 Use error_mark_node to more easily detect leaking VN_TOP.
7803 All default-defs are varying, not VN_TOP. Mark them visited.
7804 (run_scc_vn): Make code match comment.
7806 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
7808 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
7809 OPTION_MASK_FLOAT128_KEYWORD.
7810 (POWERPC_MASKS): Likewise.
7811 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
7812 support for the -mfloat128-type option, and make -mfloat128
7813 default on PowerPC Linux systems. Define or undefine
7814 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
7815 Define __float128 to be __ieee128 if IEEE 128-bit support is
7816 enabled, or undefine it.
7817 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
7818 Delete defining __FLOAT128_TYPE__.
7819 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
7820 -mfloat128-type option and make -mfloat128 default on PowerPC
7822 (TARGET_FLOAT128_TYPE): Likewise.
7823 (-mfloat128-type): Likewise.
7824 * config/rs6000/rs6000.c (rs6000_option_override_internal):
7825 Delete the -mfloat128-type option and make -mfloat128 default on
7826 PowerPC Linux systems. Always use __ieee128 to be the keyword for
7827 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
7828 128-bit floating point is enabled. Change tests from using
7829 -mfloat128-type to -mfloat128.
7830 (rs6000_mangle_type): Use the correct mangling for the __float128
7831 type even if normal long double is restricted to 64-bits.
7832 (floatn_mode): Enable the _Float128 type by default on VSX Linux
7834 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
7835 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
7836 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
7838 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
7839 documentation for -mfloat128.
7841 2017-09-06 Olivier Hainque <hainque@adacore.com>
7843 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
7845 2017-09-06 Wish Wu <wishwu007@gmail.com>
7846 Jakub Jelinek <jakub@redhat.com>
7848 * asan.c (initialize_sanitizer_builtins): Add
7849 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
7850 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
7851 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
7852 BT_FN_VOID_UINT64_PTR variables.
7853 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
7854 (BT_FN_VOID_UINT16_UINT16): Likewise.
7855 (BT_FN_VOID_UINT32_UINT32): Likewise.
7856 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
7857 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
7858 (BT_FN_VOID_UINT64_PTR): Likewise.
7859 * common.opt (flag_sanitize_coverage): New variable.
7860 (fsanitize-coverage=trace-pc): Remove.
7861 (fsanitize-coverage=): Add.
7862 * flag-types.h (enum sanitize_coverage_code): New enum.
7863 * fold-const.c (fold_range_test): Disable non-short-circuit
7864 optimization if flag_sanitize_coverage.
7865 (fold_truth_andor): Likewise.
7866 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
7867 * opts.c (COVERAGE_SANITIZER_OPT): Define.
7868 (coverage_sanitizer_opts): New array.
7869 (get_closest_sanitizer_option): Add OPTS argument, handle also
7870 OPT_fsanitize_coverage_.
7871 (parse_sanitizer_options): Adjusted to also handle
7872 OPT_fsanitize_coverage_.
7873 (common_handle_option): Add OPT_fsanitize_coverage_.
7874 * sancov.c (instrument_comparison, instrument_switch): New function.
7875 (sancov_pass): Add trace-cmp support.
7876 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
7877 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
7878 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
7879 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
7880 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
7881 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
7882 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
7883 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
7884 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
7885 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
7887 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
7889 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
7890 error. Only quit immediately if parsing is complete.
7891 (BEGIN): Initialize fatal_err and parse_done.
7892 (begin fpu, end fpu): Check number of arguments.
7893 (begin arch, end arch): Likewise.
7894 (begin cpu, end cpu): Likewise.
7895 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
7896 (optalias): Likewise.
7898 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
7900 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
7901 * config/arm/arm-isa.h: Delete. Move definitions to ...
7902 * arm-cpus.in: ... here. Use new feature and fgroup values.
7903 * config/arm/arm.c (arm_option_override): Use lower case for feature
7905 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
7906 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
7907 * config/arm/parsecpu.awk (END): Add new command 'isa'.
7909 (print_isa_bits_for): New function.
7910 (gen_isa): New function.
7911 (gen_comm_data): Use print_isa_bits_for.
7912 (define feature): New keyword.
7913 (define fgroup): New keyword.
7914 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
7915 (arm-isa.h): Add rule to generate file.
7916 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
7917 case for feature bit names.
7919 2017-09-06 Richard Biener <rguenther@suse.de>
7921 * tree-ssa-pre.c (NECESSARY): Remove.
7922 (create_expression_by_pieces): Do not touch pass-local flags.
7923 (insert_into_preds_of_block): Likewise.
7924 (do_pre_regular_insertion): Likewise.
7925 (eliminate_insert): Likewise.
7926 (eliminate_dom_walker::before_dom_children): Likewise.
7927 (fini_eliminate): Do not look at inserted_exprs.
7928 (mark_operand_necessary): Remove.
7929 (remove_dead_inserted_code): Replace with simple work-list
7930 algorithm based on inserted_exprs and SSA uses.
7931 (pass_pre::execute): Re-order fini_eliminate and
7932 remove_dead_inserted_code.
7934 2017-09-06 Olivier Hainque <hainque@adacore.com>
7936 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
7937 for VxWorks 7. Adjust surrounding comments.
7939 2017-09-06 Richard Biener <rguenther@suse.de>
7941 * gimple-ssa-strength-reduction.c
7942 (find_candidates_dom_walker::before_dom_children): Also allow
7945 2017-09-06 Richard Biener <rguenther@suse.de>
7947 PR tree-optimization/82108
7948 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
7949 for gap in the non-permutation SLP case.
7951 2017-09-06 Martin Jambor <mjambor@suse.cz>
7953 PR tree-optimization/82078
7954 * tree-sra.c (sort_and_splice_var_accesses): Move call to
7955 add_access_to_work_queue...
7956 (build_accesses_from_assign): ...here.
7957 (propagate_all_subaccesses): Make sure racc is the group
7958 representative, if there is one.
7960 2017-09-06 Jakub Jelinek <jakub@redhat.com>
7963 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
7966 2017-09-06 Richard Biener <rguenther@suse.de>
7968 * gimple-ssa-strength-reduction.c
7969 (find_candidates_dom_walker::before_doom_children): Use a
7970 type and not a mode check.
7972 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
7975 * config/arm/predicates.md (arm_general_adddi_operand): Create new
7977 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
7979 2017-09-05 Jeff Law <law@redhat.com>
7981 PR tree-optimization/64910
7982 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
7983 cases where we have 3 or more operands.
7985 2017-09-05 Jakub Jelinek <jakub@redhat.com>
7988 * omp-low.c (lower_omp_for): Recompute tree invariant if
7989 gimple_omp_for_initial/final is ADDR_EXPR.
7992 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
7993 into gimple val before gimplification fo the COND_EXPR.
7995 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
7997 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
7998 REGION_COPY argument.
7999 (thread_through_all_blocks): Remove unused argument to
8000 duplicate_thread_path.
8002 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8003 Alan Hayward <alan.hayward@arm.com>
8004 David Sherwood <david.sherwood@arm.com>
8006 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
8007 Take a scalar_mode rather than a machine_mode.
8008 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8009 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
8010 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8011 (aarch64_gen_adjusted_ldpstp): Likewise.
8012 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
8014 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8015 Alan Hayward <alan.hayward@arm.com>
8016 David Sherwood <david.sherwood@arm.com>
8018 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
8019 Take a scalar_int_mode instead of a machine_mode.
8020 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8021 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8022 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8023 (aarch64_simd_attr_length_rglist): Delete.
8024 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
8025 a scalar_int_mode instead of a machine_mode.
8026 (aarch64_add_offset): Likewise.
8027 (aarch64_internal_mov_immediate): Likewise
8028 (aarch64_add_constant_internal): Likewise.
8029 (aarch64_add_constant): Likewise.
8030 (aarch64_movw_imm): Likewise.
8031 (aarch64_rtx_arith_op_extract_p): Likewise.
8032 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8033 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8034 Remove assert that the mode isn't a vector.
8035 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8036 (aarch64_expand_mov_immediate): Update calls after above changes.
8037 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
8038 (aarch64_and_bitmask_imm): Check for scalar integer modes.
8039 (aarch64_move_imm): Likewise.
8040 (aarch64_can_const_movi_rtx_p): Likewise.
8041 (aarch64_strip_extend): Likewise.
8042 (aarch64_extr_rtx_p): Likewise.
8043 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
8044 a CONST_INT when the mode parameter is VOIDmode.
8045 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
8047 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8049 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
8050 * stor-layout.c (bitwise_mode_for_mode): Likewise.
8051 (bitwise_type_for_mode): Update accordingly.
8053 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8055 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
8056 * stor-layout.c (mode_for_size_tree): Likewise.
8057 (mode_for_array): Update accordingly.
8058 (layout_decl): Likewise.
8059 (compute_record_mode): Likewise. Only set the mode once.
8061 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8063 * target.def (get_mask_mode): Change return type to opt_mode.
8065 * doc/tm.texi: Regenerate.
8066 * targhooks.h (default_get_mask_mode): Return an opt_mode.
8067 * targhooks.c (default_get_mask_mode): Likewise.
8068 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
8069 * optabs-query.c (can_vec_mask_load_store_p): Update use of
8070 targetm.get_mask_mode.
8071 * tree.c (build_truth_vector_type): Likewise.
8073 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8075 * machmode.h (mode_for_vector): Return an opt_mode.
8076 * stor-layout.c (mode_for_vector): Likewise.
8077 (mode_for_int_vector): Update accordingly.
8078 (layout_type): Likewise.
8079 * config/i386/i386.c (emit_memmov): Likewise.
8080 (ix86_expand_set_or_movmem): Likewise.
8081 (ix86_expand_vector_init): Likewise.
8082 (ix86_get_mask_mode): Likewise.
8083 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
8085 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
8086 * expmed.c (extract_bit_field_1): Likewise.
8087 * expr.c (expand_expr_real_2): Likewise.
8088 * optabs-query.c (can_vec_perm_p): Likewise.
8089 (can_vec_mask_load_store_p): Likewise.
8090 * optabs.c (expand_vec_perm): Likewise.
8091 * targhooks.c (default_get_mask_mode): Likewise.
8092 * tree-vect-stmts.c (vectorizable_store): Likewise.
8093 (vectorizable_load): Likewise.
8094 (get_vectype_for_scalar_type_and_size): Likewise.
8096 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8098 * machmode.h (mode_for_int_vector): New function.
8099 * stor-layout.c (mode_for_int_vector): Likewise.
8100 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
8101 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
8102 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
8103 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
8104 (s390_expand_vcond): Likewise.
8106 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8108 * machmode.h (opt_machine_mode): New type.
8109 (opt_mode<T>): Allow construction from anything that can be
8111 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
8112 (mode_for_size): Return an opt_machine_mode.
8113 * stor-layout.c (mode_for_size): Likewise.
8114 (mode_for_size_tree): Update call accordingly.
8115 (bitwise_mode_for_mode): Likewise.
8116 (make_fract_type): Likewise.
8117 (make_accum_type): Likewise.
8118 * caller-save.c (replace_reg_with_saved_mem): Update call
8120 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8121 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8122 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8123 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8124 * expmed.c (extract_bit_field_1): Likewise.
8125 * reload.c (get_secondary_mem): Likewise.
8126 * varasm.c (assemble_integer): Likewise.
8127 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
8130 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8132 * machmode.h (decimal_float_mode_for_size): New function.
8133 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
8134 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
8135 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
8136 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
8137 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
8138 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
8140 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8142 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
8143 (get_builtin_sync_mode): Likewise.
8144 (expand_ifn_atomic_compare_exchange): Likewise.
8145 (expand_builtin_atomic_clear): Likewise.
8146 (expand_builtin_atomic_test_and_set): Likewise.
8147 (fold_builtin_atomic_always_lock_free): Likewise.
8148 * calls.c (compute_argument_addresses): Likewise.
8149 (emit_library_call_value_1): Likewise.
8150 (store_one_arg): Likewise.
8151 * combine.c (combine_instructions): Likewise.
8152 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
8153 * config/arm/arm.c (arm_function_value): Likewise.
8154 (aapcs_allocate_return_reg): Likewise.
8155 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
8156 * config/i386/i386.c (construct_container): Likewise.
8157 (ix86_gimplify_va_arg): Likewise.
8158 (ix86_expand_sse_cmp): Likewise.
8159 (emit_memmov): Likewise.
8160 (emit_memset): Likewise.
8161 (expand_small_movmem_or_setmem): Likewise.
8162 (ix86_expand_pextr): Likewise.
8163 (ix86_expand_pinsr): Likewise.
8164 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
8165 * config/microblaze/microblaze.c (microblaze_block_move_straight):
8167 * config/mips/mips.c (mips_function_value_1) Likewise.
8168 (mips_block_move_straight): Likewise.
8169 (mips_expand_ins_as_unaligned_store): Likewise.
8170 * config/powerpcspe/powerpcspe.c
8171 (rs6000_darwin64_record_arg_advance_flush): Likewise.
8172 (rs6000_darwin64_record_arg_flush): Likewise.
8173 * config/rs6000/rs6000.c
8174 (rs6000_darwin64_record_arg_advance_flush): Likewise.
8175 (rs6000_darwin64_record_arg_flush): Likewise.
8176 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
8177 (sparc_function_value_1): Likewise.
8178 * config/spu/spu.c (adjust_operand): Likewise.
8179 (spu_emit_branch_or_set): Likewise.
8180 (arith_immediate_p): Likewise.
8181 * emit-rtl.c (gen_lowpart_common): Likewise.
8182 * expr.c (expand_expr_real_1): Likewise.
8183 * function.c (assign_parm_setup_block): Likewise.
8184 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8185 * reload1.c (alter_reg): Likewise.
8186 * stor-layout.c (mode_for_vector): Likewise.
8187 (layout_type): Likewise.
8189 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8191 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
8192 (spu_convert_move): Likewise.
8193 * lower-subreg.c (resolve_simple_move): Likewise.
8195 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8198 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
8199 define_insn to a define_expand.
8200 (altivec_vsum2sws_direct): New define_insn.
8201 (altivec_vsumsws): Convert from a define_insn to a define_expand.
8203 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8205 * config/arm/arm.c (arm_option_params_internal): Improve setting of
8208 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
8213 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
8214 realign stack if stack alignment needed is less than incoming
8217 2017-09-05 Marek Polacek <polacek@redhat.com>
8220 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
8223 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8225 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
8227 2017-09-05 Richard Biener <rguenther@suse.de>
8229 PR tree-optimization/82084
8230 * fold-const.c (can_native_encode_string_p): Handle wide characters.
8232 2017-09-05 Richard Biener <rguenther@suse.de>
8234 PR tree-optimization/82102
8235 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
8237 2017-09-05 Martin Liska <mliska@suse.cz>
8239 PR tree-optimization/82032
8240 * tree-cfg.c (generate_range_test): New function.
8241 * tree-cfg.h (generate_range_test): Declared here.
8242 * tree-cfgcleanup.c (convert_single_case_switch): New function.
8243 (cleanup_control_expr_graph): Use it.
8244 * tree-switch-conversion.c (try_switch_expansion): Remove
8246 (emit_case_nodes): Use generate_range_test.
8248 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
8251 * config/i386/i386.md (*<btsc><mode>_mask): Add
8252 TARGET_USE_BT to insn constraint.
8253 (*btr<mode>_mask): Ditto.
8255 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
8257 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
8258 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
8260 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
8263 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
8264 TARGET_NEON and TARGET_IWMMXT.
8265 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
8266 TARGET_NEON and TARGET_IWMMXT.
8267 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
8269 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
8271 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
8272 (ix86_rewrite_tls_address): Ditto.
8273 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
8274 (ix86_rewrite_tls_address_1): Ditto.
8275 (ix86_rewrite_tls_address): Ditto.
8276 * config/i386/predicates.md (tls_address_pattern): New predicate.
8277 * config/i386/i386.md (TLS address splitter): New splitter.
8279 2017-09-04 Richard Biener <rguenther@suse.de>
8281 PR tree-optimization/82084
8282 * fold-const.h (can_native_encode_string_p): Declare.
8283 * fold-const.c (can_native_encode_string_p): Factor out from ...
8284 (native_encode_string): ... here.
8285 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
8286 vectorizing stores from constants we later cannot handle.
8288 2017-09-04 Marek Polacek <polacek@redhat.com>
8291 * doc/invoke.texi: Update -Wtautological-compare documentation.
8293 2017-09-04 Jeff Law <law@redhat.com>
8295 PR tree-optimization/64910
8296 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
8297 swap the first and last operand if the last is a constant.
8299 2017-09-04 Marek Polacek <polacek@redhat.com>
8302 * convert.c (do_narrow): When sanitizing signed integer overflows,
8303 bail out for signed types.
8304 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
8306 2017-09-04 Richard Biener <rguenther@suse.de>
8308 PR tree-optimization/82060
8309 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
8310 Move devirtualization after stmt folding and before EH/AB/noreturn
8311 cleanup to get the stmt refs canonicalized. Use a bool instead
8312 of gimple_modified_p since that doesn't work for NOPs. Schedule
8313 NOPs generated by folding for removal.
8315 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8316 Alan Hayward <alan.hayward@arm.com>
8317 David Sherwood <david.sherwood@arm.com>
8319 * coretypes.h (pad_direction): New enum.
8320 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
8321 (FUNCTION_ARG_PADDING): Likewise.
8322 * target.def (function_arg_padding): New hook.
8323 * targhooks.h (default_function_arg_padding): Declare.
8324 * targhooks.c (default_function_arg_padding): New function.
8325 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
8326 (TARGET_FUNCTION_ARG_PADDING): ...this.
8327 * doc/tm.texi: Regenerate.
8328 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
8329 instead of direction.
8330 (compute_argument_addresses): Likewise.
8331 (load_register_parameters): Likewise.
8332 (emit_library_call_value_1): Likewise.
8333 (store_one_arg): Use targetm.calls.function_arg_padding instead
8334 of FUNCTION_ARG_PADDING.
8335 (must_pass_in_stack_var_size_or_pad): Likewise.
8336 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
8337 (emit_group_store): Likewise.
8338 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
8339 instead of FUNCTION_ARG_PADDING.
8340 (emit_push_insn): Likewise, and propagate enum change throughout
8342 * function.h (direction): Delete.
8343 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
8345 * function.c (assign_parm_find_stack_rtl): Likewise.
8346 (assign_parm_setup_block_p): Likewise.
8347 (assign_parm_setup_block): Likewise.
8348 (gimplify_parameters): Likewise.
8349 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
8350 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
8352 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
8353 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8354 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
8355 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
8356 (aarch64_function_arg_padding): ...this new function.
8357 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
8358 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8359 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
8360 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8361 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
8362 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8363 (arm_pad_arg_upward): Replace with...
8364 (arm_function_arg_padding): ...this new function.
8365 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
8367 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
8368 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
8369 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8370 (ia64_hpux_function_arg_padding): Replace with...
8371 (ia64_function_arg_padding): ...this new function. Use pad_direction
8372 instead of direction. Check for TARGET_HPUX.
8373 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
8374 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8375 (iq2000_function_arg_padding): New function.
8376 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
8377 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
8378 (mips_function_arg_padding): ...this new function.
8379 (mips_pad_reg_upward): Update accordingly.
8380 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8381 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
8382 targetm.calls.function_arg_padding.
8383 (FUNCTION_ARG_PADDING): Delete.
8384 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8385 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
8386 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8387 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
8388 (nios2_block_reg_padding): Return pad_direction instead of direction.
8389 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
8390 instead of direction.
8391 (nios2_function_arg_padding): Likewise. Make static.
8392 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8393 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
8394 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
8395 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
8396 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8397 (pa_function_arg_padding): Make static. Return pad_direction instead
8399 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
8400 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8401 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
8402 instead of direction. Use targetm.calls.function_arg_padding.
8403 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
8404 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
8405 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
8406 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
8407 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
8409 (function_arg_padding): Rename to...
8410 (rs6000_function_arg_padding): ...this. Make static. Return
8411 pad_direction instead of direction.
8412 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
8413 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
8414 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8415 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
8416 instead of direction. Use targetm.calls.function_arg_padding.
8417 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
8418 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
8419 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
8420 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
8421 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8422 (function_arg_padding): Rename to...
8423 (rs6000_function_arg_padding): ...this. Make static. Return
8424 pad_direction instead of direction.
8425 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
8426 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
8427 * config/s390/s390.c (s390_function_arg_padding): New function.
8428 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8429 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
8430 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
8431 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8432 (function_arg_padding): Rename to...
8433 (sparc_function_arg_padding): ...this. Make static. Return
8434 pad_direction instead of direction.
8435 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
8436 * config/spu/spu.c (spu_function_arg_padding): New function.
8437 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8438 * system.h (FUNCTION_ARG_PADDING): Poison.
8440 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8441 Alan Hayward <alan.hayward@arm.com>
8442 David Sherwood <david.sherwood@arm.com>
8444 * target.def (modes_tieable_p): New hook.
8445 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
8446 (TARGET_MODES_TIEABLE_P): ...this.
8447 * doc/tm.texi.in: Regenerate.
8448 * hooks.h (hook_bool_mode_mode_true): Declare.
8449 * hooks.c (hook_bool_mode_mode_true): New function.
8450 * combine.c (subst): Use targetm.modes_tieable_p instead of
8452 * dse.c (find_shift_sequence): Likewise.
8453 * expmed.c (extract_low_bits): Likewise.
8454 * lower-subreg.c: Include target.h.
8455 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
8457 * rtlanal.c (rtx_cost): Likewise.
8458 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
8459 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
8460 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
8461 (TARGET_MODES_TIEABLE_P): Redefine.
8462 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
8463 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
8464 (TARGET_MODES_TIEABLE_P): Redefine.
8465 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
8466 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
8467 (arc_modes_tieable_p): New function.
8468 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
8469 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
8470 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
8471 (arm_modes_tieable_p): Make static.
8472 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
8473 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
8474 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
8475 (TARGET_MODES_TIEABLE_P): Redefine.
8476 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
8477 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
8478 (TARGET_MODES_TIEABLE_P): Redefine.
8479 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
8480 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
8481 (cr16_modes_tieable_p): New function.
8482 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
8483 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
8484 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
8485 (TRULY_NOOP_TRUNCATION): Update comment.
8486 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
8487 (TRULY_NOOP_TRUNCATION): Update comment.
8488 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
8489 (frv_modes_tieable_p): New function.
8490 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
8491 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
8492 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
8493 (TARGET_MODES_TIEABLE_P): Redefine.
8494 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
8495 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
8496 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
8497 (TARGET_MODES_TIEABLE_P): Redefine.
8498 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
8499 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
8500 (ia64_modes_tieable_p): New function.
8501 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
8502 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
8503 (iq2000_modes_tieable_p): New function.
8504 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
8505 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
8506 (lm32_modes_tieable_p): New function.
8507 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
8508 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
8509 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
8510 (TARGET_MODES_TIEABLE_P): Redefine.
8511 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
8512 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
8513 (m32r_modes_tieable_p): New function.
8514 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
8515 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
8516 (m68k_modes_tieable_p): New function.
8517 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
8518 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
8519 (mcore_modes_tieable_p): New function.
8520 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
8521 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
8523 (TARGET_MODES_TIEABLE_P): Redefine.
8524 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
8525 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
8526 * config/mips/mips.c (mips_modes_tieable_p): Make static.
8527 (TARGET_MODES_TIEABLE_P): Redefine.
8528 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
8529 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
8530 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
8531 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
8532 (mn10300_modes_tieable_p): ...this and make static.
8533 (TARGET_MODES_TIEABLE_P): Redefine.
8534 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
8535 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
8536 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
8537 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
8538 (msp430_modes_tieable_p): Make static.
8539 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
8540 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
8541 (TARGET_MODES_TIEABLE_P): Redefine.
8542 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
8543 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
8544 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
8545 (TARGET_MODES_TIEABLE_P): Redefine.
8546 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
8547 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
8548 * config/pa/pa.c (pa_modes_tieable_p): Make static.
8549 (TARGET_MODES_TIEABLE_P): Redefine.
8550 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
8551 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
8552 (pdp11_modes_tieable_p): New function.
8553 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
8554 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
8555 (rs6000_modes_tieable_p): New function.
8556 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
8557 * config/powerpcspe/powerpcspe.md: Update comment.
8558 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
8559 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
8560 (TARGET_MODES_TIEABLE_P): Redefine.
8561 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
8562 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
8563 (rl78_modes_tieable_p): New function.
8564 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
8565 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
8566 (rs6000_modes_tieable_p): New function.
8567 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
8568 * config/rs6000/rs6000.md: Update comment.
8569 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
8570 * config/rx/rx.c (rx_modes_tieable_p): New function.
8571 (TARGET_MODES_TIEABLE_P): Redefine.
8572 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
8573 * config/s390/s390.c (s390_modes_tieable_p): New function.
8574 (TARGET_MODES_TIEABLE_P): Redefine.
8575 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
8576 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
8577 (sh_modes_tieable_p): New function.
8578 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
8579 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
8580 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
8581 (sparc_modes_tieable_p): Make static.
8582 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
8583 * config/spu/spu.c (spu_modes_tieable_p): New function.
8584 (TARGET_MODES_TIEABLE_P): Redefine.
8585 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
8586 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
8587 (TARGET_MODES_TIEABLE_P): Redefine.
8588 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
8589 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
8590 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
8591 * config/v850/v850.c (v850_modes_tieable_p): New function.
8592 (TARGET_MODES_TIEABLE_P): Redefine.
8593 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
8594 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
8595 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
8596 (visium_modes_tieable_p): New function.
8597 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
8598 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
8599 (xtensa_modes_tieable_p): New function.
8600 * system.h (MODES_TIEABLE_P): Poison.
8602 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8603 Alan Hayward <alan.hayward@arm.com>
8604 David Sherwood <david.sherwood@arm.com>
8606 * target.def (hard_regno_mode_ok): New hook.
8607 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
8608 (TARGET_HARD_REGNO_MODE_OK): ...this.
8609 * doc/tm.texi.in: Regenerate.
8610 * hooks.h (hook_bool_uint_mode_true): Declare.
8611 * hooks.c (hook_bool_uint_mode_true): New function.
8612 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
8614 * genpreds.c (write_insn_preds_c): Add an include of target.h.
8615 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
8616 instead of HARD_REGNO_MODE_OK.
8617 * caller-save.c: Include target.h.
8618 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
8620 * combine.c (can_combine_p): Likewise.
8621 (combinable_i3pat): Likewise.
8622 (can_change_dest_mode): Likewise.
8623 * expr.c (init_expr_target): Likewise.
8624 (convert_move): Likewise.
8625 (convert_modes): Likewise.
8626 * ira.c (setup_prohibited_class_mode_regs): Likewise.
8627 (setup_prohibited_mode_move_regs): Likewise.
8628 * ira.h (target_ira): Likewise.
8629 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8630 * lra-constraints.c (process_alt_operands): Likewise.
8631 (split_reg): Likewise.
8632 * recog.c (peep2_find_free_register): Likewise.
8633 * ree.c (combine_reaching_defs): Likewise.
8634 * regcprop.c (maybe_mode_change): Likewise.
8635 * reginfo.c (init_reg_sets_1): Likewise.
8636 (choose_hard_reg_mode): Likewise.
8637 (simplifiable_subregs): Likewise.
8638 * regrename.c (check_new_reg_p): Likewise.
8639 * reload.c (find_valid_class): Likewise.
8640 (find_valid_class_1): Likewise.
8641 (reload_inner_reg_of_subreg): Likewise.
8642 (push_reload): Likewise.
8643 (combine_reloads): Likewise.
8644 (find_dummy_reload): Likewise.
8645 (find_reloads): Likewise.
8646 * reload1.c (find_reg): Likewise.
8647 (set_reload_reg): Likewise.
8648 (allocate_reload_reg): Likewise.
8649 (choose_reload_regs): Likewise.
8650 (reload_adjust_reg_for_temp): Likewise.
8651 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
8652 (simplify_subreg_regno): Likewise.
8653 * sel-sched.c (init_regs_for_mode): Likewise.
8654 * varasm.c (make_decl_rtl): Likewise.
8655 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
8656 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
8658 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
8659 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
8660 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8661 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
8662 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
8663 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8664 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
8665 (arc_mode_class): Delete.
8666 (HARD_REGNO_MODE_OK): Delete.
8667 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8668 (arc_hard_regno_mode_ok): Rename old array to...
8669 (arc_hard_regno_mode_ok_modes): ...this.
8670 (arc_conditional_register_usage): Update accordingly.
8671 (arc_mode_class): Make static.
8672 (arc_hard_regno_mode_ok): New function.
8673 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
8674 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
8675 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8676 (arm_hard_regno_mode_ok): Make static.
8677 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
8679 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
8680 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
8681 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
8683 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8684 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
8685 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
8686 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
8687 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
8688 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8689 * config/bfin/predicates.md (valid_reg_operand): Use
8690 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
8691 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
8692 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
8693 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8694 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
8695 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
8696 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8697 (cr16_hard_regno_mode_ok): Make static and return a bool.
8698 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
8699 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8700 (cris_hard_regno_mode_ok): New function.
8701 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
8702 (epiphany_mode_class): Delete.
8703 (HARD_REGNO_MODE_OK): Delete.
8704 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
8705 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8706 (hard_regno_mode_ok): Rename to...
8707 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
8708 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
8709 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
8711 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
8712 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
8713 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8714 (frv_hard_regno_mode_ok): Make static and return a bool.
8715 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
8717 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
8718 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
8719 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
8720 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
8722 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8723 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
8724 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
8725 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
8727 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8728 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
8729 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8730 (ia64_hard_regno_mode_ok): New function.
8731 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
8732 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8733 (iq2000_hard_regno_mode_ok): New function.
8734 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
8735 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8736 (lm32_hard_regno_mode_ok): New function.
8737 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
8738 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
8739 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
8740 instead of HARD_REGNO_MODE_OK.
8741 (m32c_hard_regno_ok): Rename to...
8742 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
8743 (m32c_cannot_change_mode_class): Update accordingly.
8744 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8745 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
8746 (m32r_mode_class): Delete.
8747 (HARD_REGNO_MODE_OK): Delete.
8748 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8749 (m32r_hard_regno_mode_ok): Rename to...
8750 (m32r_hard_regno_modes): ...this.
8751 (m32r_mode_class): Make static.
8752 (m32r_hard_regno_mode_ok): New function.
8753 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
8754 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
8755 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8756 (m68k_hard_regno_mode_ok): Make static.
8757 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
8758 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8759 (mcore_hard_regno_mode_ok): New function.
8760 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
8761 (HARD_REGNO_MODE_OK): Delete.
8762 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
8764 (microblaze_hard_regno_mode_ok_p): ...this and make static.
8765 (microblaze_hard_regno_mode_ok): New function.
8766 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8767 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
8768 (mips_hard_regno_mode_ok): Delete.
8769 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
8770 (mips_hard_regno_mode_ok_p): ...this and make static.
8771 (mips_hard_regno_mode_ok_p): Rename to...
8772 (mips_hard_regno_mode_ok_uncached): ...this.
8773 (mips_hard_regno_mode_ok): New function.
8774 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
8775 of HARD_REGNO_MODE_OK.
8776 (mips_option_override): Update after above name changes.
8777 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8778 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
8779 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
8780 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
8781 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
8782 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8783 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
8784 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
8785 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
8786 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8787 (msp430_hard_regno_mode_ok): Make static and return a bool.
8788 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
8789 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
8790 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
8792 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8793 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
8794 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
8795 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
8796 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
8797 (PA_HARD_REGNO_MODE_OK): ...this
8798 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
8799 (PA_HARD_REGNO_MODE_OK): ...this.
8800 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8801 (pa_hard_regno_mode_ok): New function.
8802 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
8803 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8804 (pdp11_hard_regno_mode_ok): New function.
8805 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
8806 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
8808 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
8810 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8811 (rs6000_hard_regno_mode_ok): Rename to...
8812 (rs6000_hard_regno_mode_ok_uncached): ...this.
8813 (rs6000_init_hard_regno_mode_ok): Update accordingly.
8814 (rs6000_hard_regno_mode_ok): New function.
8815 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
8816 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
8817 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
8818 (riscv_hard_regno_mode_ok): ...this and make static.
8819 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8820 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
8821 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
8822 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8823 (rl78_hard_regno_mode_ok): Make static and return bool.
8824 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
8825 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
8827 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
8828 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8829 (rs6000_hard_regno_mode_ok): Rename to...
8830 (rs6000_hard_regno_mode_ok_uncached): ...this.
8831 (rs6000_init_hard_regno_mode_ok): Update accordingly.
8832 (rs6000_hard_regno_mode_ok): New function.
8833 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
8834 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
8835 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8836 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
8837 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
8838 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
8839 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8840 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
8841 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
8842 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8843 (sh_hard_regno_mode_ok): Make static.
8844 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
8845 instead of HARD_REGNO_MODE_OK.
8846 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
8847 (sparc_mode_class): Delete.
8848 (HARD_REGNO_MODE_OK): Delete.
8849 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8850 (hard_regno_mode_classes): Make static.
8851 (sparc_mode_class): Likewise.
8852 (sparc_hard_regno_mode_ok): New function.
8853 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
8854 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
8855 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
8857 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8858 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
8859 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
8860 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
8861 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
8862 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8863 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
8864 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
8865 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8866 (visium_hard_regno_mode_ok): New function.
8867 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
8868 instead of HARD_REGNO_MODE_OK.
8869 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
8870 (HARD_REGNO_MODE_OK): Delete.
8871 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
8872 (xtensa_hard_regno_mode_ok_p): ...this and make static.
8873 (xtensa_option_override): Update accordingly.
8874 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8875 (xtensa_hard_regno_mode_ok): New function.
8876 * system.h (HARD_REGNO_MODE_OK): Poison.
8878 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8879 Alan Hayward <alan.hayward@arm.com>
8880 David Sherwood <david.sherwood@arm.com>
8882 * target.def (hard_regno_call_part_clobbered): New hook.
8883 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
8884 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
8885 * doc/tm.texi: Regenerate.
8886 * hooks.h (hook_bool_uint_mode_false): Declare.
8887 * hooks.c (hook_bool_uint_mode_false): New function.
8888 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8889 * cselib.c (cselib_process_insn): Use
8890 targetm.hard_regno_call_part_clobbered instead of
8891 HARD_REGNO_CALL_PART_CLOBBERED.
8892 * ira-conflicts.c (ira_build_conflicts): Likewise.
8893 * ira-costs.c (ira_tune_allocno_costs): Likewise.
8894 * lra-constraints.c (need_for_call_save_p): Likewise.
8895 * lra-lives.c: Include target.h.
8896 (check_pseudos_live_through_calls): Use
8897 targetm.hard_regno_call_part_clobbered instead of
8898 HARD_REGNO_CALL_PART_CLOBBERED.
8899 * regcprop.c: Include target.h.
8900 (copyprop_hardreg_forward_1): Use
8901 targetm.hard_regno_call_part_clobbered instead of
8902 HARD_REGNO_CALL_PART_CLOBBERED.
8903 * reginfo.c (choose_hard_reg_mode): Likewise.
8904 * regrename.c (check_new_reg_p): Likewise.
8905 * reload.c (find_equiv_reg): Likewise.
8906 * reload1.c (emit_reload_insns): Likewise.
8907 * sched-deps.c (deps_analyze_insn): Likewise.
8908 * sel-sched.c (init_regs_for_mode): Likewise.
8909 (mark_unavailable_hard_regs): Likewise.
8910 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
8911 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8912 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
8914 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8915 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8916 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
8918 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
8920 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8921 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8922 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
8924 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8925 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8926 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
8928 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8929 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
8931 * config/powerpcspe/powerpcspe.c
8932 (rs6000_hard_regno_call_part_clobbered): New function.
8933 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8934 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8935 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
8937 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8938 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8939 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
8941 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8942 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8943 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
8945 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8946 Alan Hayward <alan.hayward@arm.com>
8947 David Sherwood <david.sherwood@arm.com>
8949 * rtl.h (subreg_memory_offset): Declare.
8950 * emit-rtl.c (subreg_memory_offset): New function.
8951 * expmed.c (store_bit_field_1): Use it.
8952 * expr.c (undefined_operand_subword_p): Likewise.
8953 * simplify-rtx.c (simplify_subreg): Likewise.
8955 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
8957 PR rtl-optimization/57448
8960 * optabs.c (expand_atomic_load): Place compiler memory barriers if
8961 using atomic_load pattern.
8962 (expand_atomic_store): Likewise.
8964 2017-09-04 Jakub Jelinek <jakub@redhat.com>
8967 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
8968 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
8969 handling. Use replace_call_with_value with NULL instead of
8970 gsi_replace, unlink_stmt_vdef and release_defs.
8972 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
8975 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
8977 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8980 * rtl.h (emit_library_call_value_1): Declare.
8981 (emit_library_call): Replace declaration with a series of overloads.
8982 Remove the parameter count argument.
8983 (emit_library_call_value): Likewise.
8984 * calls.c (emit_library_call_value_1): Make global. Replace varargs
8985 with an "rtx_mode_t *".
8986 (emit_library_call_value): Delete.
8987 (emit_library_call): Likewise.
8988 * asan.c (asan_emit_stack_protection): Update calls accordingly.
8989 (asan_emit_allocas_unpoison): Likewise.
8990 * builtins.c (expand_builtin_powi): Likewise.
8991 (expand_asan_emit_allocas_unpoison): Likewise.
8992 * cfgexpand.c (expand_main_function): Likewise.
8993 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
8994 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
8995 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
8996 * config/arm/arm.c (arm_trampoline_init): Likewise.
8997 (arm_call_tls_get_addr): Likewise.
8998 (arm_expand_divmod_libfunc): Likewise.
8999 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
9000 (smulsi3_highpart): Likewise.
9001 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
9002 (c6x_expand_compare): Likewise.
9003 (c6x_expand_movmem): Likewise.
9004 * config/frv/frv.c (frv_trampoline_init): Likewise.
9005 * config/i386/i386.c (ix86_trampoline_init): Likewise.
9006 (ix86_expand_divmod_libfunc): Likewise.
9007 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
9008 (ia64_expand_compare): Likewise.
9009 (ia64_profile_hook): Likewise.
9010 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
9011 (nonlocal_goto): Likewise.
9012 (restore_stack_nonlocal): Likewise.
9013 * config/m32r/m32r.c (block_move_call): Likewise.
9014 (m32r_trampoline_init): Likewise.
9015 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
9016 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
9017 (m68k_call_m68k_read_tp): Likewise.
9018 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
9019 (microblaze_expand_divide): Likewise.
9020 * config/mips/mips.h (mips_args): Likewise.
9021 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
9022 (MIPS_ICACHE_SYNC): Likewise.
9023 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
9024 (nios2_trampoline_init): Likewise.
9025 * config/pa/pa.c (hppa_tls_call): Likewise.
9026 (pa_trampoline_init): Likewise.
9027 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
9028 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
9029 (expand_strn_compare): Likewise.
9030 (rs6000_generate_compare): Likewise.
9031 (rs6000_expand_float128_convert): Likewise.
9032 (output_profile_hook): Likewise.
9033 (rs6000_trampoline_init): Likewise.
9034 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
9035 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
9036 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
9037 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
9038 (rs6000_generate_compare): Likewise.
9039 (rs6000_expand_float128_convert): Likewise.
9040 (output_profile_hook): Likewise.
9041 (rs6000_trampoline_init): Likewise.
9042 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
9043 * config/sh/sh.c (sh_trampoline_init): Likewise.
9044 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
9045 (sparc_emit_float_lib_cmp): Likewise.
9046 (sparc32_initialize_trampoline): Likewise.
9047 (sparc64_initialize_trampoline): Likewise.
9048 (sparc_profile_hook): Likewise.
9049 * config/spu/spu.c (ea_load_store): Likewise.
9050 * config/spu/spu.md (floatunssidf2): Likewise.
9051 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
9052 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
9053 * config/visium/visium.c (expand_block_move_4): Likewise.
9054 (expand_block_move_2): Likewise.
9055 (expand_block_move_1): Likewise.
9056 (expand_block_set_4): Likewise.
9057 (expand_block_set_2): Likewise.
9058 (expand_block_set_1): Likewise.
9059 (visium_trampoline_init): Likewise.
9060 (visium_profile_hook): Likewise.
9061 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
9062 (xtensa_setup_frame_addresses): Likewise.
9063 (xtensa_trampoline_init): Likewise.
9064 * except.c (sjlj_emit_function_enter): Likewise.
9065 (sjlj_emit_function_exit): Likewise.
9066 * explow.c (allocate_dynamic_stack_space): Likewise.
9067 (probe_stack_range): Likewise.
9068 * expr.c (convert_mode_scalar): Likewise.
9069 * optabs.c (expand_binop): Likewise.
9070 (expand_twoval_binop_libfunc): Likewise.
9071 (expand_unop): Likewise.
9072 (prepare_cmp_insn): Likewise.
9073 (prepare_float_lib_cmp): Likewise.
9074 (expand_float): Likewise.
9075 (expand_fix): Likewise.
9076 (expand_fixed_convert): Likewise.
9077 (maybe_emit_sync_lock_test_and_set): Likewise.
9078 (expand_atomic_compare_and_swap): Likewise.
9079 (expand_mem_thread_fence): Likewise.
9080 (expand_atomic_fetch_op): Likewise.
9082 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
9084 * doc/generic.texi (OpenACC): Adjust URL.
9085 * doc/invoke.texi (C Dialect Options): Ditto.
9087 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
9089 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
9090 predicate for operand 1. Add (m,<S>) constraint.
9091 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
9092 Prevent memory operand 1 with register operand 2.
9094 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
9096 PR rtl-optimization/82024
9097 * combine.c (try_combine): If the combination result is a PARALLEL,
9098 and we only need to retain the SET in there that would be placed
9099 at I2, check that we can place that at I3 instead, before doing so.
9101 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9104 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
9106 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
9107 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
9110 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
9111 Jeff Law <law@redhat.com>
9113 * varasm.c (bss_initializer_p): Do not put constants into .bss
9114 (categorize_decl_for_section): Handle bss_initializer_p returning
9115 false when DECL_INITIAL is NULL.
9117 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9120 * config/s390/s390.c (s390_can_inline_p): New function.
9122 2017-09-01 Jeff Law <law@redhat.com>
9124 PR tree-optimization/82052
9125 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
9126 Always initialize the returned slot after a hash table miss
9127 when INSERT is true.
9129 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
9131 * config/s390/s390.md (mem_signal_fence): Remove.
9132 * doc/md.texi (mem_signal_fence): Remove.
9133 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
9135 * target-insns.def (mem_signal_fence): Remove.
9137 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9140 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
9143 * asan.c (create_odr_indicator): Strip name encoding from assembler
9144 name before appending it after __odr_asan_.
9146 2017-09-01 Martin Liska <mliska@suse.cz>
9148 PR tree-optimization/82059
9149 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
9150 frequency only when an edge is redirected.
9152 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9154 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
9155 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
9156 (arc_conditional_register_usage): Remove ARC600 lp_count
9158 (arc_file_start): Emit Tag_ARC_CPU_variation.
9159 (arc_can_use_doloop_p): New conditions to use ZOLs.
9160 (hwloop_fail): New function.
9161 (hwloop_optimize): Likewise.
9162 (hwloop_pattern_reg): Likewise.
9163 (arc_doloop_hooks): New struct, to be used with reorg_loops.
9164 (arc_reorg_loops): New function, calls reorg_loops.
9165 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
9166 (arc600_corereg_hazard): Remove ZOL checking, case handled by
9168 (arc_loop_hazard): Remove function, functionality moved into
9170 (arc_hazard): Remove arc_loop_hazard call.
9171 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
9172 into hwloop_optimize.
9173 (arc_label_align): Remove ZOL handling.
9174 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
9175 * config/arc/arc.md (doloop_begin): Remove pattern.
9176 (doloop_begin_i): Likewise.
9177 (doloop_end_i): Likewise.
9178 (doloop_fallback): Likewise.
9179 (doloop_fallback_m): Likewise.
9180 (doloop_end): Reimplement expand.
9181 (arc_lp): New pattern for LP instruction.
9182 (loop_end): New pattern.
9183 (loop_fail): Likewise.
9184 (decrement_and_branch_until_zero): Likewise.
9185 * config/arc/arc.opt (mlpc-width): New option.
9186 * doc/invoke.texi (mlpc-width): Document option.
9188 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9190 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
9191 (arc_ccfsm_advance): Fix checking for delay slots.
9192 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
9194 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9196 * config/arc/arc.md (movqi_insn): Add stores to save constant long
9198 (movhi_insn): Update store instruction constraint which are saving
9199 6-bit short immediates.
9200 (movsi_insn): Consider also short scaled load operations.
9201 (zero_extendhisi2_i): Use Usd constraint instead of T.
9202 (extendhisi2_i): Add q constraint.
9203 (arc_clzsi2): Add type and length attributes.
9204 (arc_ctzsi2): Likewise.
9205 * config/arc/constraints.md (Usc): Update constraint, the
9206 assembler can parse two relocations for a single instruction.
9208 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9210 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
9211 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
9213 2017-08-31 Olivier Hainque <hainque@adacore.com>
9215 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
9216 match as powerpc-wrs-vxworks*.
9218 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
9220 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
9221 register constraint for by-element operand.
9222 (aarch64_mls_elt_merge<mode>): Likewise.
9224 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9226 * config/arc/arc.c (arc_can_follow_jump): Check for short
9229 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9231 * config.gcc: Use g.opt for arc.
9232 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
9233 functionality moved to ...
9234 (legitimate_scaled_address_p): New function, ...here.
9235 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
9236 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
9237 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
9239 (arc_override_options): Handle G option.
9240 (arc_output_pic_addr_const): Correct function definition.
9241 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
9242 (arc_decl_anon_ns_mem_p): Delete.
9243 (arc_in_small_data_p): Overhaul this function to take into
9244 consideration the value given via G option.
9245 (arc_rewrite_small_data_1): Renamed and corrected old
9246 arc_rewrite_small_data function.
9247 (arc_rewrite_small_data): New function.
9248 (small_data_pattern): Don't use pic_offset_table_rtx.
9249 * config/arc/arc.h (CC1_SPEC): Recognize G option.
9250 * config/arc/simdext.md (movmisalignv2hi): Use
9251 prepare_move_operands function.
9253 (movmisalign*): Likewise.
9254 * doc/invoke.texi (ARC options): Document -G option.
9256 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9258 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
9260 * config/arc/arc.c (arc_print_operand): Output scalled address for
9261 sdata whenever is possible.
9262 (arc_in_small_data_p): Allow sdata for 64bit datum when double
9263 load/stores are available.
9264 (compact_sda_memory_operand): Check for the alignment required by
9265 code density instructions.
9266 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
9268 * config/arc/constraints.md (Usd): Update constraint.
9269 (Us0): New constraint.
9270 (Usc): Update constraint.
9272 2017-08-31 Richard Biener <rguenther@suse.de>
9275 * dwarf2out.c (dwarf2out_early_global_decl): Process each
9278 2017-08-31 Tamar Christina <tamar.christina@arm.com>
9280 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
9281 Resize type_signature.
9283 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9284 Alan Hayward <alan.hayward@arm.com>
9285 David Sherwood <david.sherwood@arm.com>
9287 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
9288 subregs whose inner modes can be stored in GPRs.
9289 (aarch64_classify_index): Likewise.
9291 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9292 Alan Hayward <alan.hayward@arm.com>
9293 David Sherwood <david.sherwood@arm.com>
9295 * config/aarch64/iterators.md (V_cmp_result): Rename to...
9296 (V_INT_EQUIV): ...this.
9297 (v_cmp_result): Rename to...
9298 (v_int_equiv): ...this.
9299 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
9300 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
9301 (copysign<mode>3): Likewise.
9302 (aarch64_simd_bsl<mode>_internal): Likewise.
9303 (aarch64_simd_bsl<mode>): Likewise.
9304 (vec_cmp<mode><mode>): Likewise.
9305 (vcond<mode><mode>): Likewise.
9306 (vcond<v_cmp_mixed><mode>): Likewise.
9307 (vcondu<mode><v_cmp_mixed>): Likewise.
9308 (aarch64_cm<optab><mode>): Likewise.
9309 (aarch64_cmtst<mode>): Likewise.
9310 (aarch64_fac<optab><mode>): Likewise.
9311 (vec_perm_const<mode>): Likewise.
9312 (vcond_mask_<mode><v_cmp_result>): Rename to...
9313 (vcond_mask_<mode><v_int_equiv>): ...this.
9314 (vec_cmp<mode><v_cmp_result>): Rename to...
9315 (vec_cmp<mode><v_int_equiv>): ...this.
9317 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9318 Alan Hayward <alan.hayward@arm.com>
9319 David Sherwood <david.sherwood@arm.com>
9321 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
9323 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
9324 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
9325 (UNSPEC_LD4_DREG): New unspecs.
9326 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
9327 (aarch64_ld2<mode>_dreg_be): Replace with...
9328 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
9330 (aarch64_ld3<mode>_dreg_le)
9331 (aarch64_ld3<mode>_dreg_be): Replace with...
9332 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
9334 (aarch64_ld4<mode>_dreg_le)
9335 (aarch64_ld4<mode>_dreg_be): Replace with...
9336 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
9339 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9341 PR tree-optimization/81987
9342 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
9343 insert an initializer in a location not dominated by the stride
9346 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
9348 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
9349 on the entire header of the finally block in the fallthru case.
9351 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
9353 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
9355 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
9357 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
9358 rs6000_emit_move_from_cr and call renamed function.
9359 (rs6000_emit_prologue): Call renamed functions.
9360 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
9361 movesi_from_cr, remove volatile CRs.
9363 2017-08-30 Jon Beniston <jon@beniston.com>
9364 Richard Biener <rguenther@suse.de>
9366 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
9367 of VECTOR_MODE_P check.
9368 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
9369 element vector types.
9371 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9373 * df.h (df_read_modify_subreg_p): Remove in favor of...
9374 * rtl.h (read_modify_subreg_p): ...this new function. Take a
9375 const_rtx instead of an rtx.
9376 * cprop.c (local_cprop_find_used_regs): Update accordingly.
9377 * df-problems.c (df_word_lr_mark_ref): Likewise.
9378 * ira-lives.c (mark_pseudo_reg_live): Likewise.
9379 (mark_pseudo_reg_dead): Likewise.
9380 (mark_ref_dead): Likewise.
9381 * reginfo.c (init_subregs_of_mode): Likewise.
9382 * sched-deps.c (sched_analyze_1): Likewise.
9383 * df-scan.c (df_def_record_1): Likewise.
9384 (df_uses_record): Likewise.
9385 (df_read_modify_subreg_p): Remove in favor of...
9386 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
9387 const_rtx instead of an rtx.
9389 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9390 Alan Hayward <alan.hayward@arm.com>
9391 David Sherwood <david.sherwood@arm.com>
9393 * rtl.h (partial_subreg_p): New function.
9394 * caller-save.c (save_call_clobbered_regs): Use it.
9395 * calls.c (expand_call): Likewise.
9396 * combine.c (combinable_i3pat): Likewise.
9397 (simplify_set): Likewise.
9398 (make_extraction): Likewise.
9399 (make_compound_operation_int): Likewise.
9400 (gen_lowpart_or_truncate): Likewise.
9401 (force_to_mode): Likewise.
9402 (make_field_assignment): Likewise.
9403 (reg_truncated_to_mode): Likewise.
9404 (record_truncated_value): Likewise.
9405 (move_deaths): Likewise.
9406 * cse.c (record_jump_cond): Likewise.
9407 (cse_insn): Likewise.
9408 * cselib.c (cselib_lookup_1): Likewise.
9409 * expmed.c (extract_bit_field_using_extv): Likewise.
9410 * function.c (assign_parm_setup_reg): Likewise.
9411 * ifcvt.c (noce_convert_multiple_sets): Likewise.
9412 * ira-build.c (create_insn_allocnos): Likewise.
9413 * lra-coalesce.c (merge_pseudos): Likewise.
9414 * lra-constraints.c (match_reload): Likewise.
9415 (simplify_operand_subreg): Likewise.
9416 (curr_insn_transform): Likewise.
9417 * lra-lives.c (process_bb_lives): Likewise.
9418 * lra.c (new_insn_reg): Likewise.
9419 (lra_substitute_pseudo): Likewise.
9420 * regcprop.c (mode_change_ok): Likewise.
9421 (maybe_mode_change): Likewise.
9422 (copyprop_hardreg_forward_1): Likewise.
9423 * reload.c (push_reload): Likewise.
9424 (find_reloads): Likewise.
9425 (find_reloads_subreg_address): Likewise.
9426 * reload1.c (alter_reg): Likewise.
9427 (eliminate_regs_1): Likewise.
9428 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9430 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
9432 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
9433 back to if statements, including unpack.
9435 2017-08-30 Martin Liska <mliska@suse.cz>
9438 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
9440 (func_checker::compare_asm_inputs_outputs): ... this function.
9441 (func_checker::compare_gimple_asm): Use the function to compare
9442 also ASM constrains.
9443 * ipa-icf-gimple.h: Rename the function.
9445 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9446 Alan Hayward <alan.hayward@arm.com>
9447 David Sherwood <david.sherwood@arm.com>
9449 * coretypes.h (complex_mode): New type.
9450 * gdbhooks.py (build_pretty_printer): Handle it.
9451 * machmode.h (complex_mode): New class.
9452 (complex_mode::includes_p): New function.
9453 (is_complex_int_mode): Likewise.
9454 (is_complex_float_mode): Likewise.
9455 * genmodes.c (get_mode_class): Handle complex mode classes.
9456 * function.c (expand_function_end): Use is_complex_int_mode.
9458 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9459 Alan Hayward <alan.hayward@arm.com>
9460 David Sherwood <david.sherwood@arm.com>
9462 * coretypes.h (scalar_mode_pod): New typedef.
9463 * gdbhooks.py (build_pretty_printer): Handle it.
9464 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
9465 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
9466 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
9467 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
9470 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9471 Alan Hayward <alan.hayward@arm.com>
9472 David Sherwood <david.sherwood@arm.com>
9474 * machmode.h (mode_for_vector): Take a scalar_mode instead
9476 * stor-layout.c (mode_for_vector): Likewise.
9477 * explow.c (promote_mode): Use as_a <scalar_mode>.
9478 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
9480 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9481 Alan Hayward <alan.hayward@arm.com>
9482 David Sherwood <david.sherwood@arm.com>
9484 * target.def (preferred_simd_mode): Take a scalar_mode
9485 instead of a machine_mode.
9486 * targhooks.h (default_preferred_simd_mode): Likewise.
9487 * targhooks.c (default_preferred_simd_mode): Likewise.
9488 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
9489 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
9490 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
9491 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
9492 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
9493 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
9494 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
9495 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
9497 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
9498 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
9499 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
9500 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
9501 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
9502 * doc/tm.texi: Regenerate.
9503 * optabs-query.c (can_vec_mask_load_store_p): Return false for
9506 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9507 Alan Hayward <alan.hayward@arm.com>
9508 David Sherwood <david.sherwood@arm.com>
9510 * target.def (scalar_mode_supported_p): Take a scalar_mode
9511 instead of a machine_mode.
9512 * targhooks.h (default_scalar_mode_supported_p): Likewise.
9513 * targhooks.c (default_scalar_mode_supported_p): Likewise.
9514 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
9515 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
9516 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
9517 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
9518 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
9519 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
9520 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
9521 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
9522 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
9523 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
9524 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
9525 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
9527 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
9528 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
9529 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
9530 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
9531 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
9533 * doc/tm.texi: Regenerate.
9535 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9536 Alan Hayward <alan.hayward@arm.com>
9537 David Sherwood <david.sherwood@arm.com>
9539 * coretypes.h (opt_scalar_mode): New typedef.
9540 * gdbhooks.py (build_pretty_printers): Handle it.
9541 * machmode.h (mode_iterator::get_2xwider): Add overload for
9543 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
9545 * expr.c (convert_mode_scalar): Likewise.
9546 * omp-low.c (omp_clause_aligned_alignment): Likewise.
9547 * optabs.c (expand_float): Likewise.
9548 (expand_fix): Likewise.
9549 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
9551 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9552 Alan Hayward <alan.hayward@arm.com>
9553 David Sherwood <david.sherwood@arm.com>
9555 * optabs.c (expand_float): Explicitly check for scalars before
9556 using a branching expansion.
9557 (expand_fix): Likewise.
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 * expr.c (convert_mode): Split scalar handling out into...
9564 (convert_mode_scalar): ...this new function. Treat the modes
9567 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9568 Alan Hayward <alan.hayward@arm.com>
9569 David Sherwood <david.sherwood@arm.com>
9571 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
9573 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
9575 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9576 Alan Hayward <alan.hayward@arm.com>
9577 David Sherwood <david.sherwood@arm.com>
9579 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
9580 rather than a machine_mode.
9581 (fixed_from_string): Likewise.
9582 (fixed_convert): Likewise.
9583 (fixed_convert_from_int): Likewise.
9584 (fixed_convert_from_real): Likewise.
9585 (real_convert_from_fixed): Likewise.
9586 * fixed-value.c (fixed_from_double_int): Likewise.
9587 (fixed_from_string): Likewise.
9588 (fixed_convert): Likewise.
9589 (fixed_convert_from_int): Likewise.
9590 (fixed_convert_from_real): Likewise.
9591 (real_convert_from_fixed): Likewise.
9592 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
9594 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9595 Alan Hayward <alan.hayward@arm.com>
9596 David Sherwood <david.sherwood@arm.com>
9598 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
9599 of separate mode class checks. Do not allow vector modes here.
9600 (immed_wide_int_const): Use as_a <scalar_mode>.
9601 * explow.c (trunc_int_for_mode): Likewise.
9602 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
9603 (wi::shwi): Likewise.
9604 (wi::min_value): Likewise.
9605 (wi::max_value): Likewise.
9606 * dwarf2out.c (loc_descriptor): Likewise.
9607 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
9610 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9611 Alan Hayward <alan.hayward@arm.com>
9612 David Sherwood <david.sherwood@arm.com>
9614 * tree.h (SCALAR_TYPE_MODE): New macro.
9615 * expr.c (expand_expr_addr_expr_1): Use it.
9616 (expand_expr_real_2): Likewise.
9617 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
9618 (fold_convert_const_fixed_from_int): Likewise.
9619 (fold_convert_const_fixed_from_real): Likewise.
9620 (native_encode_fixed): Likewise
9621 (native_encode_complex): Likewise
9622 (native_encode_vector): Likewise.
9623 (native_interpret_fixed): Likewise.
9624 (native_interpret_real): Likewise.
9625 (native_interpret_complex): Likewise.
9626 (native_interpret_vector): Likewise.
9627 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
9628 (simd_clone_adjust_argument_types): Likewise.
9629 (simd_clone_init_simd_arrays): Likewise.
9630 (simd_clone_adjust): Likewise.
9631 * stor-layout.c (layout_type): Likewise.
9632 * tree.c (build_minus_one_cst): Likewise.
9633 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
9634 * tree-inline.c (estimate_move_cost): Likewise.
9635 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
9636 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
9637 (vectorizable_reduction): Likewise.
9638 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
9639 (vect_recog_mixed_size_cond_pattern): Likewise.
9640 (check_bool_pattern): Likewise.
9641 (adjust_bool_pattern): Likewise.
9642 (search_type_for_mask_1): Likewise.
9643 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
9644 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
9645 (vectorizable_load): Likewise.
9646 (vectorizable_store): Likewise.
9647 * ubsan.c (ubsan_encode_value): Likewise.
9648 * varasm.c (output_constant): Likewise.
9650 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9651 Alan Hayward <alan.hayward@arm.com>
9652 David Sherwood <david.sherwood@arm.com>
9654 * coretypes.h (scalar_mode): New class.
9655 * machmode.h (scalar_mode): Likewise.
9656 (scalar_mode::includes_p): New function.
9657 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
9658 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
9659 * genmodes.c (get_mode_class): Handle remaining scalar modes.
9660 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
9661 * expmed.c (store_bit_field_1): Likewise.
9662 (extract_bit_field_1): Likewise.
9663 * expr.c (write_complex_part): Likewise.
9664 (read_complex_part): Likewise.
9665 (emit_move_complex_push): Likewise.
9666 (expand_expr_real_2): Likewise.
9667 * function.c (assign_parm_setup_reg): Likewise.
9668 (assign_parms_unsplit_complex): Likewise.
9669 * optabs.c (expand_binop): Likewise.
9670 * rtlanal.c (subreg_get_info): Likewise.
9671 * simplify-rtx.c (simplify_immed_subreg): Likewise.
9672 * varasm.c (output_constant_pool_2): Likewise.
9674 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9675 Alan Hayward <alan.hayward@arm.com>
9676 David Sherwood <david.sherwood@arm.com>
9678 * expmed.c (extract_high_half): Use scalar_int_mode and remove
9680 (expmed_mult_highpart_optab): Likewise.
9681 (expmed_mult_highpart): Likewise.
9683 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9684 Alan Hayward <alan.hayward@arm.com>
9685 David Sherwood <david.sherwood@arm.com>
9687 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
9688 instead of a machine_mode.
9689 (builtin_memset_read_str): Likewise.
9690 * builtins.c (c_readstr): Likewise.
9691 (builtin_memcpy_read_str): Likewise.
9692 (builtin_strncpy_read_str): Likewise.
9693 (builtin_memset_read_str): Likewise.
9694 (builtin_memset_gen_str): Likewise.
9695 (expand_builtin_signbit): Use scalar_int_mode for local variables.
9696 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
9697 instead of a machine_mode.
9698 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
9700 (make_extraction): Likewise.
9701 (try_widen_shift_mode): Take and return scalar_int_modes instead
9703 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
9704 a scalar_int_mode instead of a machine_mode.
9705 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
9706 (avr_addr_space_pointer_mode): Likewise.
9707 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
9708 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
9709 (msp430_unwind_word_mode): Likewise.
9710 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
9711 (spu_addr_space_pointer_mode): Likewise.
9712 (spu_addr_space_address_mode): Likewise.
9713 (spu_libgcc_cmp_return_mode): Likewise.
9714 (spu_libgcc_shift_count_mode): Likewise.
9715 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
9716 (rl78_addr_space_pointer_mode): Likewise.
9717 (fl78_unwind_word_mode): Likewise.
9718 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
9720 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
9721 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
9722 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
9723 (mips_valid_pointer_mode): Likewise.
9724 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
9725 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
9726 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
9728 (ft32_addr_space_address_mode): Likewise.
9729 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
9730 scalar_int_mode instead of a machine_mode.
9731 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
9733 (m32c_addr_space_address_mode): Likewise.
9734 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
9735 (rs6000_eh_return_filter_mode): Likewise.
9736 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
9737 (rs6000_eh_return_filter_mode): Likewise.
9738 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
9739 (s390_libgcc_shift_count_mode): Likewise.
9740 (s390_unwind_word_mode): Likewise.
9741 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
9743 * target.def (mode_rep_extended): Likewise.
9744 (valid_pointer_mode): Likewise.
9745 (addr_space.valid_pointer_mode): Likewise.
9746 (eh_return_filter_mode): Return a scalar_int_mode rather than
9748 (libgcc_cmp_return_mode): Likewise.
9749 (libgcc_shift_count_mode): Likewise.
9750 (unwind_word_mode): Likewise.
9751 (addr_space.pointer_mode): Likewise.
9752 (addr_space.address_mode): Likewise.
9753 * doc/tm.texi: Regenerate.
9754 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
9756 (do_jump): Use scalar_int_mode for local variables.
9757 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
9758 rather than a machine_mode.
9759 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
9760 (scompare_loc_descriptor_wide): Likewise.
9761 (scompare_loc_descriptor_narrow): Likewise.
9762 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
9764 * except.c (sjlj_emit_dispatch_table): Likewise.
9765 (expand_builtin_eh_copy_values): Likewise.
9766 * explow.c (convert_memory_address_addr_space_1): Likewise.
9767 Take a scalar_int_mode rather than a machine_mode.
9768 (convert_memory_address_addr_space): Take a scalar_int_mode rather
9769 than a machine_mode.
9770 (memory_address_addr_space): Use scalar_int_mode for local variables.
9771 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
9772 rather than a machine_mode.
9773 * expmed.c (mask_rtx): Likewise.
9774 (init_expmed_one_conv): Likewise.
9775 (expand_mult_highpart_adjust): Likewise.
9776 (extract_high_half): Likewise.
9777 (expmed_mult_highpart_optab): Likewise.
9778 (expmed_mult_highpart): Likewise.
9779 (expand_smod_pow2): Likewise.
9780 (expand_sdiv_pow2): Likewise.
9781 (emit_store_flag_int): Likewise.
9782 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
9784 (extract_low_bits): Likewise.
9785 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
9787 * expr.c (pieces_addr::adjust): Likewise.
9788 (can_store_by_pieces): Likewise.
9789 (store_by_pieces): Likewise.
9790 (clear_by_pieces_1): Likewise.
9791 (expand_expr_addr_expr_1): Likewise.
9792 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
9793 (expand_expr_real_1): Likewise.
9794 (try_casesi): Likewise.
9795 * final.c (shorten_branches): Likewise.
9796 * fold-const.c (fold_convert_const_int_from_fixed): Change the
9797 type of "mode" to machine_mode.
9798 * internal-fn.c (expand_arith_overflow_result_store): Take a
9799 scalar_int_mode rather than a machine_mode.
9800 (expand_mul_overflow): Use scalar_int_mode for local variables.
9801 * loop-doloop.c (doloop_modify): Likewise.
9802 (doloop_optimize): Likewise.
9803 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
9804 than a machine_mode.
9805 (expand_doubleword_shift_condmove): Likewise.
9806 (expand_doubleword_shift): Likewise.
9807 (expand_doubleword_clz): Likewise.
9808 (expand_doubleword_popcount): Likewise.
9809 (expand_doubleword_parity): Likewise.
9810 (expand_absneg_bit): Use scalar_int_mode for local variables.
9811 (prepare_float_lib_cmp): Likewise.
9812 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
9813 rather than a machine_mode.
9814 (convert_memory_address_addr_space): Likewise.
9815 (get_mode_bounds): Likewise.
9816 (get_address_mode): Return a scalar_int_mode rather than a
9818 * rtlanal.c (get_address_mode): Likewise.
9819 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
9820 than a machine_mode.
9821 * targhooks.c (default_mode_rep_extended): Likewise.
9822 (default_valid_pointer_mode): Likewise.
9823 (default_addr_space_valid_pointer_mode): Likewise.
9824 (default_eh_return_filter_mode): Return a scalar_int_mode rather
9825 than a machine_mode.
9826 (default_libgcc_cmp_return_mode): Likewise.
9827 (default_libgcc_shift_count_mode): Likewise.
9828 (default_unwind_word_mode): Likewise.
9829 (default_addr_space_pointer_mode): Likewise.
9830 (default_addr_space_address_mode): Likewise.
9831 * targhooks.h (default_eh_return_filter_mode): Likewise.
9832 (default_libgcc_cmp_return_mode): Likewise.
9833 (default_libgcc_shift_count_mode): Likewise.
9834 (default_unwind_word_mode): Likewise.
9835 (default_addr_space_pointer_mode): Likewise.
9836 (default_addr_space_address_mode): Likewise.
9837 (default_mode_rep_extended): Take a scalar_int_mode rather than
9839 (default_valid_pointer_mode): Likewise.
9840 (default_addr_space_valid_pointer_mode): Likewise.
9841 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
9843 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
9844 rather than a machine_mode.
9845 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
9846 for local variables.
9847 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
9848 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
9849 than a machine_mode.
9851 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9852 Alan Hayward <alan.hayward@arm.com>
9853 David Sherwood <david.sherwood@arm.com>
9855 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
9856 the mode argument to scalar_int_mode.
9857 (do_jump_by_parts_zero_rtx): Likewise.
9858 (do_jump_by_parts_equality_rtx): Likewise.
9859 (do_jump_by_parts_greater): Take a mode argument.
9860 (do_jump_by_parts_equality): Likewise.
9861 (do_jump_1): Update calls accordingly.
9863 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9864 Alan Hayward <alan.hayward@arm.com>
9865 David Sherwood <david.sherwood@arm.com>
9867 * is-a.h (safe_dyn_cast): New function.
9868 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
9869 (jump_table_for_label): Likewise.
9870 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
9871 instead of an rtx_insn *.
9872 (shorten_branches): Use dyn_cast instead of LABEL_P and
9873 JUMP_TABLE_DATA_P. Use jump_table_for_label and
9874 rtx_jump_table_data::get_data_mode.
9875 (final_scan_insn): Likewise.
9877 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9878 Alan Hayward <alan.hayward@arm.com>
9879 David Sherwood <david.sherwood@arm.com>
9881 * combine.c (try_combine): Use is_a <scalar_int_mode> when
9882 trying to combine a full-register integer set with a subreg
9885 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9886 Alan Hayward <alan.hayward@arm.com>
9887 David Sherwood <david.sherwood@arm.com>
9889 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
9890 that is always either address_mode or pointer_mode.
9892 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9893 Alan Hayward <alan.hayward@arm.com>
9894 David Sherwood <david.sherwood@arm.com>
9896 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
9897 when the two are known to be equal.
9899 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9900 Alan Hayward <alan.hayward@arm.com>
9901 David Sherwood <david.sherwood@arm.com>
9903 * simplify-rtx.c (simplify_const_unary_operation): Use
9904 is_a <scalar_int_mode> instead of checking for a nonzero
9905 precision. Forcibly convert op_mode to a scalar_int_mode
9906 in that case. More clearly differentiate the operand and
9907 result modes and use the former when deciding what the value
9908 of a count-bits operation should be. Use is_int_mode instead
9909 of checking for a MODE_INT. Remove redundant check for whether
9910 this mode has a zero precision.
9912 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9913 Alan Hayward <alan.hayward@arm.com>
9914 David Sherwood <david.sherwood@arm.com>
9916 * optabs.c (widen_leading): Change the type of the mode argument
9917 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
9918 (widen_bswap): Likewise.
9919 (expand_parity): Likewise.
9920 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
9921 (expand_ffs): Likewise.
9922 (epand_unop): Check for scalar integer modes before calling the
9925 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9926 Alan Hayward <alan.hayward@arm.com>
9927 David Sherwood <david.sherwood@arm.com>
9929 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
9931 (expand_expr_real_1): Update call accordingly.
9933 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9934 Alan Hayward <alan.hayward@arm.com>
9935 David Sherwood <david.sherwood@arm.com>
9937 * expmed.c (store_bit_field_using_insv): Add op0_mode and
9938 value_mode arguments. Use scalar_int_mode internally.
9939 (store_bit_field_1): Rename the new integer mode from imode
9940 to op0_mode and use it instead of GET_MODE (op0). Update calls
9941 to store_split_bit_field, store_bit_field_using_insv and
9942 store_fixed_bit_field.
9943 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
9944 Use scalar_int_mode internally. Use a bit count rather than a mode
9945 when calculating the largest bit size for get_best_mode.
9946 Update calls to store_split_bit_field and store_fixed_bit_field_1.
9947 (store_fixed_bit_field_1): Add mode and value_mode arguments.
9948 Remove assertion that OP0 has a scalar integer mode.
9949 (store_split_bit_field): Add op0_mode and value_mode arguments.
9950 Update calls to extract_fixed_bit_field.
9951 (extract_bit_field_using_extv): Add an op0_mode argument.
9952 Use scalar_int_mode internally.
9953 (extract_bit_field_1): Rename the new integer mode from imode to
9954 op0_mode and use it instead of GET_MODE (op0). Update calls to
9955 extract_split_bit_field, extract_bit_field_using_extv and
9956 extract_fixed_bit_field.
9957 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
9958 to extract_split_bit_field and extract_fixed_bit_field_1.
9959 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
9960 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
9962 (extract_split_bit_field): Add an op0_mode argument. Update call
9963 to extract_fixed_bit_field.
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 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
9970 * explow.c (hard_function_value): Likewise.
9971 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
9972 convert_to_mode call outside the loop.
9973 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
9974 for the mode iterator. Require the mode specified by max_pieces
9976 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
9978 (copy_blkmode_to_reg): Likewise.
9979 (set_storage_via_setmem): Likewise.
9980 * optabs.c (prepare_cmp_insn): Likewise.
9981 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
9982 * stor-layout.c (finish_bitfield_representative): Likewise.
9984 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9985 Alan Hayward <alan.hayward@arm.com>
9986 David Sherwood <david.sherwood@arm.com>
9988 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
9989 * expr.c (convert_move): Use them.
9990 (convert_modes): Likewise.
9991 (store_expr_with_bounds): Likewise.
9993 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9994 Alan Hayward <alan.hayward@arm.com>
9995 David Sherwood <david.sherwood@arm.com>
9997 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
9998 parameter for the mode of "x". Remove the "known_x", "known_mode"
9999 and "known_ret" arguments. Change the type of the mode argument
10000 to scalar_int_mode.
10001 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
10002 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
10003 (reg_num_sign_bit_copies_for_combine): Likewise.
10004 * rtlanal.c (nonzero_bits1): Likewise.
10005 (num_sign_bit_copies1): Likewise.
10006 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
10007 (reg_num_sign_bit_copies_general): Likewise.
10008 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
10009 (reg_nonzero_bits_general): Likewise.
10011 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10012 Alan Hayward <alan.hayward@arm.com>
10013 David Sherwood <david.sherwood@arm.com>
10015 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
10016 than in subroutines. Return 1 for non-integer modes.
10017 (cached_num_sign_bit_copies): Change the type of the mode parameter
10018 to scalar_int_mode.
10019 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
10020 classes. Handle CONST_INT_P first and then check whether X also
10021 has a scalar integer mode. Check the same thing for inner registers
10022 of a SUBREG and for values that are being extended or truncated.
10024 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10025 Alan Hayward <alan.hayward@arm.com>
10026 David Sherwood <david.sherwood@arm.com>
10028 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
10029 in subroutines. Return the mode mask for non-integer modes.
10030 (cached_nonzero_bits): Change the type of the mode parameter
10031 to scalar_int_mode.
10032 (nonzero_bits1): Likewise. Remove early exit for other mode
10033 classes. Handle CONST_INT_P first and then check whether X
10034 also has a scalar integer mode.
10036 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10037 Alan Hayward <alan.hayward@arm.com>
10038 David Sherwood <david.sherwood@arm.com>
10040 * expr.c (widest_int_mode_for_size): Make the comment match the code.
10041 Return a scalar_int_mode and assert that the size is greater than
10043 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
10044 (op_by_pieces_d::op_by_pieces_d): Likewise.
10045 (op_by_pieces_d::run): Likewise.
10046 (can_store_by_pieces): Likewise.
10048 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10049 Alan Hayward <alan.hayward@arm.com>
10050 David Sherwood <david.sherwood@arm.com>
10052 * combine.c (extract_left_shift): Add a mode argument and update
10054 (make_compound_operation_int): Change the type of the mode parameter
10055 to scalar_int_mode and update the call to extract_left_shift.
10057 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10058 Alan Hayward <alan.hayward@arm.com>
10059 David Sherwood <david.sherwood@arm.com>
10061 * combine.c (simplify_and_const_int): Change the type of the mode
10062 parameter to scalar_int_mode.
10063 (simplify_and_const_int_1): Likewise. Update recursive call.
10065 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10066 Alan Hayward <alan.hayward@arm.com>
10067 David Sherwood <david.sherwood@arm.com>
10069 * combine.c (simplify_compare_const): Check that the mode is a
10070 scalar_int_mode (rather than VOIDmode) before testing its
10072 (simplify_comparison): Move COMPARISON_P handling out of the
10073 loop and restrict the latter part of the loop to scalar_int_modes.
10074 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
10075 and when considering SUBREG_REGs. Use is_int_mode instead of
10076 checking GET_MODE_CLASS against MODE_INT.
10078 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10079 Alan Hayward <alan.hayward@arm.com>
10080 David Sherwood <david.sherwood@arm.com>
10082 * combine.c (try_widen_shift_mode): Move check for equal modes to...
10083 (simplify_shift_const_1): ...here. Use scalar_int_mode for
10084 shift_unit_mode and for modes involved in scalar shifts.
10086 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10087 Alan Hayward <alan.hayward@arm.com>
10088 David Sherwood <david.sherwood@arm.com>
10090 * combine.c (force_int_to_mode): New function, split out from...
10091 (force_to_mode): ...here. Keep xmode up-to-date and use it
10092 instead of GET_MODE (x).
10094 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10095 Alan Hayward <alan.hayward@arm.com>
10096 David Sherwood <david.sherwood@arm.com>
10098 * optabs-query.h (extraction_insn::struct_mode): Change type to
10099 opt_scalar_int_mode and update comment.
10100 (extraction_insn::field_mode): Change type to scalar_int_mode.
10101 (extraction_insn::pos_mode): Likewise.
10102 * combine.c (make_extraction): Update accordingly.
10103 * optabs-query.c (get_traditional_extraction_insn): Likewise.
10104 (get_optab_extraction_insn): Likewise.
10105 * recog.c (simplify_while_replacing): Likewise.
10106 * expmed.c (narrow_bit_field_mem): Change the type of the mode
10107 parameter to opt_scalar_int_mode.
10109 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10110 Alan Hayward <alan.hayward@arm.com>
10111 David Sherwood <david.sherwood@arm.com>
10113 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
10114 to a scalar_int_mode instead of a machine_mode.
10115 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
10116 (get_best_mode): Return a boolean and use a pointer argument to store
10117 the selected mode. Replace the limit mode parameter with a bit limit.
10118 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
10119 for the values returned by bit_field_mode_iterator::next_mode.
10120 (store_bit_field): Update call to get_best_mode.
10121 (store_fixed_bit_field): Likewise.
10122 (extract_fixed_bit_field): Likewise.
10123 * expr.c (optimize_bitfield_assignment_op): Likewise.
10124 * fold-const.c (optimize_bit_field_compare): Likewise.
10125 (fold_truth_andor_1): Likewise.
10126 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
10127 Update for new type of m_mode.
10128 (get_best_mode): As above.
10130 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10131 Alan Hayward <alan.hayward@arm.com>
10132 David Sherwood <david.sherwood@arm.com>
10134 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
10135 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
10136 (store_bit_field): Check is_a <scalar_int_mode> before calling
10137 strict_volatile_bitfield_p.
10138 (extract_bit_field): Likewise.
10140 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10141 Alan Hayward <alan.hayward@arm.com>
10142 David Sherwood <david.sherwood@arm.com>
10144 * target.def (cstore_mode): Return a scalar_int_mode.
10145 * doc/tm.texi: Regenerate.
10146 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
10147 * targhooks.h (default_cstore_mode): Likewise.
10148 * targhooks.c (default_cstore_mode): Likewise, using a forced
10150 * expmed.c (emit_cstore): Expect the target of the cstore to be
10153 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10154 Alan Hayward <alan.hayward@arm.com>
10155 David Sherwood <david.sherwood@arm.com>
10157 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
10159 (niter_desc): Likewise mode.
10160 (iv_analyze): Add a mode parameter.
10162 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
10163 and change its type to scalar_int_mode.
10164 * loop-iv.c: Update commentary at head of file.
10165 (iv_constant): Pass the mode paraeter before the rtx it describes
10166 and change its type to scalar_int_mode. Remove VOIDmode handling.
10167 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
10168 (iv_extend): Likewise.
10169 (shorten_into_mode): Likewise.
10170 (iv_add): Use scalar_int_mode.
10171 (iv_mult): Likewise.
10172 (iv_shift): Likewise.
10173 (canonicalize_iv_subregs): Likewise.
10174 (get_biv_step_1): Pass the outer_mode parameter before the rtx
10175 it describes and change its mode to scalar_int_mode. Also change
10176 the type of the returned inner_mode to scalar_int_mode.
10177 (get_biv_step): Likewise, turning outer_mode from a pointer
10178 into a direct parameter. Update call to get_biv_step_1.
10179 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
10180 iv_constant and get_biv_step.
10181 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
10182 and change its type to scalar_int_mode. Don't initialise iv->mode
10183 to VOIDmode and remove later checks for its still being VOIDmode.
10184 Update calls to iv_analyze_op and iv_analyze_expr. Check
10185 is_a <scalar_int_mode> when changing the mode under consideration.
10186 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
10187 Update call to iv_analyze_expr.
10188 (iv_analyze_op): Add a mode parameter. Reject subregs whose
10189 inner register is not also a scalar_int_mode. Update call to
10191 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
10192 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
10193 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
10194 separate mode class checks. Update calls to iv_analyze. Remove
10195 fix-up of VOIDmodes after iv_analyze_biv.
10196 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
10197 don't have a scalar_int_mode. Update call to biv_p.
10199 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10200 Alan Hayward <alan.hayward@arm.com>
10201 David Sherwood <david.sherwood@arm.com>
10203 * cfgexpand.c (convert_debug_memory_address): Use
10204 as_a <scalar_int_mode>.
10205 * combine.c (expand_compound_operation): Likewise.
10206 (make_extraction): Likewise.
10207 (change_zero_ext): Likewise.
10208 (simplify_comparison): Likewise.
10209 * cse.c (cse_insn): Likewise.
10210 * dwarf2out.c (minmax_loc_descriptor): Likewise.
10211 (mem_loc_descriptor): Likewise.
10212 (loc_descriptor): Likewise.
10213 * expmed.c (init_expmed_one_mode): Likewise.
10214 (synth_mult): Likewise.
10215 (emit_store_flag_1): Likewise.
10216 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
10217 of a comparison with size.
10218 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
10219 (reduce_to_bit_field_precision): Likewise.
10220 * function.c (expand_function_end): Likewise.
10221 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
10222 * loop-doloop.c (doloop_modify): Likewise.
10223 * optabs.c (expand_binop): Likewise.
10224 (expand_unop): Likewise.
10225 (expand_copysign_absneg): Likewise.
10226 (prepare_cmp_insn): Likewise.
10227 (maybe_legitimize_operand): Likewise.
10228 * recog.c (const_scalar_int_operand): Likewise.
10229 * rtlanal.c (get_address_mode): Likewise.
10230 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10231 (simplify_cond_clz_ctz): Likewise.
10232 * tree-nested.c (get_nl_goto_field): Likewise.
10233 * tree.c (build_vector_type_for_mode): Likewise.
10234 * var-tracking.c (use_narrower_mode): Likewise.
10236 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10237 Alan Hayward <alan.hayward@arm.com>
10238 David Sherwood <david.sherwood@arm.com>
10240 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
10241 * builtins.c (expand_builtin_signbit): Use it.
10242 * cfgexpand.c (expand_debug_expr): Likewise.
10243 * dojump.c (do_jump): Likewise.
10244 (do_compare_and_jump): Likewise.
10245 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
10246 * expmed.c (make_tree): Likewise.
10247 * expr.c (expand_expr_real_2): Likewise.
10248 (expand_expr_real_1): Likewise.
10249 (try_casesi): Likewise.
10250 * fold-const-call.c (fold_const_call_ss): Likewise.
10251 * fold-const.c (unextend): Likewise.
10252 (extract_muldiv_1): Likewise.
10253 (fold_single_bit_test): Likewise.
10254 (native_encode_int): Likewise.
10255 (native_encode_string): Likewise.
10256 (native_interpret_int): Likewise.
10257 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
10258 * internal-fn.c (expand_addsub_overflow): Likewise.
10259 (expand_neg_overflow): Likewise.
10260 (expand_mul_overflow): Likewise.
10261 (expand_arith_overflow): Likewise.
10262 * match.pd: Likewise.
10263 * stor-layout.c (layout_type): Likewise.
10264 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10265 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
10266 * tree-ssanames.c (get_range_info): Likewise.
10267 * tree-switch-conversion.c (array_value_type) Likewise.
10268 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
10269 (vect_recog_divmod_pattern): Likewise.
10270 (vect_recog_mixed_size_cond_pattern): Likewise.
10271 * tree-vrp.c (extract_range_basic): Likewise.
10272 (simplify_float_conversion_using_ranges): Likewise.
10273 * tree.c (int_fits_type_p): Likewise.
10274 * ubsan.c (instrument_bool_enum_load): Likewise.
10275 * varasm.c (mergeable_string_section): Likewise.
10276 (narrowing_initializer_constant_valid_p): Likewise.
10277 (output_constant): Likewise.
10279 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10280 Alan Hayward <alan.hayward@arm.com>
10281 David Sherwood <david.sherwood@arm.com>
10283 * machmode.h (NARROWEST_INT_MODE): New macro.
10284 * expr.c (alignment_for_piecewise_move): Use it instead of
10285 GET_CLASS_NARROWEST_MODE (MODE_INT).
10286 (push_block): Likewise.
10287 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
10289 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10291 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10292 Alan Hayward <alan.hayward@arm.com>
10293 David Sherwood <david.sherwood@arm.com>
10295 * postreload.c (move2add_valid_value_p): Change the type of the
10296 mode parameter to scalar_int_mode.
10297 (move2add_use_add2_insn): Add a mode parameter and use it instead
10299 (move2add_use_add3_insn): Likewise.
10300 (reload_cse_move2add): Update accordingly.
10302 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10303 Alan Hayward <alan.hayward@arm.com>
10304 David Sherwood <david.sherwood@arm.com>
10306 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
10308 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
10309 * optabs.c (expand_unop): Likewise.
10311 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10312 Alan Hayward <alan.hayward@arm.com>
10313 David Sherwood <david.sherwood@arm.com>
10315 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
10316 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
10317 (popcount_loc_descriptor): Likewise.
10318 (bswap_loc_descriptor): Likewise.
10319 (rotate_loc_descriptor): Likewise.
10320 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
10321 calling the functions above.
10323 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10324 Alan Hayward <alan.hayward@arm.com>
10325 David Sherwood <david.sherwood@arm.com>
10327 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
10329 (try_combine): Likewise.
10330 (simplify_if_then_else): Likewise.
10331 * cse.c (cse_insn): Likewise.
10332 * dwarf2out.c (mem_loc_descriptor): Likewise.
10333 * emit-rtl.c (gen_lowpart_common): Likewise.
10334 * simplify-rtx.c (simplify_truncation): Likewise.
10335 (simplify_binary_operation_1): Likewise.
10336 (simplify_const_relational_operation): Likewise.
10337 (simplify_ternary_operation): Likewise.
10338 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
10340 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10341 Alan Hayward <alan.hayward@arm.com>
10342 David Sherwood <david.sherwood@arm.com>
10344 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
10345 * reload.c (push_reload): Likewise.
10346 (find_reloads): Likewise.
10348 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10349 Alan Hayward <alan.hayward@arm.com>
10350 David Sherwood <david.sherwood@arm.com>
10352 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
10353 (make_compound_operation_int): Likewise.
10354 (change_zero_ext): Likewise.
10355 * expr.c (convert_move): Likewise.
10356 (convert_modes): Likewise.
10357 * fwprop.c (forward_propagate_subreg): Likewise.
10358 * loop-iv.c (get_biv_step_1): Likewise.
10359 * optabs.c (widen_operand): Likewise.
10360 * postreload.c (move2add_valid_value_p): Likewise.
10361 * recog.c (simplify_while_replacing): Likewise.
10362 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10363 (simplify_binary_operation_1): Likewise. Remove redundant
10364 mode equality check.
10366 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10367 Alan Hayward <alan.hayward@arm.com>
10368 David Sherwood <david.sherwood@arm.com>
10370 * combine.c (combine_simplify_rtx): Add checks for
10371 is_a <scalar_int_mode>.
10372 (simplify_if_then_else): Likewise.
10373 (make_field_assignment): Likewise.
10374 (simplify_comparison): Likewise.
10375 * ifcvt.c (noce_try_bitop): Likewise.
10376 * loop-invariant.c (canonicalize_address_mult): Likewise.
10377 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10379 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10380 Alan Hayward <alan.hayward@arm.com>
10381 David Sherwood <david.sherwood@arm.com>
10383 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
10384 is_a <scalar_int_mode> instead of != BLKmode.
10386 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10387 Alan Hayward <alan.hayward@arm.com>
10388 David Sherwood <david.sherwood@arm.com>
10390 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
10391 instead of != VOIDmode.
10392 * combine.c (if_then_else_cond): Likewise.
10393 (change_zero_ext): Likewise.
10394 * dwarf2out.c (mem_loc_descriptor): Likewise.
10395 (loc_descriptor): Likewise.
10396 * rtlanal.c (canonicalize_condition): Likewise.
10397 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
10399 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10400 Alan Hayward <alan.hayward@arm.com>
10401 David Sherwood <david.sherwood@arm.com>
10403 * simplify-rtx.c (simplify_binary_operation_1): Use
10404 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
10406 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10407 Alan Hayward <alan.hayward@arm.com>
10408 David Sherwood <david.sherwood@arm.com>
10410 * wide-int.h (int_traits<unsigned char>) New class.
10411 (int_traits<unsigned short>) Likewise.
10412 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
10413 Use GET_MODE_UNIT_PRECISION and remove redundant test for
10415 * combine.c (set_nonzero_bits_and_sign_copies): Use
10416 is_a <scalar_int_mode>.
10417 (find_split_point): Likewise.
10418 (combine_simplify_rtx): Likewise.
10419 (simplify_logical): Likewise.
10420 (expand_compound_operation): Likewise.
10421 (expand_field_assignment): Likewise.
10422 (make_compound_operation): Likewise.
10423 (extended_count): Likewise.
10424 (change_zero_ext): Likewise.
10425 (simplify_comparison): Likewise.
10426 * dwarf2out.c (scompare_loc_descriptor): Likewise.
10427 (ucompare_loc_descriptor): Likewise.
10428 (minmax_loc_descriptor): Likewise.
10429 (mem_loc_descriptor): Likewise.
10430 (loc_descriptor): Likewise.
10431 * expmed.c (init_expmed_one_mode): Likewise.
10432 * lra-constraints.c (lra_constraint_offset): Likewise.
10433 * optabs.c (prepare_libcall_arg): Likewise.
10434 * postreload.c (move2add_note_store): Likewise.
10435 * reload.c (operands_match_p): Likewise.
10436 * rtl.h (load_extend_op): Likewise.
10437 * rtlhooks.c (gen_lowpart_general): Likewise.
10438 * simplify-rtx.c (simplify_truncation): Likewise.
10439 (simplify_unary_operation_1): Likewise.
10440 (simplify_binary_operation_1): Likewise.
10441 (simplify_const_binary_operation): Likewise.
10442 (simplify_const_relational_operation): Likewise.
10443 (simplify_subreg): Likewise.
10444 * stor-layout.c (bitwise_mode_for_mode): Likewise.
10445 * var-tracking.c (adjust_mems): Likewise.
10446 (prepare_call_arguments): Likewise.
10448 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10449 Alan Hayward <alan.hayward@arm.com>
10450 David Sherwood <david.sherwood@arm.com>
10452 * machmode.h (is_int_mode): New fuction.
10453 * combine.c (find_split_point): Use it.
10454 (combine_simplify_rtx): Likewise.
10455 (simplify_if_then_else): Likewise.
10456 (simplify_set): Likewise.
10457 (simplify_shift_const_1): Likewise.
10458 (simplify_comparison): Likewise.
10459 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
10460 * cse.c (notreg_cost): Likewise.
10461 (cse_insn): Likewise.
10462 * cselib.c (cselib_lookup_1): Likewise.
10463 * dojump.c (do_jump_1): Likewise.
10464 (do_compare_rtx_and_jump): Likewise.
10465 * dse.c (get_call_args): Likewise.
10466 * dwarf2out.c (rtl_for_decl_init): Likewise.
10467 (native_encode_initializer): Likewise.
10468 * expmed.c (emit_store_flag_1): Likewise.
10469 (emit_store_flag): Likewise.
10470 * expr.c (convert_modes): Likewise.
10471 (store_field): Likewise.
10472 (expand_expr_real_1): Likewise.
10473 * fold-const.c (fold_read_from_constant_string): Likewise.
10474 * gimple-ssa-sprintf.c (get_format_string): Likewise.
10475 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
10476 * optabs.c (expand_binop): Likewise.
10477 (expand_unop): Likewise.
10478 (expand_abs_nojump): Likewise.
10479 (expand_one_cmpl_abs_nojump): Likewise.
10480 * simplify-rtx.c (mode_signbit_p): Likewise.
10481 (val_signbit_p): Likewise.
10482 (val_signbit_known_set_p): Likewise.
10483 (val_signbit_known_clear_p): Likewise.
10484 (simplify_relational_operation_1): Likewise.
10485 * tree.c (vector_type_mode): Likewise.
10487 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10488 Alan Hayward <alan.hayward@arm.com>
10489 David Sherwood <david.sherwood@arm.com>
10491 * machmode.h (smallest_mode_for_size): Fix formatting.
10492 (smallest_int_mode_for_size): New function.
10493 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
10494 instead of smallest_mode_for_size.
10495 * combine.c (make_extraction): Likewise.
10496 * config/arc/arc.c (arc_expand_movmem): Likewise.
10497 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
10498 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
10499 * config/s390/s390.c (s390_expand_insv): Likewise.
10500 * config/sparc/sparc.c (assign_int_registers): Likewise.
10501 * config/spu/spu.c (spu_function_value): Likewise.
10502 (spu_function_arg): Likewise.
10503 * coverage.c (get_gcov_type): Likewise.
10504 (get_gcov_unsigned_t): Likewise.
10505 * dse.c (find_shift_sequence): Likewise.
10506 * expmed.c (store_bit_field_1): Likewise.
10507 * expr.c (convert_move): Likewise.
10508 (store_field): Likewise.
10509 * internal-fn.c (expand_arith_overflow): Likewise.
10510 * optabs-query.c (get_best_extraction_insn): Likewise.
10511 * optabs.c (expand_twoval_binop_libfunc): Likewise.
10512 * stor-layout.c (layout_type): Likewise.
10513 (initialize_sizetypes): Likewise.
10514 * targhooks.c (default_get_mask_mode): Likewise.
10515 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
10517 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10518 Alan Hayward <alan.hayward@arm.com>
10519 David Sherwood <david.sherwood@arm.com>
10521 * machmode.h (opt_mode::else_blk): New function.
10522 (int_mode_for_mode): Declare.
10523 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
10524 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
10526 * cfgexpand.c (expand_debug_expr): Likewise.
10527 * combine.c (gen_lowpart_or_truncate): Likewise.
10528 (gen_lowpart_for_combine): Likewise.
10529 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
10530 * config/avr/avr.c (avr_to_int_mode): Likewise.
10531 (avr_out_plus_1): Likewise.
10532 (avr_out_plus): Likewise.
10533 (avr_out_round): Likewise.
10534 * config/i386/i386.c (ix86_split_to_parts): Likewise.
10535 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
10536 (s390_expand_vcond): Likewise.
10537 * config/spu/spu.c (spu_split_immediate): Likewise.
10538 (spu_expand_mov): Likewise.
10539 * dse.c (get_stored_val): Likewise.
10540 * expmed.c (store_bit_field_1): Likewise.
10541 (convert_extracted_bit_field): Use int_mode_for_mode instead of
10543 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
10544 (extract_low_bits): Likewise.
10545 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
10546 handling rather than repeating the check.
10547 (emit_group_store): Likewise.
10548 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
10549 * optabs.c (expand_absneg_bit): Likewise.
10550 (expand_copysign_absneg): Likewise.
10551 (expand_copysign_bit): Likewise.
10552 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
10553 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
10554 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
10555 * var-tracking.c (prepare_call_arguments): Likewise.
10556 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
10557 int_mode_for_mode instead of mode_for_size.
10558 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
10560 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10561 Alan Hayward <alan.hayward@arm.com>
10562 David Sherwood <david.sherwood@arm.com>
10564 * machmode.h (int_mode_for_size): New function.
10565 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
10566 instead of mode_for_size.
10567 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
10569 * combine.c (expand_field_assignment): Use int_mode_for_size
10570 instead of mode_for_size.
10571 (make_extraction): Likewise.
10572 (simplify_shift_const_1): Likewise.
10573 (simplify_comparison): Likewise.
10574 * dojump.c (do_jump): Likewise.
10575 * dwarf2out.c (mem_loc_descriptor): Likewise.
10576 * emit-rtl.c (init_derived_machine_modes): Likewise.
10577 * expmed.c (flip_storage_order): Likewise.
10578 (convert_extracted_bit_field): Likewise.
10579 * expr.c (copy_blkmode_from_reg): Likewise.
10580 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
10581 * internal-fn.c (expand_mul_overflow): Likewise.
10582 * lower-subreg.c (simple_move): Likewise.
10583 * optabs-libfuncs.c (init_optabs): Likewise.
10584 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10585 * tree.c (vector_type_mode): Likewise.
10586 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
10587 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
10588 * tree-vect-generic.c (expand_vector_parallel): Likewise.
10589 * tree-vect-stmts.c (vectorizable_load): Likewise.
10590 (vectorizable_store): Likewise.
10592 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10593 Alan Hayward <alan.hayward@arm.com>
10594 David Sherwood <david.sherwood@arm.com>
10596 * coretypes.h (pod_mode): New type.
10597 (scalar_int_mode_pod): New typedef.
10598 * machmode.h (pod_mode): New class.
10599 (int_n_data_t::m): Change type to scalar_int_mode_pod.
10600 * genmodes.c (emit_mode_int_n): Update accordingly.
10601 * lower-subreg.h (target_lower_subreg): Change type to
10602 scalar_int_mode_pod.
10603 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
10604 scalar_int_mode_pod.
10606 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10607 Alan Hayward <alan.hayward@arm.com>
10608 David Sherwood <david.sherwood@arm.com>
10610 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
10611 machine_mode to scalar_int_mode.
10612 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
10613 (rs6000_option_override_internal): Remove cast to int.
10614 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
10615 machine_mode to scalar_int_mode.
10616 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
10617 (rs6000_option_override_internal): Remove cast to int.
10618 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
10619 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
10621 * config/s390/s390.c (s390_expand_builtin): Likewise.
10622 * coretypes.h (scalar_int_mode): New type.
10623 (opt_scalar_int_mode): New typedef.
10624 * machmode.h (scalar_int_mode): New class.
10625 (scalar_int_mode::includes_p): New function.
10626 (byte_mode): Change type to scalar_int_mode.
10627 (word_mode): Likewise.
10628 (ptr_mode): Likewise.
10629 * emit-rtl.c (byte_mode): Likewise.
10630 (word_mode): Likewise.
10631 (ptr_mode): Likewise.
10632 (init_derived_machine_modes): Update accordingly.
10633 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
10634 and MODE_PARTIAL_INT.
10635 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
10636 opt_scalar_int_mode.
10638 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10639 Alan Hayward <alan.hayward@arm.com>
10640 David Sherwood <david.sherwood@arm.com>
10642 * target.def (libgcc_floating_mode_supported_p): Take a
10644 * doc/tm.texi: Regenerate.
10645 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
10647 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
10648 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
10651 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10652 Alan Hayward <alan.hayward@arm.com>
10653 David Sherwood <david.sherwood@arm.com>
10655 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
10656 * doc/tm.texi: Regenerate.
10657 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
10658 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
10659 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
10660 * targhooks.h (default_floatn_mode): Likewise.
10661 * targhooks.c (default_floatn_mode): Likewise.
10662 * tree.c (build_common_tree_nodes): Update accordingly.
10664 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10665 Alan Hayward <alan.hayward@arm.com>
10666 David Sherwood <david.sherwood@arm.com>
10668 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
10669 (mode_iterator::iterate_p): Likewise.
10670 (mode_iterator::get_wider): Likewise.
10671 * expr.c (init_expr_target): Use opt_scalar_float_mode.
10673 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10674 Alan Hayward <alan.hayward@arm.com>
10675 David Sherwood <david.sherwood@arm.com>
10677 * coretypes.h (opt_scalar_float_mode): New typedef.
10678 * machmode.h (float_mode_for_size): New function.
10679 * emit-rtl.c (double_mode): Delete.
10680 (init_emit_once): Use float_mode_for_size.
10681 * stor-layout.c (layout_type): Likewise.
10682 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
10684 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10685 Alan Hayward <alan.hayward@arm.com>
10686 David Sherwood <david.sherwood@arm.com>
10688 * output.h (assemble_real): Take a scalar_float_mode.
10689 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
10690 * config/arm/arm.md (consttable_4): Likewise.
10691 (consttable_8): Likewise.
10692 (consttable_16): Likewise.
10693 * config/mips/mips.md (consttable_float): Likewise.
10694 * config/s390/s390.c (s390_output_pool_entry): Likewise.
10695 * varasm.c (assemble_real): Take a scalar_float_mode.
10696 (output_constant_pool_2): Update accordingly.
10697 (output_constant): Likewise.
10699 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10700 Alan Hayward <alan.hayward@arm.com>
10701 David Sherwood <david.sherwood@arm.com>
10703 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
10704 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
10705 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
10706 (native_encode_real): Likewise.
10707 (native_interpret_real): Likewise.
10708 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
10709 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10711 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10712 Alan Hayward <alan.hayward@arm.com>
10713 David Sherwood <david.sherwood@arm.com>
10715 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
10716 <scalar_float_mode>. Simplify.
10717 (gen_extend_conv_libfunc): Likewise.
10719 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10720 Alan Hayward <alan.hayward@arm.com>
10721 David Sherwood <david.sherwood@arm.com>
10723 * coretypes.h (scalar_float_mode): New type.
10724 * machmode.h (mode_traits::from_int): Use machine_mode if
10725 USE_ENUM_MODES is defined.
10726 (is_a): New function.
10728 (dyn_cast): Likewise.
10729 (scalar_float_mode): New class.
10730 (scalar_float_mode::includes_p): New function.
10731 (is_float_mode): Likewise.
10732 * gdbhooks.py (MachineModePrinter): New class.
10733 (build_pretty_printer): Use it for scalar_float_mode.
10734 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
10735 (format_helper::format_helper): Turn into a template.
10736 * genmodes.c (get_mode_class): New function.
10737 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
10738 or machine_mode if none.
10739 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
10740 as_a <scalar_float_mode>.
10741 * dwarf2out.c (mem_loc_descriptor): Likewise.
10742 (insert_float): Likewise.
10743 (add_const_value_attribute): Likewise.
10744 * simplify-rtx.c (simplify_immed_subreg): Likewise.
10745 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
10746 (expand_unop): Update accordingly.
10747 (expand_abs_nojump): Likewise.
10748 (expand_copysign_absneg): Take a scalar_float_mode.
10749 (expand_copysign_bit): Likewise.
10750 (expand_copysign): Update accordingly.
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 * coretypes.h (opt_mode): New class.
10757 * machmode.h (opt_mode): Likewise.
10758 (opt_mode::else_void): New function.
10759 (opt_mode::require): Likewise.
10760 (opt_mode::exists): Likewise.
10761 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
10762 (GET_MODE_2XWIDER_MODE): Likewise.
10763 (mode_iterator::get_wider): Update accordingly.
10764 (mode_iterator::get_2xwider): Likewise.
10765 (mode_iterator::get_known_wider): Likewise, turning into a template.
10766 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
10767 forcing a wider mode to exist.
10768 * config/cr16/cr16.h (LONG_REG_P): Likewise.
10769 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
10770 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
10771 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
10772 * lower-subreg.c (init_lower_subreg): Likewise.
10773 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
10774 on the final iteration.
10775 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
10776 a wider mode exists before asking for a move pattern.
10777 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
10778 forcing a wider mode to exist.
10779 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
10780 returning false if no such mode exists.
10781 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
10782 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
10783 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
10784 Avoid checking for a MODE_INT if we already know the mode is not a
10786 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
10787 forcing a wider mode to exist.
10788 (expmed_mult_highpart_optab): Likewise.
10789 (expmed_mult_highpart): Likewise.
10790 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
10792 * lto-streamer-in.c (lto_input_mode_table): Likewise.
10793 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
10794 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
10795 * internal-fn.c (expand_mul_overflow): Update use of
10796 GET_MODE_2XWIDER_MODE.
10797 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10798 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
10799 GET_MODE_WIDER_MODE.
10800 (convert_plusminus_to_widen): Likewise.
10801 * tree-switch-conversion.c (array_value_type): Likewise.
10802 * var-tracking.c (emit_note_insn_var_location): Likewise.
10803 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10804 Return false inside rather than outside the loop if no wider mode
10806 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
10807 and GET_MODE_2XWIDER_MODE
10808 (can_compare_p): Use else_void.
10809 * gdbhooks.py (OptMachineModePrinter): New class.
10810 (build_pretty_printer): Use it for opt_mode.
10812 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10813 Alan Hayward <alan.hayward@arm.com>
10814 David Sherwood <david.sherwood@arm.com>
10816 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
10817 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
10819 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10820 Alan Hayward <alan.hayward@arm.com>
10821 David Sherwood <david.sherwood@arm.com>
10823 * machmode.h (mode_traits): New structure.
10824 (get_narrowest_mode): New function.
10825 (mode_iterator::start): Likewise.
10826 (mode_iterator::iterate_p): Likewise.
10827 (mode_iterator::get_wider): Likewise.
10828 (mode_iterator::get_known_wider): Likewise.
10829 (mode_iterator::get_2xwider): Likewise.
10830 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
10831 (FOR_EACH_MODE): Likewise.
10832 (FOR_EACH_MODE_FROM): Likewise.
10833 (FOR_EACH_MODE_UNTIL): Likewise.
10834 (FOR_EACH_WIDER_MODE): Likewise.
10835 (FOR_EACH_2XWIDER_MODE): Likewise.
10836 * builtins.c (expand_builtin_strlen): Use new mode iterators.
10837 * combine.c (simplify_comparison): Likewise
10838 * config/i386/i386.c (type_natural_mode): Likewise.
10839 * cse.c (cse_insn): Likewise.
10840 * dse.c (find_shift_sequence): Likewise.
10841 * emit-rtl.c (init_derived_machine_modes): Likewise.
10842 (init_emit_once): Likewise.
10843 * explow.c (hard_function_value): Likewise.
10844 * expmed.c (extract_fixed_bit_field_1): Likewise.
10845 (extract_bit_field_1): Likewise.
10846 (expand_divmod): Likewise.
10847 (emit_store_flag_1): Likewise.
10848 * expr.c (init_expr_target): Likewise.
10849 (convert_move): Likewise.
10850 (alignment_for_piecewise_move): Likewise.
10851 (widest_int_mode_for_size): Likewise.
10852 (emit_block_move_via_movmem): Likewise.
10853 (copy_blkmode_to_reg): Likewise.
10854 (set_storage_via_setmem): Likewise.
10855 (compress_float_constant): Likewise.
10856 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10857 * optabs-query.c (get_best_extraction_insn): Likewise.
10858 * optabs.c (expand_binop): Likewise.
10859 (expand_twoval_unop): Likewise.
10860 (expand_twoval_binop): Likewise.
10861 (widen_leading): Likewise.
10862 (widen_bswap): Likewise.
10863 (expand_parity): Likewise.
10864 (expand_unop): Likewise.
10865 (prepare_cmp_insn): Likewise.
10866 (prepare_float_lib_cmp): Likewise.
10867 (expand_float): Likewise.
10868 (expand_fix): Likewise.
10869 (expand_sfix_optab): Likewise.
10870 * postreload.c (move2add_use_add2_insn): Likewise.
10871 * reg-stack.c (reg_to_stack): Likewise.
10872 * reginfo.c (choose_hard_reg_mode): Likewise.
10873 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
10874 * stor-layout.c (mode_for_size): Likewise.
10875 (smallest_mode_for_size): Likewise.
10876 (mode_for_vector): Likewise.
10877 (finish_bitfield_representative): Likewise.
10878 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
10879 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
10880 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10881 * var-tracking.c (prepare_call_arguments): Likewise.
10883 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10884 Alan Hayward <alan.hayward@arm.com>
10885 David Sherwood <david.sherwood@arm.com>
10887 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
10888 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
10889 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
10890 * machmode.h (mode_size): Move earlier in file.
10891 (mode_precision): Likewise.
10892 (mode_inner): Likewise.
10893 (mode_nunits): Likewise.
10894 (mode_unit_size): Likewise.
10895 (unit_unit_precision): Likewise.
10896 (mode_wider): Likewise.
10897 (mode_2xwider): Likewise.
10898 (machine_mode): New class.
10899 (mode_to_bytes): New function.
10900 (mode_to_bits): Likewise.
10901 (mode_to_precision): Likewise.
10902 (mode_to_inner): Likewise.
10903 (mode_to_unit_size): Likewise.
10904 (mode_to_unit_precision): Likewise.
10905 (mode_to_nunits): Likewise.
10906 (GET_MODE_SIZE): Use mode_to_bytes.
10907 (GET_MODE_BITSIZE): Use mode_to_bits.
10908 (GET_MODE_PRECISION): Use mode_to_precision.
10909 (GET_MODE_INNER): Use mode_to_inner.
10910 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
10911 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
10912 (GET_MODE_NUNITS): Use mode_to_nunits.
10913 * system.h (ALWAYS_INLINE): New macro.
10914 * config/powerpcspe/powerpcspe-c.c
10915 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
10916 int for arg1_mode and arg2_mode.
10918 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10919 Alan Hayward <alan.hayward@arm.com>
10920 David Sherwood <david.sherwood@arm.com>
10922 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
10923 Prefix mode names with E_ in case statements.
10924 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10925 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
10926 (aarch64_split_simd_move): Likewise.
10927 (aarch64_gen_storewb_pair): Likewise.
10928 (aarch64_gen_loadwb_pair): Likewise.
10929 (aarch64_gen_store_pair): Likewise.
10930 (aarch64_gen_load_pair): Likewise.
10931 (aarch64_get_condition_code_1): Likewise.
10932 (aarch64_constant_pool_reload_icode): Likewise.
10933 (get_rsqrte_type): Likewise.
10934 (get_rsqrts_type): Likewise.
10935 (get_recpe_type): Likewise.
10936 (get_recps_type): Likewise.
10937 (aarch64_gimplify_va_arg_expr): Likewise.
10938 (aarch64_simd_container_mode): Likewise.
10939 (aarch64_emit_load_exclusive): Likewise.
10940 (aarch64_emit_store_exclusive): Likewise.
10941 (aarch64_expand_compare_and_swap): Likewise.
10942 (aarch64_gen_atomic_cas): Likewise.
10943 (aarch64_emit_bic): Likewise.
10944 (aarch64_emit_atomic_swap): Likewise.
10945 (aarch64_emit_atomic_load_op): Likewise.
10946 (aarch64_evpc_trn): Likewise.
10947 (aarch64_evpc_uzp): Likewise.
10948 (aarch64_evpc_zip): Likewise.
10949 (aarch64_evpc_ext): Likewise.
10950 (aarch64_evpc_rev): Likewise.
10951 (aarch64_evpc_dup): Likewise.
10952 (aarch64_gen_ccmp_first): Likewise.
10953 (aarch64_gen_ccmp_next): Likewise.
10954 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
10955 (alpha_emit_xfloating_libcall): Likewise.
10956 (emit_insxl): Likewise.
10957 (alpha_arg_type): Likewise.
10958 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
10959 (arc_preferred_simd_mode): Likewise.
10960 (arc_secondary_reload): Likewise.
10961 (get_arc_condition_code): Likewise.
10962 (arc_print_operand): Likewise.
10963 (arc_legitimate_constant_p): Likewise.
10964 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10965 * config/arc/arc.md (casesi_load): Likewise.
10966 (casesi_compact_jump): Likewise.
10967 * config/arc/predicates.md (proper_comparison_operator): Likewise.
10968 (cc_use_register): Likewise.
10969 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10970 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
10971 (arm_init_iwmmxt_builtins): Likewise.
10972 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
10973 (neon_expand_vector_init): Likewise.
10974 (arm_attr_length_move_neon): Likewise.
10975 (maybe_get_arm_condition_code): Likewise.
10976 (arm_emit_vector_const): Likewise.
10977 (arm_preferred_simd_mode): Likewise.
10978 (arm_output_iwmmxt_tinsr): Likewise.
10979 (thumb1_output_casesi): Likewise.
10980 (thumb2_output_casesi): Likewise.
10981 (arm_emit_load_exclusive): Likewise.
10982 (arm_emit_store_exclusive): Likewise.
10983 (arm_expand_compare_and_swap): Likewise.
10984 (arm_evpc_neon_vuzp): Likewise.
10985 (arm_evpc_neon_vzip): Likewise.
10986 (arm_evpc_neon_vrev): Likewise.
10987 (arm_evpc_neon_vtrn): Likewise.
10988 (arm_evpc_neon_vext): Likewise.
10989 (arm_validize_comparison): Likewise.
10990 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
10991 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
10992 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
10993 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
10994 (c6x_preferred_simd_mode): Likewise.
10995 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
10996 (epiphany_rtx_costs): Likewise.
10997 * config/epiphany/predicates.md (proper_comparison_operator):
10999 * config/frv/frv.c (condexec_memory_operand): Likewise.
11000 (frv_emit_move): Likewise.
11001 (output_move_single): Likewise.
11002 (output_condmove_single): Likewise.
11003 (frv_hard_regno_mode_ok): Likewise.
11004 (frv_matching_accg_mode): Likewise.
11005 * config/h8300/h8300.c (split_adds_subs): Likewise.
11006 (h8300_rtx_costs): Likewise.
11007 (h8300_print_operand): Likewise.
11008 (compute_mov_length): Likewise.
11009 (output_logical_op): Likewise.
11010 (compute_logical_op_length): Likewise.
11011 (compute_logical_op_cc): Likewise.
11012 (h8300_shift_needs_scratch_p): Likewise.
11013 (output_a_shift): Likewise.
11014 (compute_a_shift_length): Likewise.
11015 (compute_a_shift_cc): Likewise.
11016 (expand_a_rotate): Likewise.
11017 (output_a_rotate): Likewise.
11018 * config/i386/i386.c (classify_argument): Likewise.
11019 (function_arg_advance_32): Likewise.
11020 (function_arg_32): Likewise.
11021 (function_arg_64): Likewise.
11022 (function_value_64): Likewise.
11023 (ix86_gimplify_va_arg): Likewise.
11024 (ix86_legitimate_constant_p): Likewise.
11025 (put_condition_code): Likewise.
11026 (split_double_mode): Likewise.
11027 (ix86_avx256_split_vector_move_misalign): Likewise.
11028 (ix86_expand_vector_logical_operator): Likewise.
11029 (ix86_split_idivmod): Likewise.
11030 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
11031 (ix86_build_const_vector): Likewise.
11032 (ix86_build_signbit_mask): Likewise.
11033 (ix86_match_ccmode): Likewise.
11034 (ix86_cc_modes_compatible): Likewise.
11035 (ix86_expand_branch): Likewise.
11036 (ix86_expand_sse_cmp): Likewise.
11037 (ix86_expand_sse_movcc): Likewise.
11038 (ix86_expand_int_sse_cmp): Likewise.
11039 (ix86_expand_vec_perm_vpermi2): Likewise.
11040 (ix86_expand_vec_perm): Likewise.
11041 (ix86_expand_sse_unpack): Likewise.
11042 (ix86_expand_int_addcc): Likewise.
11043 (ix86_split_to_parts): Likewise.
11044 (ix86_vectorize_builtin_gather): Likewise.
11045 (ix86_vectorize_builtin_scatter): Likewise.
11046 (avx_vpermilp_parallel): Likewise.
11047 (inline_memory_move_cost): Likewise.
11048 (ix86_tieable_integer_mode_p): Likewise.
11049 (x86_maybe_negate_const_int): Likewise.
11050 (ix86_expand_vector_init_duplicate): Likewise.
11051 (ix86_expand_vector_init_one_nonzero): Likewise.
11052 (ix86_expand_vector_init_one_var): Likewise.
11053 (ix86_expand_vector_init_concat): Likewise.
11054 (ix86_expand_vector_init_interleave): Likewise.
11055 (ix86_expand_vector_init_general): Likewise.
11056 (ix86_expand_vector_set): Likewise.
11057 (ix86_expand_vector_extract): Likewise.
11058 (emit_reduc_half): Likewise.
11059 (ix86_emit_i387_round): Likewise.
11060 (ix86_mangle_type): Likewise.
11061 (ix86_expand_round_sse4): Likewise.
11062 (expand_vec_perm_blend): Likewise.
11063 (canonicalize_vector_int_perm): Likewise.
11064 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
11065 (expand_vec_perm_1): Likewise.
11066 (expand_vec_perm_interleave3): Likewise.
11067 (expand_vec_perm_even_odd_pack): Likewise.
11068 (expand_vec_perm_even_odd_1): Likewise.
11069 (expand_vec_perm_broadcast_1): Likewise.
11070 (ix86_vectorize_vec_perm_const_ok): Likewise.
11071 (ix86_expand_vecop_qihi): Likewise.
11072 (ix86_expand_mul_widen_hilo): Likewise.
11073 (ix86_expand_sse2_abs): Likewise.
11074 (ix86_expand_pextr): Likewise.
11075 (ix86_expand_pinsr): Likewise.
11076 (ix86_preferred_simd_mode): Likewise.
11077 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
11078 * config/i386/sse.md (*andnot<mode>3): Likewise.
11079 (<mask_codefor><code><mode>3<mask_name>): Likewise.
11080 (*<code><mode>3): Likewise.
11081 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
11082 (ia64_expand_atomic_op): Likewise.
11083 (ia64_arg_type): Likewise.
11084 (ia64_mode_to_int): Likewise.
11085 (ia64_scalar_mode_supported_p): Likewise.
11086 (ia64_vector_mode_supported_p): Likewise.
11087 (expand_vec_perm_broadcast): Likewise.
11088 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
11089 (iq2000_function_arg_advance): Likewise.
11090 (iq2000_function_arg): Likewise.
11091 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
11092 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
11093 (m68k_libcall_value): Likewise.
11094 (m68k_function_value): Likewise.
11095 (sched_attr_op_type): Likewise.
11096 * config/mcore/mcore.c (mcore_output_move): Likewise.
11097 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
11099 (microblaze_function_arg): Likewise.
11100 * config/mips/mips.c (mips16_build_call_stub): Likewise.
11101 (mips_print_operand): Likewise.
11102 (mips_mode_ok_for_mov_fmt_p): Likewise.
11103 (mips_vector_mode_supported_p): Likewise.
11104 (mips_preferred_simd_mode): Likewise.
11105 (mips_expand_vpc_loongson_even_odd): Likewise.
11106 (mips_expand_vec_unpack): Likewise.
11107 (mips_expand_vi_broadcast): Likewise.
11108 (mips_expand_vector_init): Likewise.
11109 (mips_expand_vec_reduc): Likewise.
11110 (mips_expand_msa_cmp): Likewise.
11111 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
11112 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
11113 (cc_flags_for_mode): Likewise.
11114 * config/msp430/msp430.c (msp430_print_operand): Likewise.
11115 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
11116 (nds32_output_casesi_pc_relative): Likewise.
11117 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11118 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
11119 (nvptx_gen_unpack): Likewise.
11120 (nvptx_gen_pack): Likewise.
11121 (nvptx_gen_shuffle): Likewise.
11122 (nvptx_gen_wcast): Likewise.
11123 (nvptx_preferred_simd_mode): Likewise.
11124 * config/pa/pa.c (pa_secondary_reload): Likewise.
11125 * config/pa/predicates.md (base14_operand): Likewise.
11126 * config/powerpcspe/powerpcspe-c.c
11127 (altivec_resolve_overloaded_builtin): Likewise.
11128 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
11130 (rs6000_preferred_simd_mode): Likewise.
11131 (output_vec_const_move): Likewise.
11132 (rs6000_expand_vector_extract): Likewise.
11133 (rs6000_split_vec_extract_var): Likewise.
11134 (reg_offset_addressing_ok_p): Likewise.
11135 (rs6000_legitimate_offset_address_p): Likewise.
11136 (rs6000_legitimize_address): Likewise.
11137 (rs6000_emit_set_const): Likewise.
11138 (rs6000_const_vec): Likewise.
11139 (rs6000_emit_move): Likewise.
11140 (spe_build_register_parallel): Likewise.
11141 (rs6000_darwin64_record_arg_recurse): Likewise.
11142 (swap_selector_for_mode): Likewise.
11143 (spe_init_builtins): Likewise.
11144 (paired_init_builtins): Likewise.
11145 (altivec_init_builtins): Likewise.
11146 (do_load_for_compare): Likewise.
11147 (rs6000_generate_compare): Likewise.
11148 (rs6000_expand_float128_convert): Likewise.
11149 (emit_load_locked): Likewise.
11150 (emit_store_conditional): Likewise.
11151 (rs6000_output_function_epilogue): Likewise.
11152 (rs6000_handle_altivec_attribute): Likewise.
11153 (rs6000_function_value): Likewise.
11154 (emit_fusion_gpr_load): Likewise.
11155 (emit_fusion_p9_load): Likewise.
11156 (emit_fusion_p9_store): Likewise.
11157 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
11158 (fusion_gpr_mem_load): Likewise.
11159 (fusion_addis_mem_combo_load): Likewise.
11160 (fusion_addis_mem_combo_store): Likewise.
11161 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
11162 (fusion_gpr_mem_load): Likewise.
11163 (fusion_addis_mem_combo_load): Likewise.
11164 (fusion_addis_mem_combo_store): Likewise.
11165 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11167 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
11168 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
11169 (rs6000_preferred_simd_mode): Likewise.
11170 (output_vec_const_move): Likewise.
11171 (rs6000_expand_vector_extract): Likewise.
11172 (rs6000_split_vec_extract_var): Likewise.
11173 (reg_offset_addressing_ok_p): Likewise.
11174 (rs6000_legitimate_offset_address_p): Likewise.
11175 (rs6000_legitimize_address): Likewise.
11176 (rs6000_emit_set_const): Likewise.
11177 (rs6000_const_vec): Likewise.
11178 (rs6000_emit_move): Likewise.
11179 (rs6000_darwin64_record_arg_recurse): Likewise.
11180 (swap_selector_for_mode): Likewise.
11181 (paired_init_builtins): Likewise.
11182 (altivec_init_builtins): Likewise.
11183 (rs6000_expand_float128_convert): Likewise.
11184 (emit_load_locked): Likewise.
11185 (emit_store_conditional): Likewise.
11186 (rs6000_output_function_epilogue): Likewise.
11187 (rs6000_handle_altivec_attribute): Likewise.
11188 (rs6000_function_value): Likewise.
11189 (emit_fusion_gpr_load): Likewise.
11190 (emit_fusion_p9_load): Likewise.
11191 (emit_fusion_p9_store): Likewise.
11192 * config/rx/rx.c (rx_gen_move_template): Likewise.
11193 (flags_from_mode): Likewise.
11194 * config/s390/predicates.md (s390_alc_comparison): Likewise.
11195 (s390_slb_comparison): Likewise.
11196 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
11197 (s390_vector_mode_supported_p): Likewise.
11198 (s390_cc_modes_compatible): Likewise.
11199 (s390_match_ccmode_set): Likewise.
11200 (s390_canonicalize_comparison): Likewise.
11201 (s390_emit_compare_and_swap): Likewise.
11202 (s390_branch_condition_mask): Likewise.
11203 (s390_rtx_costs): Likewise.
11204 (s390_secondary_reload): Likewise.
11205 (__SECONDARY_RELOAD_CASE): Likewise.
11206 (s390_expand_cs): Likewise.
11207 (s390_preferred_simd_mode): Likewise.
11208 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
11209 * config/sh/sh.c (sh_print_operand): Likewise.
11210 (dump_table): Likewise.
11211 (sh_secondary_reload): Likewise.
11212 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11213 * config/sh/sh.md (casesi_worker_1): Likewise.
11214 (casesi_worker_2): Likewise.
11215 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
11216 (fcc_comparison_operator): Likewise.
11217 * config/sparc/sparc.c (sparc_expand_move): Likewise.
11218 (emit_soft_tfmode_cvt): Likewise.
11219 (sparc_preferred_simd_mode): Likewise.
11220 (output_cbranch): Likewise.
11221 (sparc_print_operand): Likewise.
11222 (sparc_expand_vec_perm_bmask): Likewise.
11223 (vector_init_bshuffle): Likewise.
11224 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11225 (spu_vector_mode_supported_p): Likewise.
11226 (spu_expand_insv): Likewise.
11227 (spu_emit_branch_or_set): Likewise.
11228 (spu_handle_vector_attribute): Likewise.
11229 (spu_builtin_splats): Likewise.
11230 (spu_builtin_extract): Likewise.
11231 (spu_builtin_promote): Likewise.
11232 (spu_expand_sign_extend): Likewise.
11233 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11234 (tilegx_simd_int): Likewise.
11235 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
11236 (tilepro_simd_int): Likewise.
11237 * config/v850/v850.c (const_double_split): Likewise.
11238 (v850_print_operand): Likewise.
11239 (ep_memory_offset): Likewise.
11240 * config/vax/vax.c (vax_rtx_costs): Likewise.
11241 (vax_output_int_move): Likewise.
11242 (vax_output_int_add): Likewise.
11243 (vax_output_int_subtract): Likewise.
11244 * config/visium/predicates.md (visium_branch_operator): Likewise.
11245 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
11246 (visium_print_operand_address): Likewise.
11247 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11248 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
11249 (xtensa_expand_conditional_branch): Likewise.
11250 (xtensa_copy_incoming_a7): Likewise.
11251 (xtensa_output_literal): Likewise.
11252 * dfp.c (decimal_real_maxval): Likewise.
11253 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11255 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11256 Alan Hayward <alan.hayward@arm.com>
11257 David Sherwood <david.sherwood@arm.com>
11259 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
11260 (mode_nunits_inline): Likewise.
11261 (mode_inner_inline): Likewise.
11262 (mode_unit_size_inline): Likewise.
11263 (mode_unit_precision_inline): Likewise.
11264 (emit_insn_modes_h): Likewise. Also emit a #define of the
11266 (emit_mode_wider): Add an E_ prefix to mode names.
11267 (emit_mode_complex): Likewise.
11268 (emit_mode_inner): Likewise.
11269 (emit_mode_adjustments): Likewise.
11270 (emit_mode_int_n): Likewise.
11271 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
11272 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
11273 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
11274 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
11275 (CRC32_BUILTIN, ENTRY): Likewise.
11276 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
11277 (aarch64_pop_regs): Likewise.
11278 (aarch64_process_components): Likewise.
11279 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
11280 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
11281 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
11282 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
11283 * config/arm/arm.c (arm_init_libfuncs): Likewise.
11284 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
11286 * config/i386/i386-builtin.def (pcmpestr): Likewise.
11287 (pcmpistr): Likewise.
11288 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
11289 * config/mmix/mmix.c (mmix_output_condition): Likewise.
11290 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
11292 * config/rl78/rl78.c (mduc_regs): Likewise.
11293 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
11294 (htm_expand_builtin): Likewise.
11295 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
11296 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
11297 * config/xtensa/xtensa.c (print_operand): Likewise.
11298 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
11299 (NUM_MODE_VECTOR_INT): Likewise.
11300 * genoutput.c (null_operand): Likewise.
11301 (output_operand_data): Likewise.
11302 * genrecog.c (print_parameter_value): Likewise.
11303 * lra.c (debug_operand_data): Likewise.
11305 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11307 * dwarf2out.c (scompare_loc_descriptor_wide)
11308 (scompare_loc_descriptor_narrow): New functions, split out from...
11309 (scompare_loc_descriptor): ...here.
11310 * expmed.c (emit_store_flag_int): New function, split out from...
11311 (emit_store_flag): ...here.
11313 2017-08-30 Richard Biener <rguenther@suse.de>
11315 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
11316 (dwarf2out_early_finish): Move setting of AT_pubnames from
11317 early debug output to early finish.
11319 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11321 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
11322 and -mdata-region to the assembler.
11324 2017-08-30 Richard Biener <rguenther@suse.de>
11326 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
11328 (gen_subprogram_die): Add DW_AT_object_pointer only early.
11329 (dwarf2out_early_global_decl): Only generate a DIE for the
11330 abstract origin if it doesn't already exist or is a declaration DIE.
11331 (resolve_addr): Do not add the linkage name twice when
11332 generating a stub DIE for the DW_TAG_GNU_call_site target.
11334 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11336 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11337 Use machine_mode rather than int for arg1_mode.
11339 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
11342 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
11343 that the second argument of the built-in functions to unpack
11344 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
11345 switch statement instead a lot of if statements.
11346 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
11347 Allow 64-bit values to be in Altivec registers as well as
11348 traditional floating point registers.
11349 (pack<mode>, FMOVE128_VSX iterator): Likewise.
11351 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
11353 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
11354 MAX_REGS_PER_ADDRESS == 1.
11356 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
11358 * config/i386/i386.opt (flag_fentry): Do not init to -1.
11359 * config/i386/i386.c (ix86_option_override_internal): Simplify
11360 setting of opts->x_flag_entry.
11362 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11363 Jakub Jelinek <jakub@redhat.com>
11364 Richard Biener <rguenther@suse.de>
11366 PR tree-optimization/81503
11367 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
11368 folded constant fits in the target type; reorder tests for clarity.
11370 2017-08-29 Martin Liska <mliska@suse.cz>
11372 * passes.def: Include pass_lower_switch.
11373 * stmt.c (dump_case_nodes): Remove and move to
11374 tree-switch-conversion.
11375 (case_values_threshold): Likewise.
11376 (expand_switch_as_decision_tree_p): Likewise.
11377 (emit_case_decision_tree): Likewise.
11378 (expand_case): Likewise.
11379 (balance_case_nodes): Likewise.
11380 (node_has_low_bound): Likewise.
11381 (node_has_high_bound): Likewise.
11382 (node_is_bounded): Likewise.
11383 (emit_case_nodes): Likewise.
11384 (struct simple_case_node): New struct.
11385 (add_case_node): Remove.
11386 (emit_case_dispatch_table): Use vector instead of case_list.
11387 (reset_out_edges_aux): Remove.
11388 (compute_cases_per_edge): Likewise.
11389 (expand_case): Build list of simple_case_node.
11390 (expand_sjlj_dispatch_table): Use it.
11391 * tree-switch-conversion.c (struct case_node): Moved from
11392 stmt.c and adjusted.
11393 (emit_case_nodes): Likewise.
11394 (node_has_low_bound): Likewise.
11395 (node_has_high_bound): Likewise.
11396 (node_is_bounded): Likewise.
11397 (case_values_threshold): Likewise.
11398 (reset_out_edges_aux): Likewise.
11399 (compute_cases_per_edge): Likewise.
11400 (add_case_node): Likewise.
11401 (dump_case_nodes): Likewise.
11402 (balance_case_nodes): Likewise.
11403 (expand_switch_as_decision_tree_p): Likewise.
11404 (emit_jump): Likewise.
11405 (emit_case_decision_tree): Likewise.
11406 (try_switch_expansion): Likewise.
11407 (do_jump_if_equal): Likewise.
11408 (emit_cmp_and_jump_insns): Likewise.
11409 (fix_phi_operands_for_edge): New function.
11410 (record_phi_operand_mapping): Likewise.
11411 (class pass_lower_switch): New pass.
11412 (pass_lower_switch::execute): New function.
11413 (make_pass_lower_switch): Likewise.
11414 (conditional_probability):
11415 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
11416 * tree-pass.h: Add make_pass_lower_switch.
11418 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11421 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
11424 2017-08-29 Richard Biener <rguenther@suse.de>
11426 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
11427 we do not add a DW_AT_inline attribute twice.
11428 (gen_subprogram_die): Remove code setting DW_AT_inline on
11429 DECL_ABSTRACT_P nodes.
11431 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
11433 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
11434 calls to internal functions.
11435 (gimplify_modify_expr): Likewise.
11436 * tree-call-cdce.c (use_internal_fn): Likewise.
11437 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
11438 (convert_to_divmod): Set the nothrow flag.
11439 * tree-if-conv.c (predicate_mem_writes): Likewise.
11440 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
11441 (vectorizable_call): Likewise.
11442 (vectorizable_store): Likewise.
11443 (vectorizable_load): Likewise.
11444 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
11445 (vect_recog_mask_conversion_pattern): Likewise.
11447 2017-08-29 Martin Liska <mliska@suse.cz>
11450 * gcc.c (driver_handle_option): Add new argument.
11451 * opts-common.c (handle_option): Pass
11452 target_option_override_hook.
11453 * opts-global.c (lang_handle_option): Add new option.
11454 (set_default_handlers): Add new argument.
11455 (decode_options): Likewise.
11456 * opts.c (target_handle_option): Likewise.
11457 (common_handle_option): Call target_option_override_hook.
11458 * opts.h (struct cl_option_handler_func): Add hook for
11459 target option override.
11460 (struct cl_option_handlers): Likewise.
11461 (set_default_handlers): Add new argument.
11462 (decode_options): Likewise.
11463 (common_handle_option): Likewise.
11464 (target_handle_option): Likewise.
11465 * toplev.c (toplev::main): Pass targetm.target_option.override
11468 2017-08-29 Richard Biener <rguenther@suse.de>
11469 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
11471 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
11472 life to the active subtree.
11474 2017-08-28 Jeff Law <law@redhat.com>
11476 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
11477 derive_equivalences.
11478 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
11479 Code moved into....
11480 (edge_info::derive_equivalences): New private member function
11482 * tree-ssa-dom.c (class edge_info): Changed from a struct
11483 to a class. Add ctor/dtor, methods and data members.
11484 (edge_info::edge_info): Renamed from allocate_edge_info.
11485 Initialize additional members.
11486 (edge_info::~edge_info): New.
11487 (free_dom_edge_info): Delete the edge info.
11488 (record_edge_info): Use new class & associated member functions.
11489 Tighten forms for testing for edge equivalences.
11490 (record_temporary_equivalences): Iterate over the simple
11491 equivalences rather than assuming there's only one per edge.
11492 (cprop_into_successor_phis): Iterate over the simple
11493 equivalences rather than assuming there's only one per edge.
11494 (optimize_stmt): Use operand_equal_p rather than pointer
11495 equality for mini-DSE code.
11497 2017-08-28 Nathan Sidwell <nathan@acm.org>
11499 * gcc.c (execute): Fold SIGPIPE handling into switch
11500 statement. Adjust internal error message.
11502 2017-08-28 Richard Biener <rguenther@suse.de>
11505 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
11506 Do nothing for removed DIEs.
11508 2017-08-28 Richard Biener <rguenther@suse.de>
11510 PR tree-optimization/81977
11511 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
11514 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
11517 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
11518 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
11519 using targetm.gen_mem_thread_fence.
11521 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
11524 * config/i386/i386.md (*<btsc><mode>): Change operand 2
11525 predicate to register_operand. Reorder operands.
11526 (*btr<mode>): Ditto.
11527 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
11528 (*btr<mode>_mask): Ditto.
11530 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
11532 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
11533 * config/rs6000/xmmintrin.h: New file.
11534 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
11536 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11539 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
11540 parameter and_insn and return it.
11541 (recombine_lvx_pattern): Insert a copy to ensure availability of
11542 the base register of the copied masking operation at the point of
11543 the instruction replacement.
11544 (recombine_stvx_pattern): Likewise.
11546 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
11548 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
11549 undocumented switches.
11550 (-mpower9-dform-vector): Likewise.
11551 (-mpower9-dform): Likewise.
11552 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
11553 comments to delete references to -mpower9-dform* switches.
11554 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
11555 Delete reference to -mpower9-dform* switches, test for
11556 -mpower9-vector instead.
11557 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
11558 (OTHER_P9_VECTOR_MASKS): Likewise.
11559 (POWERPC_MASKS): Likewise.
11560 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
11561 tests against -mpower9-dform* to -mpower9-vector. Delete code
11562 that checked for -mpower9-dform* consistancy with other options.
11563 Add test for -mpower9-misc to enable other power9 switches.
11564 (rs6000_init_hard_regno_mode_ok): Likewise.
11565 (rs6000_option_override_internal): Likewise.
11566 (rs6000_emit_prologue): Likewise.
11567 (rs6000_emit_epilogue): Likewise.
11568 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
11569 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
11570 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
11572 (emit_fusion_p9_store): Likewise.
11573 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
11574 resetting these macros if the assembler does not support ISA 3.0
11576 (TARGET_P9_DFORM_VECTOR): Likewise.
11577 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
11578 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
11580 2017-08-25 Alan Modra <amodra@gmail.com>
11582 PR rtl-optimization/81747
11583 * cse.c (cse_extended_basic_block): Don't attempt to record
11584 equivalences for degenerate conditional branches that branch
11585 to their fall-through.
11587 2017-08-24 Martin Sebor <msebor@redhat.com>
11589 PR middle-end/81908
11590 * gimple-fold.c (size_must_be_zero_p): New function.
11591 (gimple_fold_builtin_memory_op): Call it.
11593 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
11595 * config/rs6000/mm_malloc.h: New file.
11597 2017-08-24 Bin Cheng <bin.cheng@arm.com>
11599 PR tree-optimization/81913
11600 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
11601 analysis when either IVs in condition can wrap.
11603 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
11605 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
11606 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
11608 2017-08-24 Richard Biener <rguenther@suse.de>
11611 * targhooks.c (default_target_can_inline_p): Properly
11612 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
11613 is present and always compare.
11614 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
11615 infer -mfpmath=sse from TARGET_SSE_P.
11616 (ix86_can_inline_p): Properly use target_option_default_node when
11617 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
11619 2017-08-24 Richard Biener <rguenther@suse.de>
11622 * dwarf2out.c (output_die): Handle flag_generate_offload like
11624 (output_comp_unit): Likewise.
11625 (gen_array_type_die): Likewise.
11626 (dwarf2out_early_finish): Likewise.
11627 (note_variable_value_in_expr): Likewise.
11628 (dwarf2out_finish): Likewise. Adjust assert.
11629 * cgraphunit.c (symbol_table::compile): Move setting of
11630 flag_generate_offload earlier ...
11631 (symbol_table::finalize_compilation_unit): ... here, before
11632 early debug finalization.
11634 2017-08-24 Richard Biener <rguenther@suse.de>
11636 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
11637 and ipa-fnsummary.h.
11638 (ix86_can_inline_p): When ix86_fpmath flags do not match
11639 check whether the callee uses FP math at all.
11641 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
11643 PR middle-end/81931
11644 * tree-ssanames.c (get_nonzero_bits): Use element_precision
11645 instead of TYPE_PRECISION.
11647 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
11648 Alan Hayward <alan.hayward@arm.com>
11649 David Sherwood <david.sherwood@arm.com>
11651 * combine.c (make_extraction): Use subreg_offset_from_lsb.
11653 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
11655 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
11657 (ix86_frame::stack_realign_allocate): New field.
11658 (struct machine_frame_state): Modify comments.
11659 (machine_frame_state::sp_realigned_fp_end): New field.
11660 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
11661 layout calculation.
11662 (sp_valid_at): Add assertion to assure no attempt to access invalid
11663 offset of a realigned stack.
11664 (fp_valid_at): Likewise.
11665 (choose_baseaddr): Modify comments.
11666 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
11667 ix86_expand_prologue.
11668 (ix86_expand_prologue): Modify stack realignment and allocation.
11669 (ix86_expand_epilogue): Modify comments.
11670 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
11671 avx2_runtime, avx512f, and avx512f_runtime.
11673 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
11675 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
11676 (mstackrealign): Do not init to -1.
11677 * config/i386/i386.c (ix86_option_override_internal):
11678 Check opts_set, not opts when setting default value of
11679 opts->x_ix86_force_align_arg_pointer.
11681 2017-08-23 Richard Biener <rguenther@suse.de>
11683 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
11684 lang_hooks.decl_printable_name.
11685 * print-rtl-function.c (print_rtx_function): Likewise.
11686 * tree-pretty-print.c (dump_function_header): Likewise.
11688 2017-08-23 Richard Biener <rguenther@suse.de>
11691 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
11692 -g0 at compile-time.
11694 2017-08-23 Tamar Christina <tamar.christina@arm.com>
11696 PR middle-end/19706
11697 * doc/sourcebuild.texi (Other hardware attributes):
11700 2017-08-23 Tamar Christina <tamar.christina@arm.com>
11702 PR middle-end/19706
11703 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
11704 Fix single-use check.
11706 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
11708 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
11710 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
11712 * doc/install.texi: Modify to add more details on running selected
11715 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
11717 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
11718 is combined with -mabi=ms.
11719 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
11722 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11724 PR tree-optimization/81488
11725 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
11726 and cached_basis fields.
11727 (MAX_SPREAD): New constant.
11728 (alloc_cand_and_find_basis): Initialize new fields.
11729 (clear_visited): New function.
11730 (create_phi_basis_1): Rename from create_phi_basis, set visited
11731 and cached_basis fields.
11732 (create_phi_basis): New wrapper function.
11733 (phi_add_costs_1): Rename from phi_add_costs, add spread
11734 parameter, set visited field, short-circuit when limits reached.
11735 (phi_add_costs): New wrapper function.
11736 (record_phi_increments_1): Rename from record_phi_increments, set
11738 (record_phi_increments): New wrapper function.
11739 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
11740 (phi_incr_cost): New wrapper function.
11741 (all_phi_incrs_profitable_1): Rename from
11742 all_phi_incrs_profitable, set visited field.
11743 (all_phi_incrs_profitable): New wrapper function.
11745 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
11746 Alan Hayward <alan.hayward@arm.com>
11747 David Sherwood <david.sherwood@arm.com>
11749 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
11750 that takes the outer and inner modes.
11751 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
11752 comparison as the canonical test for a paradoxical subreg.
11753 * combine.c (simplify_set): Use paradoxical_subreg_p.
11754 (make_extraction): Likewise.
11755 (force_to_mode): Likewise.
11756 (rtx_equal_for_field_assignment_p): Likewise.
11757 (gen_lowpart_for_combine): Likewise.
11758 (simplify_comparison): Likewise.
11759 * cse.c (equiv_constant): Likewise.
11760 * expmed.c (store_bit_field_1): Likewise.
11761 * final.c (alter_subreg): Likewise.
11762 * fwprop.c (propagate_rtx): Likewise.
11763 (forward_propagate_subreg): Likewise.
11764 * ira-conflicts.c (ira_build_conflicts): Likewise.
11765 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
11766 * lra-constraints.c (curr_insn_transform): Likewise.
11767 (split_reg): Likewise.
11768 * lra-eliminations.c (move_plus_up): Likewise.
11769 (lra_eliminate_regs_1): Likewise.
11770 * recog.c (general_operand): Likewise.
11771 * ree.c (combine_reaching_defs): Likewise.
11772 * reload.c (push_reload): Likewise.
11773 (find_reloads): Likewise.
11774 * reload1.c (elimination_effects): Likewise.
11775 (compute_reload_subreg_offset): Likewise.
11776 (choose_reload_regs): Likewise.
11777 * rtlanal.c (subreg_lsb_1): Likewise.
11778 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11779 (simplify_subreg): Likewise.
11780 * var-tracking.c (track_loc_p): Likewise.
11781 * emit-rtl.c (byte_lowpart_offset): Likewise.
11782 (paradoxical_subreg_p): Delete out-of-line definition.
11784 2017-08-22 Jeff Law <law@redhat.com>
11786 PR tree-optimization/81741
11787 PR tree-optimization/71947
11788 * tree-ssa-dom.c: Include tree-inline.h.
11789 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
11790 equivalences if one is more expensive to compute than the other.
11791 * tree-ssa-scopedtables.h (class const_or_copies): Make
11792 record_const_or_copy_raw method private.
11793 (class avail_exprs_stack): New method simplify_binary_operation.
11794 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
11795 avail_exprs_stack::simplify_binary_operation as needed.
11796 (avail_exprs_stack::simplify_binary_operation): New function.
11798 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
11800 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
11801 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
11802 (DOT_SYMBOLS): Likewise.
11803 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
11804 (RELOCATABLE_NEEDS_FIXUP): Likewise.
11805 (RS6000_ABI_NAME): Likewise.
11806 (TARGET_CMODEL): Likewise.
11807 (TOC_SECTION_ASM_OP): Likewise.
11808 (SET_CMODEL): New macro.
11809 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
11811 2017-08-22 Richard Biener <rguenther@suse.de>
11813 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
11814 to allow for free-lang-data replacements similar to verify_type_variant.
11816 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
11817 Alan Hayward <alan.hayward@arm.com>
11818 David Sherwood <david.sherwood@arm.com>
11820 * config/aarch64/aarch64.md (casesi): Use DImode rather than
11821 VOIDmode for the LABEL_REF.
11823 2017-08-22 Richard Biener <rguenther@suse.de>
11825 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
11827 2017-08-22 Richard Biener <rguenther@suse.de>
11829 * common.opt (feliminate-dwarf2-dups): Ignore.
11830 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
11831 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
11832 same_die_p_wrap, compute_section_prefix,
11833 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
11834 (comdat_symbol_id, comdat_symbol_number): Likewise.
11835 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
11837 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
11838 (output_die): Mark unreachable path unreachable.
11839 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
11840 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
11841 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
11842 (dwarf2out_early_finish): Likewise.
11844 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
11846 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
11848 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
11851 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
11852 not VAR_P. Filter attribute warnings with OPT_Wattributes.
11853 (avr_attribute_table) <io, io_low, address>: Initialize
11854 .decl_required with true.
11856 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
11858 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
11859 undocumented debugging options.
11860 (-mvsx-scalar-double): Likewise.
11861 (-mallow-df-permute): Likewise.
11862 (-mvectorize-builtins): Likewise.
11863 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
11864 (rs6000_builtin_vectorized_function): Likewise.
11865 (rs6000_builtin_md_vectorized_function): Likewise.
11866 (rs6000_opt_vars): Likewise.
11868 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
11871 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
11872 (*btrq_imm): Rename from *btrq.
11873 (*btcq_imm): Rename from *btcq.
11874 (btsc): New code attribute.
11875 (*<btsc><mode>): New insn pattern.
11876 (*btr<mode>): Ditto.
11877 (*<btsc><mode>_mask): New insn_and_split pattern.
11878 (*btr<mode>_mask): Ditto.
11880 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11881 Alan Hayward <alan.hayward@arm.com>
11882 David Sherwood <david.sherwood@arm.com>
11884 * function.c (pad_below): Simplify padding calculation.
11886 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11887 Alan Hayward <alan.hayward@arm.com>
11888 David Sherwood <david.sherwood@arm.com>
11890 * target.def (function_prologue): Remove frame size argument.
11891 (function_epilogue): Likewise.
11892 * doc/tm.texi: Regenerate.
11893 * final.c (final_start_function): Update call to function_prologue.
11894 (final_end_function): Update call to function_epilogue.
11895 (default_function_pro_epilogue): Remove frame size argument.
11896 * output.h (default_function_pro_epilogue): Likewise.
11897 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
11898 (arm_output_function_prologue): Likewise.
11899 * config/frv/frv.c (frv_function_prologue): Likewise.
11900 (frv_function_epilogue): Likewise.
11901 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
11902 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
11903 (ia64_output_function_epilogue): Likewise.
11904 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
11905 (m32r_output_function_epilogue): Likewise.
11906 * config/microblaze/microblaze.c (microblaze_function_prologue)
11907 (microblaze_function_epilogue): Likewise.
11908 * config/mips/mips.c (mips_output_function_prologue): Likewise.
11909 (mips_output_function_epilogue): Likewise.
11910 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
11911 (mmix_target_asm_function_epilogue): Likewise.
11912 * config/msp430/msp430.c (msp430_start_function): Likewise.
11913 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
11914 (nds32_asm_function_epilogue): Likewise.
11915 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
11916 * config/pa/pa.c (pa_output_function_prologue): Likewise.
11917 (pa_output_function_epilogue): Likewise.
11918 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
11919 (rs6000_output_function_epilogue): Likewise.
11920 * config/rl78/rl78.c (rl78_start_function): Likewise.
11921 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
11922 (rs6000_output_function_epilogue): Likewise.
11923 * config/rx/rx.c (rx_output_function_prologue): Likewise.
11924 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
11925 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
11926 (sparc_asm_function_epilogue): Likewise.
11928 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11930 * tree.h (type_has_mode_precision_p): New function.
11931 * convert.c (convert_to_integer_1): Use it.
11932 * expr.c (expand_expr_real_2): Likewise.
11933 (expand_expr_real_1): Likewise.
11934 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
11935 * match.pd: Likewise.
11936 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
11937 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
11938 * tree-tailcall.c (process_assignment): Likewise.
11939 * tree-vect-loop.c (vectorizable_reduction): Likewise.
11940 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
11941 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
11942 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11943 (vectorizable_assignment): Likewise.
11944 (vectorizable_shift): Likewise.
11945 (vectorizable_operation): Likewise.
11946 * tree-vrp.c (register_edge_assert_for_2): Likewise.
11948 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
11950 * match.pd: Add pow (C, x) simplification.
11952 2017-08-21 Richard Biener <rguenther@suse.de>
11954 PR tree-optimization/81900
11955 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
11956 for blocks with abnormal predecessors.
11957 (compute_antic): Do not set visited flag prematurely.
11959 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
11962 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
11964 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11966 * stor-layout.h (vector_type_mode): Move to...
11967 * tree.h (vector_type_mode): ...here.
11968 * stor-layout.c (vector_type_mode): Move to...
11969 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
11971 2017-08-21 Richard Biener <rguenther@suse.de>
11973 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
11974 register_external_die hooks.
11975 (debug_false_tree_charstarstar_uhwistar): Declare.
11976 (debug_nothing_tree_charstar_uhwi): Likewise.
11977 * debug.c (do_nothing_debug_hooks): Adjust.
11978 (debug_false_tree_charstarstar_uhwistar): New do nothing.
11979 (debug_nothing_tree_charstar_uhwi): Likewise.
11980 * dbxout.c (dbx_debug_hooks): Adjust.
11981 (xcoff_debug_hooks): Likewise.
11982 * sdbout.c (sdb_debug_hooks): Likewise.
11983 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
11984 * dwarf2out.c (macinfo_label_base): New global.
11985 (dwarf2out_register_external_die): New function for the
11986 register_external_die hook.
11987 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
11988 (dwarf2_debug_hooks): Use them.
11989 (dwarf2_lineno_debug_hooks): Adjust.
11990 (struct die_struct): Add with_offset flag.
11991 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
11992 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
11993 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
11994 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
11995 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
11996 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
11997 defining section names for the early LTO debug variants.
11998 (reset_indirect_string): New helper.
11999 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
12000 (print_dw_val): Add support for offsetted symbol references.
12001 (get_ultimate_context): Split out from is_cxx.
12002 (is_cxx): Use get_ultimate_context.
12003 (is_fortran): Add decl overload.
12004 (compute_comp_unit_symbol): Split out worker from
12005 compute_section_prefix.
12006 (compute_section_prefix): Call compute_comp_unit_symbol and
12007 set comdat_type_p here.
12008 (output_die): Skip DIE symbol output for the LTO added one.
12009 Handle DIE symbol references with offset.
12010 (output_comp_unit): Guard section name mangling properly.
12011 For LTO debug sections emit a symbol at the section beginning
12012 which we use to refer to its DIEs.
12013 (add_abstract_origin_attribute): For DIEs registered via
12014 dwarf2out_register_external_die directly refer to the early
12015 DIE rather than indirectly through the shadow one we created.
12016 Remove obsolete call to dwarf2out_abstract_function for
12017 non-function/block origins.
12018 (gen_array_type_die): When generating early LTO debug do
12019 not emit DW_AT_string_length.
12020 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
12021 late when in LTO. As suggested place a gcc_unreachable for
12022 the DECL_ABSTRACT_P case.
12023 (gen_subprogram_die): Avoid another specification DIE
12024 for early built declarations/definitions for the late LTO case.
12025 (gen_variable_die): Add type references for late duplicated VLA dies
12027 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
12028 we have the abstract instance already.
12029 (process_scope_var): Adjust decl DIE contexts in LTO which
12030 first puts them in limbo.
12031 (gen_decl_die): Do not generate type DIEs late apart from
12032 types for VLAs or for decls we do not yet have a DIE. Do not
12033 call dwarf2out_abstract_function late.
12034 (dwarf2out_early_global_decl): Make sure to create DIEs
12035 for abstract instances of a decl first.
12036 (dwarf2out_late_global_decl): Adjust comment.
12037 (output_macinfo_op): With multiple macro sections use
12038 macinfo_label_base to distinguish labels.
12039 (output_macinfo): Likewise. Update macinfo_label_base.
12040 Pass in the line info label.
12041 (note_variable_value_in_expr): When generating LTO resolve
12042 all variable values here by generating DIEs as needed.
12043 (init_sections_and_labels): Add early LTO debug flag parameter
12044 and generate different sections and names if set. Add generation
12045 counter for the labels so we can have multiple of them.
12046 (reset_dies): Helper to allow DIEs to be output multiple times.
12047 (dwarf2out_finish): When outputting DIEs to the fat part of an
12048 LTO object first reset DIEs.
12049 (dwarf2out_early_finish): Output early DIEs when generating LTO.
12050 (modified_type_die): Check for decl_ultimate_origin being self
12052 (gen_type_die_with_usage): Likewise.
12053 (gen_typedef_die): Allow decl_ultimate_origin being self.
12054 (set_decl_abstract_flags): Remove.
12055 (set_block_abstract_flags): Likewise.
12056 (dwarf2out_abstract_function): Treat the early generated DIEs
12057 as the abstract copy and only add DW_AT_inline and
12058 DW_AT_artificial here and call set_decl_origin_self.
12059 If the DIE has an abstract origin don't do anything.
12060 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
12061 if we have none yet (Go fails to build one, PR78628).
12062 (variably_modified_type_p): Prevent endless recursion for Ada
12063 cyclic pointer types.
12064 * lto-streamer-in.c: Include debug.h.
12065 (dref_queue): New global.
12066 (lto_read_tree_1): Stream in DIE references.
12067 (lto_input_tree): Register DIE references.
12068 (input_function): Stream DECL_DEBUG_ARGS.
12069 * lto-streamer-out.c: Include debug.h.
12070 (lto_write_tree_1): Output DIE references.
12071 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
12072 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
12073 (output_function): Stream DECL_DEBUG_ARGS.
12074 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
12075 Stream DECL_ABSTRACT_ORIGIN.
12076 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
12077 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
12078 DECL_CONTEXT for file-scope decls.
12079 * lto-streamer.h (struct dref_entry): Declare.
12080 (dref_queue): Likewise.
12081 * cfgexpand.c (pass_expand::execute): Do not call the
12082 outlining_inline_function hook here.
12083 * lto-wrapper.c (debug_obj): New global.
12084 (tool_cleanup): Unlink it if required.
12085 (debug_objcopy): New function.
12086 (run_gcc): Handle early debug sections in the IL files by
12087 extracting them to separate files, partially linkin them and
12088 feeding the result back as result to the linker.
12089 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
12090 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
12091 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
12092 sections into a separate segment.
12093 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
12095 (darwin_asm_dwarf_section): Likewise.
12096 (darwin_asm_output_dwarf_offset): Likewise.
12097 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
12099 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12100 Alan Hayward <alan.hayward@arm.com>
12101 David Sherwood <david.sherwood@arm.com>
12103 * read-md.h (md_reader::record_potential_iterator_use): Replace
12104 pointer argument with an rtx and an index.
12105 * read-rtl.c (iterator_group::apply_iterator): Likewise.
12106 (apply_mode_iterator): Likewise.
12107 (apply_code_iterator): Likewise.
12108 (apply_int_iterator): Likewise.
12109 (apply_subst_iterator): Likewise.
12110 (record_iterator_use): Likewise.
12111 (record_attribute_use): Likewise.
12112 (md_reader::record_potential_iterator_use): Likewise. Update calls
12113 to record_iterator_use and apply_iterator.
12114 (iterator_use): Replace ptr with x and index.
12115 (attribute_use): Likewise.
12116 (apply_attribute_uses): Update calls to apply_iterator.
12117 (apply_iterators): Likewise. Update initialization of iterator_use.
12118 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
12119 and record_potential_iterator_use.
12120 (rtx_reader::read_rtx_operand): Likewise.
12122 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12123 Alan Hayward <alan.hayward@arm.com>
12124 David Sherwood <david.sherwood@arm.com>
12126 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
12129 2017-08-21 Richard Biener <rguenther@suse.de>
12131 PR middle-end/81884
12132 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
12133 at struct end conservatively when comparing common bases.
12135 2017-08-21 Richard Biener <rguenther@suse.de>
12137 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
12138 (mem_ref_in_stmt): Remove.
12139 (determine_max_movement): Use ref index to get at the reference.
12140 (invariantness_dom_walker::before_dom_children): Deal with
12141 lim data already initialized.
12142 (gather_mem_refs_stmt): Initialize lim data and record ref index.
12144 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12146 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
12147 (TARGET_ISA_ROUND): Ditto.
12148 (TARGET_ROUND): Ditto.
12149 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
12150 * config/i386/i386.md: Ditto.
12151 * config/i386/sse.md: Ditto.
12152 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
12153 with OPTION_MASK_ISA_SSE4_1.
12155 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12158 * doc/extend.texi (x86 Built-in Functions): Correct the name of
12159 __builtin_ia32_lzcnt_u16.
12161 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
12164 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
12165 (rs6000_set_current_function): Rewrite function to use it.
12167 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
12170 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
12171 and TYPE_WARN_IF_NOT_ALIGN.
12172 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
12173 (handle_warn_if_not_align): New.
12174 (place_union_field): Call handle_warn_if_not_align.
12175 (place_field): Call handle_warn_if_not_align.
12176 Copy TYPE_WARN_IF_NOT_ALIGN.
12177 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
12178 (layout_type): Likewise.
12179 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
12181 (tree_decl_common): Add warn_if_not_align.
12182 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
12183 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
12184 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
12185 (DECL_WARN_IF_NOT_ALIGN): Likewise.
12186 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
12187 * doc/extend.texi: Document warn_if_not_aligned attribute.
12188 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
12190 2017-08-17 Martin Liska <mliska@suse.cz>
12193 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
12194 (get_data_dependence): Use it as pointer type.
12195 (distribute_loop): Likewise.
12197 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12199 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
12200 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
12201 (p8_vmrgow_<mode>_direct): New define_insn.
12202 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
12203 handle endianness for vmrgew and vmrgow permute patterns.
12205 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
12207 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
12208 * config/rs6000/rs6000-cpus.def: Remove comment.
12209 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
12210 (POWERPC_MASKS): Likewise.
12211 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
12212 use of TARGET_VSX_TIMODE.
12213 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
12214 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
12215 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
12216 (rs6000_option_override_internal): Remove dead code.
12217 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
12218 (rs6000_legitimize_reload_address): Likewise.
12219 (rs6000_legitimate_address_p): Likewise.
12220 (rs6000_opt_masks): Delete "vsx-timode".
12221 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
12222 from function comment.
12223 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
12224 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
12225 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
12227 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
12228 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
12229 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
12230 (VSX_TI): Likewise.
12232 (define_peephole2): Likewise.
12234 2017-08-17 Martin Sebor <msebor@redhat.com>
12237 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
12238 past the end of an array.
12239 (test_pp_format): Add test cases.
12241 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
12243 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
12244 missing ECF_NOTHROW flags.
12246 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
12249 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
12250 operands residing in integer registers.
12251 (*vsx_le_perm_load_<mode>): Likewise.
12252 (*vsx_le_perm_store_<mode>): Likewise.
12253 (define_peephole2): Add peepholes to optimize the above.
12255 2017-08-17 Marek Polacek <polacek@redhat.com>
12257 PR middle-end/81814
12258 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
12259 to mimic what shorten_compare did. Change the return type to bool.
12260 (fold_cond_expr_with_comparison): Update call to
12261 operand_equal_for_comparison_p.
12262 (fold_ternary_loc): Likewise.
12264 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
12266 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
12268 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
12270 2017-08-17 Richard Biener <rguenther@suse.de>
12272 * tree-ssa-structalias.c (solve_graph): When propagating
12273 to successors update the graphs succ edges and avoid duplicate work.
12275 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
12278 * config/i386/i386.c (ix86_option_override_internal): Save target
12279 specific options after ix86_stack_protector_guard_reg was changed.
12281 2017-08-17 Richard Biener <rguenther@suse.de>
12283 PR tree-optimization/81827
12284 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
12285 (new_var_info): Initialize it conservatively.
12286 (get_call_vi): Mark register vars.
12287 (new_scalar_tmp_constraint_exp): Likewise.
12288 (handle_rhs_call): Likewise.
12289 (handle_const_call): Likewise.
12290 (create_function_info_for): Likewise.
12291 (solve_constraints): Sort varinfos to separate register from
12292 non-register vars to pack points-to solution bitmaps during
12295 2017-08-17 Marek Polacek <polacek@redhat.com>
12297 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
12299 2017-08-17 Richard Biener <rguenther@suse.de>
12301 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
12302 to true when overflow is undefined and we saturated the result.
12304 2017-08-17 Alan Modra <amodra@gmail.com>
12307 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
12308 Don't use store multiple if only one reg needs saving.
12309 (interesting_frame_related_regno): New function.
12310 (rs6000_frame_related): Don't emit frame info for regs that
12312 (rs6000_emit_epilogue): Likewise.
12314 2017-08-16 Nathan Sidwell <nathan@acm.org>
12316 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
12317 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
12318 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
12319 * tree.c (free_lang_data_in_type): Use else-if chain. Always
12320 clear TYPE_LANG_1. Remove obsolete member-function stripping.
12321 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
12322 (verify_type): Adjust for TYPE_BINFO move.
12323 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
12324 process TYPE_BINFO directly.
12325 (hash_tree): Likewise.
12326 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
12328 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
12331 2017-08-16 David Malcolm <dmalcolm@redhat.com>
12333 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
12335 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
12338 * config/i386/i386.md (*anddi_1_btr): Change predicates of
12339 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
12340 Add ix86_binary_operator_ok to insn constraint.
12341 (*iordi_1_bts): Ditto.
12342 (*xordi_1_btc): Ditto.
12343 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
12344 Update corresponding peephole2 pattern.
12348 2017-08-16 Bin Cheng <bin.cheng@arm.com>
12350 PR tree-optimization/81832
12351 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
12352 copy loop header which has IFN_LOOP_DIST_ALIAS call.
12354 2017-08-16 Marek Polacek <polacek@redhat.com>
12357 * fold-const.c (fold_indirect_ref_1): Restore original behavior
12358 regarding size_zero_node.
12360 2017-08-16 Martin Liska <mliska@suse.cz>
12363 * config.gcc: Respect previously set extra_objs in case
12366 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
12368 PR tree-optimization/81835
12369 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
12370 the phi SSA_NAME. Check that the condition in a COND_EXPR does
12371 not depend on the phi.
12373 2017-08-16 Alan Modra <amodra@gmail.com>
12375 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
12378 2017-08-16 Alan Modra <amodra@gmail.com>
12380 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
12381 (save_reg_p): ..into this. Update all callers.
12382 (first_reg_to_save): Simplify.
12384 2017-08-16 Alan Modra <amodra@gmail.com>
12386 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
12389 2017-08-15 Joseph Myers <joseph@codesourcery.com>
12393 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
12394 constant count if that count is less than 32.
12396 2017-08-15 Nathan Sidwell <nathan@acm.org>
12398 * gcc.c (execute): Emit friendlier message if inferior is killed
12399 by an external cause.
12401 2017-08-15 Richard Biener <rguenther@suse.de>
12403 PR tree-optimization/81790
12404 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
12405 CONSTRUCTORs from simplifying and VN.
12407 2017-08-14 Martin Sebor <msebor@redhat.com>
12409 * builtin-attrs.def: Add comments.
12411 2017-08-14 Martin Sebor <msebor@redhat.com>
12414 * doc/extend.texi (attribute nonstring): Document new attribute.
12416 2017-08-14 Martin Sebor <msebor@redhat.com>
12419 * tree-diagnostic.c (default_tree_printer): Handle %G.
12420 * gimple-pretty-print.h (percent_G_format): Declare new function.
12421 * gimple-pretty-print.c (percent_G_format): Define.
12422 * tree-pretty-print.c (percent_K_format): Add argument.
12424 2017-08-14 Martin Sebor <msebor@redhat.com>
12426 PR translation/79998
12427 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
12428 Remove a stray space.
12430 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
12433 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
12434 (*iordi_1_bts): Ditto.
12435 (*xordi_1_btc): Ditto.
12437 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12440 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
12441 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12443 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
12444 quoted strings, and make more translator-friendly.
12445 (darwin_rs6000_override_options): Likewise.
12446 (rs6000_option_override_internal): Likewise.
12447 (rs6000_return_in_memory): Fix overlong line.
12448 (init_cmulative_args): Use quoted strings, and make more
12449 translator-friendly.
12450 (rs6000_pass_by_reference): Fix overlong line.
12451 (def_builtin): Use quoted strings.
12452 (altivec_expand_predicate_builtin): Use quoted strings, and make
12453 more translator-friendly.
12454 (htm_expand_builtin): Use quoted strings.
12455 (cpu_expand_builtin): Use quoted strings, and make more
12456 translator-friendly.
12457 (altivec_expand_builtin): Likewise.
12458 (paired_expand_predicate_builtin): Likewise.
12459 (rs6000_invalid_builtin): Likewise.
12460 (builtin_function_type): Use quoted strings.
12461 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
12462 more translator-friendly.
12463 (rs6000_trampoline_init): Likewise.
12464 (rs6000_handle_altivec_attribute): Likewise.
12465 (rs6000_inner_target_options): Use quoted strings.
12466 (rs6000_disable_incompatible_switches): Likewise.
12467 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
12468 strings, and make more translator-friendly.
12469 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
12471 2017-08-14 Bin Cheng <bin.cheng@arm.com>
12473 PR tree-optimization/81799
12474 * tree-loop-distribution.c (version_loop_by_alias_check): Force
12475 cond_expr to simple gimple operand.
12477 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
12479 PR middle-end/46932
12480 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
12482 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
12486 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
12487 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
12488 TARGET_GASISR_PROLOGUES.
12489 * config/avr/avr.c (avr_option_override): Same.
12490 (avr_pass_pre_proep::execute): Same.
12492 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
12495 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
12496 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
12498 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
12500 * config/i386/i386.md (*load_tp_<mode>): Redefine as
12501 define_insn_and_split. Split to a memory load from 0 in
12502 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
12503 using PTR mode iterator.
12504 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
12505 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
12506 (*add_tp_<mode>): Redefine as define_insn_and_split.
12507 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
12508 address space. Merge with *add_tp_x32 using PTR mode iterator.
12509 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
12510 Split to an add with a memory load from 0 in
12511 DEFAULT_TLS_SEG_REG address space.
12513 2017-08-12 Andrew Pinski <apinski@cavium.com>
12515 * config/aarch64/aarch64-option-extensions.def (rdma):
12516 Fix feature string to what Linux prints out in /proc/cpuinfo.
12518 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
12521 * dwarf2out.c (modified_type_die): For C typedef types that have
12522 an ultimate origin, process the ultimate origin instead of the
12524 (gen_typedef_die): Assert that input DECLs have no ultimate
12526 (gen_type_die_with_usage): For typedef variants that have an
12527 ultimate origin, just call gen_decl_die on the original DECL.
12528 (process_scope_var): Avoid creating DIEs for local typedefs and
12529 concrete static variables.
12531 2017-08-12 Alan Modra <amodra@gmail.com>
12535 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
12536 match gnu-user.h startfile.
12537 (ENDFILE_LINUX_SPEC): Similarly.
12539 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
12542 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
12544 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
12546 2017-08-11 Tamar Christina <tamar.christina@arm.com>
12547 * config/aarch64/aarch64.md (mov<mode>): Change.
12548 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
12549 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
12550 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
12552 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
12554 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
12555 for storage order barriers.
12557 2017-08-11 Martin Liska <mliska@suse.cz>
12560 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
12561 variables of void type.
12563 2017-08-11 Martin Liska <mliska@suse.cz>
12565 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
12566 TARGET_SUPPORTS_ALIASES.
12567 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
12568 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
12569 (optimize_weakref): Likewise.
12570 * symtab.c (symtab_node::noninterposable_alias): Likewise.
12571 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
12572 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
12574 2017-08-11 Martin Liska <mliska@suse.cz>
12577 * config/i386/i386.c (make_resolver_func): Do complete
12578 refactoring of the function.
12580 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
12583 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
12584 * config/i386/i386.c (ix86_stack_protect_guard): Use
12585 ix86_stack_protect_guard_symbol_str to generate varible declaration.
12586 * doc/invoke.texi (x86 Options): Document
12587 -mstack-protector-guard-symbol= option.
12589 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
12591 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
12592 * config/i386/i386.c (ix86_split_stack_guard): New function.
12593 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
12594 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
12595 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
12596 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
12597 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
12598 (split_stack_space_check): Call ix86_split_stack_guard.
12600 2017-08-10 Martin Sebor <msebor@redhat.com>
12602 * print-tree.c (print_node): Print location using the established
12604 Replace spaces with colons.
12605 (debug_raw, debug): Ditto.
12607 2017-08-10 Martin Sebor <msebor@redhat.com>
12610 * pretty-print.c (pp_format): Correct the handling of %s precision.
12612 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
12615 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
12617 (ix86_finalize_stack_frame_flags): This. Also clear
12618 frame_pointer_needed if -fno-omit-frame-pointer is used without
12620 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
12621 with ix86_finalize_stack_frame_flags.
12622 (ix86_expand_epilogue): Likewise.
12623 (ix86_expand_split_stack_prologue): Likewise.
12624 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
12626 2017-08-10 Martin Liska <mliska@suse.cz>
12629 * c-attribs.c (handle_target_attribute):
12630 Report warning for an empty string argument of target attribute.
12632 2017-08-09 Jakub Jelinek <jakub@redhat.com>
12635 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
12637 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
12638 or DECL_NONLOCAL labels.
12639 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
12640 or DECL_NONLOCAL labels here.
12642 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
12644 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
12645 to indicate when early gimple folding has been disabled.
12646 (rs6000_gimple_fold_builtin): Add debug content.
12647 (rs6000_invalid_builtin): Fix whitespace.
12648 (rs6000_expand_builtin): Fix whitespace.
12649 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
12651 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
12654 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
12655 SAVE_MULTIPLE if not all the registers that saves, should be saved.
12657 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
12659 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
12660 (qdf24xx): Likewise.
12661 * config/aarch64/aarch64.md: Include falkor.md.
12662 * config/aarch64/falkor.md: New.
12664 2017-08-09 Marek Polacek <polacek@redhat.com>
12667 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
12668 * diagnostic.c (emit_diagnostic): Add a comment.
12669 (emit_diagnostic_valist): New function.
12671 2017-08-09 Marek Polacek <polacek@redhat.com>
12674 * input.c (make_location): New overload.
12675 * input.h (make_location): Declare.
12677 2017-08-08 Alan Modra <amodra@gmail.com>
12678 H.J. Lu <hongjiu.lu@intel.com>
12681 * gcc.c (NO_PIE_SPEC): Delete.
12682 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
12684 (LINK_PIE_SPEC): ..to here.
12685 (LINK_COMMAND_SPEC): Support -no-pie.
12686 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
12687 chain of crtbegin*.o selection, update for PIE_SPEC changes and
12689 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
12690 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
12691 (ENDFILE_CRTEND_SPEC): Similarly.
12693 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
12696 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
12697 (mstack-protector-guard-offset=): Ditto.
12698 * config/i386/i386.c (ix86_option_override): Handle
12699 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
12701 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
12702 ix86_stack_protect_guard_offset variables.
12703 (TARGET_STACK_PROTECT_GUARD): Always define.
12704 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
12705 and -mstack-protector-guard-offset= options.
12707 2017-08-08 Bin Cheng <bin.cheng@arm.com>
12709 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
12710 boundary case for the last candidate.
12712 2017-08-08 Bin Cheng <bin.cheng@arm.com>
12714 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
12715 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
12717 2017-08-08 Tamar Christina <tamar.christina@arm.com>
12719 PR middle-end/19706
12720 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
12721 * config/aarch64/aarch64-builtins.c
12722 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
12723 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
12724 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
12726 2017-08-08 Tamar Christina <tamar.christina@arm.com>
12727 Andrew Pinski <pinskia@gmail.com>
12729 PR middle-end/19706
12730 * internal-fn.def (XORSIGN): New.
12731 * optabs.def (xorsign_optab): New.
12732 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
12733 (convert_expand_mult_copysign): New.
12734 (pass_optimize_widening_mul::execute): Call
12735 convert_expand_mult_copysign.
12737 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12739 PR tree-optimization/81354
12740 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
12741 Insert on edges rather than explicitly creating landing pads.
12742 (analyze_candidates_and_replace): Commit edge inserts.
12744 2017-08-08 Richard Biener <rguenther@suse.de>
12746 PR middle-end/81719
12747 * tree-ssa-loop-niter.c: Include tree-dfa.h.
12748 (expand_simple_operations): Also look through ADDR_EXPRs with
12749 MEM_REF bases treating them as POINTER_PLUS_EXPR.
12751 2017-08-08 Richard Biener <rguenther@suse.de>
12753 PR tree-optimization/81723
12754 * tree-vect-slp.c (struct bst_traits): New hash traits.
12755 (bst_fail): New global.
12756 (vect_build_slp_tree_2): New worker, split out from ...
12757 (vect_build_slp_tree): ... this now wrapping it with using
12758 bst_fail set to cache SLP tree build fails. Properly handle
12760 (vect_analyze_slp_instance): Allocate and free bst_fail.
12762 2017-08-08 Martin Liska <mliska@suse.cz>
12765 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
12766 LABEL_DECLs that can be from a different function.
12768 2017-08-08 Bin Cheng <bin.cheng@arm.com>
12770 PR tree-optimization/81744
12771 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
12772 loop's number of iterations.
12774 2017-08-08 Martin Liska <mliska@suse.cz>
12776 * asan.c: Include header files.
12777 * attribs.c (build_decl_attribute_variant): New function moved
12779 (build_type_attribute_qual_variant): Likewise.
12780 (cmp_attrib_identifiers): Likewise.
12781 (simple_cst_list_equal): Likewise.
12782 (omp_declare_simd_clauses_equal): Likewise.
12783 (attribute_value_equal): Likewise.
12784 (comp_type_attributes): Likewise.
12785 (build_type_attribute_variant): Likewise.
12786 (lookup_ident_attribute): Likewise.
12787 (remove_attribute): Likewise.
12788 (merge_attributes): Likewise.
12789 (merge_type_attributes): Likewise.
12790 (merge_decl_attributes): Likewise.
12791 (merge_dllimport_decl_attributes): Likewise.
12792 (handle_dll_attribute): Likewise.
12793 (attribute_list_equal): Likewise.
12794 (attribute_list_contained): Likewise.
12795 * attribs.h (lookup_attribute): New function moved from tree.[ch].
12796 (lookup_attribute_by_prefix): Likewise.
12797 * bb-reorder.c: Include header files.
12798 * builtins.c: Likewise.
12799 * calls.c: Likewise.
12800 * cfgexpand.c: Likewise.
12801 * cgraph.c: Likewise.
12802 * cgraphunit.c: Likewise.
12803 * convert.c: Likewise.
12804 * dwarf2out.c: Likewise.
12805 * final.c: Likewise.
12806 * fold-const.c: Likewise.
12807 * function.c: Likewise.
12808 * gimple-expr.c: Likewise.
12809 * gimple-fold.c: Likewise.
12810 * gimple-pretty-print.c: Likewise.
12811 * gimple.c: Likewise.
12812 * gimplify.c: Likewise.
12813 * hsa-common.c: Likewise.
12814 * hsa-gen.c: Likewise.
12815 * internal-fn.c: Likewise.
12816 * ipa-chkp.c: Likewise.
12817 * ipa-cp.c: Likewise.
12818 * ipa-devirt.c: Likewise.
12819 * ipa-fnsummary.c: Likewise.
12820 * ipa-inline.c: Likewise.
12821 * ipa-visibility.c: Likewise.
12823 * lto-cgraph.c: Likewise.
12824 * omp-expand.c: Likewise.
12825 * omp-general.c: Likewise.
12826 * omp-low.c: Likewise.
12827 * omp-offload.c: Likewise.
12828 * omp-simd-clone.c: Likewise.
12829 * opts-global.c: Likewise.
12830 * passes.c: Likewise.
12831 * predict.c: Likewise.
12832 * sancov.c: Likewise.
12833 * sanopt.c: Likewise.
12834 * symtab.c: Likewise.
12835 * toplev.c: Likewise.
12836 * trans-mem.c: Likewise.
12837 * tree-chkp.c: Likewise.
12838 * tree-eh.c: Likewise.
12839 * tree-into-ssa.c: Likewise.
12840 * tree-object-size.c: Likewise.
12841 * tree-parloops.c: Likewise.
12842 * tree-profile.c: Likewise.
12843 * tree-ssa-ccp.c: Likewise.
12844 * tree-ssa-live.c: Likewise.
12845 * tree-ssa-loop.c: Likewise.
12846 * tree-ssa-sccvn.c: Likewise.
12847 * tree-ssa-structalias.c: Likewise.
12848 * tree-ssa.c: Likewise.
12849 * tree-streamer-in.c: Likewise.
12850 * tree-vectorizer.c: Likewise.
12851 * tree-vrp.c: Likewise.
12852 * tsan.c: Likewise.
12853 * ubsan.c: Likewise.
12854 * varasm.c: Likewise.
12855 * varpool.c: Likewise.
12856 * tree.c: Remove functions moved to attribs.[ch].
12857 * tree.h: Likewise.
12858 * config/aarch64/aarch64.c: Add attrs.h header file.
12859 * config/alpha/alpha.c: Likewise.
12860 * config/arc/arc.c: Likewise.
12861 * config/arm/arm.c: Likewise.
12862 * config/avr/avr.c: Likewise.
12863 * config/bfin/bfin.c: Likewise.
12864 * config/c6x/c6x.c: Likewise.
12865 * config/cr16/cr16.c: Likewise.
12866 * config/cris/cris.c: Likewise.
12867 * config/darwin.c: Likewise.
12868 * config/epiphany/epiphany.c: Likewise.
12869 * config/fr30/fr30.c: Likewise.
12870 * config/frv/frv.c: Likewise.
12871 * config/ft32/ft32.c: Likewise.
12872 * config/h8300/h8300.c: Likewise.
12873 * config/i386/winnt.c: Likewise.
12874 * config/ia64/ia64.c: Likewise.
12875 * config/iq2000/iq2000.c: Likewise.
12876 * config/lm32/lm32.c: Likewise.
12877 * config/m32c/m32c.c: Likewise.
12878 * config/m32r/m32r.c: Likewise.
12879 * config/m68k/m68k.c: Likewise.
12880 * config/mcore/mcore.c: Likewise.
12881 * config/microblaze/microblaze.c: Likewise.
12882 * config/mips/mips.c: Likewise.
12883 * config/mmix/mmix.c: Likewise.
12884 * config/mn10300/mn10300.c: Likewise.
12885 * config/moxie/moxie.c: Likewise.
12886 * config/msp430/msp430.c: Likewise.
12887 * config/nds32/nds32-isr.c: Likewise.
12888 * config/nds32/nds32.c: Likewise.
12889 * config/nios2/nios2.c: Likewise.
12890 * config/nvptx/nvptx.c: Likewise.
12891 * config/pa/pa.c: Likewise.
12892 * config/pdp11/pdp11.c: Likewise.
12893 * config/powerpcspe/powerpcspe.c: Likewise.
12894 * config/riscv/riscv.c: Likewise.
12895 * config/rl78/rl78.c: Likewise.
12896 * config/rx/rx.c: Likewise.
12897 * config/s390/s390.c: Likewise.
12898 * config/sh/sh.c: Likewise.
12899 * config/sol2.c: Likewise.
12900 * config/sparc/sparc.c: Likewise.
12901 * config/spu/spu.c: Likewise.
12902 * config/stormy16/stormy16.c: Likewise.
12903 * config/tilegx/tilegx.c: Likewise.
12904 * config/tilepro/tilepro.c: Likewise.
12905 * config/v850/v850.c: Likewise.
12906 * config/vax/vax.c: Likewise.
12907 * config/visium/visium.c: Likewise.
12908 * config/xtensa/xtensa.c: Likewise.
12910 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
12913 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
12914 constraints since the -mupper-regs-* switches have been
12916 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
12917 into a vector from a double word element that was extracted from
12918 another vector, and eliminate extra XXPERMDI instructions.
12919 (vsx_concat_<mode>_2): Likewise.
12920 (vsx_concat_<mode>_3): Likewise.
12921 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
12922 concat to allow optimizing inserts from previous extracts.
12924 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
12926 * config/i386/i386.c (ix86_stack_protect_guard): Generate
12927 memory reference to a SSP offset in TLS address space.
12928 (ix86_print_operand) <case '@'>: Remove.
12929 (ix86_print_operand_punct_valid_p): Remove '@' code.
12930 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
12931 UNSPEC_SP_TLS_TEST.
12932 (stack_tls_protect_set_<mode>): Remove.
12933 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
12934 (stack_tls_protect_test_<mode>): Remove.
12935 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
12937 2017-08-07 Olivier Hainque <hainque@adacore.com>
12940 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
12942 2017-08-07 Douglas Rupp <rupp@adacore.com>
12944 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
12945 variable was referenced as multidir in command.
12947 2017-08-07 Jakub Jelinek <jakub@redhat.com>
12950 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
12953 2017-08-07 Martin Liska <mliska@suse.cz>
12955 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
12956 * config/rl78/rl78.c: Add include of attribs.h.
12957 * config/sh/sh.c: Likewise.
12958 * config/v850/v850.c: Likewise.
12960 2017-08-07 Tom de Vries <tom@codesourcery.com>
12962 PR middle-end/78266
12963 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
12965 2017-08-07 Martin Liska <mliska@suse.cz>
12967 * config/mips/mips.c: Include attribs.h.
12969 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
12972 * doc/invoke.texi: Document change in behvaior for -Ofast for
12975 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
12977 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
12979 (aarch64_pop_regs): Likewise.
12980 (aarch64_gen_load_pair): Likewise.
12981 (aarch64_save_callee_saves): Likewise.
12982 (aarch64_restore_callee_saves): Likewise.
12984 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
12986 * config/i386/i386.c: Revert the last change.
12988 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
12991 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
12993 (ix86_finalize_stack_frame_flags): This. Also clear
12994 frame_pointer_needed if -fno-omit-frame-pointer is used without
12996 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
12997 with ix86_finalize_stack_frame_flags.
12998 (ix86_expand_epilogue): Likewise.
12999 (ix86_expand_split_stack_prologue): Likewise.
13001 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13004 * config/i386/i386.c (get_builtin_code_for_version): Set priority
13005 to P_AES for Westmere.
13007 2017-08-07 Jonathan Yong <10walls@gmail.com>
13009 * config/i386/mingw.opt (fset-stack-executable): Removed.
13010 * config/i386/cygming.opt (fset-stack-executable): Moved
13012 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
13014 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
13016 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
13018 2017-08-07 Marek Polacek <polacek@redhat.com>
13020 PR middle-end/81737
13021 * fold-const.c (fold_indirect_ref_1): Check type_domain.
13023 2017-08-07 Martin Liska <mliska@suse.cz>
13025 * attribs.h (canonicalize_attr_name): New function.
13026 (cmp_attribs): Move from c-format.c and adjusted.
13027 (is_attribute_p): Moved from tree.h.
13028 * tree-inline.c: Add new includes.
13029 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
13030 (private_is_attribute_p): Remove.
13031 (private_lookup_attribute): Likewise.
13032 (private_lookup_attribute_by_prefix): Simplify.
13033 (remove_attribute): Use is_attribute_p.
13034 * tree.h: Remove removed declarations.
13036 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13038 PR middle-end/81698
13039 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
13040 instead of computing it in the function. Formatting fix.
13041 (expand_case): Don't rely on default_edge being the first edge,
13042 clear it if removing it, pass default_edge to
13043 emit_case_dispatch_table.
13044 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
13047 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
13049 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
13050 insn in the function, emit NOP after the insn.
13052 2017-08-06 Tom de Vries <tom@codesourcery.com>
13054 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
13057 2017-08-06 Tom de Vries <tom@codesourcery.com>
13059 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
13062 2017-08-04 Yury Gribov <tetra2005@gmail.com>
13064 PR tree-optimization/57371
13065 * match.pd: New pattern.
13067 2017-08-04 Marek Polacek <polacek@redhat.com>
13069 PR middle-end/81695
13070 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
13071 perform the computation in offset_int.
13073 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13075 PR tree-optimization/81136
13076 * tree-vectorizer.h: Include tree-hash-traits.h.
13077 (vec_base_alignments): New typedef.
13078 (vec_info): Add a base_alignments field.
13079 (vect_record_base_alignments): Declare.
13080 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
13082 (DR_IS_CONDITIONAL_IN_STMT): New macro.
13083 (create_data_ref): Add an is_conditional_in_stmt argument.
13084 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
13085 the is_conditional_in_stmt field.
13086 (data_ref_loc): Add an is_conditional_in_stmt field.
13087 (get_references_in_stmt): Set the is_conditional_in_stmt field.
13088 (find_data_references_in_stmt): Update call to create_data_ref.
13089 (graphite_find_data_references_in_stmt): Likewise.
13090 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
13091 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
13092 (vect_record_base_alignment): New function.
13093 (vect_record_base_alignments): Likewise.
13094 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
13095 for nested statements even if we fail to compute a misalignment.
13096 Use pooled base alignments for unconditional references.
13097 (vect_find_same_alignment_drs): Compare base addresses instead
13099 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
13100 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
13102 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13104 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
13105 Add an explicit name for the enum. Use auto_vec for slp_instances
13106 and grouped_stores.
13107 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
13109 (_bb_vec_info): Add a constructor and destructor.
13110 (vinfo_for_stmt): Return NULL for uids of -1 as well.
13111 (destroy_loop_vec_info): Delete.
13112 (vect_destroy_datarefs): Likewise.
13113 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
13114 (vec_info::vec_info): New function.
13115 (vec_info::~vec_info): Likewise.
13116 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
13117 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
13118 destroy_loop_vec_info.
13119 * tree-vect-loop.c (new_loop_vec_info): Replace with...
13120 (_loop_vec_info::_loop_vec_info): ...this.
13121 (destroy_loop_vec_info): Replace with...
13122 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
13123 the stmt_vec_infos. Leave handling of vec_info information to its
13124 destructor. Remove explicit vector releases.
13125 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
13126 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
13127 * tree-vect-slp.c (new_bb_vec_info): Replace with...
13128 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
13129 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
13130 (destroy_bb_vec_info): Replace with...
13131 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
13132 information to its destructor.
13133 (vect_slp_analyze_bb_1): Use new and delete instead of
13134 new_bb_vec_info and destroy_bb_vec_info.
13135 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
13138 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13140 * tree-data-ref.h (subscript): Add access_fn field.
13141 (data_dependence_relation): Add could_be_independent_p.
13142 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
13143 (same_access_functions): Move to tree-data-ref.c.
13144 * tree-data-ref.c (ref_contains_union_access_p): New function.
13145 (access_fn_component_p): Likewise.
13146 (access_fn_components_comparable_p): Likewise.
13147 (dr_analyze_indices): Add a reference to access_fn_component_p.
13148 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
13150 (constant_access_functions): Likewise.
13151 (add_other_self_distances): Likewise.
13152 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
13153 (initialize_data_dependence_relation): Use XCNEW and remove
13154 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
13155 of access functions that have the same type. Allow the
13156 subsequence to end with different bases in some circumstances.
13157 Record the chosen access functions in SUB_ACCESS_FN.
13158 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
13159 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
13160 (subscript_dependence_tester_1): Likewise dra and drb.
13161 (build_classic_dist_vector): Update calls accordingly.
13162 (subscript_dependence_tester): Likewise.
13163 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
13164 DDR_COULD_BE_INDEPENDENT_P.
13165 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
13166 comp_alias_ddrs instead of may_alias_ddrs.
13167 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
13169 (vect_analyze_data_ref_dependence): Use it if
13170 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
13171 distance vectors if that fails.
13172 (dependence_distance_ge_vf): New function.
13173 (vect_prune_runtime_alias_test_list): Use it. Don't clear
13174 LOOP_VINFO_MAY_ALIAS_DDRS.
13176 2017-08-04 Richard Biener <rguenther@suse.de>
13178 PR middle-end/81705
13179 * fold-const.c (fold_binary_loc): Properly restrict
13180 minus_var0 && minus_var1 case when associating undefined overflow
13183 2017-08-04 Tom de Vries <tom@codesourcery.com>
13185 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
13187 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13189 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13190 Don't start diagnostic messages with a capital letter.
13191 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13193 (rs6000_invalid_builtin): Likewise.
13194 (rs6000_trampoline_init): Likewise.
13196 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13199 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
13200 after setting changeable df flags.
13202 2017-08-03 Richard Biener <rguenther@suse.de>
13204 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
13205 up if the use is in USE - X.
13207 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
13209 * toplev.c (dumpfile.h): New include.
13210 (internal_error_reentered): New static function. Use it...
13211 (internal_error_function): ...here to handle reentered internal_error.
13213 2017-08-03 Richard Biener <rguenther@suse.de>
13215 PR middle-end/81148
13216 * fold-const.c (split_tree): Add minus_var and minus_con
13217 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
13218 here but always use minus_*.
13219 (associate_trees): Assert we never associate with MINUS_EXPR
13220 and NULL first operand. Do not recurse for PLUS_EXPR operands
13221 when associating as MINUS_EXPR either.
13222 (fold_binary_loc): Track minus_var and minus_con.
13224 2017-08-03 Tom de Vries <tom@codesourcery.com>
13227 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
13228 ACCEL_COMPILER, apply finish_options on
13229 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
13231 2017-08-03 Tom de Vries <tom@codesourcery.com>
13234 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
13235 function_entry_patch_area_size > 0.
13237 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13240 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
13241 instead of 10??LU, perform unit multiplication in wide_int,
13242 don't change alloc_object_size_limit if the limit is larger
13245 PR tree-optimization/81655
13246 PR tree-optimization/81588
13247 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
13248 the case when ranges[i].low and high are 1 for unsigned type with
13251 PR middle-end/81052
13252 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
13253 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
13255 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13257 * tree-vrp.h: Add include guard.
13259 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
13262 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
13263 (ud2): New insn pattern.
13264 * config/i386/i386.c (ix86_expand_epilogue):
13265 For naked functions, generate ud2 instead of trap insn.
13267 2017-08-02 Marek Polacek <polacek@redhat.com>
13270 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
13272 2017-08-02 Tom de Vries <tom@codesourcery.com>
13273 Cesar Philippidis <cesar@codesourcery.com>
13275 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
13276 Add missing edge probabilities.
13278 2017-08-02 Tamar Christina <tamar.christina@arm.com>
13280 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
13281 Correct endianness.
13283 2017-08-02 Jakub Jelinek <jakub@redhat.com>
13285 PR middle-end/79499
13286 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
13287 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
13288 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
13289 prologue_seq sequences - if any.
13291 2017-08-02 Richard Biener <rguenther@suse.de>
13293 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
13294 via vectors if supported, integer extracts via punning if supported
13295 or otherwise vector extracts.
13297 2017-08-02 Richard Biener <rguenther@suse.de>
13299 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
13301 (bitmap_insert_into_set): ... this.
13303 2017-08-02 Richard Biener <rguenther@suse.de>
13305 PR tree-optimization/81633
13307 2015-08-17 Alan Hayward <alan.hayward@arm.com>
13309 PR tree-optimization/71752
13310 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
13312 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
13314 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
13315 (machine_function::call_ms2sysv_pad_out): Remove field.
13316 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
13317 (ix86_compute_frame_layout): Likewise.
13319 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
13322 * config/i386/i386.c (ix86_set_func_type): Disallow naked
13323 attribute with interrupt attribute.
13325 2017-08-01 Andrew Pinski <apinski@cavium.com>
13327 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
13328 BIT_INSERT_EXPR's operand 1
13329 to see if the types precision matches.
13331 2017-08-01 Martin Liska <mliska@suse.cz>
13333 PR middle-end/70140
13334 * builtins.c (expand_builtin_memcpy_args): Remove.
13335 (expand_builtin_memcpy): Call newly added function
13336 expand_builtin_memory_copy_args.
13337 (expand_builtin_memcpy_with_bounds): Likewise.
13338 (expand_builtin_mempcpy): Remove last argument.
13339 (expand_builtin_mempcpy_with_bounds): Likewise.
13340 (expand_builtin_memory_copy_args): New function created from
13341 expand_builtin_mempcpy_args with small modifications.
13342 (expand_builtin_mempcpy_args): Remove.
13343 (expand_builtin_stpcpy): Remove unused argument.
13344 (expand_builtin): Likewise.
13345 (expand_builtin_with_bounds): Likewise.
13347 2017-08-01 Martin Liska <mliska@suse.cz>
13350 Make mempcpy more optimal (PR middle-end/70140).
13352 2017-08-01 Jakub Jelinek <jakub@redhat.com>
13355 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
13356 __builtin_vec_cmpne verify both arguments are compatible vectors
13357 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
13358 verify arg1_type is a pointer or array type. For __builtin_vec_st,
13359 move computation of aligned to after checking the argument types.
13363 * config/rs6000/vsx.md (vextract_fp_from_shorth,
13364 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
13367 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
13368 Doug Rupp <rupp@adacore.com>
13369 Olivier Hainque <hainque@adacore.com>
13371 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
13372 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
13374 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
13375 for TARGET_OS_CPP_BUILTIN.
13376 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
13377 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
13379 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
13380 passing required abi options to the assembler for EABI configurations.
13381 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
13382 of .text.hot and .text.unlikely sections for kernel modules when
13383 using ARM style exceptions.
13384 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
13385 options. Add EXTRA_CC1_SPEC.
13386 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
13388 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
13390 (ARM_TARGET2_DWARF_FORMAT): Define.
13391 * config/arm/t-vxworks: Adjust multilib control to removal of the
13392 Diab command line options.
13394 2017-08-01 Martin Liska <mliska@suse.cz>
13396 PR gcov-profile/81561
13397 * gcov.c (unblock): Make unblocking safe as we need to preserve
13398 index correspondence of blocks and block_lists.
13400 2017-08-01 Richard Biener <rguenther@suse.de>
13402 PR tree-optimization/81181
13403 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
13404 (compute_antic): ... end of iteration here.
13406 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
13408 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
13409 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
13410 (ftree-slp-vectorize): Likewise.
13411 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
13412 can no longer be set independent of flag_tree_loop_vectorize.
13413 * omp-general.c (emp_max_vf): Likewise.
13414 * opts.c (enable_fdo_optimizations): Remove references to
13415 flag_tree_vectorize, these are now implicit.
13416 (common_handle_option): Remove handling for OPT_ftree_vectorize,
13417 and leave it for the options machinery.
13419 2017-08-01 Martin Liska <mliska@suse.cz>
13421 PR middle-end/70140
13422 * builtins.c (expand_builtin_memcpy_args): Remove.
13423 (expand_builtin_memcpy): Call newly added function
13424 expand_builtin_memory_copy_args.
13425 (expand_builtin_memcpy_with_bounds): Likewise.
13426 (expand_builtin_mempcpy): Remove last argument.
13427 (expand_builtin_mempcpy_with_bounds): Likewise.
13428 (expand_builtin_memory_copy_args): New function created from
13429 expand_builtin_mempcpy_args with small modifications.
13430 (expand_builtin_mempcpy_args): Remove.
13431 (expand_builtin_stpcpy): Remove unused argument.
13432 (expand_builtin): Likewise.
13433 (expand_builtin_with_bounds): Likewise.
13435 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
13438 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
13439 print "ds:" only for immediates in generic address space.
13441 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
13444 * config/i386/i386.c (ix86_funciton_naked): New prototype.
13445 (ix86_function_ok_for_sibcall): Return false for naked functions.
13447 2017-08-01 Richard Biener <rguenther@suse.de>
13449 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
13450 (compute_antic): Seed worklist with exit block predecessors.
13451 * cfganal.c (dfs_find_deadend): For a cycle return the source
13452 of the edge closing it.
13454 2017-08-01 Tamar Christina <tamar.christina@arm.com>
13456 * config/aarch64/aarch64.c
13457 (aarch64_can_const_movi_rtx_p): Move 0 check.
13459 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13461 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
13462 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
13464 * match.pd: Ditto in address comparison pattern.
13466 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13468 PR tree-optimization/81627
13469 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
13470 closed ssa form for store-store chain.
13472 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13474 PR tree-optimization/81620
13475 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
13476 for store-store chain.
13478 2017-08-01 Jakub Jelinek <jakub@redhat.com>
13480 PR tree-optimization/81588
13481 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
13482 ranges[i].in_p, invert comparison code ccode. For >/>=,
13483 swap rhs1 and rhs2 and comparison code unconditionally,
13484 for </<= don't do that. Don't swap rhs1/rhs2 again if
13485 ranges[i].in_p, instead invert comparison code ccode if
13486 opcode or oe->rank is BIT_IOR_EXPR.
13489 * optabs.def (vec_extract_optab, vec_init_optab): Change from
13490 a direct optab to conversion optab.
13491 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
13492 with GET_MODE_INNER as last argument instead of optab_handler.
13493 * expmed.c (extract_bit_field_1): Likewise. Use vector from
13494 vector extraction if possible and optab is available.
13495 * expr.c (store_constructor): Use convert_optab_handler instead
13496 of optab_handler. Use vector initialization from smaller
13497 vectors if possible and optab is available.
13498 * tree-vect-stmts.c (vectorizable_load): Likewise.
13499 * doc/md.texi (vec_extract, vec_init): Document that the optabs
13500 now have two modes.
13501 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
13502 of vec_init from half-sized vectors with the same element mode.
13503 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
13504 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
13505 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
13506 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
13507 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
13508 after mode in gen_vec_extract* calls.
13509 (vec_extract<mode>): Renamed to ...
13510 (vec_extract<mode><ssescalarmodelower>): ... this.
13511 (vec_extract<mode><ssehalfvecmodelower>): New expander.
13512 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
13513 element mode after mode in gen_vec_init* calls.
13514 (VEC_INIT_HALF_MODE): New mode iterator.
13515 (vec_init<mode>): Renamed to ...
13516 (vec_init<mode><ssescalarmodelower>): ... this.
13517 (vec_init<mode><ssehalfvecmodelower>): New expander.
13518 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
13519 (vec_extractv2sfsf): ... this.
13520 (vec_initv2sf): Renamed to ...
13521 (vec_initv2sfsf): ... this.
13522 (vec_extractv2si): Renamed to ...
13523 (vec_extractv2sisi): ... this.
13524 (vec_initv2si): Renamed to ...
13525 (vec_initv2sisi): ... this.
13526 (vec_extractv4hi): Renamed to ...
13527 (vec_extractv4hihi): ... this.
13528 (vec_initv4hi): Renamed to ...
13529 (vec_initv4hihi): ... this.
13530 (vec_extractv8qi): Renamed to ...
13531 (vec_extractv8qiqi): ... this.
13532 (vec_initv8qi): Renamed to ...
13533 (vec_initv8qiqi): ... this.
13534 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
13535 (vec_init<mode>): Renamed to ...
13536 (vec_init<mode><VEC_base_l>): ... this.
13537 (vec_extract<mode>): Renamed to ...
13538 (vec_extract<mode><VEC_base_l>): ... this.
13539 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
13540 (vec_initv2sfsf): ... this.
13541 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
13542 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
13543 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
13544 element mode after mode in gen_vec_init* calls.
13545 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
13546 (vec_init<mode><Vel>): ... this.
13547 (vec_extract<mode>): Renamed to ...
13548 (vec_extract<mode><Vel>): ... this.
13549 * config/aarch64/iterators.md (Vel): New mode attribute.
13550 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
13551 Add element mode after mode in gen_vec_extract* calls.
13552 * config/s390/vector.md (non_vec_l): New mode attribute.
13553 (vec_extract<mode>): Renamed to ...
13554 (vec_extract<mode><non_vec_l>): ... this.
13555 (vec_init<mode>): Renamed to ...
13556 (vec_init<mode><non_vec_l>): ... this.
13557 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
13558 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
13560 * config/arm/iterators.md (V_elem_l): New mode attribute.
13561 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
13562 (vec_extract<mode><V_elem_l>): ... this.
13563 (vec_extractv2di): Renamed to ...
13564 (vec_extractv2didi): ... this.
13565 (vec_init<mode>): Renamed to ...
13566 (vec_init<mode><V_elem_l>): ... this.
13567 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
13568 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
13569 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
13570 Add element mode after gen_vec_extract* calls.
13571 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
13572 (vec_init<mode><unitmode>): ... this.
13573 (vec_extract<mode>): Renamed to ...
13574 (vec_extract<mode><unitmode>): ... this.
13575 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
13576 (vec_init<mode><unitmode>): ... this.
13577 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
13578 (vec_initv2sfsf): ... this.
13579 (vec_extractv2sf): Renamed to ...
13580 (vec_extractv2sfsf): ... this.
13581 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
13582 Add element mode after gen_vec_extract* calls.
13583 * config/mips/mips.md (unitmode): New mode iterator.
13584 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
13585 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
13586 * config/spu/spu.md (inner_l): New mode attribute.
13587 (vec_init<mode>): Renamed to ...
13588 (vec_init<mode><inner_l>): ... this.
13589 (vec_extract<mode>): Renamed to ...
13590 (vec_extract<mode><inner_l>): ... this.
13591 * config/sparc/sparc.md (veltmode): New mode iterator.
13592 (vec_init<VMALL:mode>): Renamed to ...
13593 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
13594 * config/ia64/vect.md (vec_initv2si): Renamed to ...
13595 (vec_initv2sisi): ... this.
13596 (vec_initv2sf): Renamed to ...
13597 (vec_initv2sfsf): ... this.
13598 (vec_extractv2sf): Renamed to ...
13599 (vec_extractv2sfsf): ... this.
13600 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
13601 (vec_init<mode>): Renamed to ...
13602 (vec_init<mode><VEC_base_l>): ... this.
13603 (vec_extract<mode>): Renamed to ...
13604 (vec_extract<mode><VEC_base_l>): ... this.
13605 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
13606 (vec_initv2sfsf): ... this.
13607 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
13608 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
13609 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
13610 gen_vec_init* calls.
13612 2017-08-01 Richard Biener <rguenther@suse.de>
13614 PR tree-optimization/81297
13615 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
13616 TREE_OVERFLOW from INTEGER_CSTs.
13618 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
13620 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
13622 2017-07-31 Carl Love <cel@us.ibm.com>
13624 * config/rs6000/rs6000-c: Add support for built-in functions
13625 vector signed char vec_xl_be (signed long long, signed char *);
13626 vector unsigned char vec_xl_be (signed long long, unsigned char *);
13627 vector signed int vec_xl_be (signed long long, signed int *);
13628 vector unsigned int vec_xl_be (signed long long, unsigned int *);
13629 vector signed long long vec_xl_be (signed long long, signed long long *);
13630 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
13631 vector signed short vec_xl_be (signed long long, signed short *);
13632 vector unsigned short vec_xl_be (signed long long, unsigned short *);
13633 vector double vec_xl_be (signed long long, double *);
13634 vector float vec_xl_be (signed long long, float *);
13635 * config/rs6000/altivec.h (vec_xl_be): Add #define.
13636 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
13637 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
13639 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
13640 (altivec_expand_builtin): Add switch statement to call
13641 altivec_expand_xl_be for each builtin.
13642 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
13643 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
13644 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
13645 __builtin_vsx_le_be_v16qi.
13646 * doc/extend.texi: Update the built-in documentation file for the
13647 new built-in functions.
13649 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
13652 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
13654 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
13656 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13658 * config.gcc: Add z14.
13659 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
13660 CPU model numbers for z13s and z14.
13661 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
13663 * config/s390/s390-opts.h (enum processor_type): Rename
13664 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
13665 * config/s390/s390.c (processor_table): Add field for CPU name to
13666 be passed to Binutils.
13667 (s390_asm_output_machine_for_arch): Use the new field in
13668 processor_table for Binutils.
13669 (s390_expand_builtin): Replace arch12 with z14.
13670 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
13671 (s390_get_sched_attrmask): Likewise.
13672 (s390_get_unit_mask): Likewise.
13673 * config/s390/s390.opt: Add z14 to processor_type enum.
13675 2017-07-31 Martin Jambor <mjambor@suse.cz>
13678 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
13679 regardless of optimization level.
13681 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
13682 Martin Liska <mliska@suse.cz>
13684 * predict.def: Remove old comment and adjust probability.
13685 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
13686 PREDICT statements.
13688 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
13691 * config/i386/i386.c (ix86_function_naked): New function.
13692 (ix86_can_use_return_insn_p): Return false for naked functions.
13693 (ix86_expand_prologue): Skip prologue for naked functions.
13694 (ix86_expand_epilogue): Skip epilogue for naked functions
13695 and emit trap instruction.
13696 (ix86_warn_func_return): New function.
13697 (ix86_attribute_table): Add "naked" attribute specification.
13698 (TARGET_WARN_FUNC_RETURN): Define.
13699 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
13701 2017-07-31 Martin Liska <mliska@suse.cz>
13703 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
13704 (dump_gimple_bb_header): Always dump BB info.
13705 (pp_cfg_jump): Do not append info about BB when dumping a jump.
13707 2017-07-31 Martin Liska <mliska@suse.cz>
13710 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
13711 also with current_function_decl non-null equality.
13713 2017-07-31 Jakub Jelinek <jakub@redhat.com>
13716 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
13717 change type to the element type, instead add eltype variable and
13718 use it where we are interested in the element type.
13720 PR tree-optimization/81603
13721 * ipa-polymorphic-call.c
13722 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
13723 offset arithmetic in offset_int, bail out if the resulting bit offset
13724 doesn't fit into shwi.
13726 2017-07-31 Martin Liska <mliska@suse.cz>
13728 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
13729 (gimplify_save_expr): Fix comment.
13731 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
13734 * config/i386/i386.c (ix86_function_arg): Update arguments for
13736 (ix86_compute_frame_layout): Set the initial stack offset to
13737 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
13738 INCOMING_FRAME_SP_OFFSET.
13739 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
13740 stack before exception handler returns.
13741 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
13742 the 'ERROR_CODE' for exception handler.
13744 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
13746 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
13747 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
13748 (ASM_OUTPUT_REG_POP): Ditto.
13749 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
13750 instead of asm_fprintf to output pure string.
13752 2017-07-29 Jakub Jelinek <jakub@redhat.com>
13754 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
13755 to imported_module_or_decl hook.
13756 (debug_nothing_tree_tree_tree_bool): Remove.
13757 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
13758 * debug.c (do_nothing_debug_hooks): Use
13759 debug_nothing_tree_tree_tree_bool_bool instead of
13760 debug_nothing_tree_tree_tree_bool.
13761 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
13762 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
13763 * sdbout.c (sdb_debug_hooks): Likewise.
13764 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
13765 (gen_namespace_die): Add DW_AT_export_symbols attribute if
13767 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
13768 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
13769 attribute, don't add anything.
13771 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13773 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
13774 (fold_build2_stat_loc): Likewise.
13775 (fold_build3_stat_loc): Likewise.
13776 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
13777 (fold_build1_loc): Remove macro.
13778 (fold_build2_loc): Likewise.
13779 (fold_build3_loc): Likewise.
13781 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13783 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
13784 (gimple_build_debug_bind_source_stat): Likewise.
13785 * gimple.h (gimple_build_debug_bind): Remove macro.
13786 (gimple_build_debug_bind_source): Likewise.
13788 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13790 * bitmap.c (bitmap_alloc): Adjust.
13791 (bitmap_gc_alloc): Likewise.
13792 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
13794 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13796 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
13797 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
13798 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
13799 (bitmap_gc_alloc_stat): Likewise.
13800 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
13802 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13804 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
13805 * rtl.h (shallow_copy_rtx): Remove macro.
13807 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13809 * emit-rtl.c (gen_raw_REG): Adjust.
13810 * gengenrtl.c (gendef): Likewise.
13811 * rtl.c (rtx_alloc_stat): Remove _stat from name.
13812 * rtl.h (rtx_alloc): Remove macro.
13814 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13816 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
13817 (build_tree_list_stat): Likewise.
13818 * tree.h (build_tree_list): Remove macro.
13819 (build_tree_list_vec): Likewise.
13821 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13823 * tree.c (make_vector_stat): Remove _stat from name.
13824 (build_vector_stat): Likewise.
13825 * tree.h (make_vector_stat): Remove macro.
13826 (build_vector_stat): Likewise.
13828 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13830 * tree.h (build_var_debug_value): Remove prototype.
13832 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13834 * tree.c (tree_cons_stat): Remove _stat from name.
13835 * tree.h (tree_cons): Remove macro.
13837 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13839 * tree.c (build_vl_exp_stat): Remove _stat from name.
13840 * tree.h (build_vl_exp): Remove macro.
13842 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13844 * tree.c (build_decl_stat): Remove _stat from name.
13845 * tree.h (build_decl): Remove macro.
13847 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13849 * gimple.c (gimple_build_with_ops_stat): Adjust.
13850 (gimple_alloc_stat): Remove _stat from name.
13851 * gimple.h (gimple_alloc): Remove macro.
13853 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13855 * tree.c (make_tree_vec_stat): Remove _stat from name.
13856 (grow_tree_vec_stat): Likewise.
13857 * tree.h (make_tree_vec_stat): Adjust prototype.
13858 (grow_tree_vec_stat): Likewise.
13859 (make_tree_vec): Remove macro.
13860 (grow_tree_vec): Likewise.
13862 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13864 * fold-const.c (fold_build1_stat_loc): Adjust.
13865 (fold_build2_stat_loc): Likewise.
13866 (fold_build3_stat_loc): Likewise.
13867 * tree.c (build0_stat): Remove _stat from name.
13868 (build1_stat): Likewise.
13869 (build2_stat): Likewise.
13870 (build3_stat): Likewise.
13871 (build4_stat): Likewise.
13872 (build5_stat): Likewise.
13873 * tree.h (build1_loc): Remove macro, and rename _stat function
13875 (build2_loc): Likewise.
13876 (build3_loc): Likewise.
13877 (build4_loc): Likewise.
13878 (build5_loc): Likewise.
13880 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13882 * tree.c (make_int_cst_stat): Remove _stat from name.
13883 * tree.h (make_int_cst_stat): Adjust prototype.
13884 (make_int_cst): Remove macro.
13886 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13888 * tree.c (make_tre_binfo_stat): Remove _stat from name.
13889 * tree.h (make_tree_binfo_stat): Adjust prototype.
13890 (make_tree_binfo): Remove.
13892 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13894 * tree.c (copy_node_stat): Rename to copy_node.
13895 (build_distinct_type_copy): Adjust.
13896 * tree.h (copy_node_stat): Adjust prototype.
13897 (copy_node): Remove macro.
13899 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13901 * tree.c (make_node_stat): rename to make_node.
13902 (build_tree_list_stat): Adjust.
13903 (build0_stat): Likewise.
13904 (build2_stat): Likewise.
13905 (build3_stat): Likewise.
13906 (build4_stat): Likewise.
13907 (build5_stat): Likewise.
13908 (build_decl_stat): Likewise.
13909 * tree.h (make_node_stat): Adjust prototype.
13910 (make_node): remove macro.
13912 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
13914 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
13915 (PPC_FEATURE2_SCV): Likewise.
13916 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
13918 2017-07-28 Tamar Christina <tamar.christina@arm.com>
13920 * config/aarch64/aarch64.c
13921 (aarch64_internal_mov_immediate): Add new special pattern.
13922 * config/aarch64/aarch64.md (*movdi_aarch64):
13923 Add reg/32bit const mov case.
13925 2017-07-28 Tamar Christina <tamar.christina@arm.com>
13926 Richard Sandiford <richard.sandiford@linaro.org>
13928 * config/aarch64/aarch64.md (mov<mode>): Generalize.
13929 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
13930 Add integer and movi cases.
13931 (movi-split-hf-df-sf split, fp16): New.
13932 (enabled): Added TARGET_FP_F16INST.
13933 * config/aarch64/iterators.md (GPF_HF): New.
13934 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
13936 2017-07-28 Tamar Christina <tamar.christina@arm.com>
13938 * config/aarch64/aarch64.c
13939 (aarch64_simd_container_mode): Add prototype.
13940 (aarch64_expand_mov_immediate): Add HI support.
13941 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
13942 (aarch64_can_const_movi_rtx_p): New.
13943 (aarch64_preferred_reload_class):
13944 Remove restrictions of using FP registers for certain SIMD operations.
13945 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
13946 (aarch64_valid_floating_const): Add integer move validation.
13947 (aarch64_simd_imm_scalar_p): Remove.
13948 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
13949 (aarch64_legitimate_constant_p): Expand list of supported cases.
13950 * config/aarch64/aarch64-protos.h
13951 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
13952 (aarch64_reinterpret_float_as_int): New.
13953 (aarch64_simd_imm_scalar_p): Remove.
13954 * config/aarch64/constraints.md (Uvi): New.
13955 (Dd): Split into Ds and new Dd.
13956 * config/aarch64/aarch64.md (*movsi_aarch64):
13958 (*movdi_aarch64): Add SIMD mov case.
13960 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13962 * tree-predcom.c: (struct chain): Handle store-store chain in which
13963 stores for elimination only store loop invariant values.
13964 (execute_pred_commoning_chain): Ditto.
13965 (prepare_initializers_chain_store_elim): Ditto.
13966 (prepare_finalizers): Ditto.
13967 (is_inv_store_elimination_chain): New function.
13968 (initialize_root_vars_store_elim_1): New function.
13970 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13972 * tree-predcom.c: Revise general description of the pass.
13973 (enum chain_type): New enum type for store elimination.
13974 (struct chain): New field supporting store elimination.
13975 (struct component): Ditto.
13976 (dump_chain): Dump store-stores chain.
13977 (release_chain): Release resources.
13978 (split_data_refs_to_components): Compute and create component
13979 contains only stores for elimination.
13980 (get_chain_last_ref_at): New function.
13981 (make_invariant_chain): Initialization.
13982 (make_rooted_chain): Specify chain type in parameter and record it.
13983 (add_looparound_copies): Skip for store-stores chain.
13984 (determine_roots_comp): Compute type of chain and pass it to
13986 (initialize_root_vars_store_elim_2): New function.
13987 (finalize_eliminated_stores): New function.
13988 (remove_stmt): Handle store for elimination.
13989 (execute_pred_commoning_chain): Execute predictive commoning on
13990 store-store chains.
13991 (determine_unroll_factor): Skip unroll for store-stores chain.
13992 (prepare_initializers_chain_store_elim): New function.
13993 (prepare_initializers_chain): Hanlde store-store chain.
13994 (prepare_finalizers_chain, prepare_finalizers): New function.
13995 (tree_predictive_commoning_loop): Return integer value indicating
13996 if loop is unrolled or lcssa form is corrupted.
13997 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
13999 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14001 * tree-predcom.c (initialize_root): Delete.
14002 (execute_pred_commoning_chain): Initialize root vars and replace
14003 reference of non-combined chain directly, rather than call above
14006 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14008 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
14009 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
14011 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14013 * tree-predcom.c (struct chain): New field init_seq.
14014 (release_chain): Release init_seq.
14015 (prepare_initializers_chain): Record intialization stmts in above
14017 (insert_init_seqs): New function.
14018 (tree_predictive_commoning_loop): Call insert_init_seqs.
14020 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14022 * tree-predcom.c (determine_roots_comp): Skip trivial components.
14024 2017-07-28 Richard Biener <rguenther@suse.de>
14026 * match.pd: Remove superfluous :c.
14027 * genmatch.c (simplify::id): Add member.
14028 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
14030 (current_id): New global.
14031 (dt_node::parent): Move from ...
14032 (dt_operand::parent): ... here. Add for_id member.
14033 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
14034 (decision_tree::find_node): Relax order requirement when
14035 merging DT_TRUE nodes to ones inbetween the current simplify
14036 and the one we try to merge with. Add diagnostic whenever
14037 we need to enforce pattern order by not merging.
14038 (decision_tree::insert): Set current_id.
14039 (decision_tree::print_node): Dump parent node and for_id.
14040 (parser::last_id): Add member.
14041 (parser::push_simplify): Assign unique id.
14042 (parser::parser): Initialize last_id.
14044 2017-07-28 Martin Liska <mliska@suse.cz>
14047 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
14048 gimple_build_debug_bind.
14050 2017-07-28 Richard Biener <rguenther@suse.de>
14052 PR tree-optimization/81502
14053 * match.pd: Add pattern combining BIT_INSERT_EXPR with
14055 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
14057 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
14058 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
14059 for BIT_FIELD_REF args.
14060 * fold-const.c (make_bit_field_ref): Likewise.
14061 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
14063 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14066 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
14067 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
14068 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
14069 Or it into SANITIZER_UNDEFINED.
14070 * ubsan.c: Include gimple-fold.h and varasm.h.
14071 (ubsan_expand_ptr_ifn): New function.
14072 (instrument_pointer_overflow): New function.
14073 (maybe_instrument_pointer_overflow): New function.
14074 (instrument_object_size): Formatting fix.
14075 (pass_ubsan::execute): Call instrument_pointer_overflow
14076 and maybe_instrument_pointer_overflow.
14077 * internal-fn.c (expand_UBSAN_PTR): New function.
14078 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
14079 * sanitizer.def (__ubsan_handle_pointer_overflow,
14080 __ubsan_handle_pointer_overflow_abort): New builtins.
14081 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
14082 * internal-fn.def (UBSAN_PTR): New internal function.
14083 * opts.c (sanitizer_opts): Add pointer-overflow.
14084 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
14085 * fold-const.c (build_range_check): Compute pointer range check in
14086 integral type if pointer arithmetics would be needed. Formatting
14089 2017-07-28 Martin Liska <mliska@suse.cz>
14092 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
14093 parameters that are of a variable-length.
14095 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
14097 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
14099 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
14100 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
14101 (CRT_CALL_STATIC_FUNCTION): Likewise.
14102 (ASM_DEFAULT_SPEC): New define.
14103 (ASM_SPEC32): Likewise.
14104 (ASM_SPEC64): Likewise.
14105 (ASM_SPEC_COMMON): Likewise.
14106 (ASM_SPEC): Likewise.
14107 (INVALID_64BIT): Likewise.
14108 (LINK_OS_DEFAULT_SPEC): Likewise.
14109 (LINK_OS_SPEC32): Likewise.
14110 (LINK_OS_SPEC64): Likewise.
14111 (POWERPC_LINUX): Likewise.
14112 (PTRDIFF_TYPE): Likewise.
14113 (RESTORE_FP_PREFIX): Likewise.
14114 (RESTORE_FP_SUFFIX): Likewise.
14115 (SAVE_FP_PREFIX): Likewise.
14116 (SAVE_FP_SUFFIX): Likewise.
14117 (SIZE_TYPE): Likewise.
14118 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
14119 (TARGET_64BIT): Likewise.
14120 (TARGET_64BIT): Likewise.
14121 (TARGET_AIX): Likewise.
14122 (WCHAR_TYPE_SIZE): Likewise.
14123 (WCHAR_TYPE): Undefine.
14124 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
14125 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
14126 (CPP_OS_RTEMS_SPEC): Delete.
14127 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
14128 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
14130 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
14132 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14134 PR tree-optimization/81578
14135 * tree-parloops.c (build_new_reduction): Bail out if
14136 reduction_code isn't one of the standard OpenMP reductions.
14137 Move the details printing after that decision.
14139 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14141 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
14142 related to reload_in_progress.
14143 (splat_input_operand): Likewise.
14144 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
14146 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
14148 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
14149 (TARGET_INSTANTIATE_DECLS): Likewise.
14150 (legitimate_indexed_address_p): Delete reload_in_progress code.
14151 (rs6000_debug_legitimate_address_p): Likewise.
14152 (rs6000_eliminate_indexed_memrefs): Likewise.
14153 (rs6000_emit_le_vsx_store): Likewise.
14154 (rs6000_emit_move_si_sf_subreg): Likewise.
14155 (rs6000_emit_move): Likewise.
14156 (register_to_reg_type): Likewise.
14157 (rs6000_pre_atomic_barrier): Likewise.
14158 (rs6000_machopic_legitimize_pic_address): Likewise.
14159 (rs6000_allocate_stack_temp): Likewise.
14160 (rs6000_address_for_fpconvert): Likewise.
14161 (rs6000_address_for_altivec): Likewise.
14162 (rs6000_secondary_memory_needed_rtx): Delete function.
14163 (rs6000_check_sdmode): Likewise.
14164 (rs6000_alloc_sdmode_stack_slot): Likewise.
14165 (rs6000_instantiate_decls): Likewise.
14166 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
14167 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
14168 Delete reload_in_progress.
14169 (*vec_reload_and_plus_<mptrsize>): Likewise.
14170 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
14171 (vsx_div_v2di): Likewise.
14172 (vsx_udiv_v2di): Likewise.
14174 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14176 * config/rs6000/rs6000.opt (mlra): Replace with stub.
14177 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
14178 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
14179 (rs6000_debug_reg_global): Delete print of LRA status.
14180 (rs6000_option_override_internal): Delete dead LRA related code.
14181 (rs6000_lra_p): Delete function.
14182 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
14184 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
14186 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
14187 * config/riscv/rtems.h: New file.
14189 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14190 Sudakshina Das <sudi.das@arm.com>
14192 * config/aarch64/aarch64.md
14193 (define_split for and<mode>3nr_compare): Move
14194 non aarch64_logical_operand to a register.
14195 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
14196 register immediate operand to a register.
14197 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
14199 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14201 PR middle-end/81564
14202 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
14204 2017-07-27 Richard Biener <rguenther@suse.de>
14206 PR tree-optimization/81573
14207 PR tree-optimization/81494
14208 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
14209 multi defuse cycle case.
14211 2017-07-27 Richard Biener <rguenther@suse.de>
14213 PR tree-optimization/81571
14214 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
14217 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
14219 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
14220 earlier and only if MASK_FPU is set. Adjust formatting.
14222 2017-07-27 Martin Liska <mliska@suse.cz>
14224 * opt-functions.awk: Add validation of value of Init.
14225 * optc-gen.awk: Pass new argument.
14227 2017-07-27 Martin Liska <mliska@suse.cz>
14229 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
14230 Fix wrong condition.
14232 2017-07-27 Martin Liska <mliska@suse.cz>
14234 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
14235 BBs and edges seen by autoFDO.
14237 2017-07-27 Richard Biener <rguenther@suse.de>
14239 PR tree-optimization/81502
14240 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
14241 with incompatible but same sized type.
14242 (execute_update_addresses_taken): Likewise.
14244 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
14246 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
14247 flag_tree_loop_vectorize rather than flag_tree_vectorize.
14249 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14252 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
14253 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
14254 Change s_operand to memory_operand.
14256 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
14258 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
14259 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
14260 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
14261 Emit instructions rather than returning an expression. Handle TFmode
14262 and KFmode by casting to TImode.
14263 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
14264 (rs6000_emit_le_vsx_store): Likewise.
14265 * config/rs6000/vsx.md (VSX_TI): New iterator.
14266 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
14267 (*vsx_le_undo_permute_<mode>): Likewise.
14268 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
14269 emit the split sequence.
14270 (*vsx_le_perm_store_<mode>): Likewise.
14272 2017-07-27 Jakub Jelinek <jakub@redhat.com>
14274 PR tree-optimization/81555
14275 PR tree-optimization/81556
14276 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
14277 if true, force CHANGED for the recursive invocation.
14278 (reassociate_bb): Remember original length of ops array, pass
14279 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
14281 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
14282 attributes for noipa attribute. For naked attribute use
14283 lookup_attribute first before lookup_attribute_spec.
14284 * final.c (rest_of_handle_final): Disable IPA RA for functions with
14286 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
14287 for functions with noipa attribute.
14288 (cgraph_externally_visible_p): Return true for functions with noipa
14290 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
14291 for functions with noipa attribute.
14292 * doc/extend.texi: Document noipa function attribute.
14293 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
14294 also for functions with noipa attribute.
14295 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
14297 2017-07-26 Andrew Pinski <apinski@cavium.com>
14299 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
14300 vec_unalign_load_cost and vec_unalign_store_cost.
14302 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14304 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
14305 -mvsx-small-integer option.
14306 (ISA_3_0_MASKS_IEEE): Likewise.
14307 (OTHER_VSX_VECTOR_MASKS): Likewise.
14308 (POWERPC_MASKS): Likewise.
14309 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
14310 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
14311 code, only testing for DImode being allowed in non-VSX floating
14313 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
14314 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
14316 (rs6000_option_override_internal): Delete -mvsx-small-integer.
14317 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
14318 TARGET_P8_VECTOR test.
14319 (rs6000_secondary_reload_simple_move): Likewise.
14320 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
14321 since TARGET_P9_VECTOR was already tested.
14322 (rs6000_opt_masks): Remove -mvsx-small-integer.
14323 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
14324 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14326 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
14327 test for TARGET_VEXTRACTUB was used, and that uses
14329 (p9 extract splitter): Likewise.
14330 (vsx_extract_<mode>_di_p9): Likewise.
14331 (vsx_extract_<mode>_store_p9): Likewise.
14332 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
14333 for TARGET_P9_VECTOR was used. Delete code that is now dead with
14334 the elimination of TARGET_VSX_SMALL_INTEGER.
14335 (vsx_extract_<mode>_p8): Likewise.
14336 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
14337 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
14338 (vsx_set_<mode>_p9): Likewise.
14339 (vsx_set_v4sf_p9): Likewise.
14340 (vsx_set_v4sf_p9_zero): Likewise.
14341 (vsx_insert_extract_v4sf_p9): Likewise.
14342 (vsx_insert_extract_v4sf_p9_2): Likewise.
14343 * config/rs6000/rs6000.md (sign extend splitter): Change
14344 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
14345 (floatsi<mode>2_lfiwax_mem): Likewise.
14346 (floatunssi<mode>2_lfiwzx_mem): Likewise.
14347 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
14348 since a test for TARGET_P9_VECTOR was used.
14349 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14350 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
14351 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14352 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
14353 TARGET_P8_VECTOR test.
14354 (fix_trunc<mode>si2_stfiwx): Likewise.
14355 (fix_trunc<mode>si2_internal): Likewise.
14356 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
14357 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14359 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
14360 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
14361 TARGET_P8_VECTOR test.
14362 (fixuns_trunc<mode>si2_stfiwx): Likewise.
14363 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
14364 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14366 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
14367 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
14368 since a test for TARGET_P9_VECTOR was used.
14369 (splitter for loading small constants): Likewise.
14371 2017-07-26 Andrew Pinski <apinski@cavium.com>
14373 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
14376 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
14379 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
14380 (fp_valid_at): Likewise.
14382 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
14384 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
14385 (qdf24xx_addrcost_table): Likewise.
14386 (cortexa57_tunings): Update to use generic_branch_cost.
14387 (cortexa72_tunings): Likewise.
14388 (cortexa73_tunings): Likewise.
14389 (qdf24xx_tunings): Likewise.
14391 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
14393 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
14394 (thunderx2t99_branch_cost): Likewise.
14395 (cortexa35_tunings): Update to use generic_branch_cost.
14396 (cortexa53_tunings): Likewise.
14397 (cortexa57_tunings): Likewise.
14398 (cortexa72_tunings): Likewise.
14399 (cortexa73_tunings): Likewise.
14400 (thunderx2t99_tunings): Likewise.
14402 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
14404 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
14405 (sparc_option_override): Honour MASK_FSMULD.
14406 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
14407 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
14408 * config/sparc/sparc.opt (mfsmuld): New option.
14409 * doc/invoke.texi (mfsmuld): Document option.
14411 2017-07-26 Marek Polacek <polacek@redhat.com>
14413 PR middle-end/70992
14414 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
14416 2017-07-26 Richard Biener <rguenther@suse.de>
14418 * gimple-match-head.c (do_valueize): Return OP if valueize
14420 (get_def): New helper to get at the def stmt of a SSA name
14421 if valueize allows.
14422 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
14423 do_valueize to get at the def stmt.
14424 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
14426 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
14428 PR middle-end/46932
14429 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
14431 2017-07-26 Martin Liska <mliska@suse.cz>
14434 * function.c (expand_function_start): Make expansion of
14435 nonlocal_goto_save_area after parm_birth_insn.
14437 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
14439 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
14440 from all CPU target flags enable members.
14442 2017-07-26 Richard Biener <rguenther@suse.de>
14444 * genmatch.c (dt_simplify::gen): Make iterator vars const.
14445 (decision_tree::gen): Make 'type' const.
14446 (write_predicate): Likewise.
14448 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
14450 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
14451 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
14452 (rs6000_option_override_internal): Likewise.
14453 (rs6000_expand_vector_set): Likewise.
14454 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
14455 (TARGET_UPPER_REGS_SF): Likewise.
14456 (TARGET_UPPER_REGS_DI): Likewise.
14457 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
14458 (TARGET_DIRECT_MOVE_64BIT): Likewise.
14459 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
14460 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14461 (Splitters for DI constants in Altivec registers): Likewise.
14462 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
14463 (vsx_set_v4sf_p9): Likewise.
14464 (vsx_set_v4sf_p9_zero): Likewise.
14465 (vsx_insert_extract_v4sf_p9): Likewise.
14466 (vsx_insert_extract_v4sf_p9_2): Likewise.
14468 2017-07-25 Carl Love <cel@us.ibm.com>
14470 * doc/extend.texi: Update the built-in documentation file for the
14471 existing built-in functions
14472 vector signed char vec_cnttz (vector signed char);
14473 vector unsigned char vec_cnttz (vector unsigned char);
14474 vector signed short vec_cnttz (vector signed short);
14475 vector unsigned short vec_cnttz (vector unsigned short);
14476 vector signed int vec_cnttz (vector signed int);
14477 vector unsigned int vec_cnttz (vector unsigned int);
14478 vector signed long long vec_cnttz (vector signed long long);
14479 vector unsigned long long vec_cnttz (vector unsigned long long);
14481 2017-07-25 Andrew Pinski <apinski@cavium.com>
14483 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
14484 accesses where the use is for the first operand of a BIT_INSERT.
14486 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
14489 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
14490 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
14492 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
14494 * config/i386/gstabs.h: Delete.
14495 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
14497 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
14499 * config/i386/i386.c (ix86_decompose_address): Do not check for
14500 register RTX when looking at index_reg or base_reg.
14501 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
14503 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
14505 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
14506 to update EH info here.
14508 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
14510 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
14512 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
14514 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
14516 2017-07-25 Torsten Duwe <duwe@suse.de>
14518 * common.opt: Introduce -fpatchable-function-entry
14519 command line option, and its variables function_entry_patch_area_size
14520 and function_entry_patch_area_start.
14521 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
14522 including a two-value parser.
14523 * target.def (print_patchable_function_entry): New target hook.
14524 * targhooks.h (default_print_patchable_function_entry): New function.
14525 * targhooks.c (default_print_patchable_function_entry): Likewise.
14526 * toplev.c (process_options): Switch off IPA-RA if
14527 patchable function entries are being generated.
14528 * varasm.c (assemble_start_function): Look at the
14529 patchable-function-entry command line switch and current
14530 function attributes and maybe generate NOP instructions by
14531 calling the print_patchable_function_entry hook.
14532 * doc/extend.texi: Document patchable_function_entry attribute.
14533 * doc/invoke.texi: Document -fpatchable_function_entry
14534 command line option.
14535 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
14537 * doc/tm.texi: Re-generate.
14539 2017-07-25 Jakub Jelinek <jakub@redhat.com>
14542 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
14543 TARGET_AVX512DQ rather than TARGET_AVX512BW.
14545 2017-07-25 Tamar Christina <tamar.christina@arm.com>
14547 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
14549 2017-07-25 Richard Biener <rguenther@suse.de>
14551 PR tree-optimization/81455
14552 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
14553 not walk in cycles when looking for guards.
14555 2017-07-25 Richard Biener <rguenther@suse.de>
14557 PR tree-optimization/81529
14558 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
14559 when optimizing backedge uses.
14561 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
14563 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
14565 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
14566 to dl_section_ref. On AIX, append an expression to subtract
14567 the size of the section length to dl_section_ref.
14569 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
14571 * configure.ac: If any of the config.* scripts fail, exit 1.
14572 * configure: Regenerate.
14574 2017-07-25 Richard Biener <rguenther@suse.de>
14576 PR middle-end/81546
14577 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
14578 of immediate uses, be more verbose on errors.
14580 2017-07-25 Richard Biener <rguenther@suse.de>
14582 PR tree-optimization/81510
14583 * tree-vect-loop.c (vect_is_simple_reduction): When the
14584 reduction stmt is not inside the loop bail out.
14586 2017-07-25 Richard Biener <rguenther@suse.de>
14588 PR tree-optimization/81303
14589 * tree-vect-loop-manip.c (vect_loop_versioning): Build
14590 profitability check against LOOP_VINFO_NITERSM1.
14592 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
14594 * domwalk.c (cmp_bb_postorder): Simplify.
14595 (sort_bbs_postorder): New function. Use it...
14596 (dom_walker::walk): ...here to optimize common cases.
14598 2017-07-25 Martin Liska <mliska@suse.cz>
14601 * ipa-visibility.c (function_and_variable_visibility): Make the
14602 redirection just on target that supports aliasing.
14603 Fix GNU coding style.
14605 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
14608 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
14610 * config/arm/rtems.h: Likewise.
14611 * config/bfin/rtems.h: Likewise.
14612 * config/i386/rtemself.h: Likewise.
14613 * config/lm32/rtems.h: Likewise.
14614 * config/m32c/rtems.h: Likewise.
14615 * config/m68k/rtemself.h: Likewise.
14616 * config/microblaze/rtems.h: Likewise.
14617 * config/mips/rtems.h: Likewise.
14618 * config/moxie/rtems.h: Likewise.
14619 * config/nios2/rtems.h: Likewise.
14620 * config/powerpcspe/rtems.h: Likewise.
14621 * config/rs6000/rtems.h: Likewise.
14622 * config/rtems.h: Likewise.
14623 * config/sh/rtems.h: Likewise.
14624 * config/sh/rtemself.h: Likewise.
14625 * config/sparc/rtemself.h: Likewise.
14627 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
14630 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
14631 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
14632 * tree-ssa-structalias.c (alias_get_name): Same.
14634 2017-07-25 Bin Cheng <bin.cheng@arm.com>
14637 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
14638 instructions if no du chain is found.
14640 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
14642 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
14644 2017-07-25 Richard Biener <rguenther@suse.de>
14646 PR middle-end/81505
14647 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
14650 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
14652 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
14653 upper-regs options.
14654 (ISA_2_7_MASKS_SERVER): Likewise.
14655 (ISA_3_0_MASKS_IEEE): Likewise.
14656 (OTHER_P8_VECTOR_MASKS): Likewise.
14657 (OTHER_VSX_VECTOR_MASKS): Likewise.
14658 (POWERPC_MASKS): Likewise.
14659 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
14660 duplicate list of options.
14661 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
14662 explicit -mupper-regs options.
14663 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
14664 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
14665 alias for -mupper-regs-df.
14666 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
14667 (rs6000_init_hard_regno_mode_ok): Likewise.
14668 (rs6000_option_override_internal): Likewise.
14669 (rs6000_opt_masks): Likewise.
14670 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
14671 options in terms of whether -mvsx or -mpower8-vector was used.
14672 (TARGET_UPPER_REGS_DI): Likewise.
14673 (TARGET_UPPER_REGS_SF): Likewise.
14674 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
14675 -mupper-regs-* options.
14677 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
14679 * passes.c (emergency_dump_function): Print some empty lines and a
14680 header before the RTL dump.
14682 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
14684 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
14686 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
14689 * config/aarch64/aarch64.c (aarch64_classify_symbol):
14690 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
14692 2017-07-24 Carl Love <cel@us.ibm.com>
14694 * config/rs6000/rs6000-c.c: Add support for built-in functions
14695 vector float vec_extract_fp32_from_shorth (vector unsigned short);
14696 vector float vec_extract_fp32_from_shortl (vector unsigned short);
14697 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
14698 vec_extract_fp_from_shortl): Add defines for the two builtins.
14699 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
14700 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
14702 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
14703 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
14704 * doc/extend.texi: Update the built-in documentation file for the
14705 new built-in function.
14707 2017-07-24 Jakub Jelinek <jakub@redhat.com>
14710 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
14712 * doc/generic.texi: Likewise.
14713 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
14714 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
14716 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
14718 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
14719 (aarch64_mls_elt_merge<mode>): Likewise.
14721 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
14723 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
14724 having __cxa_atexit.
14726 2017-07-23 Michael Collison <michael.collison@arm.com>
14728 * config/arm/arm.c (arm_option_override): Deprecate
14729 use of -mstructure-size-boundary.
14730 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
14731 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
14733 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14736 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
14737 Reduce cost estimate for direct moves.
14739 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
14742 * config/i386/i386.c (ix86_option_override_internal): Disable
14743 BMI, BMI2 and TBM instructions for -m16.
14745 2017-07-21 Carl Love <cel@us.ibm.com>
14747 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14748 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14749 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14750 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14751 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14752 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
14753 VMULOSW): New enum "unspec" values.
14754 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
14755 altivec_vmulosw): New patterns.
14756 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14757 VMULOSW): Add definitions.
14759 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
14761 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
14762 (qdf24xx): Likewise.
14763 * config/aarch64/aarch64-options-extensions.def (rdma); New.
14764 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
14765 (AARCH64_FL_V8_1): Renumber.
14766 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
14767 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
14768 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
14769 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
14770 rdma to feature modifiers list.
14772 2017-07-21 Yury Gribov <tetra2005@gmail.com>
14774 PR middle-end/56727
14775 * ipa-visibility (function_and_variable_visibility): Convert
14776 recursive PLT call to direct call if appropriate.
14778 2017-07-21 Andrew Pinski <apinski@cavium.com>
14780 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
14781 operand 1 to see if the types precision matches.
14782 * fold-const.c (operand_equal_p): Likewise.
14784 2017-07-21 Richard Biener <rguenther@suse.de>
14786 PR tree-optimization/81303
14787 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
14788 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
14789 (vect_peeling_hash_get_lowest_cost): Adjust.
14790 (vect_enhance_data_refs_alignment): Likewise. Use
14791 vect_get_peeling_costs_all_drs to compute the penalty for no
14792 peeling to match up costs.
14794 2017-07-21 Richard Biener <rguenther@suse.de>
14796 PR tree-optimization/81500
14797 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
14798 we didn't identify a reduction path.
14800 2017-07-21 Tom de Vries <tom@codesourcery.com>
14801 Cesar Philippidis <cesar@codesourcery.com>
14803 PR gcov-profile/81442
14804 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
14807 2017-07-21 Tom de Vries <tom@codesourcery.com>
14810 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
14812 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
14813 nvptx_override_options_after_change.
14815 2017-07-21 Ulrich Drepper <drepper@redhat.com>
14817 * dwarf2out.c (output_file_names): Avoid double testing for
14818 dwarf_version >= 5.
14820 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
14822 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
14824 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
14826 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
14828 (try_crossjump_to_edge): Do not punt on partitioned functions.
14830 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
14832 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
14833 Put all BBs reachable only via paths crossing cold region to cold
14835 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
14837 2016-07-21 Richard Biener <rguenther@suse.de>
14839 PR tree-optimization/81303
14840 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
14841 into account prologue and epilogue iterations when raising
14842 min_profitable_iters to sth at least covering one vector iteration.
14844 2017-07-21 Tamar Christina <tamar.christina@arm.com>
14846 * config/arm/arm.c (arm_test_cpu_arch_dat):
14849 2017-07-20 Nathan Sidwell <nathan@acm.org>
14851 Remove TYPE_METHODS.
14852 * tree.h (TYPE_METHODS): Delete.
14853 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
14854 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
14855 (dbxout_type_methods): Scan TYPE_FIELDS.
14856 (dbxout_type): Don't check TYPE_METHODS here.
14857 * function.c (use_register_for_decl): Always ignore register for
14858 class types when not optimizing.
14859 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
14860 * tree.c (free_lang_data_in_type): Stitch out member functions and
14861 templates from TYPE_FIELDS.
14862 (build_distinct_type_copy, verify_type_variant,
14863 verify_type): Member fns are on TYPE_FIELDS.
14864 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
14865 * tree-pretty-print.c (dump_generic_node): Likewise.
14867 2017-07-20 Jakub Jelinek <jakub@redhat.com>
14870 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
14871 V2TImode and V4TImode.
14872 (ix86_expand_vector_extract): Likewise.
14873 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
14874 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
14875 (ssescalarmode): Handle V4TImode and V2TImode.
14876 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
14877 (*vec_extractv2ti, *vec_extractv4ti): New insns.
14878 (VEXTRACTI128_MODE): New mode iterator.
14879 (splitter for *vec_extractv?ti first element): New.
14880 (VEC_INIT_MODE): New mode iterator.
14881 (vec_init<mode>): Consolidate 3 expanders into one using
14882 VEC_INIT_MODE mode iterator.
14884 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
14886 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
14887 non_spilled_static_chain_regno_p.
14889 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
14891 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
14893 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
14895 * bb-reorder.c (connect_traces): Allow copying of blocks within
14898 2017-07-20 Richard Biener <rguenther@suse.de>
14900 * gimple.h (gimple_phi_result): Add gphi * overload.
14901 (gimple_phi_result_ptr): Likewise.
14902 (gimple_phi_arg): Likewise. Adjust index assert to only
14903 allow actual argument accesses rather than all slots available
14905 (gimple_phi_arg_def): Add gphi * overload.
14906 * tree-phinodes.c (make_phi_node): Initialize only actual
14908 (resize_phi_node): Clear memory not covered by old node,
14909 do not initialize excess argument slots.
14910 (reserve_phi_args_for_new_edge): Initialize new argument slot
14913 2017-07-20 Bin Cheng <bin.cheng@arm.com>
14915 PR tree-optimization/81388
14917 2016-07-21 Bin Cheng <bin.cheng@arm.com>
14919 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
14920 by removing computation of may_be_zero.
14922 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
14923 Tom de Vries <tom@codesourcery.com>
14925 PR middle-end/81030
14926 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
14927 when gimple level profile disagrees with what RTL expander did.
14929 2017-07-20 Richard Biener <rguenther@suse.de>
14931 PR tree-optimization/61171
14932 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
14933 (vect_analyze_stmt): Add slp instance parameter.
14934 (vectorizable_reduction): Likewise.
14935 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
14936 (vect_is_simple_reduction): Deal with chains not detected
14937 as SLP reduction chain, specifically not properly associated
14938 chains containing a mix of plus/minus.
14939 (get_reduction_op): Remove.
14940 (get_initial_defs_for_reduction): Simplify, pass in whether
14941 this is a reduction chain, pass in the SLP node for the PHIs.
14942 (vect_create_epilog_for_reduction): Get the SLP instance as
14944 (vectorizable_reduction): Get the SLP instance as arg.
14945 During analysis remember the SLP node with the PHIs in the
14946 instance. Simplify getting at the vectorized reduction PHIs.
14947 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
14948 through SLP instance.
14949 (vect_slp_analyze_operations): Likewise.
14950 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
14951 (vect_transform_stmt): Likewise.
14953 2017-07-20 Tom de Vries <tom@codesourcery.com>
14955 PR tree-optimization/81489
14956 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
14957 read of phi arg location to before loop that modifies phi.
14959 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14961 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
14964 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
14966 PR middle-end/81331
14967 * except.c (execute): Fix ordering issue.
14969 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
14971 PR rtl-optimization/81423
14972 * combine.c (make_compound_operation_int): Don't try to optimize
14973 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
14975 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
14977 PR rtl-optimization/81423
14978 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
14979 with a constant that is -1 in the truncated to mode.
14981 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
14983 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
14984 (determine_unlikely_bbs): ... here.
14985 * predict.h (propagate_unlikely_bbs_forward): Declare.
14986 * cfgexpand.c (pass_expand::execute): Use it.
14987 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
14989 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
14990 propagate_unlikely_bbs_forward.
14992 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
14994 PR middle-end/81331
14995 * except.c (maybe_add_nop_after_section_switch): New function.
14998 2017-07-19 Tom de Vries <tom@codesourcery.com>
15000 * gimple.h (gimple_phi_set_arg): Make assert more strict.
15002 2017-07-19 Tom de Vries <tom@codesourcery.com>
15004 * gimple.h (gimple_phi_arg): Make assert more strict.
15006 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
15008 * config.gcc (powerpc*-*-*): Add mmintrin.h.
15009 * config/rs6000/mmintrin.h: New file.
15010 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
15012 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15014 PR tree-optimization/81346
15015 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
15017 2017-07-19 Tom de Vries <tom@codesourcery.com>
15019 * config/nvptx/nvptx.md (VECIM): Add V2DI.
15021 2017-07-19 Tom de Vries <tom@codesourcery.com>
15023 * config/nvptx/nvptx-modes.def: Add V2DImode.
15024 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
15025 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
15026 (nvptx_output_mov_insn): Handle lack of mov.b128.
15027 (nvptx_print_operand): Handle 'H' and 'L' codes.
15028 (nvptx_vector_mode_supported): Allow V2DImode.
15029 (nvptx_preferred_simd_mode): New function.
15030 (nvptx_data_alignment): New function.
15031 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
15032 nvptx_preferred_simd_mode.
15033 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
15035 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
15037 2017-07-19 Tom de Vries <tom@codesourcery.com>
15039 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
15040 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
15041 (nvptx_vector_mode_supported): New function. Allow V2SImode.
15042 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
15043 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
15044 (mov<VECIM>_insn): New define_insn.
15045 (define_expand "mov<VECIM>): New define_expand.
15047 2017-07-19 Tom de Vries <tom@codesourcery.com>
15049 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
15051 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15053 PR tree-optimization/81346
15054 * fold-const.h (fold_div_compare, range_check_type): Declare.
15055 * fold-const.c (range_check_type): New function.
15056 (build_range_check): Use range_check_type.
15057 (fold_div_compare): No longer static, rewritten into
15058 a match.pd helper function.
15059 (fold_comparison): Don't call fold_div_compare here.
15060 * match.pd (X / C1 op C2): New optimization using fold_div_compare
15061 as helper function.
15063 2017-07-19 Nathan Sidwell <nathan@acm.org>
15065 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
15066 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
15067 * tree.c (find_decls_types_r, verify_type): Use
15068 TYPE_{MIN,MAX}_VALUE_RAW.
15069 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
15070 (hash_tree): Likewise.
15071 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
15073 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
15076 2017-07-18 Tom de Vries <tom@codesourcery.com>
15078 PR middle-end/81464
15079 * omp-expand.c (expand_omp_for_static_chunk): Handle
15080 equal-argument loop exit phi.
15082 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
15085 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
15086 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
15087 operand 2 predicate.
15088 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
15089 operand 2 predicate.
15090 (ror,rol -> rorx splitters): Use const_int_operand as
15091 operand 2 predicate.
15093 2017-06-18 Richard Biener <rguenther@suse.de>
15095 PR tree-optimization/81410
15096 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
15097 the gap in the ! slp_perm SLP case after each group.
15099 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15101 PR middle-end/81463
15102 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
15105 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15107 PR middle-end/81462
15108 * predict.c (set_even_probabilities): Cleanup; do not affect
15109 probabilities that are already known.
15110 (combine_predictions_for_bb): Call even when count is set.
15112 2017-07-18 Nathan Sidwell <nathan@acm.org>
15114 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
15117 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15120 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
15121 optimization for loop niter analysis.
15123 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
15126 * config/avr/avr.c (avr_optimize_casesi): Don't use
15127 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
15129 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
15131 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
15132 body_cost_vec from _vect_peel_extended_info.
15133 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
15134 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
15137 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15139 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
15141 2017-07-18 Richard Biener <rguenther@suse.de>
15143 PR tree-optimization/80620
15144 PR tree-optimization/81403
15145 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
15146 info when re-using a VN table entry.
15148 2017-07-18 Richard Biener <rguenther@suse.de>
15150 PR tree-optimization/81418
15151 * tree-vect-loop.c (vectorizable_reduction): Properly compute
15152 vectype_in. Verify that with lane-reducing reduction operations
15153 we have a single def-use cycle.
15155 2017-07-17 Carl Love <cel@us.ibm.com>
15157 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
15159 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15160 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15161 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15162 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15163 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15164 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15165 VMULOSW): New enum "unspec" values.
15166 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
15167 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
15168 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15169 altivec_vmulosw): New patterns.
15170 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15171 VMULOSW): Add definitions.
15172 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
15174 * config/alpha/alpha.c: Include predict.h.
15176 2017-07-17 Yury Gribov <tetra2005@gmail.com>
15178 * tree-vrp.c (compare_assert_loc): Fix comparison function
15179 to return predictable results.
15181 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15183 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
15185 (subdi3): Likewise.
15186 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
15187 * doc/invoke.texi (mexpand-adddi): Update text.
15189 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15191 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
15192 that also clobbers the CC register. The old expand code is moved
15194 (*arc_clzsi2): ... here.
15195 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
15196 the CC register. The old expand code is moved to ...
15197 (arc_ctzsi2): ... here.
15199 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15201 * config/arc/arc.opt (mindexed-loads): Use initial value
15202 TARGET_INDEXED_LOADS_DEFAULT.
15203 (mauto-modify-reg): Use initial value
15204 TARGET_AUTO_MODIFY_REG_DEFAULT.
15205 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15206 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15207 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15208 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15210 2017-07-17 Martin Liska <mliska@suse.cz>
15213 * opts.c (finish_options): Do not allow -fgnu-tm
15214 w/ -fsanitize={kernel-,}address. Say sorry.
15216 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15219 * tree-loop-distribution.c (classify_partition): Only assert on
15220 numer of iterations.
15221 (merge_dep_scc_partitions): Delete prameter. Update function call.
15222 (distribute_loop): Remove code handling loop with unknown niters.
15223 (pass_loop_distribution::execute): Skip loop with unknown niters.
15225 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15228 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
15229 function sort_partitions_by_post_order.
15231 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15233 PR tree-optimization/81374
15234 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
15235 the max index of basic blocks, rather than number of basic blocks.
15237 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15239 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
15241 (arc_legitimate_pic_operand_p): Likewise.
15242 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
15244 (arc_needs_pcl_p): Likewise.
15245 (arc_legitimate_pc_offset_p): Likewise.
15246 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
15247 function is also used in constrains.md.
15248 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
15249 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
15250 PLUS. Only return true/false in known cases, otherwise assert.
15251 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
15252 is already called in arc_legitimate_constant_p.
15253 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
15255 (LEGITIMATE_PIC_OPERAND_P): Use
15256 arc_raw_symbolic_reference_mentioned_p function.
15257 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
15262 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15263 Andrew Burgess <andrew.burgess@embecosm.com>
15265 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
15266 (arc_return_address_register): New function.
15267 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
15268 (arc_handle_fndecl_attribute): Add naked attribute.
15269 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15270 (TARGET_WARN_FUNC_RETURN): Likewise.
15271 (arc_allocate_stack_slots_for_args): New function.
15272 (arc_warn_func_return): Likewise.
15273 (machine_function): Change type fn_type.
15274 (arc_compute_function_type): Consider new naked function type,
15275 change function return type.
15276 (arc_must_save_register): Adapt to handle new
15277 arc_compute_function_type's return type.
15278 (arc_expand_prologue): Likewise.
15279 (arc_expand_epilogue): Likewise.
15280 (arc_return_address_regs): Delete.
15281 (arc_return_address_register): New function.
15282 (arc_epilogue_uses): Use above function.
15283 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
15284 (arc_function_type): Change encoding, add naked type.
15285 (ARC_INTERRUPT_P): Change to handle the new encoding.
15286 (ARC_FAST_INTERRUPT_P): Likewise.
15287 (ARC_NORMAL_P): Define.
15288 (ARC_NAKED_P): Likewise.
15289 (arc_compute_function_type): Delete prototype.
15290 * config/arc/arc.md (in_ret_delay_slot): Use
15291 arc_return_address_register function.
15292 (simple_return): Likewise.
15293 (p_return_i): Likewise.
15295 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15297 PR tree-optimization/81428
15298 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
15299 can't be built for those types.
15301 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
15303 Remove stuff dead since r239246.
15305 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
15306 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
15307 (avr_inform_devices): Remove dead stuff.
15309 2017-07-17 Tamar Christina <tamar.christina@arm.com>
15311 * config/arm/arm_neon.h: Fix softp typo.
15313 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15315 PR tree-optimization/81365
15316 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
15317 aggregate moves onto bb predecessor edges, make sure there are no
15318 loads that could alias the lhs in between the start of bb and the
15321 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
15324 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
15325 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
15326 [LSHIFTRT, outer_code = TRUNCATE]: Same.
15328 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15330 PR tree-optimization/81396
15331 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
15332 (init_symbolic_number): Initialize it to 1.
15333 (perform_symbolic_merge): Add n_ops from both operands into the new
15335 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
15336 without base_addr as useless if they need more than one operation.
15337 (bswap_replace): Handle !bswap case for NULL base_addr.
15339 2017-07-17 Tom de Vries <tom@codesourcery.com>
15342 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
15345 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
15347 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
15348 conditional builtin define __FIX_LEON3FT_B2BST.
15350 2017-07-17 Daniel Cederman <cederman@gaisler.com>
15352 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
15353 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
15356 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
15358 PR rtl-optimization/81424
15359 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
15360 to remove potential trapping from operands if -fnon-call-exceptions.
15362 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15364 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
15365 profile_proability for scalling.
15366 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
15368 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15370 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
15372 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15374 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
15375 fixpoint arithmetics.
15377 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15379 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
15380 fixpoint arithmetics.
15382 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15384 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
15385 fixpoint arithmetics.
15387 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15389 * profile-count.h (profile_probability::from_reg_br_prob_note,
15390 profile_probability::to_reg_br_prob_note): New functions.
15391 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
15392 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
15393 * predict.c (probability_reliable_p): Update.
15394 (edge_probability_reliable_p): Update.
15395 (br_prob_note_reliable_p): Update.
15396 (invert_br_probabilities): Update.
15397 (add_reg_br_prob_note): New function.
15398 (combine_predictions_for_insn): Update.
15399 * asan.c (asan_clear_shadow): Update.
15400 * cfgbuild.c (compute_outgoing_frequencies): Update.
15401 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
15402 (update_br_prob_note): Update.
15403 (rtl_verify_edges): Update.
15404 (purge_dead_edges): Update.
15405 (fixup_reorder_chain): Update.
15406 * emit-rtl.c (try_split): Update.
15407 * ifcvt.c (cond_exec_process_insns): Update.
15408 (cond_exec_process_if_block): Update.
15409 (dead_or_predicable): Update.
15410 * internal-fn.c (expand_addsub_overflow): Update.
15411 (expand_neg_overflow): Update.
15412 (expand_mul_overflow): Update.
15413 * loop-doloop.c (doloop_modify): Update.
15414 * loop-unroll.c (compare_and_jump_seq): Update.
15415 * optabs.c (emit_cmp_and_jump_insn_1): Update.
15416 * predict.h: Update.
15417 * reorg.c (mostly_true_jump): Update.
15419 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
15420 * config/alpha/alpha.c (emit_unlikely_jump): Update.
15421 * config/arc/arc.c: (emit_unlikely_jump): Update.
15422 * config/arm/arm.c: (emit_unlikely_jump): Update.
15423 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
15424 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
15425 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
15426 (ix86_print_operand): Update.
15427 (ix86_split_fp_branch): Update.
15428 (predict_jump): Update.
15429 * config/ia64/ia64.c (ia64_print_operand): Update.
15430 * config/mmix/mmix.c (mmix_print_operand): Update.
15431 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
15432 (rs6000_expand_split_stack_prologue): Update.
15433 * config/rs6000/rs6000.c: Update.
15434 * config/s390/s390.c (s390_expand_vec_strlen): Update.
15435 (s390_expand_vec_movstr): Update.
15436 (s390_expand_cs_tdsi): Update.
15437 (s390_expand_split_stack_prologue): Update.
15438 * config/sh/sh.c (sh_print_operand): Update.
15439 (expand_cbranchsi4): Update.
15440 (expand_cbranchdi4): Update.
15441 * config/sparc/sparc.c (output_v9branch): Update.
15442 * config/spu/spu.c (get_branch_target): Update.
15443 (ea_load_store_inline): Update.
15444 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
15445 * config/tilepro/tilepro.c: Update.
15447 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
15449 * gimplify.c (mostly_copy_tree_r): Revert latest change.
15450 (gimplify_save_expr): Likewise.
15452 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15454 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
15456 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15458 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
15460 * timevar.def (TV_IPA_FNSUMMARY): Define.
15462 2017-07-16 Daniel Cederman <cederman@gaisler.com>
15464 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
15465 to back store errata sensitive sequence from being generated.
15466 (sqrtdf2_fix): Likewise.
15468 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15470 * tree-ssa-threadupdate.c (compute_path_counts,
15471 update_joiner_offpath_counts): Use profile_probability.
15473 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
15476 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15478 * config/arm/arm-c.c (arm_cpu_builtins): Define
15479 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
15481 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
15483 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15484 array entries to represent __ieee128 versions of the
15485 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
15486 scalar_extract_sig, and scalar_insert_exp built-in functions.
15487 (altivec_resolve_overloaded_builtin): Add special case handling
15488 for the __builtin_scalar_insert_exp function, as represented by
15489 the P9V_BUILTIN_VEC_VSIEDP constant.
15490 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
15491 exponent support for __ieee128 argument.
15492 (VSESQP): Add scalar extract signature support for __ieee128
15494 (VSTDCNQP): Add scalar test negative support for __ieee128
15496 (VSIEQP): Add scalar insert exponent support for __int128 argument
15497 with __ieee128 result.
15498 (VSIEQPF): Add scalar insert exponent support for __ieee128
15499 argument with __ieee128 result.
15500 (VSTDCQP): Add scalar test data class support for __ieee128
15502 (VSTDCNQP): Add overload support for scalar test negative with
15503 __ieee128 argument.
15504 (VSTDCQP): Add overload support for scalar test data class
15505 __ieee128 argument.
15506 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
15507 UNSPEC_VSX_SXSIGDP.
15508 (UNSPEC_VSX_SIEXPQP): New constant.
15509 (xsxexpqp): New insn for VSX scalar extract exponent quad
15511 (xsxsigqp): New insn for VSX scalar extract significand quad
15513 (xsiexpqpf): New insn for VSX scalar insert exponent quad
15514 precision with floating point argument.
15515 (xststdcqp): New expand for VSX scalar test data class quad
15517 (xststdcnegqp): New expand for VSX scalar test negative quad
15519 (xststdcqp): New insn to match expansions for VSX scalar test data
15520 class quad precision and VSX scalar test negative quad precision.
15521 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
15522 special case operand checking to enforce that second operand of
15523 VSX scalar test data class with quad precision argument is a 7-bit
15525 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
15526 prototypes and descriptions of __ieee128 versions of
15527 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
15528 scalar_test_data_class, and scalar_test_neg built-in functions.
15530 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15532 PR tree-optimization/81162
15533 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
15534 replace a negate with an add.
15536 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
15538 * doc/invoke.texi (arm/-mcpu): Document +crypto.
15540 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15542 * config/arm/arm-c.c (arm_cpu_builtins): Define
15543 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
15545 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15547 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
15548 (armv8-r): Set ARM Cortex-R52 as default CPU.
15549 * config/arm/arm-tables.opt: Regenerate.
15550 * config/arm/arm-tune.md: Regenerate.
15551 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
15553 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
15554 extension for -mcpu=cortex-r52.
15556 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15558 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
15559 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
15560 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
15561 (fp-armv8): Define it as FP_ARMv8 only.
15562 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
15563 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
15565 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
15566 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
15567 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
15568 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
15569 than TARGET_FPU_ARMV8.
15570 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
15571 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
15572 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
15574 * config/arm/neon.md (neon_vrint): Likewise.
15575 (neon_vcvt): Likewise.
15576 (neon_<fmaxmin_op><mode>): Likewise.
15577 (<fmaxmin><mode>3): Likewise.
15578 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
15579 * config/arm/predicates.md (arm_cond_move_operator): Check against
15580 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
15582 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
15584 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
15585 to top of function.
15587 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15589 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
15590 loop in comment with memset.
15592 2017-07-14 Martin Liska <mliska@suse.cz>
15594 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
15595 * dwarf2out.c (is_java): Remove the function.
15596 (output_pubname): Remove usage of the function.
15597 (lower_bound_default): Remove usage of DW_LANG_Java.
15598 (gen_compile_unit_die): Likewise.
15599 * gcc.c: Remove compiler defaults for .java and .zip files.
15600 * gimple-expr.c (remove_suffix): Change as there's no longer
15601 extension than 4-letter one.
15602 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
15603 (gimplify_save_expr): Likewise.
15604 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
15605 as it's possible even for other languages than Java.
15606 * langhooks.h (struct lang_hooks): Remove Java from a comment.
15607 * lto-opts.c (lto_write_options): Remove reference to Java.
15608 * opts.c (strip_off_ending): Update file extension handling.
15609 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
15610 * tree-eh.c (lower_resx): Likewise.
15611 * tree.c (free_lang_data_in_type): Remove dead code.
15612 (find_decls_types_r): Likewise.
15613 (build_common_builtin_nodes): Remove Java from a comment.
15614 (verify_type): Remove dead code.
15615 * varasm.c (assemble_external): Remove Java from a comment.
15617 2017-07-14 Martin Liska <mliska@suse.cz>
15619 * opts.c (finish_options): Add quotes.
15620 (common_handle_option): Likewise.
15622 2017-07-14 Martin Liska <mliska@suse.cz>
15624 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
15625 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
15626 Remove N_SO_PASCAL.
15627 * dwarf2out.c (lower_bound_default): Do not handle
15629 (gen_compile_unit_die): Likewise.
15630 * gcc.c: Remove default extension binding for GNU Pascal.
15631 * stmt.c: Remove Pascal language from a comment.
15632 * xcoffout.c: Likewise.
15634 2017-07-13 David Malcolm <dmalcolm@redhat.com>
15637 * diagnostic-show-locus.c (fixit_cmp): New function.
15638 (layout::layout): Sort m_fixit_hints.
15639 (column_range::column_range): Assert that the values are valid.
15640 (struct char_span): New struct.
15641 (correction::overwrite): New method.
15642 (struct source_line): New struct.
15643 (line_corrections::add_hint): Add assertions. Reimplement memcpy
15644 calls in terms of classes source_line and char_span, and
15645 correction::overwrite.
15646 (selftest::test_overlapped_fixit_printing_2): New function.
15647 (selftest::diagnostic_show_locus_c_tests): Call it.
15649 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
15651 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
15652 early if there is no lhs.
15654 2017-07-13 Martin Liska <mliska@suse.cz>
15656 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
15657 (gen_reference_type_die): Likewise.
15658 * stor-layout.c: Remove Pascal-related comment.
15660 2017-07-13 Martin Liska <mliska@suse.cz>
15662 * opts.c (finish_options): Add quotes to error messages.
15663 (parse_sanitizer_options): Likewise.
15665 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
15667 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
15669 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
15671 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
15673 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
15675 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
15677 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
15679 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
15682 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
15683 provides the hardware capability bits, define the macro
15684 __BUILTIN_CPU_SUPPORTS__.
15685 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
15686 if GLIBC does not provide the hardware capability bits. Add a
15687 gcc_unreachable call if the built-in cpu function is neither
15688 __builtin_cpu_is nor __builtin_cpu_supports.
15689 (rs6000_get_function_versions_dispatcher): Change the warning
15690 that an old GLIBC is used which does not export the capability
15691 bits to be an error.
15692 * doc/extend.texi (target_clones attribute): Document the
15693 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
15694 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
15695 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
15696 the macros defined by GCC if the newer GLIBC is available.
15698 2017-07-12 Jeff Law <law@redhat.com>
15700 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
15701 remaining includes slightly.
15702 * config/riscv/riscv-builtins.c: Include profile-count.h.
15704 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
15707 * config/avr/avr.c (avr_set_current_function): In diagnostic
15708 messages: Quote keywords and (parts of) identifiers.
15709 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
15712 2017-07-12 Carl Love <cel@us.ibm.com>
15714 * config/rs6000/rs6000-c.c: Add support for built-in functions
15715 vector bool char vec_revb (vector bool char);
15716 vector bool short vec_revb (vector short char);
15717 vector bool int vec_revb (vector bool int);
15718 vector bool long long vec_revb (vector bool long long);
15719 * doc/extend.texi: Update the built-in documentation file for the
15720 new built-in functions.
15722 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15724 * config/s390/s390.md: Remove movcc splitter.
15726 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15728 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
15729 load/store on condition.
15731 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
15734 * config/avr/avr.c (avr_encode_section_info)
15735 [progmem && !TREE_READONLY]: Error if progmem object needs
15738 2017-07-11 Michael Collison <michael.collison@arm.com>
15740 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
15743 2017-07-11 Carl Love <cel@us.ibm.com>
15745 * config/rs6000/rs6000-c.c: Add support for builtins
15746 vector unsigned int vec_parity_lsbb (vector signed int);
15747 vector unsigned int vec_parity_lsbb (vector unsigned int);
15748 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
15749 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
15750 vector unsigned long long vec_parity_lsbb (vector signed long long);
15751 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
15752 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
15753 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
15754 * doc/extend.texi: Update the built-in documentation file for the
15755 new built-in functions.
15757 2017-07-11 David Malcolm <dmalcolm@redhat.com>
15759 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
15760 (layout::m_primary_loc): New field.
15761 (layout::layout): Initialize new field. Move location filtering
15762 logic from here to...
15763 (layout::maybe_add_location_range): ...this new method. Add
15764 support for filtering to just the lines already specified by other
15766 (layout::will_show_line_p): New method.
15767 (gcc_rich_location::add_location_if_nearby): New method.
15768 (selftest::test_add_location_if_nearby): New test function.
15769 (selftest::diagnostic_show_locus_c_tests): Call it.
15770 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
15773 2017-07-11 Tom de Vries <tom@codesourcery.com>
15775 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
15776 (bb_first_real_insn): New function.
15777 (nvptx_single): Add extra initialization of broadcasted condition
15780 2017-07-11 Nathan Sidwell <nathan@acm.org>
15782 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
15784 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
15786 * doc/extend.texi (AVR Function Attributes): Remove weblink to
15787 Binutils doc as TEXI will mess them up.
15788 * doc/invoke.texi (AVR Options): Same here.
15790 2017-07-11 Daniel Cederman <cederman@gaisler.com>
15792 * config/sparc/sparc.opt (mfix-ut700): New option.
15793 (mfix-gr712rc): Likewise.
15794 (sparc_fix_b2bst): New variable.
15795 * doc/invoke.texi (SPARC options): Document them.
15796 (ARM options): Fix warnings.
15797 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
15798 instructions to prevent sequences that can trigger the store-store
15799 errata for certain LEON3FT processors.
15800 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
15801 (sparc_option_override): Set sparc_fix_b2bst appropriately.
15802 * config/sparc/sparc.md (fix_b2bst): New attribute.
15803 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
15805 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
15808 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
15810 (*rsqrtsf2_sse): Ditto.
15812 (div<mode>3): Macroize insn from divdf3 and divsf3
15813 using MODEF mode iterator.
15815 2017-07-10 Martin Sebor <msebor@redhat.com>
15817 PR tree-optimization/80397
15818 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
15819 instead of testing for equality to INTEGER_TYPE.
15821 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
15823 * config.gcc: Remove uclibc from arc target spec.
15825 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
15827 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
15829 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15832 * lto-wrapper.c (remove_option): New function.
15833 (merge_and_complain): Merge PIC/PIE options more realistically.
15835 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
15837 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
15841 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
15842 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
15843 * config.in: Regenerate.
15844 * configure: Regenerate.
15845 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
15846 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
15847 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
15848 (TARGET_GASISR_PROLOGUES): ...target mask.
15849 * common/config/avr/avr-common.c
15850 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
15851 Set -mgas-isr-prologues.
15852 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
15853 INSERT_PASS_BEFORE for it.
15854 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
15855 * config/avr/avr.c (avr_option_override)
15856 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
15857 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
15858 (avr_attribute_table) <no_gccisr>: Add new function attribute.
15859 (avr_set_current_function) <is_no_gccisr>: Init machine field.
15860 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
15862 (make_avr_pass_pre_proep): New function.
15863 (emit_push_sfr) <treg>: Add argument to function and use it
15864 instead of TMP_REG.
15865 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
15866 and set machine->gasisr.yes.
15867 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
15868 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
15869 __gcc_isr.n_pushed to .L__stack_usage.
15870 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
15871 (avr_asm_final_postscan_insn): ...this new static function.
15872 * config/avr/avr.h (machine_function)
15873 <is_no_gccisr, use_L__stack_usage>: New fields.
15874 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
15875 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
15876 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
15877 (gasisr, *gasisr): New expander and insn.
15878 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
15879 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
15880 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
15882 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
15884 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
15887 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
15889 Move jump-tables out of .text again.
15892 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
15893 (ASM_OUTPUT_ADDR_VEC): New function.
15894 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
15895 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
15896 INSN_ADDRESSes as asm comment.
15897 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
15898 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
15899 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
15900 * config/avr/avr.md (*tablejump): Adjust comment.
15901 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
15902 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
15904 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
15905 (avr_output_addr_vec): New proto.
15906 (avr_log_t) <insn_addresses>: New field.
15908 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
15911 * config/i386/i386.c (ix86_function_arg_advance): Set
15912 outgoing_args_on_stack to true if there are outgoing arguments
15914 (ix86_function_arg): Likewise.
15915 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
15916 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
15917 * config/i386/i386.h (machine_function): Add
15918 outgoing_args_on_stack.
15920 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
15922 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
15923 supporting pthreds.
15924 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
15926 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
15928 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
15929 (REAL_H): Remove $(MACHMODE_H).
15930 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
15932 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
15933 $(MACHMODE_H) and double-int.h.
15934 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
15936 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
15939 2017-07-07 Andrew Pinski <apinski@cavium.com>
15941 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
15942 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
15944 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15946 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
15947 Add warning if GCC was not configured to link against a GLIBC that
15948 exports the hardware capability bits.
15949 (make_resolver_func): Make resolver function private and not a
15950 COMDAT function. Create the name with clone_function_name instead
15951 of make_unique_name.
15954 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
15955 correct operand in doing the split.
15957 2017-07-07 Carl Love <cel@us.ibm.com>
15959 * config/rs6000/rs6000-c: Add support for built-in function
15960 vector unsigned short vec_pack_to_short_fp32 (vector float,
15962 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
15963 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
15964 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
15965 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
15966 (convert_4f32_8i16): Add define_expand.
15967 * doc/extend.texi: Update the built-in documentation file for the
15968 new built-in function.
15970 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15972 * config/sparc/m8.md: New file.
15973 * config/sparc/sparc.md: Include m8.md.
15975 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15977 * config/sparc/sparc.opt: New option -mvis4b.
15978 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
15979 (sparc_option_override): Handle VIS4B.
15980 (enum sparc_builtins): Define
15981 SPARC_BUILTIN_DICTUNPACK{8,16,32},
15982 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
15983 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
15984 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
15985 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
15986 (check_constant_argument): New function.
15987 (sparc_vis_init_builtins): Define builtins
15988 __builtin_vis_dictunpack{8,16,32},
15989 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
15990 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
15991 __builtin_vis_fpcmpde{8,16,32}shl and
15992 __builtin_vis_fpcmpur{8,16,32}shl.
15993 (sparc_expand_builtin): Check that the constant operands to
15994 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
15995 constant and in range.
15996 * config/sparc/sparc-c.c (sparc_target_macros): Handle
15998 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
15999 (SPARC_IMM5_P): Likewise.
16000 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
16001 (enabled): Handle vis4b.
16002 (UNSPEC_DICTUNPACK): New unspec.
16003 (UNSPEC_FPCMPSHL): Likewise.
16004 (UNSPEC_FPUCMPSHL): Likewise.
16005 (UNSPEC_FPCMPDESHL): Likewise.
16006 (UNSPEC_FPCMPURSHL): Likewise.
16007 (cpu_feature): New CPU feature `vis4b'.
16008 (dictunpack{8,16,32}): New insns.
16009 (FPCSMODE): New mode iterator.
16010 (fpcscond): New code iterator.
16011 (fpcsucond): Likewise.
16012 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
16013 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
16014 (fpcmpde{8,16,32}{si,di}shl): Likewise.
16015 (fpcmpur{8,16,32}{si,di}shl): Likewise.
16016 * config/sparc/constraints.md: Define constraints `q' for unsigned
16017 2-bit integer constants and `t' for unsigned 5-bit integer
16019 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
16021 (imm5_operand_dictunpack16): Likewise.
16022 (imm5_operand_dictunpack32): Likewise.
16023 (imm2_operand): Likewise.
16024 * doc/invoke.texi (SPARC Options): Document -mvis4b.
16025 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
16026 ditunpack* and fpcmp*shl builtins.
16028 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16030 * config.gcc: Handle m8 in --with-{cpu,tune} options.
16031 * config.in: Add HAVE_AS_SPARC6 define.
16032 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
16034 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
16036 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
16037 (CPP_CPU_SPEC): Handle m8.
16038 (ASM_CPU_SPEC): Likewise.
16039 * config/sparc/sparc-opts.h (enum processor_type): Add
16041 * config/sparc/sparc.c (m8_costs): New struct.
16042 (sparc_option_override): Handle TARGET_CPU_m8.
16043 (sparc32_initialize_trampoline): Likewise.
16044 (sparc64_initialize_trampoline): Likewise.
16045 (sparc_issue_rate): Likewise.
16046 (sparc_register_move_cost): Likewise.
16047 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
16048 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
16049 (ASM_CPU64_DEFAULT_SPEC): Likewise.
16050 (CPP_CPU_SPEC): Handle M8.
16051 (ASM_CPU_SPEC): Likewise.
16052 (AS_M8_FLAG): Define.
16053 * config/sparc/sparc.md: Add m8 to the cpu attribute.
16054 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
16055 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
16057 * configure: Regenerate.
16058 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
16061 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16063 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
16065 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
16066 ("*movdi_insn_sp32"): Do not set v3pipe.
16067 ("*movsi_insn"): Likewise.
16068 ("*movdi_insn_sp64"): Likewise.
16069 ("*movsf_insn"): Likewise.
16070 ("*movdf_insn_sp32"): Likewise.
16071 ("*movdf_insn_sp64"): Likewise.
16072 ("*zero_extendsidi2_insn_sp64"): Likewise.
16073 ("*sign_extendsidi2_insn"): Likewise.
16074 ("*mov<VM32:mode>_insn"): Likewise.
16075 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16076 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16077 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16078 ("<vlop:code><VL:mode>3"): Likewise.
16079 ("*not_<vlop:code><VL:mode>3"): Likewise.
16080 ("*nand<VL:mode>_vis"): Likewise.
16081 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
16082 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
16083 ("one_cmpl<VL:mode>2"): Likewise.
16084 ("faligndata<VM64:mode>_vis"): Likewise.
16085 ("alignaddrsi_vis"): Likewise.
16086 ("alignaddrdi_vis"): Likweise.
16087 ("alignaddrlsi_vis"): Likewise.
16088 ("alignaddrldi_vis"): Likewise.
16089 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16090 ("bmaskdi_vis"): Likewise.
16091 ("bmasksi_vis"): Likewise.
16092 ("bshuffle<VM64:mode>_vis"): Likewise.
16093 ("cmask8<P:mode>_vis"): Likewise.
16094 ("cmask16<P:mode>_vis"): Likewise.
16095 ("cmask32<P:mode>_vis"): Likewise.
16096 ("pdistn<P:mode>_vis"): Likewise.
16097 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16099 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16101 * config/sparc/sparc.md ("subtype"): New insn attribute.
16102 ("*wrgsr_sp64"): Set insn subtype.
16103 ("*rdgsr_sp64"): Likewise.
16104 ("alignaddrsi_vis"): Likewise.
16105 ("alignaddrdi_vis"): Likewise.
16106 ("alignaddrlsi_vis"): Likewise.
16107 ("alignaddrldi_vis"): Likewise.
16108 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16109 ("fexpand_vis"): Likewise.
16110 ("fpmerge_vis"): Likewise.
16111 ("faligndata<VM64:mode>_vis"): Likewise.
16112 ("bshuffle<VM64:mode>_vis"): Likewise.
16113 ("cmask8<P:mode>_vis"): Likewise.
16114 ("cmask16<P:mode>_vis"): Likewise.
16115 ("cmask32<P:mode>_vis"): Likewise.
16116 ("fchksm16_vis"): Likewise.
16117 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
16118 ("fmean16_vis"): Likewise.
16119 ("fp<plusminus_insn>64_vis"): Likewise.
16120 ("<plusminus_insn>v8qi3"): Likewise.
16121 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16122 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
16123 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
16124 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
16125 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
16126 ("*movqi_insn"): Likewise.
16127 ("*movhi_insn"): Likewise.
16128 ("*movsi_insn"): Likewise.
16129 ("movsi_pic_gotdata_op"): Likewise.
16130 ("*movdi_insn_sp32"): Likewise.
16131 ("*movdi_insn_sp64"): Likewise.
16132 ("movdi_pic_gotdata_op"): Likewise.
16133 ("*movsf_insn"): Likewise.
16134 ("*movdf_insn_sp32"): Likewise.
16135 ("*movdf_insn_sp64"): Likewise.
16136 ("*zero_extendhisi2_insn"): Likewise.
16137 ("*zero_extendqihi2_insn"): Likewise.
16138 ("*zero_extendqisi2_insn"): Likewise.
16139 ("*zero_extendqidi2_insn"): Likewise.
16140 ("*zero_extendhidi2_insn"): Likewise.
16141 ("*zero_extendsidi2_insn_sp64"): Likewise.
16142 ("ldfsr"): Likewise.
16143 ("prefetch_64"): Likewise.
16144 ("prefetch_32"): Likewise.
16145 ("tie_ld32"): Likewise.
16146 ("tie_ld64"): Likewise.
16147 ("*tldo_ldub_sp32"): Likewise.
16148 ("*tldo_ldub1_sp32"): Likewise.
16149 ("*tldo_ldub2_sp32"): Likewise.
16150 ("*tldo_ldub_sp64"): Likewise.
16151 ("*tldo_ldub1_sp64"): Likewise.
16152 ("*tldo_ldub2_sp64"): Likewise.
16153 ("*tldo_ldub3_sp64"): Likewise.
16154 ("*tldo_lduh_sp32"): Likewise.
16155 ("*tldo_lduh1_sp32"): Likewise.
16156 ("*tldo_lduh_sp64"): Likewise.
16157 ("*tldo_lduh1_sp64"): Likewise.
16158 ("*tldo_lduh2_sp64"): Likewise.
16159 ("*tldo_lduw_sp32"): Likewise.
16160 ("*tldo_lduw_sp64"): Likewise.
16161 ("*tldo_lduw1_sp64"): Likewise.
16162 ("*tldo_ldx_sp64"): Likewise.
16163 ("*mov<VM32:mode>_insn"): Likewise.
16164 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16165 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16167 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16169 * config/sparc/sparc.md ("type"): New insn type viscmp.
16170 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
16172 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
16173 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
16174 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16175 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
16177 ("n7_vis_logical_11cycle"): Likewise.
16178 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
16179 * config/sparc/niagara2.md ("niag3_vis": Likewise.
16180 * config/sparc/niagara.md ("niag_vis"): Likewise.
16181 * config/sparc/ultra3.md ("us3_fga"): Likewise.
16182 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
16184 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16186 * config/sparc/sparc.md: New instruction type `bmask'.
16187 (bmaskdi_vis): Use the `bmask' type.
16188 (bmasksi_vis): Likewise.
16189 * config/sparc/ultra3.md (us3_array): Likewise.
16190 * config/sparc/niagara7.md (n7_array): Likewise.
16191 * config/sparc/niagara4.md (n4_array): Likewise.
16192 * config/sparc/niagara2.md (niag2_vis): Likewise.
16193 (niag3_vis): Likewise.
16194 * config/sparc/niagara.md (niag_vis): Likewise.
16196 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16198 * ipa-comdats.c: Remove optimize check from gate.
16199 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
16200 for functions not optimized.
16201 (ipa_fn_summary_read): Skip optimize check.
16202 (ipa_fn_summary_write): Likewise.
16203 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
16205 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
16207 (can_inline_edge_p): Check flag_pcc_struct_return for match.
16208 (check_callers): Give up on caller which is not optimized.
16209 (inline_small_functions): Likewise.
16210 (ipa_inline): Do not give up when not optimizing.
16211 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
16212 away unoptimizes cdtors.
16213 (whole_program_function_and_variable_visibility): Do
16214 ipa_discover_readonly_nonaddressable_vars in LTO mode.
16215 * ipa.c (process_references): Do not check optimize.
16216 (symbol_table::remove_unreachable_nodes): Update optimize check.
16217 (set_writeonly_bit): Update optimize check.
16218 (pass_ipa_cdtor_merge::gate): Do not check optimize.
16219 (pass_ipa_single_use::gate): Remove.
16221 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16223 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
16224 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
16225 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
16226 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
16227 permute_load, permute_store, adjust_extract, adjust_splat,
16228 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
16229 replace_swap_with_copy, dump_swap_insn_table,
16230 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
16231 recombine_lvx_pattern, recombine_stvx_pattern,
16232 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
16233 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
16234 to file rs6000-p8swap.c.
16235 * config/rs6000/rs6000-p8swap.c: New file.
16236 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
16237 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
16238 and rs6000*-*-* targets.
16240 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16242 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
16244 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16246 * lto-wrapper.c (merge_and_complain): Do not merge
16247 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
16248 fsigned_zeros, ftrapping_math, fwrapv.
16249 (append_compiler_options): Do not track these options.
16250 (append_linker_options): Likewie
16252 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16254 * cgraphunit.c (cgraph_node::finalize_function): When
16255 !flag_toplevel_reorde set no_reorder flag.
16256 (varpool_node::finalize_decl): Likewise.
16257 (symbol_table::compile): Drop no toplevel reorder path.
16259 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16261 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
16262 edges; zero probability is not better than uninitialized.
16264 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
16266 * asan.h (asan_sanitize_allocas_p): Declare.
16267 * asan.c (asan_sanitize_allocas_p): New function.
16268 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
16269 (handle_builtin_alloca): Likewise.
16270 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
16271 if !asan_sanitize_allocas_p.
16272 * params.def (asan-instrument-allocas): Add new option.
16273 * params.h (ASAN_PROTECT_ALLOCAS): Define.
16274 * opts.c (common_handle_option): Disable allocas sanitization for
16277 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
16279 * asan.c: Include gimple-fold.h.
16280 (get_last_alloca_addr): New function.
16281 (handle_builtin_stackrestore): Likewise.
16282 (handle_builtin_alloca): Likewise.
16283 (asan_emit_allocas_unpoison): Likewise.
16284 (get_mem_refs_of_builtin_call): Add new parameter, remove const
16285 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
16286 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
16287 (instrument_builtin_call): Pass gimple iterator to
16288 get_mem_refs_of_builtin_call.
16289 (last_alloca_addr): New global.
16290 * asan.h (asan_emit_allocas_unpoison): Declare.
16291 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
16292 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
16293 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
16294 if function calls alloca.
16295 * gimple-fold.c (replace_call_with_value): Remove static keyword.
16296 * gimple-fold.h (replace_call_with_value): Declare.
16297 * internal-fn.c: Include asan.h.
16298 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
16299 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
16301 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16303 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
16304 (C_SELFTEST_FLAGS): New.
16305 (CPP_SELFTEST_FLAGS): New.
16306 (SELFTEST_DEPS): New, from deps of s-selftest.
16307 (C_SELFTEST_DEPS): New, from deps of s-selftest.
16308 (CPP_SELFTEST_DEPS): New.
16309 (selftest): Add dependency on s-selftest-c++.
16310 (s-selftest): Rename to...
16311 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
16312 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
16313 than SELFTEST_FLAGS.
16314 (selftest-gdb): Rename to...
16315 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
16317 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
16318 (selftest-valgrind): Rename to...
16319 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
16321 (selftest-valgrind): Reintroduce as an alias for
16322 selftest-c-valgrind.
16323 (s-selftest-c++): New.
16324 (selftest-c++-gdb): New.
16325 (selftest-c++-valgrind): New.
16327 2017-07-06 Olivier Hainque <hainque@adacore.com>
16329 * gcc.c (process_command): When deciding if undefined variables
16330 should be ignored when processing specs, accept "gcc -v" as well.
16332 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16334 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
16335 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
16337 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
16339 * config/arm/arm-cpus.in (armv8-r): Add new entry.
16340 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
16341 * config/arm/arm-tables.opt: Regenerate.
16342 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
16344 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
16346 2017-07-06 Carl Love <cel@us.ibm.com>
16348 * ChangeLog: Clean up from mid air collision
16350 2017-07-06 Carl Love <cel@us.ibm.com>
16352 * config/rs6000/rs6000-c.c: Add support for built-in functions
16353 vector signed int vec_subc (vector signed int, vector signed int);
16354 vector signed __int128 vec_subc (vector signed __int128,
16355 vector signed __int128);
16356 vector unsigned __int128 vec_subc (vector unsigned __int128,
16357 vector unsigned __int128);
16358 vector signed int vec_sube (vector signed int, vector signed int,
16359 vector signed int);
16360 vector unsigned int vec_sube (vector unsigned int,
16361 vector unsigned int,
16362 vector unsigned int);
16363 vector signed __int128 vec_sube (vector signed __int128,
16364 vector signed __int128,
16365 vector signed__int128);
16366 vector unsigned __int128 vec_sube (vector unsigned __int128,
16367 vector unsigned __int128,
16368 vector unsigned __int128);
16369 vector signed int vec_subec (vector signed int, vector signed int,
16370 vector signed int);
16371 vector unsigned int vec_subec (vector unsigned int,
16372 vector unsigned int,
16373 vector unsigned int);
16374 vector signed __int128 vec_subec (vector signed __int128,
16375 vector signed __int128,
16376 vector signed__int128);
16377 vector unsigned __int128 vec_subec (vector unsigned __int128,
16378 vector unsigned __int128,
16379 vector unsigned __int128);
16380 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
16381 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
16382 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
16383 BU_ALTIVEC_OVERLOAD_X definitions.
16384 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
16385 * doc/extend.texi: Update the built-in documentation file for the new
16386 built-in functions.
16388 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16391 * diagnostic-show-locus.c (layout::layout): Use start and finish
16392 spelling location for the start and finish of each range.
16393 * genmatch.c (linemap_client_expand_location_to_spelling_point):
16394 Add unused aspect param.
16395 * input.c (expand_location_1): Add "aspect" param, and use it
16396 to access the correct part of the location.
16397 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
16399 (expand_location_to_spelling_point): Likewise.
16400 (linemap_client_expand_location_to_spelling_point): Add "aspect"
16401 param, and pass it to expand_location_1.
16403 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
16405 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
16406 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
16407 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
16408 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
16409 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
16410 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
16411 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
16412 _mm_maskz_getmant_ss): New intrinsics.
16413 (__builtin_ia32_getexpss128_mask): Changed to ...
16414 __builtin_ia32_getexpss128_round ... this.
16415 (__builtin_ia32_getexpsd128_mask): Changed to ...
16416 __builtin_ia32_getexpsd128_round ... this.
16417 * config/i386/i386-builtin-types.def
16418 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
16419 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
16420 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
16421 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
16422 __builtin_ia32_getmantss_mask_round): New builtins.
16423 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
16424 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
16425 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
16426 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
16427 * config/i386/sse.md
16428 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
16429 avx512f_sgetexp<mode><mask_scalar_name>
16430 <round_saeonly_scalar_name> ... this.
16431 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
16432 %0, %1, %2<round_saeonly_op3>}): Changed to ...
16433 vgetexp<ssescalarmodesuffix>
16434 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16435 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
16436 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
16437 avx512f_vgetmant<mode><mask_scalar_name>
16438 <round_saeonly_scalar_name> ... this.
16439 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
16440 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
16441 vgetmant<ssescalarmodesuffix>
16442 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
16443 %0<mask_scalar_operand4>, %1, %2
16444 <round_saeonly_scalar_mask_op4>, %3} ... this.
16445 * config/i386/subst.md (mask_scalar_operand4,
16446 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
16447 round_saeonly_scalar_nimm_predicate): New subst attributes.
16449 2017-07-06 Julia Koval <julia.koval@intel.com>
16451 * config/i386/i386.c (ix86_erase_embedded_rounding):
16452 Remove code for old rounding pattern.
16454 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
16456 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
16458 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
16460 * doc/sourcebuild.texi (Test Directives, Variants of
16461 dg-require-support): Add documentation for dg-require-stack-check.
16463 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
16465 * config/i386/subst.md (mask_scalar, round_scalar,
16466 round_saeonly_scalar): New meta-templates.
16467 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
16468 round_scalar_mask_operand3, round_scalar_mask_op3,
16469 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
16470 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
16471 round_saeonly_scalar_constraint,
16472 round_saeonly_scalar_prefix): New subst attribute.
16473 * config/i386/sse.md
16474 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
16475 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
16476 <round_scalar_name> ... this.
16477 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
16478 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
16479 <round_scalar_name> ... this.
16480 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
16481 <sse>_vm<code><mode>3<mask_scalar_name>
16482 <round_saeonly_scalar_name> ... this.
16483 (v<plusminus_mnemonic><ssescalarmodesuffix>
16484 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
16485 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
16486 v<plusminus_mnemonic><ssescalarmodesuffix>
16487 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16488 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
16489 (v<multdiv_mnemonic><ssescalarmodesuffix>
16490 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
16491 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
16492 v<multdiv_mnemonic><ssescalarmodesuffix>
16493 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16494 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
16495 (v<maxmin_float><ssescalarmodesuffix>
16496 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
16497 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
16498 v<maxmin_float><ssescalarmodesuffix>
16499 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16500 %0<mask_scalar_operand3>, %1, %<iptr>2
16501 <round_saeonly_scalar_mask_op3>} ... this.
16503 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
16505 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
16506 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
16508 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
16509 Alan Hayward <alan.hayward@arm.com>
16510 David Sherwood <david.sherwood@arm.com>
16512 * combine.c (simplify_if_then_else): Remove "enum" before
16514 * compare-elim.c (can_eliminate_compare): Likewise.
16515 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
16517 (aarch64_lookup_simd_builtin_type): Likewise.
16518 (aarch64_simd_builtin_type): Likewise.
16519 (aarch64_init_simd_builtin_types): Likewise.
16520 (aarch64_simd_expand_args): Likewise.
16521 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
16523 (aarch64_reverse_mask): Likewise.
16524 (aarch64_simd_emit_reg_reg_move): Likewise.
16525 (aarch64_gen_adjusted_ldpstp): Likewise.
16526 (aarch64_ccmp_mode_to_code): Likewise.
16527 (aarch64_operands_ok_for_ldpstp): Likewise.
16528 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16529 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
16531 (aarch64_min_divisions_for_recip_mul): Likewise.
16532 (aarch64_reassociation_width): Likewise.
16533 (aarch64_get_condition_code_1): Likewise.
16534 (aarch64_simd_emit_reg_reg_move): Likewise.
16535 (aarch64_simd_attr_length_rglist): Likewise.
16536 (aarch64_reverse_mask): Likewise.
16537 (aarch64_operands_ok_for_ldpstp): Likewise.
16538 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16539 (aarch64_gen_adjusted_ldpstp): Likewise.
16540 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
16542 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
16543 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
16544 (arm_lookup_simd_builtin_type): Likewise.
16545 (arm_simd_builtin_type): Likewise.
16546 (arm_init_simd_builtin_types): Likewise.
16547 (arm_expand_builtin_args): Likewise.
16548 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
16549 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
16550 (ft32_setup_incoming_varargs): Likewise.
16551 (ft32_function_arg): Likewise.
16552 (ft32_function_arg_advance): Likewise.
16553 (ft32_pass_by_reference): Likewise.
16554 (ft32_arg_partial_bytes): Likewise.
16555 (ft32_valid_pointer_mode): Likewise.
16556 (ft32_addr_space_pointer_mode): Likewise.
16557 (ft32_addr_space_legitimate_address_p): Likewise.
16558 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
16560 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
16561 (ix86_emit_outlined_ms2sysv_restore): Likewise.
16562 (iamcu_alignment): Likewise.
16563 (canonicalize_vector_int_perm): Likewise.
16564 (ix86_noce_conversion_profitable_p): Likewise.
16565 (ix86_mpx_bound_mode): Likewise.
16566 (ix86_operands_ok_for_move_multiple): Likewise.
16567 * config/microblaze/microblaze-protos.h
16568 (microblaze_expand_conditional_branch_reg): Likewise.
16569 * config/microblaze/microblaze.c
16570 (microblaze_expand_conditional_branch_reg): Likewise.
16571 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
16573 (rs6000_reassociation_width): Likewise.
16574 (rs6000_invalid_binary_op): Likewise.
16575 (fusion_p9_p): Likewise.
16576 (emit_fusion_p9_load): Likewise.
16577 (emit_fusion_p9_store): Likewise.
16578 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
16580 (riscv_hard_regno_mode_ok_p): Likewise.
16581 (riscv_address_insns): Likewise.
16582 (riscv_split_symbol): Likewise.
16583 (riscv_legitimize_move): Likewise.
16584 (riscv_function_value): Likewise.
16585 (riscv_hard_regno_nregs): Likewise.
16586 (riscv_expand_builtin): Likewise.
16587 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
16588 (riscv_build_integer): Likewise.
16589 (riscv_split_integer): Likewise.
16590 (riscv_legitimate_constant_p): Likewise.
16591 (riscv_cannot_force_const_mem): Likewise.
16592 (riscv_regno_mode_ok_for_base_p): Likewise.
16593 (riscv_valid_base_register_p): Likewise.
16594 (riscv_valid_offset_p): Likewise.
16595 (riscv_valid_lo_sum_p): Likewise.
16596 (riscv_classify_address): Likewise.
16597 (riscv_legitimate_address_p): Likewise.
16598 (riscv_address_insns): Likewise.
16599 (riscv_load_store_insns): Likewise.
16600 (riscv_force_binary): Likewise.
16601 (riscv_split_symbol): Likewise.
16602 (riscv_force_address): Likewise.
16603 (riscv_legitimize_address): Likewise.
16604 (riscv_move_integer): Likewise.
16605 (riscv_legitimize_const_move): Likewise.
16606 (riscv_legitimize_move): Likewise.
16607 (riscv_address_cost): Likewise.
16608 (riscv_subword): Likewise.
16609 (riscv_output_move): Likewise.
16610 (riscv_canonicalize_int_order_test): Likewise.
16611 (riscv_emit_int_order_test): Likewise.
16612 (riscv_function_arg_boundary): Likewise.
16613 (riscv_pass_mode_in_fpr_p): Likewise.
16614 (riscv_pass_fpr_single): Likewise.
16615 (riscv_pass_fpr_pair): Likewise.
16616 (riscv_get_arg_info): Likewise.
16617 (riscv_function_arg): Likewise.
16618 (riscv_function_arg_advance): Likewise.
16619 (riscv_arg_partial_bytes): Likewise.
16620 (riscv_function_value): Likewise.
16621 (riscv_pass_by_reference): Likewise.
16622 (riscv_setup_incoming_varargs): Likewise.
16623 (riscv_print_operand): Likewise.
16624 (riscv_elf_select_rtx_section): Likewise.
16625 (riscv_save_restore_reg): Likewise.
16626 (riscv_for_each_saved_reg): Likewise.
16627 (riscv_register_move_cost): Likewise.
16628 (riscv_hard_regno_mode_ok_p): Likewise.
16629 (riscv_hard_regno_nregs): Likewise.
16630 (riscv_class_max_nregs): Likewise.
16631 (riscv_memory_move_cost): Likewise.
16632 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
16633 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
16634 (rl78_addr_space_address_mode): Likewise.
16635 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16637 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
16638 (rs6000_reassociation_width): Likewise.
16639 (rs6000_invalid_binary_op): Likewise.
16640 (fusion_p9_p): Likewise.
16641 (emit_fusion_p9_load): Likewise.
16642 (emit_fusion_p9_store): Likewise.
16643 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
16644 (ok_for_simple_move_operands): Likewise.
16645 (ok_for_simple_move_strict_operands): Likewise.
16646 (ok_for_simple_arith_logic_operands): Likewise.
16647 (visium_legitimize_reload_address): Likewise.
16648 (visium_select_cc_mode): Likewise.
16649 (output_cbranch): Likewise.
16650 (visium_split_double_move): Likewise.
16651 (visium_expand_copysign): Likewise.
16652 (visium_expand_int_cstore): Likewise.
16653 (visium_expand_fp_cstore): Likewise.
16654 * config/visium/visium.c (visium_pass_by_reference): Likewise.
16655 (visium_function_arg): Likewise.
16656 (visium_function_arg_advance): Likewise.
16657 (visium_libcall_value): Likewise.
16658 (visium_setup_incoming_varargs): Likewise.
16659 (visium_legitimate_constant_p): Likewise.
16660 (visium_legitimate_address_p): Likewise.
16661 (visium_legitimize_address): Likewise.
16662 (visium_secondary_reload): Likewise.
16663 (visium_register_move_cost): Likewise.
16664 (visium_memory_move_cost): Likewise.
16665 (prepare_move_operands): Likewise.
16666 (ok_for_simple_move_operands): Likewise.
16667 (ok_for_simple_move_strict_operands): Likewise.
16668 (ok_for_simple_arith_logic_operands): Likewise.
16669 (visium_function_value_1): Likewise.
16670 (rtx_ok_for_offset_p): Likewise.
16671 (visium_legitimize_reload_address): Likewise.
16672 (visium_split_double_move): Likewise.
16673 (visium_expand_copysign): Likewise.
16674 (visium_expand_int_cstore): Likewise.
16675 (visium_expand_fp_cstore): Likewise.
16676 (visium_split_cstore): Likewise.
16677 (visium_select_cc_mode): Likewise.
16678 (visium_split_cbranch): Likewise.
16679 (output_cbranch): Likewise.
16680 (visium_print_operand_address): Likewise.
16681 * expmed.c (flip_storage_order): Likewise.
16682 * expmed.h (emit_cstore): Likewise.
16683 (flip_storage_order): Likewise.
16684 * genrecog.c (validate_pattern): Likewise.
16685 * hsa-gen.c (gen_hsa_addr): Likewise.
16686 * internal-fn.c (expand_arith_overflow): Likewise.
16687 * ira-color.c (allocno_copy_cost_saving): Likewise.
16688 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16689 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
16690 (process_invariant_for_inheritance): Likewise.
16691 * lra-eliminations.c (move_plus_up): Likewise.
16692 * omp-low.c (lower_oacc_reductions): Likewise.
16693 * simplify-rtx.c (simplify_subreg): Likewise.
16694 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
16695 (TARGET_CHKP_BOUND_MODE): Likewise..
16696 * targhooks.c (default_chkp_bound_mode): Likewise.
16697 (default_setup_incoming_vararg_bounds): Likewise.
16698 * targhooks.h (default_chkp_bound_mode): Likewise.
16699 (default_setup_incoming_vararg_bounds): Likewise.
16700 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
16701 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
16702 (have_whole_vector_shift): Likewise.
16703 * tree-vect-stmts.c (vectorizable_load): Likewise.
16704 * doc/tm.texi: Regenerate.
16706 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
16708 Graceful degrade if Binutils PR21472 is not available.
16711 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
16712 .rodata in flash test fails.
16713 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
16714 * confgure: Regenerate.
16715 * config.in: Regenerate.
16716 * config/avr/avr.c (avr_asm_named_section)
16717 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
16718 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
16719 (avr_asm_init_sections): Same.
16721 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16723 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
16724 (fma<VH:mode>4_intrinsic): Likewise.
16725 (*fmsub<VCVTF:mode>4): Likewise.
16726 (*fmsub<VH:mode>4_intrinsic): Likewise.
16728 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
16731 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
16732 Don't depend on "optimize > 0".
16733 (out_movhi_r_mr, out_movqi_mr_r): Same.
16734 (out_movhi_mr_r, out_movqi_r_mr): Same.
16735 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
16736 io_address_operand on "optimize > 0".
16738 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16740 * tree-loop-distribution.c: Add general explanantion on the pass.
16741 (generate_loops_for_partition): Mark distributed loop.
16742 (pg_add_dependence_edges): New parameter. Handle alias data
16743 dependence specially and record it in the parameter if asked.
16744 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
16745 (init_partition_graph_vertices, add_partition_graph_edge): New.
16746 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
16747 (free_partition_graph_vdata, build_partition_graph): New.
16748 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
16749 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
16750 (data_ref_segment_size, latch_dominated_by_data_ref): New.
16751 (compute_alias_check_pairs, version_loop_by_alias_check): New.
16752 (version_for_distribution_p, finalize_partitions): New.
16753 (distribute_loop): Handle alias data dependence specially. Factor
16754 out loop fusion code as functions and call these functions.
16756 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16758 * tree-loop-distribution.c (classify_partition): New parameter and
16759 better handle reduction statement.
16760 (rdg_build_partitions): Revise comment.
16761 (distribute_loop): Compute statements in all partitions and pass it
16762 to classify_partition.
16764 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16766 * tree-loop-distribution.c (enum partition_type): New.
16767 (struct partition): New field type.
16768 (partition_merge_into): Add parameter. Update partition type.
16769 (data_dep_in_cycle_p, update_type_for_merge): New functions.
16770 (build_rdg_partition_for_vertex): Compute partition type.
16771 (rdg_build_partitions): Dump partition type.
16772 (distribute_loop): Update calls to partition_merge_into.
16774 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16776 * tree-loop-distribution.c (struct ddr_hasher): New.
16777 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
16779 (classify_partition): Call get_data_dependence.
16780 (pg_add_dependence_edges): Ditto.
16781 (distribute_loop): Release data dependence hash table.
16783 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16785 * tree-loop-distribution.c (ref_base_address): Delete.
16786 (similar_memory_accesses): Rename ...
16787 (share_memory_accesses): ... to this. Check if partitions access
16788 the same memory reference.
16789 (distribute_loop): Call share_memory_accesses.
16791 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16793 * tree-loop-distribution.c (struct partition): New field recording
16794 its data reference.
16795 (partition_alloc, partition_free): Init and release data refs.
16796 (partition_merge_into): Merge data refs.
16797 (build_rdg_partition_for_vertex): Collect data refs for partition.
16798 (pg_add_dependence_edges): Change parameters from vector to bitmap.
16800 (distribute_loop): Remve data refs from vertice data of partition
16803 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16805 * tree-loop-distribution.c (params.h): Include header file.
16806 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
16807 (datarefs_vec): New global var.
16808 (create_rdg_vertices): Use datarefs_vec directly.
16809 (free_rdg): Don't free data references.
16810 (build_rdg): Update use. Don't free data references.
16811 (distribute_loop): Compute global variable for data references.
16812 Bail out if there are too many data references.
16814 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16816 * tree-loop-distribution.c (loop_nest): New global var.
16817 (build_rdg): Use loop directly, rather than loop nest.
16818 (pg_add_dependence_edges): Remove loop nest parameter. Use global
16820 (distribute_loop): Compute global variable loop nest. Update use.
16822 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16824 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
16825 (partition_merge_into): New parameter. Dump reason for fusion.
16826 (distribute_loop): Update use of partition_merge_into.
16828 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16830 * tree-loop-distribution.c (bb_top_order_index): New.
16831 (bb_top_order_index_size, bb_top_order_cmp): New.
16832 (stmts_from_loop): Use topological order.
16833 (pass_loop_distribution::execute): Compute and release topological
16834 order for basic blocks.
16836 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16838 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
16841 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16843 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
16844 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
16845 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
16846 * internal-fn.def (LOOP_DIST_ALIAS): New.
16847 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
16848 (fold_loop_internal_call): ... this.
16849 (vect_loop_dist_alias_call): New function.
16850 (set_uid_loop_bbs): Call fold_loop_internal_call.
16851 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
16854 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
16857 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
16858 Require dead FLAGS_REG at the beginning of a peephole.
16860 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
16863 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
16864 arguments in the call to __builtin_ia32_sbb_u32.
16865 (_subborrow_u64): Swap _X and _Y arguments in the call to
16866 __builtin_ia32_sbb_u64.
16868 2017-07-04 Jakub Jelinek <jakub@redhat.com>
16871 * tree-vrp.c (compare_assert_loc): Turn into a function template
16872 with stable template parameter. Only test if a->e is NULL,
16873 !a->e == !b->e has been verified already. Use e == NULL or
16874 e != NULL instead of e or ! e tests. If stable is true, don't use
16875 iterative_hash_expr, on the other side allow a or b or both NULL
16876 and sort the NULLs last.
16877 (process_assert_insertions): Sort using compare_assert_loc<false>
16878 instead of compare_assert_loc, later sort using
16879 compare_assert_loc<true> before calling process_assert_insertions_for
16880 in a loop. Use break instead of continue once seen NULL pointer.
16882 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
16884 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
16885 Cortex-R7 and Cortex-R8 processors.
16887 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
16889 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
16890 uninitialized while src is not.
16892 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
16894 * common/config/arm/arm-common.c: Adjust include path for
16896 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
16897 (arm-cpu.h): Create in build directory. Adjust dependency rules.
16898 (arm-cpu-data.h): Likewise.
16899 (arm-cpu-cdata.h): Likewise.
16900 * config/arm/arm-cpu.h: Delete.
16901 * config/arm/arm-cpu-cdata.h: Delete.
16902 * config/arm/arm-cpu-data.h: Delete.
16904 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
16906 * config/arm/arm-cpus.in (cortex-a55): New.
16907 (cortex-a75): Likewise.
16908 (cortex-a75.cortex-a55): Likewise.
16909 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
16911 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
16912 * config/arm/arm-cpu-cdata.h: Regenerate.
16913 * config/arm/arm-cpu-data.h: Regenerate.
16914 * config/arm/arm-cpu.h: Regenerate.
16915 * config/arm/arm-tables.opt: Regenerate.
16916 * config/arm/arm-tune.md: Regenerate.
16918 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
16920 * haifa-sched.c (sched_create_recovery_edges): Update profile.
16922 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
16924 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
16927 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
16929 PR tree-optimization/81292
16930 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
16931 full_string_p, also call adjust_related_strinfos if the adjustment
16932 is simple, otherwise invalidate related strinfos.
16934 2017-07-04 Martin Liska <mliska@suse.cz>
16937 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
16938 newly created variable as DECL_IGNORED_P.
16940 2017-07-04 Martin Liska <mliska@suse.cz>
16943 * ipa-inline.c (inline_small_functions):
16944 Use xstrdup_for_dump.
16946 2017-07-04 Tom de Vries <tom@codesourcery.com>
16948 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
16950 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
16953 * config/darwin.c (darwin_function_switched_text_sections):
16956 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
16958 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
16960 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
16962 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
16964 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16966 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
16967 min_profitable_iters, and th as inclusive lower bounds.
16968 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
16969 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
16970 for min_profitable_iters and min_profitable_estimate.
16971 (vect_transform_loop): Treat th as an inclusive lower bound.
16972 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
16974 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
16977 * config/darwin.c (darwin_function_switched_text_sections):
16978 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
16979 in two pieces, and suppress the use of buf.
16981 2017-07-03 Nathan Sidwell <nathan@acm.org>
16983 * hash-table.h (hash_table_mod1): Fix indentation.
16985 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16987 PR middle-end/81290
16988 * predict.c (force_edge_cold): Be more careful about propagation
16990 * profile-count.h (profile_probability::guessed,
16991 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
16993 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
16995 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
16997 * doc/invoke.texi (rcpc architecture extension): Document it.
16999 2017-07-03 Richard Biener <rguenther@suse.de>
17001 PR tree-optimization/60510
17002 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
17003 the scalar reduction PHI and use it.
17004 (vectorizable_reduction): Properly guard the single_defuse_cycle
17005 path for non-SLP reduction chains where we cannot use it.
17006 Rework reduc_def/index and vector type deduction. Rework
17007 vector operand gathering during reduction op code-gen.
17008 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
17009 chains dissolve the chain and leave it to non-SLP reduction
17012 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17014 * tree-data-ref.h (dr_alignment): Declare.
17015 * tree-data-ref.c (dr_alignment): New function.
17016 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
17017 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
17019 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
17021 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17023 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
17024 and base_misalignment fields.
17025 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
17026 * tree-data-ref.c: Include builtins.h.
17027 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
17028 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
17029 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
17030 * tree-vect-data-refs.c: Include tree-cfg.h.
17031 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
17032 fields instead of calculating an alignment here.
17033 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
17034 innermost_loop_behavior fields.
17036 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17038 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
17040 (DR_STEP_ALIGNMENT): New macro.
17041 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
17042 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
17043 (create_data_ref): Print it.
17044 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
17045 to tell whether the step preserves vector (mis)alignment.
17046 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17047 Move the check for an integer step and generalise to all INTEGER_CST.
17048 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
17049 Print the outer step alignment.
17051 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17053 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
17054 with offset_alignment.
17055 (DR_ALIGNED_TO): Delete.
17056 (DR_OFFSET_ALIGNMENT): New macro.
17057 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
17058 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
17059 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
17060 (create_data_ref): Likewise.
17061 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17062 (vect_analyze_data_refs): Likewise.
17063 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
17064 creating dummy innermost behavior.
17066 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17068 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
17069 with a "innermost_loop_behavior *" and refeence tree.
17070 * tree-data-ref.c (dr_analyze_innermost): Likewise.
17071 (create_data_ref): Update call accordingly.
17072 * tree-predcom.c (find_looparound_phi): Likewise.
17074 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17076 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
17077 fields with dr_wrt_vec_loop.
17078 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
17079 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
17080 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
17081 (vect_dr_behavior): New function.
17082 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17083 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
17084 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
17085 track whether the step preserves the misalignment.
17086 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17087 Use vect_dr_behavior.
17088 (vect_setup_realignment): Update call accordingly.
17089 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
17090 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
17091 call to vect_create_addr_base_for_vector_ref.
17092 (vect_create_cond_for_align_checks): Likewise.
17093 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
17094 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
17095 (vect_recog_mask_conversion_pattern): Likewise.
17096 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
17097 (new_stmt_vec_info): Remove redundant zeroing.
17099 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17101 * common/config/arm/arm-common.c (arm_be8_option): New function.
17102 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
17103 (ISA_ARMv6): Add isa_bit_be8.
17104 * config/arm/arm.h (arm_be8_option): Add prototype.
17105 (BE8_SPEC_FUNCTION): New define.
17106 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
17107 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
17108 (mlittle-endian): Similarly.
17109 (mbe8, mbe32): New options.
17110 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
17111 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
17113 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17115 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
17117 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17119 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
17120 (cleanup_tree_cfg_bb): Use it.
17121 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
17123 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
17125 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17128 * loop-doloop.c (add_test): Update profile.
17130 2017-07-03 Martin Liska <mliska@suse.cz>
17133 * sanopt.c (rewrite_usage_of_param): New function.
17134 (sanitize_rewrite_addressable_params): Likewise.
17135 (pass_sanopt::execute): Call rewrite_usage_of_param.
17137 2017-07-03 Richard Biener <rguenther@suse.de>
17139 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
17140 back to using VIEW_CONVERT_EXPR.
17142 2017-07-03 Martin Liska <mliska@suse.cz>
17145 * doc/extend.texi: Document when a resolver function is
17146 generated for target_clones.
17148 2017-07-03 Martin Liska <mliska@suse.cz>
17150 * asan.c (asan_emit_stack_protection): Unpoison just red zones
17151 and shadow memory of auto variables which are subject of
17152 use-after-scope sanitization.
17153 (asan_expand_mark_ifn): Add do set only when is_poison.
17155 2016-07-03 Richard Biener <rguenther@suse.de>
17157 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
17159 (vect_force_simple_reduction): Record reduction def -> phi mapping.
17160 (vectorizable_reduction): Perform reduction PHI creation when
17161 visiting a reduction PHI and adjust and simplify code generation
17162 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
17163 (vect_transform_loop): Visit reduction PHIs.
17164 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
17165 defs into the SLP tree.
17166 (vect_build_slp_tree): Reduction defs terminate the recursion.
17167 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
17169 (vect_get_vec_defs_for_stmt_copy): Export.
17170 (vect_get_vec_defs): Likewise.
17171 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
17173 (vect_get_vec_defs_for_stmt_copy): Declare.
17174 (vect_get_vec_defs): Likewise.
17176 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17178 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
17179 parameter with a "loop" parameter and use it instead of the
17180 loop containing DR_STMT. Don't check simple_iv when doing
17181 BB analysis. Describe the two analysis modes in the comment.
17183 2017-07-03 Tom de Vries <tom@codesourcery.com>
17185 PR tree-optimization/69468
17186 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
17187 (find_same_succ_bb): Handle ignore_edge_flags.
17189 2017-07-03 Tom de Vries <tom@codesourcery.com>
17191 PR tree-optimization/81192
17192 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
17194 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
17196 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
17198 2017-07-03 Tom de Vries <tom@codesourcery.com>
17200 PR tree-optimization/81192
17201 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
17202 BB_SAME_SUCC (bb) == NULL.
17204 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17206 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
17209 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17211 * dumpfile.c: Include profile-count.h
17212 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
17214 (insert_cond_bb): Update profile.
17215 * tree-cfg.h (insert_cond_bb): Update prototype.
17216 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
17217 * tree-dump.c: Do not include tree-cfg.
17219 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17221 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
17223 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17225 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
17228 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17230 * tree-complex.c (expand_complex_div_wide): update profile.
17232 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17233 Alan Hayward <alan.hayward@arm.com>
17234 David Sherwood <david.sherwood@arm.com>
17236 * Makefile.in (MACHMODE_H): Remove insn-modes.h
17237 (CORETYPES_H): New define.
17238 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
17239 (insn-modes-inline.h, s-modes-inline-h): New rules.
17240 (generated_files): Add insn-modes-inline.h.
17241 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
17242 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
17243 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
17244 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
17245 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
17246 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
17247 (build/gencodes.o, build/genconditions.o): Likewise.
17248 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
17249 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
17250 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
17251 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
17252 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
17253 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
17254 * coretypes.h: Include everything up to real.h for generators.
17255 Include insn-modes.h first. Include wide-int-print.h after
17256 wide-int.h. Include insn-modes-inline.h and then machmode.h.
17257 * machmode.h: Don't include insn-modes.h here.
17258 * function-tests.c: Remove includes of signop.h, machmode.h,
17259 double-int.h and wide-int.h.
17261 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
17263 * optc-save-gen.awk: Likewise.
17264 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
17265 * godump.c: Remove include of wide-int-print.h.
17266 * pretty-print.h: Likewise.
17267 * wide-int-print.cc: Likewise.
17268 * wide-int.cc: Likewise.
17269 * hash-map-tests.c: Remove include of signop.h.
17270 * hash-set-tests.c: Likewise.
17271 * rtl-tests.c: Likewise.
17272 * mkconfig.sh: Remove include of machmode.h.
17273 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
17275 (emit_insn_modes_inline_h): ...this new function. Emit the code
17276 into an insn-modes-inline.h header file, adding appropriate
17277 include guards and end comments.
17278 (emit_insn_modes_c_header): Remove include of machmode.h.
17279 (emit_min_insn_modes_c_header): Include coretypes.h rather than
17281 (main): Handle -i flag and call emit_insn_modes_inline_h when
17284 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17286 * tree-ssa-strlen.c (strinfo): Rename the length field to
17287 nonzero_chars. Add a full_string_p field.
17288 (compare_nonzero_chars, zero_length_string_p): New functions.
17289 (get_addr_stridx): Add an offset_out parameter.
17290 Use compare_nonzero_chars.
17291 (get_stridx): Update accordingly. Use compare_nonzero_chars.
17292 (new_strinfo): Update after above changes to strinfo.
17293 (set_endptr_and_length): Set full_string_p.
17294 (get_string_length): Update after above changes to strinfo.
17295 (unshare_strinfo): Update call to new_strinfo.
17296 (maybe_invalidate): Likewise.
17297 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
17298 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
17299 as a uhwi instead of an shwi. Update after above changes to
17300 strinfo and new_strinfo.
17301 (zero_length_string): Assert that chainsi contains full strings.
17302 Use zero_length_string_p. Update call to new_strinfo.
17303 (adjust_related_strinfos): Update after above changes to strinfo.
17304 Copy full_string_p from origsi.
17305 (adjust_last_stmt): Use zero_length_string_p.
17306 (handle_builtin_strlen): Update after above changes to strinfo and
17307 new_strinfo. Install the lhs as the string length if the previous
17308 entry didn't describe a full string.
17309 (handle_builtin_strchr): Update after above changes to strinfo
17311 (handle_builtin_strcpy): Likewise.
17312 (handle_builtin_strcat): Likewise.
17313 (handle_builtin_malloc): Likewise.
17314 (handle_pointer_plus): Likewise.
17315 (handle_builtin_memcpy): Likewise. Track nonzero characters
17316 that aren't necessarily followed by a nul terminator.
17317 (handle_char_store): Likewise.
17319 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17321 PR tree-optimization/80769
17322 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
17323 for malloc and calloc. Document the new invariant that all related
17324 strinfos have delayed lengths or none do.
17325 (verify_related_strinfos): Move earlier in file.
17326 (set_endptr_and_length): New function, split out from...
17327 (get_string_length): ...here. Also set the lengths of related
17329 (zero_length_string): Assert that chainsi has known (rather than
17331 (adjust_related_strinfos): Likewise.
17333 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17335 PR tree-optimization/81136
17336 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
17337 assert that two references with the same misalignment have the same
17338 compile-time misalignment if those compile-time misalignments
17341 2017-07-01 Andi Kleen <ak@linux.intel.com>
17343 * print-tree.c (print_node): Print all attributes.
17345 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17347 * cfg.c (scale_bbs_frequencies): New function.
17348 * cfg.h (scale_bbs_frequencies): Declare it.
17349 * cfgloopanal.c (single_likely_exit): Cleanup.
17350 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
17352 (scale_loop_profile): Likewise.
17353 (loop_version): Likewise.
17354 (create_empty_loop_on_edge): Update.
17355 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
17356 scale_loop_frequencies, scale_loop_profile, loopify,
17357 loop_version): Update prototypes.
17358 * modulo-sched.c (sms_schedule): Update.
17359 * predict.c (unlikely_executed_edge_p): Also check probability.
17360 (probably_never_executed_edge_p): Fix typo.
17361 * tree-if-conv.c (version_loop_for_if_conversion): Update.
17362 * tree-parloops.c (gen_parallel_loop): Update.
17363 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
17364 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
17365 * tree-ssa-loop-split.c (split_loop): Update.
17366 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
17367 * tree-vect-loop-manip.c (vect_do_peeling): Update.
17368 (vect_loop_versioning): Update.
17369 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
17371 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17373 * trans-mem.c (split_bb_make_tm_edge): Update profile.
17375 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17377 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
17378 to keep profile consistent.
17380 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17382 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
17383 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
17384 * profile-count.h (max_safe_multiplier): Make unsigned.
17385 (profile_count::guessed_zero): New.
17387 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17389 * bb-reorder.c (fix_up_crossing_landing_pad,
17390 fix_crossing_conditional_branches): Use make_single_succ_edge
17391 to keep profile consistent.
17393 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17395 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
17398 2017-07-01 Jakub Jelinek <jakub@redhat.com>
17401 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
17402 the right scopes, make sure cond_jump isn't preserved between multiple
17403 iterations. Search for fallthru edge whenever there are 3+ edges and
17404 use find_fallthru_edge for it.
17406 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17408 Patch by Alexander Monakov <amonakov@ispras.ru>
17409 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
17410 probabilities consistently.
17412 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17414 * pa.c (pa_expand_compare_and_swap_loop): Update call of
17415 emit_cmp_and_jump_insns.
17417 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17420 * tree-inline.c (expand_call_inline): Combine profile statuses.
17422 2017-06-30 Andrew Pinski <apinski@cavium.com>
17424 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
17425 fold_stmt returned true.
17427 2017-06-30 Nathan Sidwell <nathan@acm.org>
17429 * ggc.h (empty_string): Delete.
17430 * cfgexpand.c (expand_asm_stmt): Use plain "".
17431 * optabs.c (expand_asm_memory_barrier): Likewise.
17432 * stringpool.c (empty_string): Delete.
17433 (digit_vector, digit_string): Delete.
17434 (ggc_alloc_string): Use plain "", don't optimize single digit
17435 strings. Use ggc_alloc_atomic.
17437 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
17439 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
17440 comparison set and one other set, use the cost of the non-comparison
17443 2017-06-30 Nathan Sidwell <nathan@acm.org>
17445 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
17448 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
17450 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
17451 loops. Remove now unneeded calls to gimple_switch_set_label() that
17452 just set removed labels to NULL_TREE.
17454 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
17456 * tree-ssanames.c (set_range_info_raw): Abstract from ...
17457 (set_range_info): ...here. Only call set_range_info_raw if domain
17459 (set_nonzero_bits): Call set_range_info_raw.
17460 * tree-ssanames.h (set_range_info_raw): New.
17462 2017-06-30 Jakub Jelinek <jakub@redhat.com>
17465 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
17466 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
17467 of nonimmediate_operand and <store_mask_constraint> instead of m
17468 for the input operand. For V8FI iterator, always split if input
17469 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
17470 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
17471 <store_mask_predicate> instead of register_operand and
17472 <store_mask_constraint> instead of v for the input operand. Make
17473 sure both operands aren't MEMs for if not <mask_applied>.
17475 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
17477 * lto-wrapper.c (copy_file) Close both file descriptors before
17480 2017-06-30 Martin Liska <mliska@suse.cz>
17483 * multiple_target.c (create_dispatcher_calls): Make ifunc
17484 also for function that don't have calls or are not referenced.
17486 2017-06-30 Richard Biener <rguenther@suse.de>
17488 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
17489 analyze the first scalar stmt. Move vector type computation
17490 for the BB case here from ...
17491 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
17492 live operation processing in the SLP case properly.
17494 2017-06-30 Richard Biener <rguenther@suse.de>
17496 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
17498 2017-06-30 Martin Liska <mliska@suse.cz>
17501 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
17502 before BUILT_IN_UNWIND_RESUME when ASAN is used.
17504 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
17506 * doc/invoke.texi (AArch64): Add missing options and remove redundant
17509 2017-06-30 Richard Biener <rguenther@suse.de>
17511 PR tree-optimization/81249
17512 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
17513 condition reduction result to original scalar type.
17515 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17517 * profile-count.h (enum profile_quality): Fix typos and whitespace
17520 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17522 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
17523 type for branch probabilities.
17525 2017-06-29 Julian Brown <julian@codesourcery.com>
17526 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17528 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
17529 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
17530 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
17531 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
17533 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17535 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
17536 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
17537 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
17538 CC usage from generic code to here.
17539 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
17540 CC usage into the target macros.
17542 2017-06-29 Maya Rashish <coypu@sdf.org>
17544 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
17547 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17549 * arm/arm-builtins.c: Include profile-count.h
17550 * except.c (sjlj_emit_function_enter): Use
17551 profile_probability::unlikely.
17553 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17555 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
17556 and tocrel_offset be pointer args rather than implicitly using
17558 (legitimate_constant_pool_address_p, rs6000_emit_move,
17559 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
17560 tocrel_offset and use in toc_relative_expr_p call.
17561 (print_operand, print_operand_address): Use static tocrel_base_oac
17562 and tocrel_offset_oac.
17563 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
17566 2017-06-29 Maya Rashish <coypu@sdf.org>
17568 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
17570 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
17572 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
17573 objects, take into account only the alignment of 'op0' and 'mode1' if
17576 2017-06-29 Steve Ellcey <sellcey@cavium.com>
17578 * ccmp.c (ccmp_tree_comparison_p): New function.
17579 (ccmp_candidate_p): Update to use above function.
17580 (get_compare_parts): New function.
17581 (expand_ccmp_next): Update to use new functions.
17582 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
17584 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
17585 take mode as argument.
17586 * ccmp.h (expand_ccmp_expr): Add mode as argument.
17587 * expr.c (expand_expr_real_1): Pass mode as argument.
17589 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
17591 * combine.c (combine_instructions): Print insns to dump_file, together
17594 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17596 * asan.c (asan_emit_stack_protection): Update.
17597 (create_cond_insert_point): Update.
17598 * auto-profile.c (afdo_propagate_circuit): Update.
17599 * basic-block.h (struct edge_def): Turn probability to
17600 profile_probability.
17601 (EDGE_FREQUENCY): Update.
17602 * bb-reorder.c (find_traces_1_round): Update.
17603 (better_edge_p): Update.
17604 (sanitize_hot_paths): Update.
17605 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
17606 (make_single_succ_edge): Update.
17607 (check_bb_profile): Update.
17608 (dump_edge_info): Update.
17609 (update_bb_profile_for_threading): Update.
17610 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
17611 probabilitycount to 0.
17612 * cfgbuild.c (compute_outgoing_frequencies): Update.
17613 * cfgcleanup.c (try_forward_edges): Update.
17614 (outgoing_edges_match): Update.
17615 (try_crossjump_to_edge): Update.
17616 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
17617 (expand_gimple_tailcall): Update.
17618 (construct_init_block): Use make_single_succ_edge.
17619 (construct_exit_block): Use make_single_succ_edge.
17620 * cfghooks.c (verify_flow_info): Update.
17621 (redirect_edge_succ_nodup): Update.
17622 (split_edge): Update.
17623 (account_profile_record): Update.
17624 * cfgloopanal.c (single_likely_exit): Update.
17625 * cfgloopmanip.c (scale_loop_profile): Update.
17626 (set_zero_probability): Remove.
17627 (duplicate_loop_to_header_edge): Update.
17628 * cfgloopmanip.h (loop_version): Update prototype.
17629 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
17630 (force_nonfallthru_and_redirect): Update.
17631 (update_br_prob_note): Update.
17632 (rtl_verify_edges): Update.
17633 (purge_dead_edges): Update.
17634 (rtl_lv_add_condition_to_bb): Update.
17635 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
17636 * cgraphunit.c (init_lowered_empty_function): Update.
17637 (cgraph_node::expand_thunk): Update.
17638 * cilk-common.c: Include profile-count.h
17639 * dojump.c (inv): Remove.
17640 (jumpifnot): Update.
17641 (jumpifnot_1): Update.
17642 (do_jump_1): Update.
17644 (do_jump_by_parts_greater_rtx): Update.
17645 (do_compare_rtx_and_jump): Update.
17646 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
17647 do_jump_1. do_compare_rtx_and_jump): Update prototype.
17648 * dwarf2cfi.c: Include profile-count.h
17649 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
17650 (sjlj_emit_dispatch_table): Likewise.
17651 * explow.c: Include profile-count.h
17652 * expmed.c (emit_store_flag_force): Update.
17653 (do_cmp_and_jump): Update.
17654 * expr.c (compare_by_pieces_d::generate): Update.
17655 (compare_by_pieces_d::finish_mode): Update.
17656 (emit_block_move_via_loop): Update.
17657 (store_expr_with_bounds): Update.
17658 (store_constructor): Update.
17659 (expand_expr_real_2): Update.
17660 (expand_expr_real_1): Update.
17661 * expr.h (try_casesi, try_tablejump): Update prototypes.
17662 * gimple-pretty-print.c (dump_probability): Update.
17663 (dump_profile): New.
17664 (dump_gimple_label): Update.
17665 (dump_gimple_bb_header): Update.
17666 * graph.c (draw_cfg_node_succ_edges): Update.
17667 * hsa-gen.c (convert_switch_statements): Update.
17668 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
17669 (find_if_case_1): Update.
17670 (find_if_case_2): Update.
17671 * internal-fn.c (expand_arith_overflow_result_store): Update.
17672 (expand_addsub_overflow): Update.
17673 (expand_neg_overflow): Update.
17674 (expand_mul_overflow): Update.
17675 (expand_vector_ubsan_overflow): Update.
17676 * ipa-cp.c (good_cloning_opportunity_p): Update.
17677 * ipa-split.c (split_function): Use make_single_succ_edge.
17678 * ipa-utils.c (ipa_merge_profiles): Update.
17679 * loop-doloop.c (add_test): Update.
17680 (doloop_modify): Update.
17681 * loop-unroll.c (compare_and_jump_seq): Update.
17682 (unroll_loop_runtime_iterations): Update.
17683 * lra-constraints.c (lra_inheritance): Update.
17684 * lto-streamer-in.c (input_cfg): Update.
17685 * lto-streamer-out.c (output_cfg): Update.
17686 * mcf.c (adjust_cfg_counts): Update.
17687 * modulo-sched.c (sms_schedule): Update.
17688 * omp-expand.c (expand_omp_for_init_counts): Update.
17689 (extract_omp_for_update_vars): Update.
17690 (expand_omp_ordered_sink): Update.
17691 (expand_omp_for_ordered_loops): Update.
17692 (expand_omp_for_generic): Update.
17693 (expand_omp_for_static_nochunk): Update.
17694 (expand_omp_for_static_chunk): Update.
17695 (expand_cilk_for): Update.
17696 (expand_omp_simd): Update.
17697 (expand_omp_taskloop_for_outer): Update.
17698 (expand_omp_taskloop_for_inner): Update.
17699 * omp-simd-clone.c (simd_clone_adjust): Update.
17700 * optabs.c (expand_doubleword_shift): Update.
17701 (expand_abs): Update.
17702 (emit_cmp_and_jump_insn_1): Update.
17703 (expand_compare_and_swap_loop): Update.
17704 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
17705 * predict.c (predictable_edge_p): Update.
17706 (edge_probability_reliable_p): Update.
17707 (set_even_probabilities): Update.
17708 (combine_predictions_for_insn): Update.
17709 (combine_predictions_for_bb): Update.
17710 (propagate_freq): Update.
17711 (estimate_bb_frequencies): Update.
17712 (force_edge_cold): Update.
17713 * profile-count.c (profile_count::dump): Add missing space into dump.
17714 (profile_count::debug): Add newline.
17715 (profile_count::differs_from_p): Explicitly convert to unsigned.
17716 (profile_count::stream_in): Update.
17717 (profile_probability::dump): New member function.
17718 (profile_probability::debug): New member function.
17719 (profile_probability::differs_from_p): New member function.
17720 (profile_probability::differs_lot_from_p): New member function.
17721 (profile_probability::stream_in): New member function.
17722 (profile_probability::stream_out): New member function.
17723 * profile-count.h (profile_count_quality): Rename to ...
17724 (profile_quality): ... this one.
17725 (profile_probability): New.
17726 (profile_count): Update.
17727 * profile.c (compute_branch_probabilities): Update.
17728 * recog.c (peep2_attempt): Update.
17729 * sched-ebb.c (schedule_ebbs): Update.
17730 * sched-rgn.c (find_single_block_region): Update.
17731 (compute_dom_prob_ps): Update.
17732 (schedule_region): Update.
17733 * sel-sched-ir.c (compute_succs_info): Update.
17734 * stmt.c (struct case_node): Update.
17735 (do_jump_if_equal): Update.
17736 (get_outgoing_edge_probs): Update.
17737 (conditional_probability): Update.
17738 (emit_case_dispatch_table): Update.
17739 (expand_case): Update.
17740 (expand_sjlj_dispatch_table): Update.
17741 (emit_case_nodes): Update.
17742 * targhooks.c: Update.
17743 * tracer.c (better_p): Update.
17744 (find_best_successor): Update.
17745 * trans-mem.c (expand_transaction): Update.
17746 * tree-call-cdce.c: Update.
17747 * tree-cfg.c (gimple_split_edge): Upate.
17748 (move_sese_region_to_fn): Upate.
17749 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
17750 * tree-eh.c (lower_resx): Upate.
17751 (cleanup_empty_eh_move_lp): Upate.
17752 * tree-if-conv.c (version_loop_for_if_conversion): Update.
17753 * tree-inline.c (copy_edges_for_bb): Update.
17754 (copy_cfg_body): Update.
17755 * tree-parloops.c (gen_parallel_loop): Update.
17756 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
17757 (gimple_gen_time_profiler): Update.
17758 * tree-ssa-dce.c (remove_dead_stmt): Update.
17759 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
17760 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
17761 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
17762 (unloop_loops): Update.
17763 (try_peel_loop): Update.
17764 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
17765 * tree-ssa-loop-split.c (connect_loops): Update.
17766 (split_loop): Update.
17767 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
17768 (hoist_guard): Update.
17769 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
17770 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
17771 (value_replacement): Update.
17772 * tree-ssa-reassoc.c (branch_fixup): Update.
17773 * tree-ssa-tail-merge.c (replace_block_by): Update.
17774 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
17775 (create_edge_and_update_destination_phis): Update.
17776 (compute_path_counts): Update.
17777 (recompute_probabilities): Update.
17778 (update_joiner_offpath_counts): Update.
17779 (freqs_to_counts_path): Update.
17780 (duplicate_thread_path): Update.
17781 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
17782 (struct switch_conv_info): Update.
17783 (gen_inbound_check): Update.
17784 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
17785 (vect_do_peeling): Update.
17786 (vect_loop_versioning): Update.
17787 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
17788 (optimize_mask_stores): Update.
17789 * ubsan.c (ubsan_expand_null_ifn): Update.
17790 * value-prof.c (gimple_divmod_fixed_value): Update.
17791 (gimple_divmod_fixed_value_transform): Update.
17792 (gimple_mod_pow2): Update.
17793 (gimple_mod_pow2_value_transform): Update.
17794 (gimple_mod_subtract): Update.
17795 (gimple_mod_subtract_transform): Update.
17796 (gimple_ic): Update.
17797 (gimple_stringop_fixed_value): Update.
17798 (gimple_stringops_transform): Update.
17799 * value-prof.h: Update.
17801 2017-06-29 Carl Love <cel@us.ibm.com>
17803 * config/rs6000/rs6000-c.c: Add support for built-in functions
17804 vector signed int vec_signed (vector float);
17805 vector signed long long vec_signed (vector double);
17806 vector signed int vec_signed2 (vector double, vector double);
17807 vector signed int vec_signede (vector double);
17808 vector signed int vec_signedo (vector double);
17809 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
17810 instruction generator.
17811 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
17812 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
17813 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
17815 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
17816 vunsignede_v2df): Add define_expands.
17817 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
17818 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
17819 VEC_UNSIGNEDO): Add definitions.
17820 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
17821 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
17822 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
17823 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
17824 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
17825 * config/rs6000/altivec.h (vec_signed, vec_signed2,
17826 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
17827 vec_unsignede, vec_unsignedo): Add builtin defines.
17828 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
17830 * doc/extend.texi: Update the built-in documentation file for the
17831 new built-in functions.
17833 2017-06-29 Richard Biener <rguenther@suse.de>
17835 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
17836 reduction chains to LOOP_VINFO_REDUCTIONS.
17837 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
17838 SLP reductions after processing reduction chains.
17840 2017-06-29 Nathan Sidwell <nathan@acm.org>
17842 * builtins.c (fold_builtin_FUNCTION): Use
17843 lang_hooks.decl_printable_name.
17845 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
17847 PR middle-end/81194
17848 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
17849 with only one label.
17850 * stmt.c (expand_case): Assert NCASES is greater than one.
17852 2017-06-29 Richard Biener <rguenther@suse.de>
17854 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
17856 (group_case_labels): Likewise.
17857 (find_taken_edge): Push sanity checking on val to workers...
17858 (find_taken_edge_cond_expr): ... here
17859 (find_taken_edge_switch_expr): ... and here, handle cases
17860 with just a default label.
17861 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
17862 (group_case_labels): Likewise.
17863 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
17864 group_case_labels does anything cleanup the CFG again.
17866 2017-06-29 Bin Cheng <bin.cheng@arm.com>
17868 PR tree-optimization/81196
17869 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
17870 exit condition comparing two IVs.
17872 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
17874 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
17875 profile to the dummy entry at the end of the list of architectures.
17876 * config/arm/arm-cpu-cdata.h: Regenerated.
17878 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17879 Michael Collison <michael.collison@arm.com>
17882 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
17884 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
17885 (*aarch64_reg_<mode>3_minus_mask): New pattern.
17886 (*aarch64_<optab>_reg_di3_mask2): New pattern.
17887 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
17888 of shift when the shift amount is masked with constant equal to
17889 the size of the mode.
17890 * config/aarch64/predicates.md (subreg_lowpart_operator): New
17893 2017-06-29 Martin Liska <mliska@suse.cz>
17895 * config/i386/i386.opt: Change range from [1,5] to [0,5].
17897 2017-06-29 Yury Gribov <tetra2005@gmail.com>
17900 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
17903 (edge_growth_cache_entry::edge_growth_cache_entry): New
17905 (reset_edge_growth_cache): Update to use constructor.
17907 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17909 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
17910 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
17911 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
17913 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
17915 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
17916 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
17918 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
17920 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
17921 (*-linux-uclibc*): Add t-uclibc tmake_file.
17922 * config/t-musl: New.
17923 * config/t-uclibc: New.
17925 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
17927 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
17929 (gen_comm_data): Emit architectural setting of arch_prof.
17930 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
17932 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
17933 (armv8-m.base, armv8-m.main): Likewise.
17934 * arm-protos.h (arm_build_target): Add profile field.
17935 (arch_option): Likewise.
17936 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
17938 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
17939 arm_active_target.profile.
17941 2017-06-28 Richard Biener <rguenther@suse.de>
17943 PR middle-end/81227
17944 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
17945 TYPE_OVERFLOW_WRAPS.
17946 * match.pd (negate_expr_p): Likewise.
17947 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
17948 fold_build2, not fold_binary.
17950 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
17952 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
17953 Convert memory address to Pmode.
17954 (aarch64_print_operand): Assert MEM operands are always Pmode.
17956 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
17959 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
17960 Remove redundant if.
17961 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
17962 * config/arm/aarch-common-protos.h
17963 (aarch_forward_to_shift_is_not_shifted_re): Remove.
17964 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
17966 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
17969 * multiple_target.c (create_dispatcher_calls): Set the default
17970 clone to be static, not public.
17972 2017-06-28 Richard Biener <rguenther@suse.de>
17974 * tree-vect-loop.c (vectorizable_reduction): Move special
17975 cond reduction IV var creation ...
17976 (vect_create_epilog_for_reduction): ... here. Remove induction_index
17977 parameter. Use STMT_VINFO_VECTYPE.
17978 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
17981 2017-06-28 Martin Liska <mliska@suse.cz>
17984 * ipa-visibility.c (non_local_p): Handle visibility.
17986 2017-06-28 Martin Liska <mliska@suse.cz>
17989 * common.opt: Add IntegerRange to various options.
17990 * opt-functions.awk (integer_range_info): New function.
17991 * optc-gen.awk: Add integer_range_info to cl_options struct.
17992 * opts-common.c (decode_cmdline_option): Handle
17993 CL_ERR_INT_RANGE_ARG.
17994 (cmdline_handle_error): Likewise.
17995 * opts.c (print_filtered_help): Show valid interval in
17996 when --help is provided.
17997 * opts.h (struct cl_option): Add range_min and range_max fields.
17998 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
18000 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
18002 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
18003 (x * C EQ/NE y * C): New transformation.
18005 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
18007 * genmultilib (combination_space): Accept '+' in option names.
18009 2017-06-28 Martin Liska <mliska@suse.cz>
18012 * asan.c (instrument_derefs): Bail out inner references
18013 that are hard register variables.
18015 2017-06-28 Jakub Jelinek <jakub@redhat.com>
18018 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
18019 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
18021 2017-06-28 Richard Biener <rguenther@suse.de>
18023 * tree-vectorizer.h (vect_get_vec_defs): Remove.
18024 (vect_get_slp_defs): Adjust.
18025 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
18027 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
18029 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
18030 get_initial_defs_for_reduction instead of vect_get_vec_defs.
18031 (vectorizable_reduction): Adjust.
18032 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
18034 (vect_get_slp_defs): Likewise.
18035 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
18036 (vectorizable_bswap): Adjust.
18037 (vectorizable_call): Likewise.
18038 (vectorizable_conversion): Likewise.
18039 (vectorizable_assignment): Likewise.
18040 (vectorizable_shift): Likewise.
18041 (vectorizable_operation): Likewise.
18042 (vectorizable_store): Likewise.
18043 (vectorizable_condition): Likewise.
18044 (vectorizable_comparison): Likewise.
18046 2017-06-28 Michael Collison <michael.collison@arm.com>
18049 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
18051 (arm_gen_movmemqi): Removed unused variable 'i'.
18052 Convert 'for' loop into 'while' loop.
18053 (arm_expand_prologue): Remove last unnecessary set of insn.
18054 (thumb_pop): Remove unused variable 'pushed_words'.
18055 (thumb_exit): Remove last unnecessary set of regs_to_pop.
18057 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18059 * config/s390/predicates.md: Use s390_rel_address_ok_p.
18060 * config/s390/s390-protos.h: Add prototype of
18061 s390_rel_address_ok_p.
18062 * config/s390/s390.c (s390_got_symbol): New function.
18063 (s390_rel_address_ok_p): New function.
18064 (legitimize_pic_address): Use s390_rel_address_ok_p.
18065 (s390_load_got): Use s390_got_symbol.
18066 (s390_option_override): Issue error if
18067 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
18068 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
18070 * config/s390/s390.opt: New option mpic-data-is-text-relative.
18072 2017-06-27 Andrew Pinski <apinski@cavium.com>
18074 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
18075 (X * copysign (1.0, X)): New pattern.
18076 (X * copysign (1.0, -X)): New pattern.
18077 (copysign (-1.0, CST)): New pattern.
18079 2017-06-27 Joseph Myers <joseph@codesourcery.com>
18081 * genmultilib (combination_space): Remove variable.
18082 Validate reuse rules against regular expression for any sequence
18083 of multilib options in any order.
18085 2017-06-27 Michael Collison <michael.collison@arm.com>
18087 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
18088 call aarch64_split_simd_combine.
18089 * (aarch64_combine_internal<mode>): Delete pattern.
18090 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
18091 Allow register and subreg operands.
18093 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18095 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
18096 specific need, just fallback on defaults.
18097 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
18099 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18100 Olivier Hainque <hainque@adacore.com>
18102 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
18104 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
18105 targets. Pick a default if no particular attempt applied.
18106 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
18109 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18111 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
18112 (x86_64-wrs-vxworks7): Likewise.
18114 2017-06-27 Marek Polacek <polacek@redhat.com>
18117 * ubsan.c (instrument_null): Check get_base_address's result for null.
18119 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18121 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
18123 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18125 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
18126 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
18127 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
18128 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
18129 New function types.
18130 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
18131 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
18132 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
18133 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
18134 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
18135 BUILT_IN_FEUPDATEENV): New builtins.
18136 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
18137 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
18138 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
18139 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
18141 (builtin_structptr_types): Adjust size.
18142 * tree.c (builtin_structptr_types): Add four entries.
18144 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18145 Olivier Hainque <hainque@adacore.com>
18147 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
18148 (TLS_SYM): New local macro, forcing reference to __tls__ on
18149 link command lines for VxWorks 7 RTPs, triggering initialization
18151 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
18152 OS features TLS support, true for RTPs on VxWorks 7.
18153 * config/vxworks.c (vxworks_override_options): Setup emutls
18156 2017-06-27 Jakub Jelinek <jakub@redhat.com>
18158 * predict.c (test_prediction_value_range): Use -1U instead of -1
18159 to avoid narrowing conversion warning.
18160 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
18161 to avoid narrowing conversion warning.
18162 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
18164 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
18166 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18168 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
18169 64bit configurations.
18170 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
18171 (SIZE_TYPE): Likewise.
18172 * config/vxworks.c (vxworks_emutls_var_fields): Use
18173 long_unsigned_type_node instead of unsigned_type_node as the offset
18174 field type, which is "pointer" mode in emutls.c.
18176 2017-06-27 Jakub Jelinek <jakub@redhat.com>
18179 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
18181 PR middle-end/81207
18182 * gimple-fold.c (replace_call_with_call_and_fold): Handle
18183 gimple_vuse copying separately from gimple_vdef copying.
18185 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18187 * value-prof.c (free_hist): Remove call to memset and the enclosing if
18190 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18191 Olivier Hainque <hainque@adacore.com>
18193 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
18194 for all vxworks7 targets.
18195 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
18196 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
18197 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
18198 variations for VX6/VX7 and 32/64bits later on in ...
18199 (VXWORKS_LIB_SPEC): Leverage new macros.
18200 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
18201 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
18203 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18205 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
18206 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
18208 2017-06-26 Carl Love <cel@us.ibm.com>
18210 * config/rs6000/rs6000-c.c: Add support for built-in functions
18211 vector bool char vec_reve (vector bool char);
18212 vector signed char vec_reve (vector signed char);
18213 vector unsigned char vec_reve (vector unsigned char);
18214 vector bool int vec_reve (vector bool int);
18215 vector signed int vec_reve (vector signed int);
18216 vector unsigned int vec_reve (vector unsigned int);
18217 vector bool long long vec_reve (vector bool long long);
18218 vector signed long long vec_reve (vector signed long long);
18219 vector unsigned long long vec_reve (vector unsigned long long);
18220 vector bool short vec_reve (vector bool short);
18221 vector signed short vec_reve (vector signed short);
18222 vector double vec_reve (vector double);
18223 vector float vec_reve (vector float);
18224 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
18225 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
18226 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
18227 (altivec_vreve): New pattern.
18228 * config/rs6000/altivec.h (vec_reve): New define.
18229 * doc/extend.texi (vec_rev): Update the built-in documentation file
18230 for the new built-in functions.
18232 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18234 PR tree-optimization/71815
18235 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
18237 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
18239 (slsr_process_phi): Likewise.
18240 (replace_uncond_cands_and_profitable_phis): Don't replace a
18241 multiply candidate with a stride of 1 (copy or cast).
18242 (phi_incr_cost): Call uses_consumed_by_stmt rather than
18244 (lowest_cost_path): Likewise.
18245 (total_savings): Likewise.
18247 2017-06-26 Richard Biener <rguenther@suse.de>
18250 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
18251 Use def_builtin_pure for all gather builtins.
18253 2017-06-26 Richard Biener <rguenther@suse.de>
18255 PR tree-optimization/81203
18256 * tree-tailcall.c (find_tail_calls): Do not move stmts into
18257 non-dominating BBs.
18259 2017-06-26 Marek Polacek <polacek@redhat.com>
18262 * doc/invoke.texi: Document -Wmultistatement-macros.
18264 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
18266 * doc/sourcebuild.texi (ARM-specific attributes): Document new
18267 arm_neon_ok_no_float_abi effective target.
18269 2017-06-26 Richard Biener <rguenther@suse.de>
18271 PR tree-optimization/80928
18272 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
18273 (copy_bbs): Set BB_DUPLICATED flag early.
18274 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
18276 (execute_on_shrinking_pred): Likewise.
18277 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
18278 BB_DUPLICATED blocks.
18279 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
18280 iterate over all PHIs considering removal of *gsi.
18282 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
18284 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
18287 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18289 * config/rs6000/rs6000-string.c: (expand_block_clear,
18290 do_load_for_compare, select_block_compare_mode,
18291 compute_current_alignment, expand_block_compare,
18292 expand_strncmp_align_check, expand_strn_compare,
18293 expand_block_move, rs6000_output_load_multiple)
18294 Move functions related to string/block move/compare
18295 to a separate file.
18296 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
18297 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
18298 for this function which is now used in two files.
18299 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
18300 * config.gcc: Add rs6000-string.o to extra_objs for
18301 targets powerpc*-*-* and rs6000*-*-*.
18303 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
18306 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
18307 32-bit, since indexed is not valid for DImode.
18308 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
18309 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
18310 (define_peephole2 for Altivec d-form load): Add 32-bit support.
18311 (define_peephole2 for Altivec d-form store): Likewise.
18314 * multiple_target.c (create_dispatcher_calls): Only create the
18315 dispatcher call if the function is the default clone of a
18316 versioned function.
18318 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
18320 PR middle-end/80902
18321 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
18322 a call, force the call to not be a tail call.
18324 2017-06-23 Jeff Law <law@redhat.com>
18326 * doc/contrib.texi: Add entry for Steven Pemberton's work on
18329 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
18331 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
18332 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
18333 handling for early expansion of vector shifts (sl,sr,sra,rl).
18334 (builtin_function_type): Add vector shift right instructions
18335 to the unsigned argument list.
18337 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
18339 rtl-optimizatoin/79286
18340 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
18341 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
18342 trap. PIC register plus a const unspec without offset can never trap.
18344 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
18346 * tree.h (builtin_structptr_type): New type.
18347 (builtin_structptr_types): Declare new array.
18348 * tree.c (builtin_structptr_types): New array.
18349 (free_lang_data, build_common_tree_nodes): Use it.
18351 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
18354 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
18357 2017-06-22 Matt Turner <mattst88@gmail.com>
18359 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
18360 Lake models to skylake case. Assume skylake for unknown
18361 models with clflushopt.
18363 2017-06-22 Jeff Law <law@redhat.com>
18365 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
18366 frame sizes that do not satisfy aarch64_uimm12_shift.
18368 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
18370 * profile-count.h (apply_probability,
18371 apply_scale, probability_in): Fix checks for zero.
18373 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18375 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
18376 * doc/cppdiropts.texi (-I @var{dir}): Document it.
18378 2016-06-22 Richard Biener <rguenther@suse.de>
18380 * tree-vect-loop.c (vect_model_reduction_cost): Handle
18381 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
18382 REDUC_MAX_EXPR support.
18383 (vectorizable_reduction): Likewise.
18384 (vect_create_epilog_for_reduction): Likewise.
18386 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
18388 * match.pd (A / (1 << B) -> A >> B): New.
18389 * generic-match-head.c: Include optabs-tree.h.
18390 * gimple-match-head.c: Likewise.
18391 * optabs-tree.h (target_supports_op_p): New.
18392 * optabs-tree.c (target_supports_op_p): New.
18394 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18396 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
18397 $gcc_cv_ld --help output.
18398 (gcc_cv_ld_demangle): Likewise.
18399 (gcc_cv_ld_eh_frame_hdr): Likewise.
18400 (gcc_cv_ld_pie): Likewise.
18401 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
18402 (gcc_cv_ld_buildid): Likewise.
18403 (gcc_cv_ld_sysroot): Likewise.
18404 (ld_bndplt_support): Likewise.
18405 (ld_pushpopstate_support): Likewise.
18406 * configure: Regenerate.
18407 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
18409 2017-06-21 Jakub Jelinek <jakub@redhat.com>
18412 * config/i386/sse.md (round<mode>2): Renumber match_dup and
18413 operands indexes to avoid gap between operands and match_dups.
18415 2017-06-21 Andrew Pinski <apinski@cavium.com>
18417 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
18418 Increment Arith_shift and Arith_shift_reg by 1.
18419 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
18421 * config/aarch64/aarch64.c (thunderx_tunings): Enable
18422 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
18423 (aarch64_strip_extend): Add new argument and test for it.
18424 (aarch64_cheap_mult_shift_p): New function.
18425 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
18426 add a cost if it is true.
18427 Update calls to aarch64_strip_extend.
18428 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
18430 2017-06-21 Andrew Pinski <apinski@cavium.com>
18432 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
18434 (thunderxt88): Likewise.
18435 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
18436 (thunderx_prefetch_tune): New variable.
18437 (thunderx2t99_prefetch_tune): Update for the correct values.
18438 (thunderxt88_tunings): New variable.
18439 (thunderx_tunings): Use thunderx_prefetch_tune instead of
18440 generic_prefetch_tune.
18441 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
18443 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18445 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
18446 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
18447 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
18448 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
18449 (aarch64_atomic_cas<mode>, GPI): Likewise.
18451 2017-06-21 Martin Liska <mliska@suse.cz>
18453 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
18454 statements on cold and hot labels.
18455 * predict.c (tree_estimate_probability_bb): Remove the
18456 prediction from this place.
18458 2017-06-21 Martin Liska <mliska@suse.cz>
18460 PR tree-optimization/79489
18461 * gimplify.c (maybe_add_early_return_predict_stmt): New
18463 (gimplify_return_expr): Call the function.
18464 * predict.c (tree_estimate_probability_bb): Remove handling
18466 * predict.def: Update comment about early return predictor.
18467 * gimple-predict.h (is_gimple_predict): New function.
18468 * predict.def: Change default value of early return to 66.
18469 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
18471 * passes.def: Put pass_strip_predict_hints to the beginning of
18474 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
18476 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
18477 FUNCTION_DECL declarations.
18478 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
18480 (dwaf2out_decl): Likewise.
18481 * godump.c (go_early_global_decl): Skip call to the real debug hook
18482 for FUNCTION_DECL declarations.
18483 * passes.c (rest_of_decl_compilation): Skip call to the
18484 early_global_decl debug hook for FUNCTION_DECL declarations, unless
18485 -fdump-go-spec is passed.
18487 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
18489 * config/i386/i386.c (struct builtin_isa): New field pure_p.
18490 Reorder for compactness.
18491 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
18492 (def_builtin_pure, def_builtin_pure2): New functions.
18493 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
18495 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
18497 * match.pd (nop_convert): New predicate.
18498 ((A +- CST1) +- CST2): Allow some NOP conversions.
18500 2017-06-21 Jakub Jelinek <jakub@redhat.com>
18503 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
18504 with ctors/dtors if GOVD_SHARED is set.
18506 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
18508 * config/aarch64/aarch64.md (movti_aarch64):
18509 Emit mov rather than orr.
18510 (movtf_aarch64): Likewise.
18511 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
18512 Emit mov rather than orr.
18514 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
18516 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
18517 Swap alternatives, make integer dup more expensive.
18519 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
18521 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
18522 Return true for non-tls symbols.
18524 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
18526 * config/aarch64/aarch64-cores.def (cortex-a55): New.
18527 (cortex-a75): Likewise.
18528 (cortex-a75.cortex-a55): Likewise.
18529 * config/aarch64/aarch64-tune.md: Regenerate.
18530 * doc/invoke.texi (-mtune): Document new values for -mtune.
18532 2017-06-21 Tom de Vries <tom@codesourcery.com>
18534 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
18535 stack_size feature.
18536 (Effective-Target Keywords, Other attributes): Suggest using
18537 dg-add-options stack_size feature to get stack limit in stack_size
18538 effective target documentation.
18540 2017-06-21 Julian Brown <julian@codesourcery.com>
18541 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18543 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
18544 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
18545 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
18547 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
18548 attribute type list for neon_multiply.
18549 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
18550 attribute type list for neon_multiply.
18551 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
18552 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
18553 attribute type list for neon_multiply.
18554 * config/arm/types.md (crypto_pmull): Add.
18555 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
18556 attribute type list.
18558 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
18560 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
18563 2017-06-20 Jakub Jelinek <jakub@redhat.com>
18565 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
18566 to make sure not to dereference a NULL cost_classes_ptr pointer.
18568 2017-06-20 Carl Love <cel@us.ibm.com>
18570 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18571 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
18572 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
18573 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
18574 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
18575 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
18576 VMULOSW): New enum "unspec" values.
18577 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
18578 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
18579 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
18580 altivec_vmulosw): New patterns.
18581 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
18582 VMULOSW): Add definitions.
18584 2017-06-20 Julia Koval <julia.koval@intel.com>
18586 * config/i386/i386.c: Fix rounding expand for new pattern.
18587 * config/i386/subst.md: Fix pattern (parallel -> unspec).
18589 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
18591 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
18592 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
18594 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
18596 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
18599 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
18601 * config/aarch64/aarch64-cores.def: Rearrange to sort by
18602 architecture, then by implementer ID.
18603 * config/aarch64/aarch64-tune.md: Regenerate.
18605 2017-06-20 Richard Biener <rguenther@suse.de>
18607 PR middle-end/81097
18608 * fold-const.c (split_tree): Fold to type before negating.
18610 2017-06-20 David Malcolm <dmalcolm@redhat.com>
18612 * diagnostic-show-locus.c
18613 (selftest::test_fixit_deletion_affecting_newline): New function.
18614 (selftest::diagnostic_show_locus_c_tests): Call it.
18616 2017-06-20 Andreas Schwab <schwab@suse.de>
18619 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
18622 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18624 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
18625 __ARM_FEATURE_COPROC according to support.
18627 2017-06-20 Jakub Jelinek <jakub@redhat.com>
18629 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
18630 Rewritten to avoid overflow for > 32-bit pointers.
18633 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
18634 by removing enum keyword.
18635 (ubsan_type_descriptor): Likewise. Formatting fix.
18638 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
18639 splitter): Require TARGET_SSE2 in the condition.
18641 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
18644 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
18645 for doing vector set of SFmode on ISA 3.0.
18646 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
18647 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
18649 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
18650 SFmode value into a V4SF variable that was extracted from another
18651 V4SF variable without converting the element to double precision
18652 and back to single precision vector format.
18653 (vsx_insert_extract_v4sf_p9_2): Likewise.
18655 2017-06-19 Jakub Jelinek <jakub@redhat.com>
18657 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
18658 in UWHI to avoid undefined overflow.
18661 * ubsan.h (enum ubsan_encode_value_phase): New.
18662 (ubsan_encode_value): Change second argument to
18663 enum ubsan_encode_value_phase with default value of
18664 UBSAN_ENCODE_VALUE_GENERIC.
18665 * ubsan.c (ubsan_encode_value): Change second argument to
18666 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
18667 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
18668 create_tmp_var_raw instead of create_tmp_var and use a
18670 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
18671 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
18672 ubsan_encode_value callers.
18675 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
18676 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
18677 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
18679 2017-06-19 Richard Biener <rguenther@suse.de>
18681 PR middle-end/81118
18682 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
18683 estimates if we changed anything.
18685 2017-06-19 Richard Biener <rguenther@suse.de>
18687 PR tree-optimization/80887
18688 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
18689 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
18690 simplified lookups, then reset mprts_hook.
18691 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
18693 (try_to_simplify): Likewise.
18695 2017-06-19 Martin Liska <mliska@suse.cz>
18698 * gimplify.c (gimplify_switch_expr):
18699 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
18701 2017-06-19 Martin Liska <mliska@suse.cz>
18703 * doc/install.texi: Document that PGO runs in 4 stages.
18705 2017-06-19 Martin Liska <mliska@suse.cz>
18708 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
18709 to dispatcher function name.
18710 * multiple_target.c (replace_function_decl): New function.
18711 (create_dispatcher_calls): Redirect both edges and references.
18713 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
18715 * profile-count.c (profile_count::dump): Dump quality.
18716 (profile_count::differs_from_p): Update for unsigned val.
18717 * profile-count.h (profile_count_quality): New enum.
18718 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
18720 2017-06-19 Richard Biener <rguenther@suse.de>
18722 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
18723 struct function as arg.
18724 (estimate_numbers_of_iterations): Export overload with loop arg.
18725 (free_numbers_of_iterations_estimates_loop): Use an overload of
18726 free_numbers_of_iterations_estimates instead.
18727 * tree-cfg.c (remove_bb): Adjust.
18728 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
18729 * tree-parloops.c (gen_parallel_loop): Likewise.
18730 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
18732 (tree_unroll_loops_completely): Likewise.
18733 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
18734 Use an overload instead and export.
18735 (estimated_loop_iterations): Adjust.
18736 (max_loop_iterations): Likewise.
18737 (likely_max_loop_iterations): Likewise.
18738 (estimate_numbers_of_iterations): Take struct function as arg
18740 (loop_exits_before_overflow): Adjust.
18741 (free_numbers_of_iterations_estimates_loop): Use an overload.
18742 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
18743 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
18745 2017-06-19 Richard Biener <rguenther@suse.de>
18748 * ipa-prop.c (find_constructor_constant_at_offset): Handle
18749 RANGE_EXPR conservatively.
18751 2017-06-16 Carl Love <cel@us.ibm.com>
18753 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18754 definitions for vec_float, vec_float2, vec_floato,
18755 vec_floate built-ins.
18756 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
18757 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
18759 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
18760 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
18761 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
18762 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
18763 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
18764 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
18765 vec_floato): Add builtin defines.
18766 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
18767 Update the built-in documentation file for the new built-in
18770 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18772 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
18773 (mthumb): Mark as the negative of -marm.
18775 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18777 * doc/invoke.texi (ARM Options, -mcpu): Document supported
18779 (ARM Options, -mtune): Document that this accepts the same
18780 extension options as -mcpu.
18781 (ARM Options, -mfpu): Document addition of -mfpu=auto.
18783 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18785 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
18786 permitted extensions.
18788 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18790 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
18791 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
18792 (armv8-m.main): Add option +nodsp.
18793 * config/arm/arm-cpu-cdata.h: Regenerated.
18795 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18797 * config/arm/t-fuchsia: New file.
18798 * config.gcc (arm*-*-fuchsia*): Use it.
18800 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18802 * config/arm/t-symbian: Rewrite for new option infrastructure.
18804 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18806 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
18807 (MULTILIB_REQUIRED): Likewise.
18809 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18811 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
18812 (MULTILIB_RESUE): Likewise.
18813 (MULTILIB_MATCHES): Likewise.
18814 (MULTLIB_REQUIRED): Likewise.
18816 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18818 * config/arm/t-rtems: Rewrite for new option framework.
18820 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18822 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
18823 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
18824 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
18825 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
18826 * config/arm/t-multilib: ... here.
18827 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
18828 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
18829 armv7-a and armv8*-a when A-profile libraries have not been built.
18830 * config/arm/t-rmprofile: Rewrite.
18832 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18834 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
18835 with a backslash. Remove the backslash after substituting unescaped
18837 * doc/fragments.texi (MULTILIB_REUSE): Document it.
18839 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18841 * config.gcc: (arm*-*-*): When building a-profile libraries, force
18842 the driver to pass through the default setting of -mfloat-abi.
18843 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
18845 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
18846 (all_feat_combs): New rule.
18847 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
18849 * config/arm/t-aprofile: Rewrite.
18851 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18853 * config/arm/arm.h (FPUTYPE_AUTO): Define.
18854 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
18855 fpu is not specified by the user/command-line.
18856 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
18857 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
18858 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
18859 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
18860 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
18861 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
18863 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18865 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
18866 * config/arm/t-arm-elf: Rewritten.
18868 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18870 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
18871 have some floating-point instructions.
18872 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
18873 (TARGET_MAYBE_HARD_FLOAT): New macro.
18874 * config/arm/arm-builtins.c (arm_init_builtins): Use
18875 TARGET_MAYBE_HARD_FLOAT.
18876 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
18878 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18880 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
18881 (configargs.h): Include it.
18882 (arm_print_hint_for_fpu_option): New function.
18883 (arm_parse_fpu_option): New function.
18884 (candidate_extension): New class.
18885 (arm_canon_for_multilib): New function.
18886 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
18887 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
18888 (ARCH_CANONICAL_SPECS): New macro.
18889 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
18891 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18893 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
18894 are set after handling multilib fragments. Set target_cpu_default2
18897 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18899 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
18901 (arm*-*-*): Set target_cpu_default2 to a quoted string.
18902 * config/arm/parsecpu.awk (check_cpu): Validate any extension
18904 (check_arch): Likewise.
18905 * config/arm/arm.c (arm_configure_build_target): Handle
18906 TARGET_CPU_DEFAULT being a string constant. Scan any feature
18907 options in the default.
18909 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18911 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
18912 when an option is an alias of another.
18913 * config/arm/parsecpu.awk (optalias): New parser token.
18914 (gen_comm_data): Mark non-alias options as such. Emit entries
18915 for extension aliases.
18916 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
18917 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
18918 (armv6kz, armv6zk, armv6t2): Likewise.
18919 (armv7): Make vfpv3-d16 an alias.
18920 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
18922 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
18923 Sort in canonical order.
18924 (armv8-a): Sort in canonical order.
18925 (armv8.1-a, armv8.2-a): Likewise.
18926 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
18928 (cortex-a9): Sort in canonical order.
18929 * config/arm/arm.c (selftests.h): Include it.
18930 (arm_test_cpu_arch_data): New function.
18931 (arm_run_self_tests): New function.
18932 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
18933 (targetm): Move declaration to the end of the file.
18934 * arm-cpu-cdata.h: Regenerated.
18936 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18938 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
18939 call to target_mode_check describing the type of option passed.
18940 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
18941 (arm_target_thumb_only): Use arm_parse_arch_option_name or
18942 arm_parse_cpu_option_name to match parameters against list of
18944 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
18945 arm_arch_core_flags data structure.
18946 * config/arm/arm-cpu_cdata.h: Regenerated.
18948 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18950 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
18952 (arm_print_hint_for_cpu_option): Likewise.
18953 (arm_print_hint_for_arch_option): Likewise.
18954 (arm_parse_cpu_option_name): Likewise.
18955 (arm_parse_arch_option_name): Likewise.
18956 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
18957 of entries in the all_fpus list.
18958 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
18959 (arm_parse_cpu_option_name): Declare.
18960 (arm_parse_arch_option_name): Declare.
18961 (arm_parse_option_features): Declare.
18962 (arm_intialize_isa): Declare.
18963 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
18965 (gen_comm_data): ... here. Make definitions non-static.
18966 * config/arm/arm-cpu-data.h: Regenerated.
18967 * config/arm/arm-cpu-cdata.h: Regenerated.
18969 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18971 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
18972 (cpu_arch_extension): New structure.
18973 (cpu_arch_option, arch_option, cpu_option): New structures.
18974 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
18975 architecture types.
18976 (gen_data): Generate new format data tables.
18977 * config/arm/arm.c (cpu_tune): New structure.
18978 (cpu_option, processors): Delete.
18979 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
18980 (arm_print_hint_for_cpu_option): ... this and ...
18981 (arm_print_hint_for_arch_option): ... this.
18982 (arm_parse_arch_cpu_name): Delete. Replace with ...
18983 (arm_parse_cpu_option_name): ... this and ...
18984 (arm_parse_arch_option_name): ... this.
18985 (arm_unrecognized_feature): Change type of target parameter to
18987 (arm_parse_arch_cpu_features): Delete. Replace with ...
18988 (arm_parse_option_features): ... this.
18989 (arm_configure_build_target): Rework to use new configuration data
18991 (arm_print_tune_info): Rework for new configuration data tables.
18992 * config/arm/arm-cpu-data.h: Regenerated.
18993 * config/arm/arm-cpu.h: Regenerated.
18995 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18997 * Makefile.in (OBJS): Move sbitmap.o from here ...
18998 (OBJS-libcommon): ... to here.
19000 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19002 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
19003 (ISA_ALL_CRYPTO): New macro.
19004 (ISA_ALL_SIMD): New macro
19005 (ISA_ALL_FP): New macro.
19006 * config/arm/arm.c (fpu_bitlist): Update initializer.
19007 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
19009 (arm9e): Add fpu. Add option for nofp
19010 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
19011 (arm926ej-s, arm1026ej-s): Likewise.
19012 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
19013 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
19014 neon-fp16, neon-vfpv4, nofp and nosimd.
19015 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
19016 (cortex-a8): Add fpu. Add option for nofp.
19017 (cortex-a9): Add fpu. Add options for nosimd and nofp.
19018 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
19019 (cortex-r4f): Add fpu.
19020 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
19021 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
19023 (cortex-r8): Likewise.
19024 (cortex-m4): Add fpu. Add option for nofp.
19025 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
19026 (cortex-a17.cortex-a7): Likewise.
19027 (cortex-a32): Add fpu. Add options for crypto and nofp.
19028 (cortex-a35, cortex-a53): Likewise.
19029 (cortex-a57): Add fpu. Add option for crypto.
19030 (cortex-a72, cortex-a73): Likewise.
19031 (exynos-m1): Likewise.
19032 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
19033 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
19034 (cortex-m33): Add fpu. Add option for nofp.
19035 * config/arm/arm-cpu-cdata.h: Regenerated
19036 * config/arm/arm-cpu-data.h: Regenerated.
19038 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19040 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
19041 (armv5te, armv5tej): Likewise.
19042 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
19043 (armv7): Add options fp and vfpv3-d16.
19044 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
19045 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
19047 (armv7ve): Likewise.
19048 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
19049 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
19050 (armv8-a): Add nocrypto option.
19051 (armv8.1-a, armv8.2-a): Likewise.
19052 (armv8-m.main): add options fp, fp.dp and nofp.
19054 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19056 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
19058 (armv8-a+crc): Delete.
19059 (armv8.1-a): Add options simd, crypto and nofp.
19060 (armv8.2-a): Add options fp16, simd, crypto and nofp.
19061 (armv8.2-a+fp16): Delete.
19062 (armv8-m.main): Add option dsp.
19063 (armv8-m.main+dsp): Delete.
19064 (cortex-a8): Add fpu. Add nofp option.
19065 (cortex-a9): Add fpu. Add nofp and nosimd options.
19066 * config/arm/parsecpu.awk (gen_data): Generate option tables and
19067 link to main cpu and architecture data structures.
19068 (gen_comm_data): Only put isa attributes from the main architecture
19070 (option): New statement for architecture and CPU entries.
19071 * arm.c (struct cpu_option): New structure.
19072 (struct processors): Add entry for options.
19073 (arm_unrecognized_feature): New function.
19074 (arm_parse_arch_cpu_name): Ignore any characters after the first
19076 (arm_parse_arch_cpu_feature): New function.
19077 (arm_configure_build_target): Separate out any CPU and architecture
19078 features and parse separately. Don't error out if -mfpu=auto is
19079 used with only an architecture string.
19080 (arm_print_asm_arch_directives): New function.
19081 (arm_file_start): Call it.
19082 * config/arm/arm-cpu-cdata.h: Regenerated.
19083 * config/arm/arm-cpu-data.h: Likewise.
19084 * config/arm/arm-tables.opt: Likewise.
19086 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19088 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
19089 assembler when it is not -mfpu=auto.
19091 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19093 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
19094 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
19095 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
19096 (ASM_CPU_SPEC): Rewrite.
19097 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
19098 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
19099 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
19100 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
19101 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
19102 copied string is NUL-terminated. Also strip any characters prefixed
19104 (arm_rewrite_selected_arch): New function.
19105 (arm_rewrite_march): New function.
19107 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19109 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
19110 (x_arm_cpu_string, x_arm_tune_string): Likewise.
19111 (march, mcpu, mtune): Convert to string-based options.
19112 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
19113 (arm_parse_arch_cpu_name): New function.
19114 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
19115 identify selected architecture or CPU.
19116 (arm_option_save): New function.
19117 (TARGET_OPTION_SAVE): Redefine.
19118 (arm_option_restore): Restore string options.
19119 (arm_option_print): Print string options.
19121 2017-06-16 Martin Sebor <msebor@redhat.com>
19123 PR tree-optimization/80933
19124 PR tree-optimization/80934
19125 * builtins.c (fold_builtin_3): Do not handle bcmp here.
19126 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
19127 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
19128 (gimple_fold_builtin): Call them.
19130 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19132 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
19133 as unlikely; update profile.
19135 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19137 * predict.c (force_edge_cold): Handle declaring edges impossible
19140 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19142 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
19144 (try_unroll_loop_completely): Fix reporting.
19146 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19148 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
19150 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
19153 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
19154 if given a non-constant argument for an intrinsic which requires a
19157 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19159 * profile.c (compare_freqs): New function.
19160 (branch_prob): Sort edge list.
19161 (find_spanning_tree): Assume that the list is priority sorted.
19163 2017-06-16 Richard Biener <rguenther@suse.de>
19165 PR tree-optimization/81090
19166 * passes.def (pass_record_bounds): Remove.
19167 * tree-pass.h (make_pass_record_bounds): Likewise.
19168 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
19169 make_pass_record_bounds): Likewise.
19170 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
19171 not free niter estimates at the beginning but at the end.
19172 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
19174 2017-06-16 Richard Biener <rguenther@suse.de>
19176 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
19177 initializer to workaround ICE in host GCC 4.8.
19179 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19181 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
19183 (clone_inlined_nodes): Update.
19185 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19187 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
19188 prefetch settings, and enable prefetching by default at -O3.
19190 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19192 * config/aarch64/aarch64.c (aarch64_override_options_internal):
19193 Set flag_prefetch_loop_arrays according to tuning data.
19195 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19197 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
19198 New tune structure.
19199 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
19200 [Unrelated to main purpose of the patch] Place the pointer field last
19201 to enable type checking errors when tune structure are wrongly merged.
19202 * config/aarch64/aarch64.c (generic_prefetch_tune,)
19203 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
19204 (thunderx2t99_prefetch_tune): New tune constants.
19205 (tune_params *_tunings): Update all tunings (no functional change).
19206 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
19207 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
19208 from tunings structures.
19210 2017-06-16 Jakub Jelinek <jakub@redhat.com>
19213 * ubsan.c (instrument_null): Add T argument, use it instead
19214 of computing it based on IS_LHS.
19215 (instrument_object_size): Likewise.
19216 (pass_ubsan::execute): Adjust instrument_null and
19217 instrument_object_size callers to pass gimple_get_lhs or
19218 gimple_assign_rhs1 result to it. Use instrument_null instead of
19219 calling get_base_address and instrument_mem_ref. Handle
19220 aggregate call arguments for object-size sanitization.
19222 2017-06-16 Yury Gribov <tetra2005@gmail.com>
19224 PR tree-optimization/81089
19225 * tree-vrp.c (is_masked_range_test): Validate operands of
19228 2017-06-15 Martin Sebor <msebor@redhat.com>
19231 * dumpfile.c (dump_register): Avoid calling memset to initialize
19232 a class with a default ctor.
19233 * gcc.c (struct compiler): Remove const qualification.
19234 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
19235 * hash-table.h: Ditto.
19236 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
19238 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
19239 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
19241 * params.h (struct param_info): Make struct members non-const.
19242 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
19243 with default initialization.
19244 * vec.h (vec_copy_construct, vec_default_construct): New helper
19246 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
19247 with vec_copy_construct.
19248 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
19249 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
19250 * doc/invoke.texi (-Wclass-memaccess): Document.
19252 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19254 * emit-rtl.h (is_leaf): Update comment about local
19255 register allocator.
19257 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19260 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
19261 for a variable to have a section before checking if the section has a
19263 Set section to.persistent if persistent attribute is set.
19264 Warn if .persistent attribute is used on an automatic variable.
19266 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
19268 PR rtl-optimization/80474
19269 * reorg.c (update_block): Do not ignore instructions in a delay slot.
19271 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
19273 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
19276 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
19278 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
19279 (casesi): Emit bounds checking as RTL.
19280 (casesi_internal_mips16_<mode>): Remove bounds checking.
19282 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
19284 * config/xtensa/xtensa.c (xtensa_option_override): Append
19285 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
19286 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
19287 xtensa_doloop_hooks): Define unconditionally.
19288 (xtensa_reorg_loops): Only call reorg_loops in the presence of
19290 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
19291 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
19292 for it in xtensa_option_override.
19293 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
19294 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
19296 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
19298 * doc/cppopts.texi: Document '-' special value to -MF.
19300 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
19302 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
19303 (cortex_a53_fconst): Likewise.
19304 (cortex_a53_fpmul): Likewise.
19305 (cortex_a53_f_load_64): Likewise.
19306 (cortex_a53_f_load_many): Likewise.
19307 (cortex_a53_advsimd_alu): Likewise.
19308 (cortex_a53_advsimd_alu_q): Likewise.
19309 (cortex_a53_advsimd_mul): Likewise.
19310 (cortex_a53_advsimd_mul_q): Likewise.
19311 (fpmac bypass): Add new bypass for fpmac-fpmac case.
19312 Add missing fmul, r2f_cvt and fconst cases.
19314 2017-06-14 Richard Biener <rguenther@suse.de>
19316 PR middle-end/81088
19317 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
19319 (fold_binary_loc): When associating do not treat pre-existing
19320 TREE_OVERFLOW on literal constants as a reason to allow
19321 TREE_OVERFLOW on associated literal constants.
19323 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19325 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
19326 (MASK_FEATURES): New macro.
19327 * config/sparc/sparc.c (sparc_option_override): Remove the special
19328 handling of -mfpu and generalize it to all MASK_FEATURES switches.
19330 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19332 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
19333 a division of 0 if non-call exceptions are enabled.
19335 2017-06-14 Andrew Pinski <apinski@cavium.com>
19336 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19339 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
19340 Improve vector initialization code gen for only variable case.
19342 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19344 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
19346 2017-06-14 Richard Biener <rguenther@suse.de>
19348 PR tree-optimization/81083
19349 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
19352 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19354 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
19355 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
19356 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
19357 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
19358 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
19359 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
19361 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19363 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
19364 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
19366 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19368 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
19370 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19372 * config/rs6000/t-rtems: Don't handle SPE.
19374 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19376 * config/rs6000/t-linux: Don't handle SPE.
19378 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19380 * config/rs6000/eabispe.h: Delete file.
19382 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19384 * config/rs6000/t-spe: Delete file.
19386 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19388 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
19389 (rs6000_legitimate_offset_address_p): Return false for anything in
19390 V2SImode or V2SFmode.
19392 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19394 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
19395 except V2SF and V2SI. Rearrange the vector modes, and add comments.
19396 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
19398 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
19399 (rs6000_legitimate_offset_address_p): Ditto.
19400 (rs6000_emit_move): Ditto.
19401 (rs6000_init_builtins): Remove V4HI_type_node.
19403 2017-06-13 Martin Liska <mliska@suse.cz>
19406 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
19407 (gate_asan): Likewise.
19408 * asan.h (asan_no_sanitize_address_p): Remove the function.
19409 (sanitize_flags_p): New function.
19410 * builtins.def: Fix coding style.
19411 * common.opt: Use renamed enum value.
19412 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
19413 * doc/extend.texi: Document no_sanitize attribute.
19414 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
19415 to SANITIZE_UNDEFINED_NONDEFAULT.
19416 * gcc.c (sanitize_spec_function): Use the renamed enum value.
19417 * gimple-fold.c (optimize_atomic_compare_exchange_p):
19418 Use sanitize_flags_p.
19419 * gimplify.c (gimplify_function_tree): Likewise.
19420 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
19421 * opts.c (parse_no_sanitize_attribute): New function.
19422 (common_handle_option): Use renamed enum value.
19423 * opts.h (parse_no_sanitize_attribute): Declare.
19424 * tree.c (sanitize_flags_p): New function.
19425 * tree.h: Declared here.
19426 * tsan.c: Use sanitize_flags_p.
19427 * ubsan.c (ubsan_expand_null_ifn): Likewise.
19428 (instrument_mem_ref): Likewise.
19429 (instrument_bool_enum_load): Likewise.
19430 (do_ubsan_in_current_function): Remove the function.
19431 (pass_ubsan::execute): Use sanitize_flags_p.
19432 * ubsan.h: Remove do_ubsan_in_current_function
19433 * tree-cfg.c (print_no_sanitize_attr_value): New function.
19434 (dump_function_to_file): Use it here.
19436 2017-06-13 Martin Jambor <mjambor@suse.cz>
19438 PR tree-optimization/80803
19439 PR tree-optimization/81063
19440 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
19441 (propagate_subaccesses_across_link): Enqueue subtree whenever
19442 necessary instead of relying on the caller.
19444 2017-06-13 Martin Jambor <mjambor@suse.cz>
19446 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
19447 that have a first_link.
19448 (sort_and_splice_var_accesses): Do not check first_link before
19450 (subtree_mark_written_and_enqueue): Likewise.
19451 (propagate_all_subaccesses): Likewise and do not stop at first
19452 parent with a first_link.
19454 2017-06-13 Martin Jambor <mjambor@suse.cz>
19456 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
19459 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19461 * match.pd: New pattern.
19463 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19465 * tree-vrp.c (is_masked_range_test): New function.
19466 (register_edge_assert_for): Determine ranges for
19469 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19471 PR tree-optimization/67328
19472 * fold-const.c (maskable_range_p): New function.
19473 (build_range_check): Generate bittests if possible.
19475 2017-06-13 Martin Liska <mliska@suse.cz>
19477 * gimple-pretty-print.c (dump_probability): Add new argument.
19478 (dump_edge_probability): Dump both probability and count.
19479 (dump_gimple_label): Likewise.
19480 (dump_gimple_bb_header): Likewise.
19482 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
19485 * config/avr/avr-devices.c: Fix indentation.
19486 * config/avr/gen-avr-mmcu-specs.c: Dito.
19488 2017-06-13 Richard Biener <rguenther@suse.de>
19490 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
19491 instead get vector type from stmt_info.
19492 (vectorizable_reduction): Adjust. Remove dead code.
19494 2017-06-13 Richard Biener <rguenther@suse.de>
19496 PR middle-end/81065
19497 * fold-const.c (extract_muldiv_1): Remove bogus distribution
19498 case of C * (x * C2 + C3).
19499 (fold_addr_of_array_ref_difference): Properly fold index difference.
19501 2017-06-12 David S. Miller <davem@davemloft.net>
19504 * config/sparc/sparc.md (return expander): Emit frame blockage if
19505 function uses alloca.
19507 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
19509 * combine.c (make_field_assignment): Check len rather than the mode
19510 precision when calling force_to_mode.
19512 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
19514 Support multilibs and devices that see flash in RAM address range.
19517 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
19518 (avr_mcu_t) <flash_pm_offset>: New field.
19519 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
19520 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
19521 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
19522 (AVR_TINY_PM_OFFSET): Remove macro.
19523 * config/avr/avr.opt (-mshort-calls): New option.
19524 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
19525 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
19526 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
19527 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
19528 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
19529 instead of avr_arch->have_jmp_call.
19530 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
19531 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
19532 avr_arch->flash_pm_offset to define.
19533 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
19534 new field flash_pm_offset. Add entry for avrxmega3.
19535 (avr_texinfo): Add entry for avrxmega3.
19536 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
19537 attiny212, attiny214,
19538 attiny412, attiny414, attiny416, attiny417,
19539 attiny814, attiny816, attiny817,
19540 attiny1614, attiny1616, attiny1617,
19541 attiny3214, attiny3216, attiny3217.
19542 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
19543 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
19544 (avr_print_operand_address) [AVR_TINY]: Same.
19545 (avr_asm_init_sections) <readonly_data_section>: Only patch
19546 callback if avr_arch->flash_pm_offset = 0.
19547 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
19548 for rodata if avr_arch->flash_pm_offset != 0.
19549 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
19550 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
19551 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
19552 (m_options): Append opt_rcall.
19553 (m_dirnames): Append dir_rcall.
19554 * config/avr/t-multilib: Regenerate.
19556 * configure.ac [target=avr]: Check whether avrxmega3 default
19557 linker description file works as needed.
19558 * configure: Regenerate.
19559 * doc/avr-mmcu.texi: Regenerate.
19560 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
19561 <__AVR_ARCH__>: Document avrxmega3 and 103.
19562 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
19563 <__AVR_SHORT_CALLS__>: Document it.
19564 <__AVR_PM_BASE_ADDRESS__>: Document it.
19565 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
19566 (AVR Variable Attributes) <progmem>: Document this is
19567 not needed for avrxmega3.
19568 (AVR Named Address Spaces) <__flash>: Dito.
19570 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
19572 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
19574 2017-06-12 Doug Rupp <rupp@adacore.com>
19576 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
19577 Append vxworks-stdint.h to the tm_file list.
19578 * config/vxworks-stdint.h: New file.
19580 2017-06-12 Martin Liska <mliska@suse.cz>
19582 PR tree-optimization/81041
19583 * tree-profile.c (gimple_gen_ic_func_profiler):
19584 Create an extra BB in profile-generate
19585 (gimple_gen_time_profiler): Likewise.
19587 2017-06-12 Jakub Jelinek <jakub@redhat.com>
19589 PR tree-optimization/81003
19590 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
19591 (update_range_test): Use it instead of force_gimple_operand_gsi.
19593 2017-06-12 Richard Biener <rguenther@suse.de>
19595 PR tree-optimization/81053
19596 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
19597 with backedge value not defined in loop. Simplify def stmt
19600 2017-06-11 Tom de Vries <tom@codesourcery.com>
19603 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
19605 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
19606 nvptx_cannot_force_const_mem.
19608 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19610 * opts.c (finish_options): Move test for flag_split_stack after
19611 it has been initialized.
19613 2017-06-11 Jason Merrill <jason@redhat.com>
19615 * tree.h (id_equal): New.
19616 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
19617 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
19618 instead of strcmp of IDENTIFIER_POINTER.
19620 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19622 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
19623 (mark_all_inlined_calls_cdtor): Fix formating.
19624 (inline_transform): Rescale profile before inlining.
19626 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19628 * cgraph.h (cgraph_edge::clone): Update prototype.
19629 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
19630 (cgraph_node::create_clone): Update.
19631 (cgraph_node::create_version_clone): Update.
19632 * tree-inline.c (copy_bb): Update.
19633 (expand_call_inline): Update.
19635 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
19637 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
19638 factored out from ...
19639 (rs6000_emit_prologue): ... here.
19641 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
19643 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
19644 factored out from ...
19645 (rs6000_emit_prologue): ... here.
19647 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19649 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
19651 (handle_missing_profiles): Fix computation of tp_first_run.
19652 (counts_to_freqs): Do not touch freqs when count is 0.
19654 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19656 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
19659 2017-06-10 Tom de Vries <tom@codesourcery.com>
19661 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
19662 attributes): Document signal effective target.
19664 2017-06-10 Tom de Vries <tom@codesourcery.com>
19666 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
19667 Document effective target stack_size.
19669 2017-06-09 David Malcolm <dmalcolm@redhat.com>
19671 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
19672 to the edit_context if they can be auto-applied.
19674 2017-06-9 Ian Lance Taylor <iant@golang.org>
19676 * opts.c (finish_options): If -fsplit-stack, disable implicit
19677 -forder-blocks-and-partition.
19678 * doc/invoke.texi (Optimize Options): Document that when using
19679 -fsplit-stack -forder-blocks-and-partition is not implicitly
19682 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
19684 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
19685 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
19686 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
19687 * builtins.def (abort, trap, unreachable): Declare cold.
19688 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
19689 * tree-core.h (ECF_COLD): New.
19690 * tree.c (set_call_expr_flags): Handle ECF_COLD.
19691 (build_common_builtin_nodes): Mark unreachable and abort as cold.
19693 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
19695 * predict.c (unlikely_executed_stmt_p): Cleanup.
19697 2017-06-09 Richard Biener <rguenther@suse.de>
19699 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
19700 model if the ref is always written to.
19702 2017-06-09 Tamar Christina <tamar.christina@arm.com>
19704 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
19706 2017-06-09 Tamar Christina <tamar.christina@arm.com>
19708 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
19711 2017-06-09 Tom de Vries <tom@codesourcery.com>
19714 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
19715 "target cannot support label values" when encountering LABEL_REF.
19717 2017-06-09 Martin Liska <mliska@suse.cz>
19719 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
19720 (gimple_gen_ic_func_profiler): Emit direct comparison
19721 of __gcov_indirect_call_callee with NULL.
19722 (gimple_gen_time_profiler): Change probability from
19723 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
19725 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
19727 * profile.c (edge_gcov_counts): Turn to pointer.
19728 (compute_branch_probabilities, compute_branch_probabilities): Update.
19729 (branch_prob): Do not clear edge_gcov_count.
19730 * profile.h (edge_gcov_counts): Turn to pointer.
19731 (edge_gcov_count): Update.
19733 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
19735 * gimple.h (gimple_check_failed): Mark cold.
19737 2017-06-09 Richard Biener <rguenther@suse.de>
19739 PR tree-optimization/66623
19740 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
19741 refactor check_reduction into two parts, properly computing
19742 whether we have to check reduction validity for outer loop
19745 2017-06-09 Richard Biener <rguenther@suse.de>
19747 PR tree-optimization/79483
19748 * graphite-scop-detection.c (order): New global.
19749 (get_order): Compute bb to order mapping that satisfies code
19750 generation constraints.
19751 (cmp_pbbs): New helper.
19752 (build_scops): Start domwalk at entry block, sort generated
19755 2017-06-09 Richard Biener <rguenther@suse.de>
19757 PR middle-end/81007
19758 * ipa-polymorphic-call.c
19759 (ipa_polymorphic_call_context::restrict_to_inner_class):
19760 Skip FIELD_DECLs with error_mark_node type.
19761 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
19764 2017-06-09 Martin Liska <mliska@suse.cz>
19766 * predict.c (struct branch_predictor): New struct.
19767 (test_prediction_value_range): New test.
19768 (predict_c_tests): New function.
19769 * selftest-run-tests.c (selftest::run_tests): Run the function.
19770 * selftest.h: Declare new tests.
19772 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
19775 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
19776 gen_add3_insn did not fail.
19777 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
19778 r0, construct that number in a temporary reg and add that reg to r0.
19779 If asked to put the result in r0 as well, fail.
19781 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
19783 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
19784 for early expansion of vec_eqv.
19786 2017-06-08 Jakub Jelinek <jakub@redhat.com>
19788 PR middle-end/81005
19789 * ubsan.c (instrument_null): Avoid pointless code temporary.
19790 (pass_ubsan::execute): Instrument aggregate arguments of calls.
19792 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
19796 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
19799 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
19802 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19804 * predict.c (unlikely_executed_edge_p): Move ahead.
19805 (probably_never_executed_edge_p): Use it.
19807 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
19809 PR middle-end/79988
19810 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
19811 gimple_call_builtin_p call.
19813 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19815 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
19816 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
19817 rtl_check_failed_type2, rtl_check_failed_code1,
19818 rtl_check_failed_code2, rtl_check_failed_code_mode,
19819 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
19820 rtvec_check_failed_bounds, rtl_check_failed_flag,
19821 _fatal_insn_not_found, _fatal_insn): Likewise.
19822 * tree.h (tree_contains_struct_check_failed,
19823 tree_check_failed, tree_not_check_failed,
19824 tree_class_check_failed, tree_range_check_failed,
19825 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
19826 tree_vec_elt_check_failed, phi_node_elt_check_failed,
19827 tree_operand_check_failed, omp_clause_check_failed,
19828 omp_clause_operand_check_failed, omp_clause_range_check_failed):
19831 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19833 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
19834 flag_branch_probabilities.
19835 * ipa-inline.c (edge_badness): Likewise.
19836 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
19837 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
19838 * predict.c (maybe_hot_frequency_p): Likewise.
19839 (probably_never_executed): Likewise.
19840 * sched-ebb.c (schedule_ebbs): Likewise.
19841 * sched-rgn.c (find_single_block_region): Likewise.
19842 * tracer.c (tail_duplicate): Likewise.
19844 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19846 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
19847 longer requires x_flag_profile_use.
19849 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19851 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
19852 instead of flag_reorder_blocks_and_partition.
19853 * dbxout.c (dbxout_function_end): Likewise.
19854 * dwarf2out.c (gen_subprogram_die): Likewise.
19855 * haifa-sched.c (sched_create_recovery_edges): Likewise.
19856 * hw-doloop.c (reorg_loops): Likewise.
19857 * varasm.c (assemble_start_function,
19858 assemble_end_function): Likewise.
19859 (decide_function_section): Do not check for
19860 flag_reorder_blocks_and_partition.
19862 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
19864 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
19866 (chkp_get_hard_register_fake_addr_expr): Ditto.
19867 (chkp_build_addr_expr): Add check for hard reg case.
19868 (chkp_parse_array_and_component_ref): Ditto.
19869 (chkp_find_bounds_1): Ditto.
19870 (chkp_process_stmt): Don't generate bounds store for
19873 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19875 * predict.c (maybe_hot_bb_p): Do not check profile status.
19876 (maybe_hot_edge_p): Likewise.
19877 (probably_never_executed): Check for zero counts even if profile
19879 (unlikely_executed_edge_p): New function.
19880 (unlikely_executed_stmt_p): New function.
19881 (unlikely_executed_bb_p): New function.
19882 (set_even_probabilities): Use unlikely predicates.
19883 (combine_predictions_for_bb): Likewise.
19884 (predict_paths_for_bb): Likewise.
19885 (predict_paths_leading_to_edge): Likewise.
19886 (determine_unlikely_bbs): New function.
19887 (estimate_bb_frequencies): Use it.
19888 (compute_function_frequency): Use zero counts even if profile is
19890 * profile-count.h: Fix typo.
19892 2017-08-08 Julia Koval <julia.koval@intel.com>
19894 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
19895 _mm512_mask_cvtsepi16_storeu_epi8,
19896 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
19897 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
19898 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
19899 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
19900 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
19901 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
19902 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
19903 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
19904 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
19905 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
19906 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
19907 __builtin_ia32_pmovuswb256mem_mask,
19908 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
19909 __builtin_ia32_pmovwb512mem_mask): New builtins.
19911 2017-08-08 Julia Koval <julia.koval@intel.com>
19913 PR target/73350,80862
19914 * config/i386/subst.md (round): Fix round pattern.
19915 * config/i386/i386.c (ix86_erase_embedded_rounding):
19916 Fix erasing rounding for the fixed pattern.
19918 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19920 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
19922 2017-06-08 Martin Liska <mliska@suse.cz>
19924 PR gcov-profile/80911
19925 * gcov.c (block_info::block_info): New constructor.
19927 2017-06-07 Carl Love <cel@us.ibm.com>
19929 * config/rs6000/rs6000-c: The return type of the following
19930 built-in functions was implemented as int not long long. Fix sign
19931 of return value for the unsigned version of vec_mulo and vec_mule.
19932 vector unsigned long long vec_bperm (vector unsigned long long,
19933 vector unsigned char)
19934 vector signed long long vec_mule (vector signed int,
19936 vector unsigned long long vec_mule (vector unsigned int,
19937 vector unsigned int)
19938 vector signed long long vec_mulo (vector signed int,
19940 vector unsigned long long vec_mulo (vector unsigned int,
19941 vector unsigned int)
19942 * doc/extend.texi: Fix the documentation for the built-in
19945 2017-06-07 Carl Love <cel@us.ibm.com>
19948 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
19951 2017-06-07 Carl Love <cel@us.ibm.com>
19953 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
19954 support, Generate doublehv for signed int/float for BE case only.
19956 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
19958 * doc/invoke.texi (mcx16): Rewrite.
19960 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19962 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
19963 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
19964 *mov<mode>_softfloat, and an anonymous splitter): Use
19965 nonimmediate_operand instead of rs6000_nonimmediate_operand.
19967 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19969 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
19971 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
19972 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
19973 (rs6000_debug_reg_global): Adjust.
19974 (rs6000_init_hard_regno_mode_ok): Adjust.
19975 (rs6000_dbx_register_number): Adjust.
19976 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
19977 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
19978 Remove SPE_ACC and SPEFSCR.
19979 (REG_ALLOC_ORDER): Ditto.
19980 (FRAME_POINTER_REGNUM): Change to 111.
19981 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
19982 (REG_CLASS_NAMES): Ditto.
19983 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
19984 (REGISTER_NAMES): Ditto.
19985 (ADDITIONAL_REG_NAMES): Ditto.
19986 (rs6000_reg_names): Ditto.
19987 * config/rs6000/rs6000.md: Renumber some register number
19990 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19992 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
19994 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
19995 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
19997 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
19998 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
19999 Delete the SPE high registers.
20000 (REG_ALLOC_ORDER): Ditto.
20001 (enum reg_class): Remove SPE_HIGH_REGS.
20002 (REG_CLASS_NAMES): Ditto.
20003 (REG_CLASS_CONTENTS): Delete the SPE high registers.
20004 (REGISTER_NAMES): Ditto.
20005 (rs6000_reg_names): Ditto.
20006 * doc/tm.texi.in: Remove SPE as example.
20007 * doc/tm.texi: Regenerate.
20009 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20011 * config/rs6000/8540.md (ppc8540_brinc): Delete.
20012 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
20013 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
20014 * config/rs6000/rs6000.md (type): Remove "brinc".
20016 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20018 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
20019 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
20020 * config/rs6000/linuxspe.h: Delete file.
20021 * config/rs6000/rs6000.md: Don't include spe.md.
20022 * config/rs6000/spe.h: Delete file.
20023 * config/rs6000/spe.md: Delete file.
20024 * config/rs6000/t-rs6000: Remove spe.md.
20026 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20028 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
20029 (reg_or_none500mem_operand): Delete.
20030 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
20031 instead of reg_or_none500mem_operand.
20033 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20035 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
20036 handling of SPE flags.
20037 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
20039 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20041 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
20043 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
20044 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
20045 paired_divv2sf3): Similar.
20046 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
20047 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
20048 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
20050 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
20051 Rename the paired_* instruction patterns.
20052 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
20054 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
20055 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
20056 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
20057 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
20058 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
20059 PAIRED_VECTOR_MODE.
20060 (struct machine_function): Delete field spe_insn_chain_scanned_p.
20061 (spe_func_has_64bit_regs_p): Delete.
20062 (spe_expand_predicate_builtin): Delete.
20063 (spe_expand_evsel_builtin): Delete.
20064 (TARGET_DWARF_REGISTER_SPAN): Do not define.
20065 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
20066 (invalid_e500_subreg): Delete.
20067 (rs6000_legitimize_address): Always force_reg op2 as well, for
20068 paired single memory accesses.
20069 (rs6000_member_type_forces_blk): Delete.
20070 (rs6000_spe_function_arg): Delete.
20071 (rs6000_expand_unop_builtin): Delete SPE handling.
20072 (rs6000_expand_binop_builtin): Ditto.
20073 (spe_expand_stv_builtin): Delete.
20074 (bdesc_2arg_spe): Delete.
20075 (spe_expand_builtin): Delete.
20076 (spe_expand_predicate_builtin): Delete.
20077 (spe_expand_evsel_builtin): Delete.
20078 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
20079 (spe_init_builtins): Delete.
20080 (spe_func_has_64bit_regs_p): Delete.
20081 (savres_routine_name): Delete "info" parameter. Adjust callers.
20082 (rs6000_emit_stack_reset): Ditto.
20083 (rs6000_dwarf_register_span): Delete.
20084 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
20085 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
20086 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
20088 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
20090 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
20091 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
20092 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
20093 mulv2sf3, divv2sf3): Delete expanders.
20095 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20097 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
20099 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20101 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
20102 * config/rs6000/rs6000.c: Ditto.
20104 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20106 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
20107 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
20108 comparison_operator.
20110 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20112 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
20113 * config/rs6000/rs6000.opt: Ditto.
20114 * config/rs6000/t-rtems: Ditto.
20116 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20118 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
20119 TARGET_E500_SINGLE by 0, simplify.
20120 * config/rs6000/rs6000.c: Ditto.
20121 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
20122 (spe_build_register_parallel): Delete.
20123 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
20124 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
20125 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
20126 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
20127 (E500_CONVERT): Delete.
20128 * config/rs6000/spe.md: Remove many patterns and all define_constants.
20130 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20132 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
20133 * config/rs6000/dfp.md: Ditto.
20134 (negdd2, *negdd2_fpr): Merge.
20135 (absdd2, *absdd2_fpr): Merge.
20136 (negtd2, *negtd2_fpr): Merge.
20137 (abstd2, *abstd2_fpr): Merge.
20138 * config/rs6000/e500.h: Delete file.
20139 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
20140 TARGET_FPRS by 1 and simplify.
20141 * config/rs6000/rs6000-c.c: Ditto.
20142 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
20143 TARGET_DF_SPE by 0.
20144 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
20146 * config/rs6000/rs6000.md: Ditto.
20147 (floatdidf2, *floatdidf2_fpr): Merge.
20148 (move_from_CR_gt_bit): Delete.
20149 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
20150 (E500_CR_IOR_COMPARE): Delete.
20151 (All patterns that require !TARGET_FPRS): Delete.
20152 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
20154 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20156 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
20158 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20160 * graphds.c (add_edge): Intitialize edge's attached data.
20161 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
20162 pointer parameter. Call pointed function on each edge during
20163 graph traversing. Skip traversing the edge when the function
20165 (graphds_dfs, graphds_scc): Ditto.
20166 (for_each_edge): New parameter. Pass the new parameter to callback
20168 * graphds.h (skip_edge_callback): New function pointer type.
20169 (graphds_dfs, graphds_scc): New function pointer parameter.
20170 (graphds_edge_callback, for_each_edge): New parameter.
20172 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20174 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
20175 out code checking if runtime alias check is possible to below ...
20176 Call the new function.
20177 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
20178 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
20180 2017-06-07 Marek Polacek <polacek@redhat.com>
20183 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
20184 TYPE_OVERFLOW_WRAPS check.
20186 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20188 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
20189 if versioning is required.
20190 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
20191 peeling with the check for versioning.
20193 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20195 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
20196 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
20197 Set true to new parameter if new ssa variable is defined.
20198 (vect_gen_vector_loop_niters): Refactor. Set range information
20199 for the new vector loop bound variable.
20200 (vect_do_peeling): Ditto.
20202 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20204 * tree-affine.c (ssa.h): Include header file.
20205 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
20206 has wrapping overflow behavior.
20208 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20210 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
20212 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20214 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
20215 (tree_to_aff_combination): ... here.
20217 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20219 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
20220 reg_pressure model function.
20221 (ivopts_global_cost_for_size): Delete.
20222 (determine_set_costs, iv_ca_recount_cost): Call new model function
20223 ivopts_estimate_reg_pressure.
20225 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20227 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
20228 expensive than udiv. Remove floating point cases from mod.
20230 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20232 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
20233 Increase idiv cost.
20235 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20237 * config/aarch64/aarch64.md
20238 (copysignsf3): Fix mask generation.
20240 2017-06-07 Jakub Jelinek <jakub@redhat.com>
20242 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
20244 (class dump_manager): Add register_dumps method.
20245 * dumpfile.c: Include langhooks.h.
20246 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
20247 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
20248 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
20249 (dump_manager::dump_register): Start with 512 entries instead of 32.
20250 (dump_manager::register_dumps): New method.
20251 * toplev.c (general_init): Instead of invoking register_dumps
20252 langhook, invoke register_dumps method on the dump manager.
20253 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
20256 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
20258 * doc/md.texi: Clarify the restrictions on a define_insn condition.
20259 Say that # requires an associated define_split to exist, and that
20260 the define_split must be suitable for use after register allocation.
20262 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20264 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
20265 (compute_outgoing_frequencies): Also initialize zero counts.
20266 (find_many_sub_basic_blocks): Do not produce uninitialized profile
20267 around loops; preserve more of profile when nothing changes.
20269 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
20271 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
20273 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
20274 * config/arm/arm-cpu-cdata.h: Regenerate.
20275 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
20276 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
20277 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
20278 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
20279 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
20281 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
20283 * config/arm/t-rmprofile: Likewise.
20284 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
20286 2017-06-06 David S. Miller <davem@davemloft.net>
20289 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
20290 blockage if function uses alloca.
20292 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20294 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
20295 New "uid" fields to hold pretty-print IDs of group and ref.
20296 Memory references are now identified as <group_id>:<ref_id>
20297 instead of using [random] addresses.
20298 (dump_mem_details): Simplify, no functional change.
20299 (dump_mem_ref): Simplify and make output more concise.
20300 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
20301 (find_or_create_group): Initialize group uid.
20302 (record_ref): Initialize ref uid. Improve debug output.
20303 (prune_group_by_reuse, should_issue_prefetch_p,)
20304 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
20305 (mark_nontemporal_store, determine_loop_nest_reuse):
20306 Improve debug output.
20308 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20310 * dbgcnt.def (prefetch): New debug counter.
20311 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
20312 (schedule_prefetches): Stop issueing prefetches if debug counter
20315 2017-06-06 Tom de Vries <tom@codesourcery.com>
20317 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
20318 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
20320 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20322 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
20323 Use aarch64_reg_or_zero predicate for operand 4.
20324 (aarch64_compare_and_swap<mode> define_insn_and_split):
20325 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
20326 (aarch64_store_exclusive<mode>): Likewise for operand 2.
20328 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
20330 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
20331 (arm_compute_save_core_reg_mask): This.
20332 (thumb1_compute_save_reg_mask): Rename into ...
20333 (thumb1_compute_save_core_reg_mask): This.
20334 (arm_compute_save_reg0_reg12_mask): Adapt comment.
20335 (arm_compute_frame_layout): Likewise.
20337 2017-06-06 Richard Biener <rguenther@suse.de>
20339 PR tree-optimization/80974
20340 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
20341 keep or clear leaders SSA info.
20343 2017-06-06 Tom de Vries <tom@codesourcery.com>
20345 * config/nvptx/nvptx.c (split_mode_p): New function.
20346 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
20348 2017-06-06 Tom de Vries <tom@codesourcery.com>
20350 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
20352 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20355 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
20358 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20360 * shrink-wrap.c (handle_simple_exit): Update profile.
20361 (try_shrink_wrapping): Upate profile.
20363 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20365 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
20366 (tree_guess_outgoing_edge_probabilities): New.
20367 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
20368 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
20370 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20372 * ipa-split.c (split_function): Initialize return bb profile.
20374 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20376 * profile.c (compute_branch_probabilities): Also initialize
20377 EXIT_BLOCK profile.
20379 2017-06-06 Richard Biener <rguenther@suse.de>
20381 PR tree-optimization/80928
20382 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
20383 (vect_analyze_loop_operations): Properly guard analysis for
20385 (vect_transform_loop): Likewise.
20386 (vect_analyze_loop_2): Also reset SLP type on PHIs.
20387 (vect_model_induction_cost): Do not cost for pure SLP.
20388 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
20389 of induction in inner loop vectorization.
20390 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
20391 (vect_get_and_check_slp_defs): Handle vect_induction_def.
20392 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
20394 (vect_analyze_slp_cost_1): Cost induction.
20395 (vect_detect_hybrid_slp_stmts): Handle PHIs.
20396 (vect_get_slp_vect_defs): Likewise.
20397 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
20398 (vect_transform_stmt): Handle SLP reductions.
20399 * tree-vectorizer.h (vectorizable_induction): Adjust.
20401 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
20403 * config/rs6000/rs6000.c (make_resolver_func): Update
20404 init_lowered_empty_function call.
20406 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
20408 * doc/invoke.texi: Document the -fprofile-abs-path option.
20409 * common.opt (fprofile-abs-path): New option.
20410 * gcov-io.h (gcov_write_filename): Declare.
20411 * gcov-io.c (gcov_write_filename): New function.
20412 * coverage.c (coverage_begin_function): Use gcov_write_filename.
20413 * profile.c (output_location): Likewise.
20415 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20417 * shring-wrap.c: Revert accidental commit.
20419 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
20421 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
20423 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20425 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
20427 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
20428 profile in callgraph edge.
20429 * profile-count.h (apply_probability): If THIS is 0, then result is 0
20430 (apply_scale): Likewise.
20431 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
20432 Also scale profile when inlining function with zero profile.
20433 (initialize_cfun): Update exit block profile even when it is zero.
20434 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
20435 when profile is read.
20437 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
20439 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
20440 (CLONE_*): New constants to define the processors we can generate
20441 code for with the target_clone attribute.
20442 (rs6000_clone_map): New array to identify which clone processors
20443 the current program is running on.
20444 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
20445 target_clone attribute.
20446 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
20447 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
20448 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
20449 (cpu_expand_builtin): Add support for target_clone attribute.
20450 (rs6000_valid_attribute_p): Allow "default" attribute.
20451 (get_decl_name): New debug function to simplify printing the
20452 current function name in debugging statements.
20453 (rs6000_clone_priority): New functions to support the target_clone
20454 attribute, and be able to generate code to switch between ISA 2.05
20455 through ISA 3.0 (power6 through power9).
20456 (rs6000_compare_version_priority): Likewise.
20457 (rs6000_get_function_versions_dispatcher): Likewise.
20458 (make_resolver_func): Likewise.
20459 (add_condition_to_bb): Likewise.
20460 (dispatch_function_versions): Likewise.
20461 (rs6000_generate_version_dispatcher_body): Likewise.
20462 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
20463 (fusion_gpr_load_p): Fix a spacing issue.
20464 * doc/extend.texi (Common Function Attributes): Document that the
20465 PowerPC supports the target_clone attribute.
20467 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
20469 * config/arm/arm.h: explain F symbol found in description of ARM
20470 register allocation in its legend.
20472 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20474 * config/mips/frame-header-opt.c: Include profile-count.h.
20475 * config/riscv/riscv.c: Include profile-count.h
20477 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20479 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
20481 (sm_set_flag_if_changed): Add bbs field.
20482 (execute_sm_if_changed_flag_set): Pass BBS.
20483 (execute_sm): Update.
20485 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20487 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
20490 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20492 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
20493 (peephole2): New peephole2 to emit the above.
20494 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
20496 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20498 * config/aarch64/aarch64.c (define_peephole2 above
20499 *sub_<shift>_<mode>): New peephole.
20501 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
20503 * config/i386/i386.c (make_resolver_func): Update.
20504 * Makefile.in: Add profile-count.h and profile-count.o
20505 * auto-profile.c (afdo_indirect_call): Update to new API.
20506 (afdo_set_bb_count): Update.
20507 (afdo_propagate_edge): Update.
20508 (afdo_propagate_circuit): Update.
20509 (afdo_calculate_branch_prob): Update.
20510 (afdo_annotate_cfg): Update.
20511 * basic-block.h: Include profile-count.h
20512 (struct edge_def): Turn count to profile_count.
20513 (struct basic_block_def): Likewie.
20514 (REG_BR_PROB_BASE): Move to profile-count.h
20515 (RDIV): Move to profile-count.h
20516 * bb-reorder.c (max_entry_count): Turn to profile_count.
20517 (find_traces): Update.
20518 (rotate_loop):Update.
20519 (connect_traces):Update.
20520 (sanitize_hot_paths):Update.
20521 * bt-load.c (migrate_btr_defs): Update.
20522 * cfg.c (RDIV): Remove.
20523 (init_flow): Use alloc_block.
20524 (alloc_block): Uninitialize count.
20525 (unchecked_make_edge): Uninitialize count.
20526 (check_bb_profile): Update.
20527 (dump_edge_info): Update.
20528 (dump_bb_info): Update.
20529 (update_bb_profile_for_threading): Update.
20530 (scale_bbs_frequencies_int): Update.
20531 (scale_bbs_frequencies_gcov_type): Update.
20532 (scale_bbs_frequencies_profile_count): New.
20533 * cfg.h (update_bb_profile_for_threading): Update.
20534 (scale_bbs_frequencies_profile_count): Declare.
20535 * cfgbuild.c (compute_outgoing_frequencies): Update.
20536 (find_many_sub_basic_blocks): Update.
20537 * cfgcleanup.c (try_forward_edges): Update.
20538 (try_crossjump_to_edge): Update.
20539 * cfgexpand.c (expand_gimple_tailcall): Update.
20540 (construct_exit_block): Update.
20541 * cfghooks.c (verify_flow_info): Update.
20542 (dump_bb_for_graph): Update.
20543 (split_edge): Update.
20544 (make_forwarder_block): Update.
20545 (duplicate_block): Update.
20546 (account_profile_record): Update.
20547 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
20548 (get_estimated_loop_iterations): Update.
20549 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
20550 (single_likely_exit): Update.
20551 * cfgloopmanip.c (scale_loop_profile): Update.
20553 (set_zero_probability): Update.
20554 (lv_adjust_loop_entry_edge): Update.
20555 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
20556 (purge_dead_edges): Update.
20557 (rtl_account_profile_record): Update.
20558 * cgraph.c (cgraph_node::create): Uninitialize count.
20559 (symbol_table::create_edge): Uninitialize count.
20560 (cgraph_update_edges_for_call_stmt_node): Update.
20561 (cgraph_edge::dump_edge_flags): Update.
20562 (cgraph_node::dump): Update.
20563 (cgraph_edge::maybe_hot_p): Update.
20564 * cgraph.h: Include profile-count.h
20565 (create_clone), create_edge, create_indirect_edge): Update.
20566 (cgraph_node): Turn count to profile_count.
20567 (cgraph_edge0: Likewise.
20568 (make_speculative, clone): Update.
20569 (create_edge): Update.
20570 (init_lowered_empty_function): Update.
20571 * cgraphclones.c (cgraph_edge::clone): Update.
20572 (duplicate_thunk_for_node): Update.
20573 (cgraph_node::create_clone): Update.
20574 * cgraphunit.c (cgraph_node::analyze): Update.
20575 (cgraph_node::expand_thunk): Update.
20576 * final.c (dump_basic_block_info): Update.
20577 * gimple-streamer-in.c (input_bb): Update.
20578 * gimple-streamer-out.c (output_bb): Update.
20579 * graphite.c (print_global_statistics): Update.
20580 (print_graphite_scop_statistics): Update.
20581 * hsa-brig.c: Include basic-block.h.
20582 * hsa-dump.c: Include basic-block.h.
20583 * hsa-gen.c (T sum_slice): Update.
20584 (convert_switch_statements):Update.
20585 * hsa-regalloc.c: Include basic-block.h.
20586 * ipa-chkp.c (chkp_produce_thunks): Update.
20587 * ipa-cp.c (struct caller_statistics): Update.
20588 (init_caller_stats): Update.
20589 (gather_caller_stats): Update.
20590 (ipcp_cloning_candidate_p): Update.
20591 (good_cloning_opportunity_p): Update.
20592 (get_info_about_necessary_edges): Update.
20593 (dump_profile_updates): Update.
20594 (update_profiling_info): Update.
20595 (update_specialized_profile): Update.
20596 (perhaps_add_new_callers): Update.
20597 (decide_about_value): Update.
20598 (ipa_cp_c_finalize): Update.
20599 * ipa-devirt.c (struct odr_type_warn_count): Update.
20600 (struct decl_warn_count): Update.
20601 (struct final_warning_record): Update.
20602 (possible_polymorphic_call_targets): Update.
20603 (ipa_devirt): Update.
20604 * ipa-fnsummary.c (redirect_to_unreachable): Update.
20605 * ipa-icf.c (sem_function::merge): Update.
20606 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
20607 * ipa-inline.c (compute_uninlined_call_time): Update.
20608 (compute_inlined_call_time): Update.
20609 (want_inline_small_function_p): Update.
20610 (want_inline_self_recursive_call_p): Update.
20611 (edge_badness): Update.
20612 (lookup_recursive_calls): Update.
20613 (recursive_inlining): Update.
20614 (inline_small_functions): Update.
20615 (dump_overall_stats): Update.
20616 (dump_inline_stats): Update.
20617 * ipa-profile.c (ipa_profile_generate_summary): Update.
20618 (ipa_propagate_frequency): Update.
20619 (ipa_profile): Update.
20620 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
20621 * ipa-utils.c (ipa_merge_profiles): Update.
20622 * loop-doloop.c (doloop_modify): Update.
20623 * loop-unroll.c (report_unroll): Update.
20624 (unroll_loop_runtime_iterations): Update.
20625 * lto-cgraph.c (lto_output_edge): Update.
20626 (lto_output_node): Update.
20627 (input_node): Update.
20628 (input_edge): Update.
20629 (merge_profile_summaries): Update.
20630 * lto-streamer-in.c (input_cfg): Update.
20631 * lto-streamer-out.c (output_cfg): Update.
20632 * mcf.c (create_fixup_graph): Update.
20633 (adjust_cfg_counts): Update.
20634 (sum_edge_counts): Update.
20635 * modulo-sched.c (sms_schedule): Update.
20636 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
20637 * predict.c (maybe_hot_count_p): Update.
20638 (probably_never_executed): Update.
20639 (dump_prediction): Update.
20640 (combine_predictions_for_bb): Update.
20641 (propagate_freq): Update.
20642 (handle_missing_profiles): Update.
20643 (counts_to_freqs): Update.
20644 (rebuild_frequencies): Update.
20645 (force_edge_cold): Update.
20646 * predict.h: Include profile-count.h
20647 (maybe_hot_count_p, counts_to_freqs): UPdate.
20648 * print-rtl-function.c: Do not include cfg.h
20649 * print-rtl.c: Include basic-block.h
20650 * profile-count.c: New file.
20651 * profile-count.h: New file.
20652 * profile.c (is_edge_inconsistent): Update.
20653 (correct_negative_edge_counts): Update.
20654 (is_inconsistent): Update.
20655 (set_bb_counts): Update.
20656 (read_profile_edge_counts): Update.
20657 (compute_frequency_overlap): Update.
20658 (compute_branch_probabilities): Update; Initialize and deinitialize
20660 (branch_prob): Update.
20661 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
20662 (edge_gcov_count): New.
20663 (bb_gcov_count): New.
20664 * shrink-wrap.c (try_shrink_wrapping): Update.
20665 * tracer.c (better_p): Update.
20666 * trans-mem.c (expand_transaction): Update.
20667 (ipa_tm_insert_irr_call): Update.
20668 (ipa_tm_insert_gettmclone_call): Update.
20669 * tree-call-cdce.c: Update.
20670 * tree-cfg.c (gimple_duplicate_sese_region): Update.
20671 (gimple_duplicate_sese_tail): Update.
20672 (gimple_account_profile_record): Update.
20673 (execute_fixup_cfg): Update.
20674 * tree-inline.c (copy_bb): Update.
20675 (copy_edges_for_bb): Update.
20676 (initialize_cfun): Update.
20677 (freqs_to_counts): Update.
20678 (copy_cfg_body): Update.
20679 (expand_call_inline): Update.
20680 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
20681 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
20682 (try_unroll_loop_completely): Update.
20683 (try_peel_loop): Update.
20684 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
20685 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
20686 * tree-ssa-loop-split.c (connect_loops): Update.
20687 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
20688 * tree-ssa-reassoc.c (branch_fixup): Update.
20689 * tree-ssa-tail-merge.c (replace_block_by): Update.
20690 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
20691 (compute_path_counts): Update.
20692 (update_profile): Update.
20693 (recompute_probabilities): Update.
20694 (update_joiner_offpath_counts): Update.
20695 (estimated_freqs_path): Update.
20696 (freqs_to_counts_path): Update.
20697 (clear_counts_path): Update.
20698 (ssa_fix_duplicate_block_edges): Update.
20699 (duplicate_thread_path): Update.
20700 * tree-switch-conversion.c (case_bit_test_cmp): Update.
20701 (struct switch_conv_info): Update.
20702 * tree-tailcall.c (decrease_profile): Update.
20703 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
20704 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
20705 * value-prof.c (check_counter): Update.
20706 (gimple_divmod_fixed_value): Update.
20707 (gimple_mod_pow2): Update.
20708 (gimple_mod_subtract): Update.
20709 (gimple_ic_transform): Update.
20710 (gimple_stringop_fixed_value): Update.
20711 * value-prof.h (gimple_ic): Update.
20713 2017-06-02 Carl Love <cel@us.ibm.com>
20715 * config/rs6000/rs6000-c: Add support for built-in functions
20716 vector double vec_doublee (vector signed int);
20717 vector double vec_doublee (vector unsigned int);
20718 vector double vec_doublee (vector float);
20719 vector double vec_doubleh (vector signed int);
20720 vector double vec_doubleh (vector unsigned int);
20721 vector double vec_doubleh (vector float);
20722 vector double vec_doublel (vector signed int);
20723 vector double vec_doublel (vector unsigned int);
20724 vector double vec_doublel (vector float);
20725 vector double vec_doubleo (vector signed int);
20726 vector double vec_doubleo (vector unsigned int);
20727 vector double vec_doubleo (vector float);.
20728 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
20729 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
20731 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
20732 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
20733 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
20735 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
20736 vec_doublel, vec_doubleh.
20737 * doc/extend.texi: Update the built-in documentation file for the
20738 new built-in functions.
20740 2017-06-02 David Malcolm <dmalcolm@redhat.com>
20743 * ipa-inline-analysis.c (free_growth_caches): Set
20744 edge_removal_hook_holder to NULL after removing it.
20746 2017-06-02 Sudakshina Das <sudi.das@arm.com>
20748 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
20749 comparision with zero.
20751 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
20752 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
20753 for early expansion of vec_min and vec_max builtins.
20754 (builtin_function_type): Add min/max unsigned variants to those
20755 identified as having unsigned arguments.
20757 2017-06-02 Olivier Hainque <hainque@adacore.com>
20759 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
20761 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20763 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
20764 Use VALL_F16 iterator rather than VALL.
20766 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20768 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
20769 Emit CBNZ inside loop when doing a strong exchange and comparing
20770 against zero. Generate the CC flags after the loop.
20772 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
20774 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
20775 (dl_section_ref): New.
20776 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
20777 On AIX, append an expression to subtract the size of the
20778 section length to dl_section_ref.
20780 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
20782 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
20783 for early expansion of vector absolute builtins.
20785 2017-06-02 Richard Biener <rguenther@suse.de>
20787 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
20788 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
20790 2017-06-02 Richard Biener <rguenther@suse.de>
20792 PR tree-optimization/80948
20793 * tree-tailcall.c (find_tail_calls): Track stmts to move in
20794 stmt order as well.
20796 2017-06-02 Richard Biener <rguenther@suse.de>
20798 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
20800 * tree-vect-stmts.c (process_use): Do not mark backedge defs
20801 for inductions as relevant.
20803 2017-06-02 Richard Biener <rguenther@suse.de>
20805 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
20806 (vectorizable_induction): ... this. Remove dead code.
20808 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
20810 * builtins. (expand_builtin_alloca): Remove second parameter and
20811 infer its value from the first parameter instead.
20812 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
20814 2017-06-02 Jakub Jelinek <jakub@redhat.com>
20816 PR rtl-optimization/80903
20817 * loop-doloop.c (add_test): Unshare sequence.
20819 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
20821 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
20823 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
20825 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
20827 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
20828 xlogue_layout::get_instance, logue_layout::xlogue_layout,
20829 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
20830 (xlogue_layout::get_stub_rtx): Make static.
20831 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
20832 (xlogue_layout::compute_stub_managed_regs): Rename to...
20833 (xlogue_layout::count_stub_managed_regs): ...this.
20834 (xlogue_layout::is_stub_managed_reg): New function.
20835 (xlogue_layout::m_stub_names): Rename to...
20836 (xlogue_layout::s_stub_names): ...this, make static.
20837 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
20838 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
20839 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
20840 xlogue_layout::s_stub_names): Instantiate statics.
20841 (stub_managed_regs): Remove.
20842 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
20843 (disable_call_ms2sysv_xlogues): Rename to...
20844 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
20845 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
20847 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
20848 change after reload_completed.
20849 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
20851 (ix86_expand_prologue): Likewise.
20852 (ix86_expand_epilogue): Likewise.
20853 (ix86_expand_split_stack_prologue): Likewise.
20854 (ix86_compute_frame_layout): Remove frame parameter ...
20855 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
20856 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
20858 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
20859 (ix86_frame): Move from here ...
20860 * config/i386/i386.h (ix86_frame): ... to here.
20861 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
20862 complete ix86_frame data structure instead. Remove some_ld_name.
20864 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
20866 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
20867 symbols that hold a DECL_VALUE_EXPR.
20869 2017-06-01 Martin Jambor <mjambor@suse.cz>
20871 PR tree-optimization/80898
20872 * tree-sra.c (process_subtree_disqualification): Removed.
20873 (disqualify_candidate): Do not acll
20874 process_subtree_disqualification.
20875 (subtree_mark_written_and_enqueue): New function.
20876 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
20877 RHS has been disqualified and re-queue LHS if necessary. Apart
20878 from that, ignore disqualified RHS.
20880 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20882 * config/s390/s390.c (s390_emit_epilogue): Disable early return
20883 address fetch for z10 or later.
20885 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20887 * config/arc/arc.md (tst_movb): Add guard when splitting.
20889 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20891 * config/arc/arc.c (arc_can_eliminate): Test against
20892 arc_frame_pointer_needed.
20894 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20896 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
20897 to prevent store reordering.
20898 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
20899 (type): Add block type.
20900 (stack_tie): Define special instruction to be used in
20903 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20905 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
20906 constraint. It is not valid for the pattern.
20907 (noncommutative_binary_comparison): Likewise.
20909 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20911 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
20914 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20916 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
20917 be used by the reg-alloc.
20919 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20921 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
20922 reg-alloc when having mul64 or mul32x16 instructions.
20923 (mulsidi3): Likewise.
20924 (umulsidi3): Likewise.
20925 (mulsi32x16): New pattern.
20926 (mulsi64): Likewise.
20927 (mulsidi64): Likewise.
20928 (umulsidi64): Likewise.
20929 (MUL32x16_REG): Define.
20930 (mul64_600): Use MUL32x16_REG.
20931 (mac64_600): Likewise.
20932 (umul64_600): Likewise.
20933 (umac64_600): Likewise.
20935 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20937 * config/arc/arc.md (mulsi3_700): Make it commutative.
20939 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
20941 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
20943 (*sign_extendsidi2_insn): Likewise for movstosw.
20945 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
20947 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
20948 the type of the input discriminant value. Convert the
20949 discriminant value of signedness vary.
20951 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
20953 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
20954 Add to -Wall section.
20956 2017-06-01 Richard Biener <rguenther@suse.de>
20958 PR middle-end/66313
20959 * fold-const.c (fold_plusminus_mult_expr): If the factored
20960 factor may be zero use a wrapping type for the inner operation.
20961 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
20962 and handle moved defs.
20963 (process_assignment): Properly guard the unary op case. Return a
20964 tri-state indicating that moving the stmt before the call may allow
20965 to continue. Pass through to_move.
20966 (find_tail_calls): Handle moving unrelated defs before
20969 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
20972 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
20973 splitter result in the canonical way.
20975 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
20977 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
20978 also for 32bit target. Update insn attributes.
20979 (zero-extendsidi2 splitter): Allow all registers for operand 1.
20981 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
20983 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
20984 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
20985 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
20986 (_mm_maskz_min_ss): New intrinsics.
20988 2017-05-31 Martin Liska <mliska@suse.cz>
20990 * tree-vect-loop.c (vect_create_epilog_for_reduction):
20991 Change comment style to one we normally use.
20992 (vectorizable_reduction): Likewise.
20993 (vectorizable_induction): Likewise.
20994 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
20995 (vectorizable_call): Likewise.
20996 (vectorizable_simd_clone_call): Likewise.
20997 (vectorizable_conversion): Likewise.
20998 (vectorizable_assignment): Likewise.
20999 (vectorizable_shift): Likewise.
21000 (vectorizable_operation): Likewise.
21001 (vectorizable_store): Likewise.
21002 (vectorizable_load): Likewise.
21003 * tree-vectorizer.h: Likewise.
21005 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
21007 * passes.c (emergency_dump_function): New.
21008 * tree-pass.h (emergency_dump_function): Declare.
21009 * plugin.c (plugins_internal_error_function): Remove.
21010 * plugin.h (plugins_internal_error_function): Remove declaration.
21011 * toplev.c (internal_error_function): New static function. Use it...
21012 (general_init): ...here.
21014 2017-05-31 Graham Markall <graham.markall@embecosm.com>
21016 * config/arc/arc.c (arc_print_operand): Handle constant operands.
21017 (arc_rtx_costs): Add costs for new patterns.
21018 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
21019 * config/arc/predicates.md: Add _1_2_3_operand predicate.
21021 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21023 * tree-ssa-strlen.c (get_next_strinfo): New function.
21024 (get_stridx_plus_constant): Use it.
21025 (zero_length_string): Likewise.
21026 (adjust_related_strinfos): Likewise.
21027 (adjust_last_stmt): Likewise.
21029 2017-05-31 Richard Biener <rguenther@suse.de>
21032 * config/i386/i386.c (ix86_expand_builtin): Remove assert
21033 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
21035 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21037 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
21038 loop_vinfo argument and use of dependence distance vectors.
21039 Check instead whether the two references differ only in their
21040 initial value and assume that they have the same alignment if the
21041 difference is a multiple of the vector alignment.
21042 (vect_analyze_data_refs_alignment): Update call accordingly.
21044 2017-05-31 Martin Liska <mliska@suse.cz>
21047 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
21049 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21051 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
21052 (create_intersect_range_checks): Move from ...
21053 * tree-data-ref.c (create_intersect_range_checks_index)
21054 (create_intersect_range_checks): ... to here.
21055 (create_runtime_alias_checks): New function factored from ...
21056 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
21057 here. Call above function.
21058 * tree-data-ref.h (create_runtime_alias_checks): New function.
21060 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21062 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
21063 segment length for dr_b and compute it in wide_int.
21065 2017-05-31 Richard Biener <rguenther@suse.de>
21067 PR tree-optimization/80906
21068 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
21069 and pass through iv_map.
21070 (copy_bb_and_scalar_dependences): Adjust.
21071 (translate_pending_phi_nodes): Likewise.
21072 (copy_loop_close_phi_args): Handle code-generating IVs instead
21075 2017-05-30 David Malcolm <dmalcolm@redhat.com>
21077 * diagnostic-color.c (color_dict): Add "type-diff".
21078 (parse_gcc_colors): Update comment.
21079 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
21080 -fdiagnostics-show-template-tree and -fno-elide-type.
21081 (GCC_COLORS): Add type-diff to example.
21083 (-fdiagnostics-show-template-tree): New.
21084 (-fno-elide-type): New.
21085 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
21086 the pp_format_decoder callback. Call any m_format_postprocessor's
21088 (pretty_printer::pretty_printer): Initialize
21089 m_format_postprocessor.
21090 (pretty_printer::~pretty_printer): Delete any
21091 m_format_postprocessor.
21092 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
21093 (class format_postprocessor): New class.
21094 (struct pretty_printer::format_decoder): Document the new parameters.
21095 (struct pretty_printer::m_format_postprocessor): New field.
21096 * tree-diagnostic.c (default_tree_printer): Update for new
21097 bool and const char ** params.
21098 * tree-diagnostic.h (default_tree_printer): Likewise.
21100 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
21102 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
21103 (lwa_operand): Delete rs6000_gen_cell_microcode test.
21104 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21105 rs6000_gen_cell_microcode code.
21106 (rs6000_final_prescan_insn): Delete.
21107 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
21108 "warn-cell-microcode" entries.
21109 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
21110 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
21111 throughout. Change cc_reg_not_micro_cr0_operand to
21112 cc_reg_not_cr0_operand throughout.
21113 (*extendhi<mode>2_noload): Delete.
21114 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
21115 (mwarn-cell-microcode): Delete.
21116 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
21117 -mgen-cell-microcode and -mwarn-cell-microcode.
21119 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
21122 * config/i386/constraints.md (Yd): New constraint.
21124 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
21125 and (?Yd, r) alternatives. Update insn attributes.
21126 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
21127 and (?*Yd, r) alternatives. Update insn attributes.
21128 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
21130 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
21132 * gimplify.c (gimplify_modify_expr): Don't create a
21133 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
21136 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
21138 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
21140 2017-05-30 Richard Biener <rguenther@suse.de>
21142 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
21143 and reduc_def fields.
21144 (STMT_VINFO_REDUC_TYPE): New define.
21145 (STMT_VINFO_REDUC_DEF): Likewise.
21146 (vect_force_simple_reduction): Adjust prototype.
21147 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
21148 (vect_is_simple_reduction): Remove check_reduction argument.
21149 (vect_force_simple_reduction): Adjust and set
21150 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21151 (vectorizable_reduction): Do not re-do reduction analysis
21152 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21153 * tree-parloops.c (gather_scalar_reductions): Adjust.
21155 2017-05-30 Richard Biener <rguenther@suse.de>
21157 PR middle-end/80901
21158 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
21161 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
21163 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
21164 Introduce unknown_misalignment parameter and remove vf.
21165 (vect_peeling_hash_get_lowest_cost):
21166 Pass unknown_misalignment parameter.
21167 (vect_enhance_data_refs_alignment):
21168 Fix unsupportable data ref treatment.
21170 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21172 * tree-vect-data-refs.c (vect_get_data_access_cost):
21173 Workaround for SLP handling.
21174 (vect_enhance_data_refs_alignment):
21175 Compute costs for doing no peeling at all, compare to the best
21176 peeling costs so far and avoid peeling if cheaper.
21178 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21180 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
21181 Return peeling info and set costs to zero for unlimited cost
21183 (vect_enhance_data_refs_alignment): Also inspect all datarefs
21184 with unknown misalignment. Compute and costs for unknown
21185 misalignment, compare them to the costs for known misalignment
21186 and choose the cheapest for peeling.
21188 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21190 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
21191 (vect_get_peeling_costs_all_drs): Create function.
21192 (vect_peeling_hash_get_lowest_cost):
21193 Use vect_get_peeling_costs_all_drs.
21194 (vect_peeling_supportable): Create function.
21195 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
21197 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21199 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
21200 DR_HAS_NEGATIVE_STEP.
21201 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
21202 (vect_enhance_data_refs_alignment): Use.
21203 (vect_duplicate_ssa_name_ptr_info): Use.
21204 * tree-vectorizer.h (dr_misalignment): Use.
21205 (known_alignment_for_access_p): Use.
21207 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21210 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
21212 (has_section_name): New function.
21214 2017-05-30 Martin Liska <mliska@suse.cz>
21217 * auto-profile.c (get_function_decl_from_block): Fix
21220 2017-05-30 Richard Biener <rguenther@suse.de>
21222 PR middle-end/80876
21223 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
21225 2017-05-30 Martin Liska <mliska@suse.cz>
21227 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
21228 * dumpfile.h (struct dump_file_info): Remove ctors.
21230 2017-05-30 Martin Liska <mliska@suse.cz>
21232 * predict.def: Fix GNU coding style.
21234 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21236 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
21237 Mark 'to' argument with ATTRIBUTE_UNUSED.
21239 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21241 * config/xtensa/xtensa.c (xtensa_emit_call): Use
21242 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
21243 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
21246 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
21248 * doc/install.texi (Options specification): Restore entry of
21249 --enable-sjlj-exceptions.
21251 2017-05-27 Michael Eager <eager@eagercon.com>
21254 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
21256 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
21258 * config/microblaze/microblaze.h
21259 (FIXED_REGISTERS): Update in macro.
21260 (CALL_USED_REGISTERS): Update in macro.
21262 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
21264 * doc/install.texi: Add links to macOS binary distributions.
21266 2017-05-27 Jakub Jelinek <jakub@redhat.com>
21270 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21272 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
21274 2017-05-26 Martin Liska <mliska@suse.cz>
21276 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
21278 2017-05-26 Martin Liska <mliska@suse.cz>
21280 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
21281 always leading ';; '.
21282 (dump_bb_info): Likewise.
21283 (brief_dump_cfg): Likewise.
21284 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
21285 * dumpfile.c: Remove usage of TDF_VERBOSE.
21286 * dumpfile.h (enum dump_kind): Likewise.
21287 (dump_gimple_bb_header): Do not use TDF_COMMENT.
21288 * print-tree.c (debug_verbose): Remove.
21289 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
21290 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
21291 * tree-diagnostic.c (default_tree_printer): Replace
21292 TDF_DIAGNOSTIC with TDF_SLIM.
21294 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21296 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
21297 in parameter loop, rather than loop_vinfo.
21298 (create_intersect_range_checks): Ditto.
21299 (vect_create_cond_for_alias_checks): Update call to above functions.
21301 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21303 PR tree-optimization/80815
21304 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
21305 for merging runtime alias checks. Handle negative DR_STEPs.
21307 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21309 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
21311 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
21312 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
21313 out code pruning runtime alias checks.
21314 * tree-data-ref.c (prune_runtime_alias_test_list): New function
21315 factored out from above.
21316 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
21318 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
21320 (prune_runtime_alias_test_list): New decalaration.
21322 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21324 * tree-vect-data-refs.c (compare_tree): Rename and move ...
21325 * tree-data-ref.c (data_ref_compare_tree): ... to here.
21326 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
21327 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
21328 (operator==, comp_dr_with_seg_len_pair): Ditto.
21329 (vect_prune_runtime_alias_test_list): Ditto.
21331 2017-05-26 Martin Liska <mliska@suse.cz>
21334 * params.def: Bound partial-inlining-entry-probability param.
21336 2017-05-26 Marek Polacek <polacek@redhat.com>
21339 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
21342 2017-05-26 Richard Biener <rguenther@suse.de>
21344 PR tree-optimization/80842
21345 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
21348 2017-05-26 Richard Biener <rguenther@suse.de>
21350 PR tree-optimization/80844
21351 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
21353 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
21355 * doc/md.texi (Machine Constraints): Update x86 family
21356 machine constraints section to match 'config/i386/constraints.md'.
21358 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
21360 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
21362 2017-05-25 Nathan Sidwell <nathan@acm.org>
21364 * doc/invoke.texi (--enable-languages): Update documentation.
21366 2017-05-25 Martin Liska <mliska@suse.cz>
21368 * dumpfile.c: Add TDF_FOLDING.
21369 * dumpfile.h (enum dump_kind): Likewise.
21370 * genmatch.c (dt_simplify::gen_1): Use it.
21372 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21374 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
21376 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21378 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
21379 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
21381 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21383 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
21384 * match.pd (X == C): Rewrite it here.
21385 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
21386 with_certain_nonzero_bits2): New predicates.
21387 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
21389 2017-05-24 Nathan Sidwell <nathan@acm.org>
21391 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
21394 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
21397 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
21399 * config/powerpcspe: New port. Files are copied from the rs6000
21400 port, with "rs6000" in filenames replaced by "powerpcspe".
21402 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
21404 PR rtl-optimization/80754
21405 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
21407 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
21409 * config/sparc/sparc.md (length): Return the correct value for -mflat
21410 sibcalls to match output_sibcall.
21412 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
21416 * config/rs6000/rs6000.c (struct machine_function): Add new field
21418 (rs6000_get_separate_components): Init that field, use it.
21419 (rs6000_components_for_bb): Use the field.
21421 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
21423 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
21425 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
21427 PR middle-end/80823
21428 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
21430 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21433 * config/s390/s390.c (s390_check_qrst_address): Check incoming
21434 address against address_operand predicate.
21435 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
21437 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
21439 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
21440 parameters passed indirectly.
21442 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
21444 * config/i386/i386.md (*movdi_internal): Remove SSE4
21445 alternative 18 (?r, *v). Update insn attributes.
21446 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
21447 Update insn attributes.
21448 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
21449 Update insn attributes.
21450 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
21451 alternative 1 (r, v). Remove isa attribute.
21452 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
21453 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
21454 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
21456 2017-05-23 Tom de Vries <tom@codesourcery.com>
21458 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
21461 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21463 * cgraphunit.c (symbol_table::process_new_functions): Update.
21464 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
21465 (inline_generate_summary): Rename to ...
21466 (ipa_fn_summary_generate): ... this one.
21467 (inline_read_summary): Rename to ...
21468 (ipa_fn_summary_read): ... this one.
21469 (inline_write_summary): Rename to ...
21470 (ipa_fn_summary_write): ... this one.
21471 (inline_free_summary): Rename to ...
21472 (ipa_free_fn_summary): ... this one.
21473 (pass_data_local_fn_summary, pass_local_fn_summary,
21474 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
21475 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
21476 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
21477 make_pass_ipa_fn_summary): New.
21478 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
21479 inline_write_summary, inline_free_summary): Remove.
21480 (ipa_free_fn_summary) : New.
21481 * ipa-inline.c (ipa_inline): Update.
21482 (pass_ipa_inline): Do not generate summaries.
21483 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
21485 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
21486 and add pass_ipa_fn_summary.
21487 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
21489 (make_pass_inline_parameters): Remove.
21491 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
21493 * omp-low.c (struct omp_context): Remove "default_kind" member.
21496 * omp-offload.c (execute_oacc_device_lower): Remove the
21497 parallelism dimensions function attributes for unparallelized
21498 OpenACC kernels constructs.
21500 2017-05-23 Martin Liska <mliska@suse.cz>
21502 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
21504 (cgraph_edge::make_speculative): Likewise.
21505 (cgraph_edge::resolve_speculation): Likewise.
21506 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
21507 (cgraph_node::dump): Likewise.
21508 * cgraph.h: Likewise.
21509 * cgraphunit.c (analyze_functions): Likewise.
21510 (symbol_table::compile): Likewise.
21511 * ipa-cp.c (print_all_lattices): Likewise.
21512 (determine_versionability): Likewise.
21513 (initialize_node_lattices): Likewise.
21514 (ipcp_verify_propagated_values): Likewise.
21515 (estimate_local_effects): Likewise.
21516 (update_profiling_info): Likewise.
21517 (create_specialized_node): Likewise.
21518 (perhaps_add_new_callers): Likewise.
21519 (decide_about_value): Likewise.
21520 (decide_whether_version_node): Likewise.
21521 (identify_dead_nodes): Likewise.
21522 (ipcp_store_bits_results): Likewise.
21523 * ipa-devirt.c (dump_targets): Likewise.
21524 (ipa_devirt): Likewise.
21525 * ipa-icf.c (sem_item::dump): Likewise.
21526 (sem_function::equals): Likewise.
21527 (sem_variable::equals): Likewise.
21528 (sem_item_optimizer::read_section): Likewise.
21529 (sem_item_optimizer::execute): Likewise.
21530 (congruence_class::dump): Likewise.
21531 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
21532 (dump_inline_summary): Likewise.
21533 (estimate_node_size_and_time): Likewise.
21534 (inline_analyze_function): Likewise.
21535 * ipa-inline-transform.c (inline_call): Likewise.
21536 * ipa-inline.c (report_inline_failed_reason): Likewise.
21537 (want_early_inline_function_p): Likewise.
21538 (edge_badness): Likewise.
21539 (update_edge_key): Likewise.
21540 (inline_small_functions): Likewise.
21541 * ipa-profile.c (ipa_profile): Likewise.
21542 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
21543 (ipa_make_edge_direct_to_target): Likewise.
21544 (remove_described_reference): Likewise.
21545 (ipa_impossible_devirt_target): Likewise.
21546 (propagate_controlled_uses): Likewise.
21547 (ipa_print_node_params): Likewise.
21548 (ipcp_transform_function): Likewise.
21549 * ipa-pure-const.c (pure_const_read_summary): Likewise.
21550 (propagate_pure_const): Likewise.
21551 * ipa-reference.c (generate_summary): Likewise.
21552 (read_write_all_from_decl): Likewise.
21553 (propagate): Likewise.
21554 (ipa_reference_read_optimization_summary): Likewise.
21555 * ipa-utils.c (ipa_merge_profiles): Likewise.
21556 * ipa.c (walk_polymorphic_call_targets): Likewise.
21557 (symbol_table::remove_unreachable_nodes): Likewise.
21558 (ipa_single_use): Likewise.
21559 * passes.c (execute_todo): Likewise.
21560 * predict.c (drop_profile): Likewise.
21561 * symtab.c (symtab_node::get_dump_name): New function.
21562 (symtab_node::dump_name): Likewise.
21563 (symtab_node::dump_asm_name): Likewise.
21564 (symtab_node::dump_references): Likewise.
21565 (symtab_node::dump_referring): Likewise.
21566 (symtab_node::dump_base): Likewise.
21567 (symtab_node::debug_symtab): Likewise.
21568 * tree-sra.c (convert_callers_for_node): Likewise.
21569 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
21570 * value-prof.c (init_node_map): Likewise.
21572 2017-05-23 Martin Liska <mliska@suse.cz>
21574 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
21575 and symtab_node::debug_symtab to symbol_table::debug.
21576 * cgraphunit.c (analyze_functions): Use the renamed function.
21577 (symbol_table::compile): Likewise.
21578 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
21579 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
21580 * passes.c (execute_todo): Likewise.
21581 * symtab.c (symbol_table::dump): New function.
21582 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
21584 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21586 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
21587 that nonconst implies exec.
21589 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21591 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
21592 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
21593 (inline_edge_summary_vec): Turn into ...
21594 (ipa_call_summaries): ... this one.
21595 (redirect_to_unreachable, edge_set_predicate,
21596 evaluate_properties_for_edge, inline_summary_alloc,
21597 reset_ipa_call_summary, reset_inline_summary,
21598 inline_summary_t::duplicate): Update.
21599 (inline_edge_duplication_hook): Turn to ...
21600 (ipa_call_summary_t::duplicate): ... this one.
21601 (inline_edge_removal_hook): Turn to ...
21602 (ipa_call_summary_t::remove): ... this one.
21603 (dump_inline_edge_summary): Turn to ...
21604 (dump_ipa_call_summary): ... this one.
21605 (estimate_function_body_sizes): Update.
21606 (inline_update_callee_summaries): Update.
21607 (remap_edge_change_prob): Update.
21608 (remap_edge_summaries): Update.
21609 (inline_merge_summary): Update.
21610 (do_estimate_edge_time): Update.
21611 (inline_generate_summary): Update.
21612 (inline_read_section): Update.
21613 (inline_read_summary): Update.
21614 (inline_free_summary): Update.
21615 * ipa-inline.c (can_inline_edge_p): Update.
21616 (compute_inlined_call_time): Update.
21617 (want_inline_small_function_p): Update.
21618 (edge_badness): Update.
21619 (early_inliner): Update.
21620 * ipa-inline.h (inline_edge_summary): Turn to ...
21621 (ipa_call_summary): ... this one.
21622 (ipa_call_summary_t): New class.
21623 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
21624 (ipa_call_summaries): New.
21625 (inline_edge_summary): Remove.
21626 (estimate_edge_growth): Update.
21627 * ipa-profile.c (ipa_propagate_frequency_1): Update.
21628 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21629 * ipa-split.c (execute_split_functions): Update.
21630 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
21632 2017-05-23 Tom de Vries <tom@codesourcery.com>
21634 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
21635 attributes): Document rdrand effective target.
21637 2017-05-23 Tom de Vries <tom@codesourcery.com>
21639 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
21640 attributes): Sort alphabetically.
21642 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
21644 * config/avr/genmultilib.awk: Use gsub instead of gensub.
21646 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
21649 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
21650 V2DF/V2DI splat into two separate patterns, one that handles
21651 registers, and the other that only handles memory. Drop support
21652 for splatting from a GPR on ISA 2.07 and then splitting the
21653 splat into direct move and splat.
21654 (vsx_splat_<mode>_reg): Likewise.
21655 (vsx_splat_<mode>_mem): Likewise.
21657 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
21659 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
21661 2017-05-22 Jakub Jelinek <jakub@redhat.com>
21663 PR middle-end/80809
21664 * omp-low.c (finish_taskreg_remap): New function.
21665 (finish_taskreg_scan): If unit size of ctx->record_type
21666 is non-constant, unshare the size expression and replace
21667 decls in it with possible outer var refs.
21669 PR middle-end/80809
21670 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
21671 GOVD_SHARED rather than GOVD_PRIVATE with it.
21672 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
21673 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
21675 PR middle-end/80853
21676 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
21677 as last argument to build_outer_var_ref for pointer bases of array
21678 section reductions.
21680 2017-05-19 Martin Sebor <msebor@redhat.com>
21682 * print-tree.c (print_node): Print DECL_READ_P flag.
21684 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21686 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
21687 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
21688 * cgraph.c: Likewise.
21689 * cgraphunit.c: Likewise.
21690 * gengtype.c: Likewise.
21691 * ipa-cp.c: Likewise.
21692 * ipa-devirt.c: Likewise.
21693 * ipa-icf.c: Likewise.
21694 * ipa-predicate.c: Likewise.
21695 * ipa-profile.c: Likewise.
21696 * ipa-prop.c: Likewise.
21697 * ipa-split.c: Likewise.
21699 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
21700 edge_predicate_pool, dump_inline_hints,
21701 inline_summary::account_size_time, redirect_to_unreachable,
21702 edge_set_predicate, set_hint_predicate,
21703 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
21704 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
21705 inline_summary_t::remove, remap_hint_predicate_after_duplication,
21706 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
21707 ipa_call_summary_t::remove, initialize_growth_caches,
21708 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
21709 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
21710 mark_modified, unmodified_parm_1, unmodified_parm,
21711 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
21712 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
21713 compute_bb_predicates, will_be_nonconstant_expr_predicate,
21714 will_be_nonconstant_predicate, record_modified_bb_info,
21715 get_minimal_bb, record_modified, param_change_prob,
21716 phi_result_unknown_predicate, predicate_for_phi_result,
21717 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
21718 estimate_function_body_sizes, compute_inline_parameters,
21719 compute_inline_parameters_for_curren, pass_data_inline_parameters,
21720 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
21721 inline_update_callee_summaries, remap_edge_change_prob,
21722 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
21723 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
21724 inline_analyze_function, inline_summary_t::insert,
21725 inline_generate_summary, read_ipa_call_summary, inline_read_section,
21726 inline_read_summary, write_ipa_call_summary, inline_write_summary,
21727 inline_free_summary): Move to ipa-fnsummary.h
21728 (predicate_t): Remove.
21729 * ipa-fnsummary.c: New file.
21730 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
21731 (enum inline_hints_vals, inline_hints, agg_position_info,
21732 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
21733 inline_summaries, ipa_call_summary, ipa_call_summary_t,
21734 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
21735 dump_inline_summary, dump_inline_hints, inline_generate_summary,
21736 inline_read_summary, inline_write_summary, inline_free_summary,
21737 inline_analyze_function, initialize_inline_failed,
21738 inline_merge_summary, inline_update_overall_summary,
21739 compute_inline_parameters): Move to ipa-fnsummary.h
21740 * ipa-fnsummary.h: New file.
21741 * ipa-inline-transform.h: Include ipa-inline.h.
21742 * ipa-inline.c: LIkewise.
21744 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21746 * ipa-inline.c (edge_badness): Use inlined_time instead of
21747 inline_summaries->get.
21749 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21751 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
21753 2017-05-22 Nathan Sidwell <nathan@acm.org>
21755 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
21756 (fdump-lang): Document 'raw' option.
21757 * dumpfile.h (TDI_tu): Delete.
21758 * dumpfile.c (dump_files): Remove translation-unit.
21759 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
21761 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
21763 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
21764 command option from $(AWK) call.
21765 * config/avr/genmultilib.awk: Simplify and rewrite so that it
21766 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
21767 [FORMAT]: Remove handling of variable.
21768 * config/avr/t-multilib: Regenerate.
21770 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21772 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
21774 (dump_inline_summary): Do not print self_time.
21775 (estimate_function_body_sizes): Do not set self_time.
21776 (compute_inline_parameters): Likewise.
21777 (inline_read_section, inline_write_summary): Do not stream self_time.
21778 * ipa-inline.h (inline_summary): Drop self_time.
21780 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21782 * ipa-inline-analysis.c (account_size_time): Rename to ...
21783 (inline_summary::account_size_time): ... this one.
21784 (reset_ipa_call_summary): Turn to ...
21785 (ipa_call_summary::reset): ... this one.
21786 (reset_inline_summary): Turn to ...
21787 (inline_summary::reset): ... this one.
21788 (inline_summary_t::remove): Update.
21789 (inline_summary_t::duplicate): Update.
21790 (ipa_call_summary_t::remove): Update.
21791 (dump_inline_summary): Update.
21792 (estimate_function_body_sizes): Update.
21793 (compute_inline_parameters): Update.
21794 (estimate_node_size_and_time): Update.
21795 (inline_merge_summary): Update.
21796 (inline_update_overall_summary): Update.
21797 (inline_read_section): Update.
21798 (inline_write_summary): Update.
21799 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
21800 add account_size_time and reset member functions.
21801 (ipa_call_summary): Add reset function.
21802 * ipa-predicate.h (predicate::operator &): Constify.
21804 2017-05-22 Richard Biener <rguenther@suse.de>
21806 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
21808 2017-05-19 Jason Merrill <jason@redhat.com>
21810 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
21812 2017-05-19 Marek Polacek <polacek@redhat.com>
21815 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
21816 TYPE_OVERFLOW_WRAPS checks.
21818 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
21820 * tree-core.h (enum omp_clause_default_kind): Add
21821 "OMP_CLAUSE_DEFAULT_PRESENT".
21822 * tree-pretty-print.c (dump_omp_clause): Handle it.
21823 * gimplify.c (enum gimplify_omp_var_data): Add
21824 "GOVD_MAP_FORCE_PRESENT".
21825 (gimplify_adjust_omp_clauses_1): Map it to
21826 "GOMP_MAP_FORCE_PRESENT".
21827 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
21829 * gimplify.c (oacc_default_clause): Clarify.
21831 2017-05-19 Nathan Sidwell <nathan@acm.org>
21833 LANG_HOOK_REGISTER_DUMPS
21834 * toplev.c (general_init): Call register dump lang hook.
21835 * doc/invoke.texi: Document -fdump-lang option family.
21836 * dumpfile.c (dump_files): Remove class dump here.
21837 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
21838 * dumpfile.h (tree_dump_index): Remove TDI_class.
21839 * langhooks-def.h (lhd_register_dumps): Declare.
21840 (LANG_HOOKS_REGISTER_DUMPS): Define.
21841 (LANG_HOOKS_INITIALIZER): Add it.
21842 * langhooks.c (lhd_register_dumps): Define.
21843 * langhooks.h (struct lang_hooks): Add register_dumps.
21845 2017-05-19 Nathan Sidwell <nathan@acm.org>
21847 * context.h (context::set_passes): New.
21848 * context.c (context::context): Do not create pass manager.
21849 * toplev.c (general_init): Create pass manager here.
21851 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
21853 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
21854 use this splitter if two add or or instructions would also work for
21855 the constant we want to generate.
21857 2017-05-19 Richard Biener <rguenther@suse.de>
21860 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
21861 predicate evaluation.
21863 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
21865 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
21867 * ipa-inline.c (want_inline_small_function_p): Do not cast to
21870 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
21872 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
21873 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
21874 (inline_edge_summary_vec): Turn into ...
21875 (ipa_call_summaries): ... this one.
21876 (redirect_to_unreachable, edge_set_predicate,
21877 evaluate_properties_for_edge, inline_summary_alloc,
21878 reset_ipa_call_summary, reset_inline_summary,
21879 inline_summary_t::duplicate): Update.
21880 (inline_edge_duplication_hook): Turn to ...
21881 (ipa_call_summary_t::duplicate): ... this one.
21882 (inline_edge_removal_hook): Turn to ...
21883 (ipa_call_summary_t::remove): ... this one.
21884 (dump_inline_edge_summary): Turn to ...
21885 (dump_ipa_call_summary): ... this one.
21886 (estimate_function_body_sizes): Update.
21887 (inline_update_callee_summaries): Update.
21888 (remap_edge_change_prob): Update.
21889 (remap_edge_summaries): Update.
21890 (inline_merge_summary): Update.
21891 (do_estimate_edge_time): Update.
21892 (inline_generate_summary): Update.
21893 (inline_read_section): Update.
21894 (inline_read_summary): Update.
21895 (inline_free_summary): Update.
21896 * ipa-inline.c (can_inline_edge_p): Update.
21897 (compute_inlined_call_time): Update.
21898 (want_inline_small_function_p): Update.
21899 (edge_badness): Update.
21900 (early_inliner): Update.
21901 * ipa-inline.h (inline_edge_summary): Turn to ...
21902 (ipa_call_summary): ... this one.
21903 (ipa_call_summary_t): New class.
21904 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
21905 (ipa_call_summaries): New.
21906 (inline_edge_summary): Remove.
21907 (estimate_edge_growth): Update.
21908 * ipa-profile.c (ipa_propagate_frequency_1): Update.
21909 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21910 * ipa-split.c (execute_split_functions): Update.
21911 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
21913 2017-05-19 Richard Biener <rguenther@suse.de>
21915 PR middle-end/80764
21916 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
21918 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
21920 * config/rs6000/rs6000.c (struct machine_function): Add field
21921 fpr_is_wrapped_separately.
21922 (rs6000_get_separate_components): Use 64 components. Handle the
21923 new FPR components.
21924 (rs6000_components_for_bb): Handle the FPR components.
21925 (rs6000_emit_prologue_components): Handle the FPR components.
21926 (rs6000_emit_epilogue_components): Handle the FPR components.
21927 (rs6000_set_handled_components): Handle the FPR components.
21928 (rs6000_emit_prologue): Don't output prologue code for those FPRs
21929 that are already separately shrink-wrapped.
21930 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
21931 that are already separately shrink-wrapped.
21933 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
21936 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
21939 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
21940 (define_peephole2 for Altivec d-form load): Add peepholes to catch
21941 cases where the register allocator uses a move and an offsettable
21942 memory operation to/from a FPR register on ISA 2.06/2.07.
21943 (define_peephole2 for Altivec d-form store): Likewise.
21945 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
21948 * config/i386/mmx.md (*mov<mode>_internal): Enable
21949 alternatives 11, 12, 13 and 14 also for 32bit targets.
21950 Remove alternatives 15, 16, 17 and 18.
21951 * config/i386/sse.md (vec_concatv2di): Change
21952 alternative (!x, *y) to (x, ?!*Yn).
21954 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
21956 * dumpfile.h (enum dump_kind): Remove stray comma.
21958 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
21960 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
21961 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
21962 predicate::num_conditions
21963 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
21964 (CHANGED): turn into predicate::changed.
21965 (agg_position_info): Move to ipa-predicate.h
21966 (add_condition, predicate::add_clause, predicate::operator &=,
21967 predicate::or_with, predicate::evaluate, predicate::probability,
21968 dump_condition, dump_clause, predicate::dump,
21969 predicate::remap_after_duplication, predicate::remap_after_inlining,
21970 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
21971 (evaluate_conditions_for_known_args): Update.
21972 (set_cond_stmt_execution_predicate): Update.
21973 * ipa-inline.h: Include ipa-predicate.h
21974 (condition, inline_param_summary, conditions, agg_position_info,
21975 predicate): Move to ipa-predicate.h
21976 * ipa-predicate.c: New file.
21977 * ipa-predicate.h: New file.
21979 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
21981 * final.c (leaf_function_p): Check we are not in a sequence.
21983 2017-05-18 Martin Liska <mliska@suse.cz>
21985 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
21986 * dumpfile.c (dump_register): Use new enum dump_kind.
21987 (get_dump_file_name): Likewise.
21988 (dump_enable_all): Likewise.
21989 (dump_switch_p_1): Likewise.
21990 (enable_rtl_dump_file): Remove usage of TDF_RTL.
21991 * dumpfile.h (enum dump_kind): New enum type.
21992 (struct dump_file_info): Create constructor and
21993 format fields and comments.
21994 * passes.c (pass_manager::register_one_dump_file):
21996 * statistics.c (statistics_early_init): Likewise.
21997 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
21998 TDF_TREE with TDF_SLIM.
21999 (gather_memory_references_ref): Likewise.
22001 2017-05-18 Martin Liska <mliska@suse.cz>
22003 * vec.h (struct vnull): Use it.
22005 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22007 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
22008 (true_predicate, false_predicate, true_predicate_p,
22009 false_predicate_p): Remove.
22010 (single_cond_predicate, not_inlined_predicate): Turn to member function
22012 (add_condition): Update.
22013 (add_clause): Turn to...
22014 (predicate::add_clause): ... this one; update; allow passing NULL
22016 (and_predicates): Turn to ...
22017 (predicate::operator &=): ... this one.
22018 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
22019 (or_predicates): Turn to ...
22020 (predicate::or_with): ... this one.
22021 (evaluate_predicate): Turn to ...
22022 (predicate::evaluate): ... this one.
22023 (predicate_probability): Turn to ...
22024 (predicate::probability): ... this one.
22025 (dump_condition): Update.
22026 (dump_predicate): Turn to ...
22027 (predicate::dump): ... this one.
22028 (account_size_time): Update.
22029 (edge_set_predicate): Update.
22030 (set_hint_predicate): UPdate.
22031 (evaluate_conditions_for_known_args): Update.
22032 (evaluate_properties_for_edge): Update.
22033 (remap_predicate_after_duplication): Turn to...
22034 (predicate::remap_after_duplication): ... this one.
22035 (remap_hint_predicate_after_duplication): Update.
22036 (inline_summary_t::duplicate): UPdate.
22037 (dump_inline_edge_summary): Update.
22038 (dump_inline_summary): Update.
22039 (set_cond_stmt_execution_predicate): Update.
22040 (set_switch_stmt_execution_predicate): Update.
22041 (compute_bb_predicates): Update.
22042 (will_be_nonconstant_expr_predicate): Update.
22043 (will_be_nonconstant_predicate): Update.
22044 (phi_result_unknown_predicate): Update.
22045 (predicate_for_phi_result): Update.
22046 (array_index_predicate): Update.
22047 (estimate_function_body_sizes): Update.
22048 (estimate_node_size_and_time): Update.
22049 (estimate_ipcp_clone_size_and_time): Update.
22050 (remap_predicate): Rename to ...
22051 (predicate::remap_after_inlining): ... this one.
22052 (remap_hint_predicate): Update.
22053 (inline_merge_summary): Update.
22054 (inline_update_overall_summary): Update.
22055 (estimate_size_after_inlining): Update.
22056 (read_predicate): Rename to ...
22057 (predicate::stream_in): ... this one.
22058 (read_inline_edge_summary): Update.
22059 (write_predicate): Rename to ...
22060 (predicate::stream_out): ... this one.
22061 (write_inline_edge_summary): Update.
22062 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
22063 (clause_t): Turn to uint32_t
22064 (predicate): Turn to class; implement constructor and operators
22066 (size_time_entry): Update.
22067 (inline_summary): Update.
22068 (inline_edge_summary): Update.
22070 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
22072 * fold-const.c (fold_binary_loc): Move transformation...
22073 * match.pd (C - X CMP X): ... here.
22075 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
22077 * config/sparc/sparc.c (sparc_option_override): Set function
22078 alignment for -mcpu=niagara7 to 64 to match the I$ line.
22079 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
22081 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
22083 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
22085 2017-05-18 Marek Polacek <polacek@redhat.com>
22088 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
22089 (pass_ubsan::execute): Call gimple_assign_single_p instead of
22090 gimple_assign_load_p.
22092 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
22094 PR middle-end/80692
22095 * real.c (do_compare): Give decimal_do_compare preference over
22096 comparing just the signs.
22098 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
22100 * doc/md.texi (Canonicalization of Instructions): Describe the
22101 canonical form of instructions that inherently set a condition
22104 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
22106 PR middle-end/80775
22107 * tree-cfg.c: Move deletion of unreachable case statements to after
22108 the merging of consecutive case labels.
22110 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
22112 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
22113 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
22114 restoring of callee-saved registers.
22116 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
22118 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
22119 * config/visium/visium.c (single_set_and_flags): Likewise.
22120 * config/visium/visium.md (Substitutions): Likewise.
22122 2017-05-17 Martin Liska <mliska@suse.cz>
22124 * cfg.c: Introduce dump_flags_t type and
22125 use it instead of int type.
22127 * cfghooks.c: Likewise.
22128 * cfghooks.h (struct cfg_hooks): Likewise.
22129 * cfgrtl.c: Likewise.
22130 * cfgrtl.h: Likewise.
22131 * cgraph.c (cgraph_node::get_body): Likewise.
22132 * coretypes.h: Likewise.
22133 * domwalk.c: Likewise.
22134 * domwalk.h: Likewise.
22135 * dumpfile.c (struct dump_option_value_info): Likewise.
22136 (dump_enable_all): Likewise.
22137 (dump_switch_p_1): Likewise.
22138 (opt_info_switch_p): Likewise.
22139 * dumpfile.h (enum tree_dump_index): Likewise.
22140 (struct dump_file_info): Likewise.
22141 * genemit.c: Likewise.
22142 * generic-match-head.c: Likewise.
22143 * gengtype.c (open_base_files): Likewise.
22144 * gimple-pretty-print.c: Likewise.
22145 * gimple-pretty-print.h: Likewise.
22146 * graph.c (print_graph_cfg): Likewise.
22147 * graphite-scop-detection.c (dot_all_sese): Likewise.
22148 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
22149 * loop-unroll.c (report_unroll): Likewise.
22150 * passes.c (pass_manager::register_one_dump_file): Likewise.
22151 * print-tree.c: Likewise.
22152 * statistics.c: Likewise.
22153 * tree-cfg.c: Likewise.
22154 * tree-cfg.h: Likewise.
22155 * tree-dfa.c: Likewise.
22156 * tree-dfa.h: Likewise.
22157 * tree-dump.c (dump_function): Likewise.
22158 * tree-dump.h (struct dump_info): Likewise.
22159 * tree-pretty-print.c: Likewise.
22160 * tree-pretty-print.h: Likewise.
22161 * tree-ssa-live.c: Likewise.
22162 * tree-ssa-live.h: Likewise.
22163 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
22164 * tree-vect-loop.c: Likewise.
22165 * tree-vect-slp.c: Likewise.
22167 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
22168 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22170 PR tree-optimization/80457
22171 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
22172 of all arguments to a statement as scalar_to_vec operations.
22173 (vectorizable_call): Adjust call to vect_model_simple_cost for
22175 (vectorizable_conversion): Likewise.
22176 (vectorizable_assignment): Likewise.
22177 (vectorizable_shift): Likewise.
22178 (vectorizable_operation): Likewise.
22179 (vectorizable_comparison): Likewise.
22180 (vect_is_simple_cond): Record the def types for operands.
22181 (vectorizable_condition): Likewise, call vect_model_simple_cost.
22182 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
22183 for statement argument count.
22185 2017-05-16 Carl Love <cel@us.ibm.com>
22187 * config/rs6000/rs6000-c: Add support for built-in functions
22188 vector unsigned long long vec_bperm (vector unsigned long long,
22189 vector unsigned char)
22190 vector signed long long vec_mule (vector signed int,
22192 vector unsigned long long vec_mule (vector unsigned int,
22193 vector unsigned int)
22194 vector signed long long vec_mulo (vector signed int,
22196 vector unsigned long long vec_mulo (vector unsigned int,
22197 vector unsigned int)
22198 vector signed char vec_sldw (vector signed char,
22199 vector signed char,
22201 vector unsigned char vec_sldw (vector unsigned char,
22202 vector unsigned char,
22204 vector signed short vec_sldw (vector signed short,
22205 vector signed short,
22207 vector unsigned short vec_sldw (vector unsigned short,
22208 vector unsigned short,
22210 vector signed int vec_sldw (vector signed int,
22213 vector unsigned int vec_sldw (vector unsigned int,
22214 vector unsigned int,
22216 vector signed long long vec_sldw (vector signed long long,
22217 vector signed long long,
22219 vector unsigned long long vec_sldw (vector unsigned long long,
22220 vector unsigned long long,
22222 * config/rs6000/rs6000-c: Add support for built-in functions
22223 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
22224 * config/rs6000/altivec.h: Add defintion for vec_sldw.
22225 * doc/extend.texi: Update the built-in documentation for the
22226 new built-in functions.
22228 2017-05-16 Marek Polacek <polacek@redhat.com>
22232 * tree.c (save_expr): Don't fold the expression.
22234 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
22236 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
22237 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
22238 and (?*y,m). Update insn attributes.
22240 2017-05-16 Martin Liska <mliska@suse.cz>
22242 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
22243 flags argument of print_gimple_stmt, print_gimple_expr,
22244 print_generic_stmt and print_generic_expr.
22245 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
22246 * coretypes.h: Likewise.
22247 * except.c (dump_eh_tree): Likewise.
22248 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
22249 * gimple-pretty-print.h: Likewise.
22250 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
22251 (backprop::push_to_worklist): Likewise.
22252 (backprop::pop_from_worklist): Likewise.
22253 (backprop::process_use): Likewise.
22254 (backprop::intersect_uses): Likewise.
22255 (note_replacement): Likewise.
22256 * gimple-ssa-store-merging.c
22257 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
22258 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
22259 (pass_store_merging::execute): Likewise.
22260 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
22261 (ssa_base_cand_dump_callback): Likewise.
22262 (dump_incr_vec): Likewise.
22263 (replace_refs): Likewise.
22264 (replace_mult_candidate): Likewise.
22265 (create_add_on_incoming_edge): Likewise.
22266 (create_phi_basis): Likewise.
22267 (insert_initializers): Likewise.
22268 (all_phi_incrs_profitable): Likewise.
22269 (introduce_cast_before_cand): Likewise.
22270 (replace_one_candidate): Likewise.
22271 * gimplify.c (gimplify_expr): Likewise.
22272 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
22273 (set_rename): Likewise.
22274 (rename_uses): Likewise.
22275 (copy_loop_phi_nodes): Likewise.
22276 (add_close_phis_to_merge_points): Likewise.
22277 (copy_loop_close_phi_args): Likewise.
22278 (copy_cond_phi_args): Likewise.
22279 (graphite_copy_stmts_from_block): Likewise.
22280 (translate_pending_phi_nodes): Likewise.
22281 * graphite-poly.c (print_pdr): Likewise.
22282 (dump_gbb_cases): Likewise.
22283 (dump_gbb_conditions): Likewise.
22284 (print_scop_params): Likewise.
22285 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
22286 (build_cross_bb_scalars_use): Likewise.
22287 (gather_bbs::before_dom_children): Likewise.
22288 * hsa-dump.c (dump_hsa_immed): Likewise.
22289 * ipa-cp.c (print_ipcp_constant_value): Likewise.
22290 (get_replacement_map): Likewise.
22291 * ipa-inline-analysis.c (dump_condition): Likewise.
22292 (estimate_function_body_sizes): Likewise.
22293 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
22294 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
22295 * ipa-prop.c (ipa_dump_param): Likewise.
22296 (ipa_print_node_jump_functions_for_edge): Likewise.
22297 (ipa_modify_call_arguments): Likewise.
22298 (ipa_modify_expr): Likewise.
22299 (ipa_dump_param_adjustments): Likewise.
22300 (ipa_dump_agg_replacement_values): Likewise.
22301 (ipcp_modif_dom_walker::before_dom_children): Likewise.
22302 * ipa-pure-const.c (check_stmt): Likewise.
22303 (pass_nothrow::execute): Likewise.
22304 * ipa-split.c (execute_split_functions): Likewise.
22305 * omp-offload.c (dump_oacc_loop_part): Likewise.
22306 (dump_oacc_loop): Likewise.
22307 * trans-mem.c (tm_log_emit): Likewise.
22308 (tm_memopt_accumulate_memops): Likewise.
22309 (dump_tm_memopt_set): Likewise.
22310 (dump_tm_memopt_transform): Likewise.
22311 * tree-cfg.c (gimple_verify_flow_info): Likewise.
22312 (print_loop): Likewise.
22313 * tree-chkp-opt.c (chkp_print_addr): Likewise.
22314 (chkp_gather_checks_info): Likewise.
22315 (chkp_get_check_result): Likewise.
22316 (chkp_remove_check_if_pass): Likewise.
22317 (chkp_use_outer_bounds_if_possible): Likewise.
22318 (chkp_reduce_bounds_lifetime): Likewise.
22319 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
22320 (chkp_mark_completed_bounds): Likewise.
22321 (chkp_register_incomplete_bounds): Likewise.
22322 (chkp_mark_invalid_bounds): Likewise.
22323 (chkp_maybe_copy_and_register_bounds): Likewise.
22324 (chkp_build_returned_bound): Likewise.
22325 (chkp_get_bound_for_parm): Likewise.
22326 (chkp_build_bndldx): Likewise.
22327 (chkp_get_bounds_by_definition): Likewise.
22328 (chkp_generate_extern_var_bounds): Likewise.
22329 (chkp_get_bounds_for_decl_addr): Likewise.
22330 * tree-chrec.c (chrec_apply): Likewise.
22331 * tree-data-ref.c (dump_data_reference): Likewise.
22332 (dump_subscript): Likewise.
22333 (dump_data_dependence_relation): Likewise.
22334 (analyze_overlapping_iterations): Likewise.
22335 * tree-inline.c (expand_call_inline): Likewise.
22336 (tree_function_versioning): Likewise.
22337 * tree-into-ssa.c (dump_defs_stack): Likewise.
22338 (dump_currdefs): Likewise.
22339 (dump_names_replaced_by): Likewise.
22340 (dump_update_ssa): Likewise.
22341 (update_ssa): Likewise.
22342 * tree-object-size.c (pass_object_sizes::execute): Likewise.
22343 * tree-parloops.c (build_new_reduction): Likewise.
22344 (try_create_reduction_list): Likewise.
22345 (ref_conflicts_with_region): Likewise.
22346 (oacc_entry_exit_ok_1): Likewise.
22347 (oacc_entry_exit_single_gang): Likewise.
22348 * tree-pretty-print.h: Likewise.
22349 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
22350 (get_scalar_evolution): Likewise.
22351 (add_to_evolution): Likewise.
22352 (get_loop_exit_condition): Likewise.
22353 (analyze_evolution_in_loop): Likewise.
22354 (analyze_initial_condition): Likewise.
22355 (analyze_scalar_evolution): Likewise.
22356 (instantiate_scev): Likewise.
22357 (number_of_latch_executions): Likewise.
22358 (gather_chrec_stats): Likewise.
22359 (final_value_replacement_loop): Likewise.
22360 (scev_const_prop): Likewise.
22361 * tree-sra.c (dump_access): Likewise.
22362 (disqualify_candidate): Likewise.
22363 (create_access): Likewise.
22364 (reject): Likewise.
22365 (maybe_add_sra_candidate): Likewise.
22366 (create_access_replacement): Likewise.
22367 (analyze_access_subtree): Likewise.
22368 (analyze_all_variable_accesses): Likewise.
22369 (sra_modify_assign): Likewise.
22370 (initialize_constant_pool_replacements): Likewise.
22371 (find_param_candidates): Likewise.
22372 (decide_one_param_reduction): Likewise.
22373 (replace_removed_params_ssa_names): Likewise.
22374 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
22375 * tree-ssa-copy.c (dump_copy_of): Likewise.
22376 (copy_prop_visit_cond_stmt): Likewise.
22377 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
22378 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
22379 (record_equivalences_from_stmt): Likewise.
22380 * tree-ssa-dse.c (compute_trims): Likewise.
22381 (delete_dead_call): Likewise.
22382 (delete_dead_assignment): Likewise.
22383 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
22384 (forward_propagate_into_cond): Likewise.
22385 (pass_forwprop::execute): Likewise.
22386 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
22387 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
22389 (move_computations_worker): Likewise.
22390 (execute_sm): Likewise.
22391 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
22392 (remove_exits_and_undefined_stmts): Likewise.
22393 (remove_redundant_iv_tests): Likewise.
22394 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
22395 (adjust_iv_update_pos): Likewise.
22396 * tree-ssa-math-opts.c (bswap_replace): Likewise.
22397 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
22398 (value_replacement): Likewise.
22399 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
22400 * tree-ssa-pre.c (print_pre_expr): Likewise.
22401 (get_representative_for): Likewise.
22402 (create_expression_by_pieces): Likewise.
22403 (insert_into_preds_of_block): Likewise.
22404 (eliminate_insert): Likewise.
22405 (eliminate_dom_walker::before_dom_children): Likewise.
22406 (eliminate): Likewise.
22407 (remove_dead_inserted_code): Likewise.
22408 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
22409 * tree-ssa-reassoc.c (get_rank): Likewise.
22410 (eliminate_duplicate_pair): Likewise.
22411 (eliminate_plus_minus_pair): Likewise.
22412 (eliminate_not_pairs): Likewise.
22413 (undistribute_ops_list): Likewise.
22414 (eliminate_redundant_comparison): Likewise.
22415 (update_range_test): Likewise.
22416 (optimize_range_tests_var_bound): Likewise.
22417 (optimize_vec_cond_expr): Likewise.
22418 (rewrite_expr_tree): Likewise.
22419 (rewrite_expr_tree_parallel): Likewise.
22420 (linearize_expr): Likewise.
22421 (break_up_subtract): Likewise.
22422 (linearize_expr_tree): Likewise.
22423 (attempt_builtin_powi): Likewise.
22424 (attempt_builtin_copysign): Likewise.
22425 (transform_stmt_to_copy): Likewise.
22426 (transform_stmt_to_multiply): Likewise.
22427 (dump_ops_vector): Likewise.
22428 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
22429 (print_scc): Likewise.
22430 (set_ssa_val_to): Likewise.
22431 (visit_reference_op_store): Likewise.
22432 (visit_use): Likewise.
22433 (sccvn_dom_walker::before_dom_children): Likewise.
22434 (run_scc_vn): Likewise.
22435 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
22437 (expr_hash_elt::print): Likewise.
22438 (const_and_copies::pop_to_marker): Likewise.
22439 (const_and_copies::record_const_or_copy_raw): Likewise.
22440 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
22441 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
22442 (dump_predicates): Likewise.
22443 (find_uninit_use): Likewise.
22444 (warn_uninitialized_phi): Likewise.
22445 (pass_late_warn_uninitialized::execute): Likewise.
22446 * tree-ssa.c (verify_vssa): Likewise.
22447 (verify_ssa): Likewise.
22448 (maybe_optimize_var): Likewise.
22449 * tree-vrp.c (dump_value_range): Likewise.
22450 (dump_all_value_ranges): Likewise.
22451 (dump_asserts_for): Likewise.
22452 (register_edge_assert_for_2): Likewise.
22453 (vrp_visit_cond_stmt): Likewise.
22454 (vrp_visit_switch_stmt): Likewise.
22455 (vrp_visit_stmt): Likewise.
22456 (vrp_visit_phi_node): Likewise.
22457 (simplify_cond_using_ranges_1): Likewise.
22458 (fold_predicate_in): Likewise.
22459 (evrp_dom_walker::before_dom_children): Likewise.
22460 (evrp_dom_walker::push_value_range): Likewise.
22461 (evrp_dom_walker::pop_value_range): Likewise.
22462 (execute_early_vrp): Likewise.
22464 2017-05-16 Richard Biener <rguenther@suse.de>
22466 * dwarf2out.c (loc_list_from_tree_1): Do not create
22467 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
22469 2017-05-16 Richard Biener <rguenther@suse.de>
22471 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
22473 (note_variable_value_in_expr): If we resolved the decl ref
22474 do not push to the stack.
22476 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
22478 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
22479 operations in fast-math mode.
22480 (vaddq_f16): Likewise.
22481 (vmul_f16): Likewise.
22482 (vmulq_f16): Likewise.
22483 (vsub_f16): Likewise.
22484 (vsubq_f16): Likewise.
22485 * config/arm/neon.md (add<mode>3): New.
22487 (fma:<VH:mode>3): New. Also remove outdated comment.
22490 2017-05-16 Martin Liska <mliska@suse.cz>
22494 * ipa-devirt.c (warn_types_mismatch): Fix typo.
22495 (odr_types_equivalent_p): Likewise.
22497 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
22499 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
22501 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
22504 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
22505 non-interunit SSE move alternatives with '?'.
22506 (zero-extendsidi peephole2): New peephole to skip intermediate
22507 general register in SSE zero-extend sequence.
22509 2017-05-15 Jeff Law <law@redhat.com>
22511 * reorg.c (relax_delay_slots): Create a new variable to hold
22512 the temporary target rather than clobbering TARGET_LABEL.
22514 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
22515 missing argument to extract_bit_field call.
22516 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
22518 2017-05-15 Martin Liska <mliska@suse.cz>
22521 * gcc.c (process_command): Do not allow empty argument of -o option.
22523 2017-05-15 Renlin Li <renlin.li@arm.com>
22525 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
22526 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
22527 * config/aarch64/constraints.md (Usf): Add long call check.
22528 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
22529 (call_value): Likewise.
22530 (sibcall): Likewise.
22531 (sibcall_value): Likewise.
22533 (call_value_insn): New.
22534 (sibcall_insn): Update rtx pattern.
22535 (sibcall_value_insn): Likewise.
22536 (call_internal): Remove.
22537 (call_value_internal): Likewise.
22538 (sibcall_internal): Likewise.
22539 (sibcall_value_internal): Likewise.
22540 (call_reg): Likewise.
22541 (call_symbol): Likewise.
22542 (call_value_reg): Likewise.
22543 (call_value_symbol): Likewise.
22545 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
22548 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
22550 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
22552 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
22553 compatible with CCGOCmode and with CCZmode.
22555 2017-05-14 Martin Sebor <msebor@redhat.com>
22557 PR middle-end/77671
22558 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
22559 (gimple_fold_builtin_snprintf): Same.
22560 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
22561 (gimple_fold_builtin_snprintf): Same.
22562 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
22563 of character types.
22564 (is_call_safe): New function.
22565 (try_substitute_return_value): Call it.
22566 (try_simplify_call): New function.
22567 (pass_sprintf_length::handle_gimple_call): Call it.
22569 2017-05-14 Martin Sebor <msebor@redhat.com>
22571 PR middle-end/80669
22572 * builtins.c (expand_builtin_stpncpy): Simplify.
22574 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
22576 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
22577 * config/i386/i386.h
22578 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
22579 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
22580 (struct machine_function): Add new members call_ms2sysv,
22581 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
22582 (struct machine_frame_state): New fields sp_realigned and
22583 sp_realigned_offset.
22584 * config/i386/i386.c
22585 (enum xlogue_stub): New enum.
22586 (enum xlogue_stub_sets): New enum.
22587 (class xlogue_layout): New class.
22588 (struct ix86_frame): New fields stack_realign_allocate_offset,
22589 stack_realign_offset and outlined_save_offset. Modify comments to
22590 detail stack layout when using out-of-line stubs.
22591 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
22592 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
22593 -mcall-ms2sysv-xlogues.
22594 (stub_managed_regs): New static variable.
22595 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
22596 registers managed by out-of-line stub.
22597 (disable_call_ms2sysv_xlogues): New function.
22598 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
22599 m->call_ms2sysv when appropriate and compute frame layout for
22601 (sp_valid_at, fp_valid_at): New inline functions.
22602 (choose_basereg): New function.
22603 (choose_baseaddr): Add align parameter, use choose_basereg and modify
22605 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
22606 Use align parameter of choose_baseaddr to generated aligned SSE movs
22608 (pro_epilogue_adjust_stack): Modify to track
22609 machine_frame_state::sp_realigned.
22610 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
22611 (ix86_nsaved_sseregs): Likewise.
22612 (ix86_emit_save_regs): Likewise.
22613 (ix86_emit_save_regs_using_mov): Likewise.
22614 (ix86_emit_save_sse_regs_using_mov): Likewise.
22615 (get_scratch_register_on_entry): Likewise.
22616 (gen_frame_set): New function.
22617 (gen_frame_load): Likewise.
22618 (gen_frame_store): Likewise.
22619 (emit_outlined_ms2sysv_save): Likewise.
22620 (emit_outlined_ms2sysv_restore): Likewise.
22621 (ix86_expand_prologue): Modify stack re-alignment code and call
22622 emit_outlined_ms2sysv_save when appropriate.
22623 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
22624 parameter rtx_insn *insn, which allows the function to be used to only
22625 generate the notes.
22626 (ix86_expand_epilogue): Modify validity checks of frame and stack
22627 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
22628 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
22629 * config/i386/predicates.md
22630 (save_multiple): New predicate.
22631 (restore_multiple): Likewise.
22632 * config/i386/sse.md
22633 (save_multiple<mode>): New pattern.
22634 (save_multiple_realign<mode>): Likewise.
22635 (restore_multiple<mode>): Likewise.
22636 (restore_multiple_and_return<mode>): Likewise.
22637 (restore_multiple_leave_return<mode>): Likewise.
22638 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
22640 2017-05-14 Julia Koval <julia.koval@intel.com>
22642 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
22643 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
22644 (__builtin_ia32_xsetbv): New builtins.
22645 * config/i386/i386.c (ix86_expand_special_args_builtin):
22647 (ix86_expand_builtin): Special expand for new intrinsics.
22648 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
22649 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
22650 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
22652 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22654 * cfganal.c (inverted_post_order_compute): Change argument type
22656 * cfganal.h (inverted_post_order_compute): Adjust prototype.
22657 * df-core.c (rest_of_handle_df_initialize): Adjust.
22658 (rest_of_handle_df_finish): Likewise.
22659 (df_analyze_1): Likewise.
22660 (df_analyze): Likewise.
22661 (loop_inverted_post_order_compute): Change argument to be a vec *.
22662 (df_analyze_loop): Adjust.
22663 (df_get_n_blocks): Likewise.
22664 (df_get_postorder): Likewise.
22665 * df.h (struct df_d): Change field to be a vec.
22666 * lcm.c (compute_laterin): Adjust.
22667 (compute_available): Likewise.
22668 * lra-lives.c (lra_create_live_ranges_1): Likewise.
22669 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
22670 * tree-ssa-pre.c (compute_antic): Likewise.
22672 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22674 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
22675 (depth_first_search::depth_first_search): Change structure init
22676 function to this constructor.
22677 (depth_first_search::add_bb): Rename function to this member.
22678 (depth_first_search::execute): Likewise.
22679 (flow_dfs_compute_reverse_finish): Adjust.
22681 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22683 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
22684 (longest_simple_path): Likewise.
22685 * shrink-wrap.c (spread_components): Likewise.
22686 (disqualify_problematic_components): Likewise.
22687 (emit_common_heads_for_components): Likewise.
22688 (emit_common_tails_for_components): Likewise.
22689 (insert_prologue_epilogue_for_components): Likewise.
22691 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22693 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
22696 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22698 * df-core.c (df_set_blocks): Start using auto_bitmap.
22699 (df_compact_blocks): Likewise.
22700 * df-problems.c (df_rd_confluence_n): Likewise.
22701 * df-scan.c (df_insn_rescan_all): Likewise.
22702 (df_process_deferred_rescans): Likewise.
22703 (df_update_entry_block_defs): Likewise.
22704 (df_update_exit_block_uses): Likewise.
22705 (df_entry_block_bitmap_verify): Likewise.
22706 (df_exit_block_bitmap_verify): Likewise.
22707 (df_scan_verify): Likewise.
22708 * lra-constraints.c (lra_constraints): Likewise.
22709 (undo_optional_reloads): Likewise.
22710 (lra_undo_inheritance): Likewise.
22711 * lra-remat.c (calculate_gen_cands): Likewise.
22712 (do_remat): Likewise.
22713 * lra-spills.c (assign_spill_hard_regs): Likewise.
22714 (spill_pseudos): Likewise.
22715 * tree-ssa-pre.c (bitmap_set_and): Likewise.
22716 (bitmap_set_subtract_values): Likewise.
22718 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22720 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
22721 management with auto_bitmap.
22722 (fix_inter_tick): Likewise.
22723 (fix_recovery_deps): Likewise.
22724 * ira.c (add_store_equivs): Likewise.
22725 (find_moveable_pseudos): Likewise.
22726 (split_live_ranges_for_shrink_wrap): Likewise.
22727 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
22728 (rtx_reuse_manager::seen_def_p): Likewise.
22729 (rtx_reuse_manager::set_seen_def): Likewise.
22730 * print-rtl.h (class rtx_reuse_manager): Likewise.
22732 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22734 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
22736 (migrate_btr_def): Likewise.
22737 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
22738 * df-core.c (loop_post_order_compute): Likewise.
22739 (loop_inverted_post_order_compute): Likewise.
22740 * hsa-common.h: Likewise.
22741 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
22742 * init-regs.c (initialize_uninitialized_regs): Likewise.
22743 * ipa-inline.c (resolve_noninline_speculation): Likewise.
22744 (inline_small_functions): Likewise.
22745 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
22746 * ira.c (combine_and_move_insns): Likewise.
22747 (build_insn_chain): Likewise.
22748 * loop-invariant.c (find_invariants): Likewise.
22749 * lower-subreg.c (propagate_pseudo_copies): Likewise.
22750 * predict.c (tree_predict_by_opcode): Likewise.
22751 (predict_paths_leading_to): Likewise.
22752 (predict_paths_leading_to_edge): Likewise.
22753 (estimate_loops_at_level): Likewise.
22754 (estimate_loops): Likewise.
22755 * shrink-wrap.c (try_shrink_wrapping): Likewise.
22756 (spread_components): Likewise.
22757 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
22758 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
22759 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
22760 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
22761 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
22762 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
22763 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
22764 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
22765 (mark_threaded_blocks): Likewise.
22766 (thread_through_all_blocks): Likewise.
22767 * tree-ssa.c (verify_ssa): Likewise.
22768 (execute_update_addresses_taken): Likewise.
22769 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
22771 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22773 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
22775 (post_order_compute): Likewise.
22776 (inverted_post_order_compute): Likewise.
22777 (pre_and_rev_post_order_compute_fn): Likewise.
22779 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22781 * genrecog.c (int_set::int_set): Explicitly construct our
22782 auto_vec base class.
22783 * vec.h (auto_vec::auto_vec): New constructor.
22785 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22787 * bitmap.h (class auto_bitmap): New constructor taking
22788 bitmap_obstack * argument.
22790 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22792 * bitmap.h (class auto_bitmap): Change type of m_bits to
22793 bitmap_head, and adjust ctor / dtor and member operators.
22795 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
22797 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
22798 when returned register mode doesn't match original mode.
22800 2017-05-12 Jeff Law <law@redhat.com>
22801 Jakub Jelinek <jakub@redhat.com>
22803 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
22804 we look for cc setter after the compare-elim changes.
22805 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
22806 within the vector to match what compare-elim now expects.
22807 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
22808 (xorsi3_flags, one_cmplsi2_flags): Likewise.
22810 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
22811 after the compare-elim changes.
22812 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
22813 the vector to match what compare-elim now expects.
22814 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
22815 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
22816 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
22817 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
22818 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
22820 * config/visium/visium.c (single_set_and_flags): Fix where
22821 we look for cc setter after the compare-elim changes.
22822 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
22823 with the vector to match what compare-elim now expects.
22824 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
22825 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
22826 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
22827 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
22828 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
22829 (neg<mode>2_insn_set_overflow): Likewise.
22831 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
22833 PR middle-end/79794
22834 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
22835 maybe_expand_insn call, set ops[0].target. If still set after call,
22836 set alt_rtl. Add extra arg to recursive calls.
22837 (extract_bit_field): Add alt_rtl argument. Pass to
22839 * expmed.h (extract_bit_field): Fix prototype.
22840 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
22841 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
22842 to extract_bit_field_calls.
22843 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
22844 Pass alt_rtl to extract_bit_field calls.
22845 * calls.c (store_unaligned_arguments_into_psuedos)
22846 load_register_parameters): Pass extra NULL to extract_bit_field calls.
22847 * optabs.c (maybe_legitimize_operand): Clear op->target when call
22849 * optabs.h (struct expand_operand): Add target bitfield.
22851 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
22853 * compare-elim.c (try_eliminate_compare): Canonicalize
22854 operation with embedded compare to
22855 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
22856 (set (reg) (operation)].
22858 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
22860 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
22863 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
22864 cost of adding a carry flag for ADC instruction.
22865 [case MINUS]: Ignore the cost of subtracting a carry flag
22866 for SBB instruction.
22868 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
22870 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
22872 * config/rs6000/bmiintrin.h: New file.
22873 * config/rs6000/bmi2intrin.h: New file.
22874 * config/rs6000/x86intrin.h: New file.
22876 2017-05-12 Jeff Law <law@redhat.com>
22878 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
22881 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
22883 PR middle-end/80707
22884 * tree-cfg.c: Remove cfg edges of unreachable case statements.
22886 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
22888 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
22889 early expansion of vector divide builtins.
22890 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
22891 builtins identified as having unsigned arguments.
22893 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
22895 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
22896 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
22897 expansion of vector logical operations (and, andc, or, xor,
22900 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
22902 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
22903 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
22905 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
22907 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
22908 early GIMPLE expansion of vector multiplies.
22910 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22912 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
22913 TARGET_HAVE_MOVT conditional.
22914 (movt splitter): Likewise.
22916 2017-05-12 Richard Biener <rguenther@suse.de>
22918 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
22919 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
22920 Fold all stmts not inplace.
22922 2017-05-12 Richard Biener <rguenther@suse.de>
22924 PR tree-optimization/80713
22925 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
22926 inserted_exprs bit for not removed stmts.
22928 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
22930 PR middle-end/69921
22931 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
22932 parallelized" attribute for parallelized OpenACC kernels.
22933 * omp-offload.c (execute_oacc_device_lower): Use it.
22935 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
22936 Set "oacc kernels" attribute.
22937 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
22938 parameter. Adjust all users.
22939 (oacc_fn_attrib_kernels_p): Remove function.
22940 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
22941 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
22942 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
22943 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
22944 assert "oacc kernels" attribute is set.
22946 2017-05-11 Carl Love <cel@us.ibm.com>
22948 * config/rs6000/rs6000-c: Add support for built-in functions
22949 vector unsigned char vec_popcnt (vector signed char)
22950 vector unsigned char vec_popcnt (vector unsigned char)
22951 vector unsigned short vec_popcnt (vector signed short)
22952 vector unsigned short vec_popcnt (vector unsigned short)
22953 vector unsigned int vec_popcnt (vector signed int)
22954 vector unsigned int vec_popcnt (vector unsigned int)
22955 vector unsigned long long vec_popcnt (vector signed long long)
22956 vector unsigned long long vec_popcnt (vector unsigned long long)
22957 vector signed long long vec_slo (vector signed long long,
22958 vector signed char)
22959 vector signed long long vec_slo (vector signed long long,
22960 vector unsigned char)
22961 vector unsigned long long vec_slo (vector unsigned long long,
22962 vector signed char)
22963 vector unsigned long long vec_slo (vector unsigned long long,
22964 vector unsigned char)
22965 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
22966 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
22967 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
22968 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
22969 * doc/extend.texi: Update the built-in documentation file for the
22970 new built-in functions.
22972 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
22974 * attribs.h (sorted_attr_string): Move machine independent
22975 functions for target clone support from the i386 port to common
22976 code. Rename ix86_function_versions to common_function_versions.
22977 Rename make_name to make_unique_name.
22978 (common_function_versions): Likewise.
22979 (make_unique_name): Likewise.
22980 (make_dispatcher_decl): Likewise.
22981 (is_function_default_version): Likewise.
22982 * attribs.c (attr_strcmp): Likewise.
22983 (sorted_attr_string): Likewise.
22984 (common_function_versions): Likewise.
22985 (make_unique_name): Likewise.
22986 (make_dispatcher_decl): Likewise.
22987 (is_function_default_version): Likewise.
22988 * config/i386/i386.c (attr_strcmp): Likewise.
22989 (sorted_attr_string): Likewise.
22990 (ix86_function_versions): Likewise.
22991 (make_name): Likewise.
22992 (make_dispatcher_decl): Likewise.
22993 (is_function_default_version): Likewise.
22994 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
22996 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22999 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23000 Account for direct move costs for vec_construct of integer
23003 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
23006 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
23007 (UNSPEC_STX_ATOMIC): Ditto.
23008 (loaddi_via_sse): New insn.
23009 (storedi_via_sse): Ditto.
23010 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
23011 Update corresponding peephole2 patterns.
23012 (atomic_storedi_fpu): Ditto.
23014 2017-05-11 Julia Koval <julia.koval@intel.com>
23016 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
23017 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
23019 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
23020 (__builtin_ia32_rsqrt14ss_mask): New builtins.
23021 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
23023 2017-05-11 Nathan Sidwell <nathan@acm.org>
23025 * graphite-poly.c: Include dumpfile.h.
23027 * dumpfle.h (dump_function): Declare here ...
23028 * tree-dump.h (dump_function): ... not here.
23029 * dumpfile.c: #include tree-cfg.h.
23030 (dump_function): Move here from ...
23031 * tree-dump.c (dump_function): ... here.
23032 * gimplify.c: #include splay-tree.h, not tree-dump.h.
23033 * graphite-poly.c: Don't include tree-dump.h.
23034 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
23035 * print-tree.c: Likewise.
23036 * stor-layout.c: Likewise.
23037 * tree-nested.c: Likewise.
23039 * dumpfile.c (dump_start): Use TDF_FLAGS.
23040 (dump_enable_all): Fix TDF_KIND check thinko.
23042 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
23044 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23045 array entries to represent two legal parameterizations of the
23046 overloaded __builtin_cmpb function, as represented by the
23047 P6_OV_BUILTIN_CMPB constant.
23048 (altivec_resolve_overloaded_builtin): Add special case handling
23049 for the __builtin_cmpb function, as represented by the
23050 P6_OV_BUILTIN_CMPB constant.
23051 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
23052 (BU_P6_64BIT_2): New macro.
23053 (BU_P6_OVERLOAD_2): New macro
23054 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
23055 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
23056 (CMPB): Add overload support to represent both 32-bit and 64-bit
23057 compare-bytes function.
23058 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
23059 support for TARGET_CMPB.
23060 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
23061 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23062 documentation of the __builtin_cmpb overloaded built-in function.
23064 2017-05-11 Richard Biener <rguenther@suse.de>
23066 PR tree-optimization/80705
23067 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
23068 bases are not vectorizable.
23070 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23072 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
23073 when counting register pressure.
23075 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23077 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
23078 (iv_ca_more_deps): Renamed to ...
23079 (iv_ca_compare_deps): ... this.
23080 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
23082 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23084 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
23086 (determine_group_iv_costs): ... here.
23087 (find_inv_vars_cb): Record inv var if it's not recorded before.
23089 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23091 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
23092 (get_shiftadd_cost): Ditto.
23094 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23096 * tree-ssa-address.c: Include header file.
23097 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
23099 (add_to_parts): Refactor.
23100 (addr_to_parts): New parameter. Update use of move_hint_to_base.
23101 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
23104 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23106 PR tree-optimization/53090
23107 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
23109 (extract_cond_operands): Detect condition with IV on both sides
23110 and return COMP_IV_EXPR_2.
23111 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
23112 (rewrite_use_compare): Simplify by removing call to function
23113 extract_cond_operands.
23115 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23117 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
23118 (extract_cond_operands): Detect condition comparing against non-
23119 invariant bound and return appropriate enum value.
23120 (find_interesting_uses_cond): Update use of extract_cond_operands.
23121 Handle its return value accordingly.
23122 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
23124 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23126 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
23127 nonlinear iv_use computation in loop invariant sensitive way.
23129 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23131 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
23132 (find_iv_candidates): Call relate_compare_use_with_all_cands.
23134 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23136 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
23137 (dump_cand): Support iv_cand.inv_exprs.
23138 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
23140 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
23143 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23145 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
23147 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
23148 as local function. Include necessary header files.
23149 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
23151 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23153 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
23155 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23157 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
23158 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
23159 RSHIFT_EXPR and BIT_NOT_EXPR.
23161 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23163 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
23164 (adjust_setup_cost): New parameter supporting round up adjustment.
23165 (struct address_cost_data): Delete.
23166 (force_expr_to_var_cost): Don't bound cost with spill_cost.
23167 (split_address_cost, ptr_difference_cost): Delete.
23168 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
23169 (struct ainc_cost_data): New struct.
23170 (get_address_cost_ainc): New function.
23171 (get_address_cost, get_computation_cost): Reimplement.
23172 (determine_group_iv_cost_address): Record inv_expr for all uses of
23174 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
23175 (iv_ca_has_deps): Reimplemented to ...
23176 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
23178 (iv_ca_extend): Call iv_ca_more_deps.
23180 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23182 * tree-ssa-address.c (struct mem_address): Move to header file.
23183 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
23184 * tree-ssa-address.h (struct mem_address): Move from C file.
23185 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
23187 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23189 * tree-affine.h (aff_combination_type): New interface.
23190 (aff_combination_zero_p): Remove static.
23191 (aff_combination_const_p): New interface.
23192 (aff_combination_singleton_var_p): New interfaces.
23194 2017-05-11 Richard Biener <rguenther@suse.de>
23196 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23197 Skip unreachable blocks and destinations.
23198 (eliminate): Move stmt removal and fixup ...
23199 (fini_eliminate): ... here. Skip inserted exprs.
23200 (pass_pre::execute): Move fini_pre after fini_eliminate.
23201 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
23202 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
23203 PRE to get rid of dead code that has invalid SSA form and
23204 split critical edges again.
23206 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23208 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
23210 2017-05-11 Richard Biener <rguenther@suse.de>
23212 * passes.c (execute_function_todo): Verify loops if they are
23213 said to be up-to-date.
23214 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
23215 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
23217 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
23220 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
23221 handle calling assemble_external ourself.
23224 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
23225 modes with zero size. Enhance comment.
23227 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23229 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
23230 built-ins for vec_xl and vec_xst with short and char pointer
23233 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
23235 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
23236 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
23237 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
23238 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
23239 (_mm_maskz_min_round_ss): New intrinsics.
23240 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
23241 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
23242 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
23243 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
23244 (__builtin_ia32_minss_mask_round): New builtins.
23245 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23246 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23247 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
23249 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
23250 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
23252 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
23255 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
23257 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
23258 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
23259 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
23260 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
23261 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
23262 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
23263 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
23264 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23265 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
23266 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
23267 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
23268 (__builtin_ia32_mulss_mask_round): New builtins.
23269 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23270 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23271 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
23273 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
23274 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
23276 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
23279 2017-05-10 Julia Koval <julia.koval@intel.com>
23281 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
23282 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
23283 (_mm256_setr_m128i): New intrinsics.
23285 2017-05-10 Julia Koval <julia.koval@intel.com>
23287 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
23288 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
23289 (_mm_maskz_rcp14_ss): New intrinsics.
23290 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
23291 (__builtin_ia32_rcp14ss_mask): New builtins.
23292 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
23294 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
23296 PR tree-optimization/51513
23297 * tree-cfg.c (gimple_seq_unreachable_p): New function.
23298 (assert_unreachable_fallthru_edge_p): Use it.
23299 (group_case_labels_stmt): Likewise.
23300 * tree-cfg.h: Prototype it.
23301 * stmt.c: Include cfghooks.h and tree-cfg.h.
23302 (emit_case_dispatch_table) <gap_label>: New local variable.
23303 Use it to fill dispatch table gaps.
23304 Test for default_label before updating probabilities.
23305 (expand_case) <default_label>: Remove unneeded initialization.
23306 Test for unreachable default case statement and remove its edge.
23307 Set default_label accordingly.
23308 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
23310 2017-05-10 Carl Love <cel@us.ibm.com>
23312 * config/rs6000/rs6000-c: Add support for built-in functions
23313 vector signed char vec_neg (vector signed char)
23314 vector signed short int vec_neg (vector short int)
23315 vector signed int vec_neg (vector signed int)
23316 vector signed long long vec_neg (vector signed long long)
23317 vector float vec_neg (vector float)
23318 vector double vec_neg (vector double)
23319 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
23321 * config/rs6000/altivec.h: Add define for vec_neg
23322 * doc/extend.texi: Update the built-in documentation for the
23323 new built-in functions.
23325 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23327 PR tree-optimization/77644
23328 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
23330 2017-05-10 Nathan Sidwell <nathan@acm.org>
23332 * dumpfile.h (TDI_lang_all): New.
23333 (TDF_KIND): New. Renumber others
23334 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
23336 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
23338 (get_dump_file_name): Adjust suffix generation.
23339 (dump_enable_all): Use TDF_KIND.
23340 * doc/invoke.texi (-fdump-lang-all): Document.
23342 * dumpfile.h: Tabify.
23344 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
23347 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
23348 Move member access before delete.
23350 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
23352 * tree-inline.c (expand_call_inline): Split block at stmt
23355 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
23358 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
23359 are now unused after splitting mov{sf,sd}_hardfloat.
23360 (f32_lr2): Likewise.
23361 (f32_lm): Likewise.
23362 (f32_lm2): Likewise.
23363 (f32_li): Likewise.
23364 (f32_li2): Likewise.
23365 (f32_lv): Likewise.
23366 (f32_sr): Likewise.
23367 (f32_sr2): Likewise.
23368 (f32_sm): Likewise.
23369 (f32_sm2): Likewise.
23370 (f32_si): Likewise.
23371 (f32_si2): Likewise.
23372 (f32_sv): Likewise.
23373 (f32_dm): Likewise.
23374 (f32_vsx): Likewise.
23375 (f32_av): Likewise.
23376 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
23377 For movsf, order stores so the VSX stores occur before the GPR
23378 store which encourages the register allocator to use a traditional
23379 FPR instead of a GPR. For movsd, order the stores so that the GPR
23380 store comes before the VSX stores to allow the power6 to work.
23381 This is due to the power6 not having a 32-bit integer store
23382 instruction from a FPR.
23383 (movsf_hardfloat): Likewise.
23384 (movsd_hardfloat): Likewise.
23386 2017-05-09 Martin Sebor <msebor@redhat.com>
23388 PR translation/80280
23389 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
23392 PR translation/80280
23393 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
23394 data member added in r247778.
23395 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
23397 2017-05-09 Nathan Sidwell <nathan@acm.org>
23399 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
23401 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
23404 2017-05-09 Marek Polacek <polacek@redhat.com>
23406 * doc/invoke.texi: Fix typo.
23408 2017-05-09 Richard Biener <rguenther@suse.de>
23410 * tree-vrp.c (vrp_val_is_max): Adjust comment.
23411 (vrp_val_is_min): Likewise.
23412 (set_value_range_to_value): Likewise.
23413 (set_value_range_to_nonnegative): Likewise.
23414 (gimple_assign_nonzero_p): Likewise.
23415 (gimple_stmt_nonzero_p): Likewise.
23416 (vrp_int_const_binop): Likewise. Remove unreachable case.
23417 (adjust_range_with_scev): Adjust comments.
23418 (compare_range_with_value): Likewise.
23419 (extract_range_from_phi_node): Likewise.
23420 (test_for_singularity): Likewise.
23422 2017-05-09 Richard Biener <rguenther@suse.de>
23424 * tree-vrp.c (get_single_symbol): Add assert that we don't
23425 get overflowed constants as invariant part.
23426 (compare_values_warnv): Add comment before the TREE_NO_WARNING
23427 checks. Use wi::cmp instead of recursing for integer constants.
23428 (compare_values): Just ignore whether we assumed undefined
23429 overflow instead of failing the compare.
23430 (extract_range_for_var_from_comparison_expr): Add comment before the
23431 TREE_NO_WARNING sets.
23432 (test_for_singularity): Likewise.
23433 (extract_range_from_comparison): Do not disable optimization
23434 when we assumed undefined overflow.
23435 (extract_range_basic): Remove init of unused var.
23437 2017-05-09 Richard Biener <rguenther@suse.de>
23439 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
23440 (extract_range_from_multiplicative_op_1): Adjust.
23441 (extract_range_from_binary_expr_1): Use int_const_binop.
23443 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
23446 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
23447 rs6000_store_data_bypass_p in seven define_bypass directives and
23448 in several comments.
23449 * config/rs6000/rs6000-protos.h: Add prototype for
23450 rs6000_store_data_bypass_p function.
23451 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
23452 function implements slightly different (rs6000-specific) semantics
23453 than store_data_bypass_p, returning false rather than aborting
23454 with assertion error when arguments do not satisfy the
23455 requirements of store data bypass.
23456 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
23457 rs6000_store_data_bypass_p.
23459 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
23461 * config/xtensa/xtensa-protos.h
23462 (xtensa_initial_elimination_offset): New declaration.
23463 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
23464 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
23465 macro definition, add case for FRAME_POINTER_REGNUM when
23466 FRAME_GROWS_DOWNWARD.
23467 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
23468 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
23469 xtensa_initial_elimination_offset.
23471 2017-05-08 Nathan Sidwell <nathan@acm.org>
23473 * doc/invoke.texi: Alphabetize -fdump options.
23475 2017-05-08 Martin Sebor <msebor@redhat.com>
23477 PR translation/80280
23478 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
23480 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
23482 * target.def (compute_frame_layout): New optional target hook.
23483 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
23484 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
23485 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
23487 * reload1.c (verify_initial_elim_offsets): Likewise.
23488 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
23489 (use_simple_return_p): Call arm_compute_frame_layout if needed.
23490 (arm_get_frame_offsets): Split up into this ...
23491 (arm_compute_frame_layout): ... and this function.
23493 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
23495 * config/aarch64/constraints.md (Usa): New constraint.
23496 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
23498 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
23500 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
23501 with_multilib_list after it has been checked.
23503 2017-05-08 Richard Biener <rguenther@suse.de>
23505 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
23506 (bitmap_set_subtract_values): Likewise.
23508 2017-05-08 Richard Biener <rguenther@suse.de>
23510 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
23511 (gimple_assign_nonzero): ... this and remove strict_overflow_p
23513 (gimple_stmt_nonzero_warnv_p): Rename to ...
23514 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
23516 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
23517 (extract_range_basic): Adjust, do not disable propagation on
23518 strict overflow sensitive simplification.
23519 (vrp_visit_cond_stmt): Likewise.
23521 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
23523 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
23524 body size unconditionally.
23526 2017-05-07 Jeff Law <law@redhat.com>
23529 2017-05-06 Jeff Law <law@redhat.com>
23530 PR tree-optimization/78496
23531 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
23534 PR tree-optimization/78496
23535 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
23536 (simplify_stmt_using_ranges): Call it.
23537 (vrp_dom_walker::before_dom_children): Extract equivalences
23538 from an ASSERT_EXPR with an equality comparison against a
23541 2017-05-06 Jeff Law <law@redhat.com>
23543 PR tree-optimization/78496
23544 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
23547 PR tree-optimization/78496
23548 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
23549 (simplify_stmt_using_ranges): Call it.
23550 (vrp_dom_walker::before_dom_children): Extract equivalences
23551 from an ASSERT_EXPR with an equality comparison against a
23554 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
23556 * lra-constraints.c (lra_copy_reg_equiv): New function.
23557 (split_reg): Use it to copy equivalence information from the
23558 original register to the spill register.
23560 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
23562 PR rtl-optimization/75964
23563 * simplify-rtx.c (simplify_const_relational_operation): Remove
23564 invalid handling of comparisons of integer ABS.
23566 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
23568 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
23569 initialize to zero.
23570 (init_regs): Remove declaration.
23571 (function_arg_advance_32): Initialize error_p as boolean variable.
23573 2017-05-05 Nathan Sidwell <nathan@acm.org>
23575 * store-motion.c (remove_reachable_equiv_notes): Reformat long
23576 lines. Use for (;;).
23578 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23580 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
23581 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
23582 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
23583 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
23584 VF=2 that require versioning.
23586 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23588 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
23591 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23593 * diagnostic.h (diagnostic_override_option_index): Convert from
23594 macro to inline function.
23596 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23598 * diagnostic.c (last_module_changed_p): New function.
23599 (set_last_module): New function.
23600 (diagnostic_report_current_module): Convert macro usage to
23601 the above functions.
23602 * diagnostic.h (diagnostic_context::last_module): Strengthen
23603 from const line_map * to const line_map_ordinary *.
23604 (diagnostic_last_module_changed): Delete macro.
23605 (diagnostic_set_last_module): Delete macro.
23607 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23609 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
23610 with diagnostic_report_diagnostic.
23611 (diagnostic_n_impl_richloc): Likewise.
23612 * diagnostic.h (report_diagnostic): Delete macro.
23613 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
23614 with diagnostic_report_diagnostic.
23615 * substring-locations.c (format_warning_va): Likewise.
23617 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23619 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
23620 save/restor of format_spec. Move option-printing code to...
23621 (print_option_information): ...this new function, and
23622 reimplement by simply printing to the pretty_printer,
23623 rather than appending to the format string.
23625 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23627 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
23628 handling logic into...
23629 (update_effective_level_from_pragmas): ...this new function.
23631 2017-05-04 Andrew Waterman <andrew@sifive.com>
23633 * config/riscv/riscv.opt (mstrict-align): New option.
23634 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
23635 (SLOW_UNALIGNED_ACCESS): Define.
23636 (riscv_slow_unaligned_access): Declare.
23637 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
23639 (riscv_slow_unaligned_access): New variable.
23640 (rocket_tune_info): Set slow_unaligned_access to true.
23641 (optimize_size_tune_info): Set slow_unaligned_access to false.
23642 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
23643 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
23644 (riscv_option_override): Set riscv_slow_unaligned_access.
23645 * doc/invoke.texi: Add -mstrict-align to RISC-V.
23647 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
23649 * config/riscv/riscv.md: Unify indentation.
23651 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
23656 * config/rs6000/rs6000.md (u code attribute): Add FIX and
23658 (extendsi<mode>2): Add support for doing sign extension via
23659 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
23660 don't have ISA 3.0 instructions.
23661 (extendsi<mode>2 splitter): Likewise.
23662 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
23663 generate the normal insns since SImode can now go in vector
23664 registers. Disallow the special UNSPECs needed for previous
23665 machines to hide SImode being used. Add new insns
23666 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
23667 (fix_trunc<mode>si2_stfiwx): Likewise.
23668 (fix_trunc<mode>si2_internal): Likewise.
23669 (fixuns_trunc<mode>si2): Likewise.
23670 (fixuns_trunc<mode>si2_stfiwx): Likewise.
23671 (fctiw<u>z_<mode>_smallint): Likewise.
23672 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
23673 of floating point to 32-bit integer from doing a direct move to
23674 the GPR registers to do a store.
23675 (fctiwz_<mode>): Break long line.
23677 2017-05-05 Bin Cheng <bin.cheng@arm.com>
23679 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
23680 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
23681 (addr_list, addr_offset_valid_p): New.
23682 (split_address_groups): Check offset validity with above function.
23683 (gt-tree-ssa-loop-ivopts.h): Include header file.
23685 2017-05-05 Nathan Sidwell <nathan@acm.org>
23687 * config.gcc (arm*-*-*): Add missing 'fi'.
23689 2017-05-05 Steve Ellcey <sellcey@cavium.com>
23691 * doc/invoke.texi (-fopt-info): Explicitly say order of options
23692 included in -fopt-info does not matter.
23693 * doc/optinfo.texi (-fopt-info): Fix description of default
23694 behavour. Explicitly say order of options included in -fopt-info
23697 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
23699 * config.gcc: Allow combinations of aprofile and rmprofile values for
23700 --with-multilib-list.
23701 * config/arm/t-multilib: New file.
23702 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
23703 variables. Remove setting of ISA and floating-point ABI in
23704 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
23705 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
23706 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
23707 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
23709 * config/arm/t-rmprofile: Likewise except for the matches changes.
23710 * doc/install.texi (--with-multilib-list): Document the combination of
23711 aprofile and rmprofile values and warn about pitfalls in doing that.
23713 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
23715 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
23716 (movdi_aarch64): Likewise.
23718 2017-05-05 Jakub Jelinek <jakub@redhat.com>
23720 PR tree-optimization/80632
23721 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
23723 (build_arrays): Initialize it for virtual phis.
23724 (fix_phi_nodes): Use it for virtual phis.
23726 PR tree-optimization/80558
23727 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
23728 [x, y] op z into [x op, y op z] for op & or | if conditions
23731 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
23732 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23735 * config/arm/arm.md (use_literal_pool): Remove.
23736 (64-bit immediate split): No longer takes cost into consideration
23737 if arm_disable_literal_pool is enabled.
23738 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
23739 used when arm_disable_literal_pool is enabled.
23740 (arm_max_const_double_inline_cost): Remove use of
23741 arm_disable_literal_pool.
23742 (push_minipool_fix): Add assert.
23743 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
23744 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
23745 (no_literal_pool_sf_immediate): New.
23747 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23749 PR tree-optimization/80613
23750 * tree-ssa-dce.c (propagate_necessity): Remove cases for
23751 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
23753 2017-05-05 Richard Biener <rguenther@suse.de>
23755 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
23757 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
23759 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
23760 of this flag from insn conditions due to removal from r247495.
23762 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
23764 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
23766 (arm_early_store_addr_dep_ptr): Likewise.
23767 * config/arm/aarch-common-protos.h
23768 (arm_early_load_addr_dep_ptr): Add prototype.
23769 (arm_early_store_addr_dep_ptr): Likewise.
23770 * config/arm/cortex-a53.md: Add new bypasses.
23772 2017-05-05 Jakub Jelinek <jakub@redhat.com>
23774 * tree.c (next_type_uid): Change type to unsigned.
23775 (type_hash_canon): Decrement back next_type_uid if
23776 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
23777 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
23780 2017-05-04 Martin Sebor <msebor@redhat.com>
23782 * builtins.c: Fix a trivial typo in a comment.
23784 PR middle-end/79234
23785 * builtins.c (check_sizes): Adjust to handle reading past the end.
23786 Avoid printing excessive upper bound of ranges. Use %E to print
23787 tree nodes instead of converting them to %wu.
23788 (expand_builtin_memchr): New function.
23789 (compute_dest_size): Rename...
23790 (compute_objsize): ...to this.
23791 (expand_builtin_memcpy): Adjust.
23792 (expand_builtin_mempcpy): Adjust.
23793 (expand_builtin_strcat): Adjust.
23794 (expand_builtin_strcpy): Adjust.
23795 (check_strncat_sizes): Adjust.
23796 (expand_builtin_strncat): Adjust.
23797 (expand_builtin_strncpy): Adjust and simplify.
23798 (expand_builtin_memset): Adjust.
23799 (expand_builtin_bzero): Adjust.
23800 (expand_builtin_memcmp): Adjust.
23801 (expand_builtin): Handle memcmp.
23802 (maybe_emit_chk_warning): Check strncat just once.
23804 2017-05-04 Martin Sebor <msebor@redhat.com>
23806 PR preprocessor/79214
23807 PR middle-end/79222
23808 PR middle-end/79223
23809 * builtins.c (check_sizes): Add inlining context and issue
23810 warnings even when -Wno-system-headers is set.
23811 (check_strncat_sizes): Same.
23812 (expand_builtin_strncat): Same.
23813 (expand_builtin_memmove): New function.
23814 (expand_builtin_stpncpy): Same.
23815 (expand_builtin): Handle memmove and stpncpy.
23817 2017-05-04 Bin Cheng <bin.cheng@arm.com>
23819 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
23820 which is not used any more.
23822 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
23824 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
23826 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
23828 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
23829 (cortexa53_tunings): Likewise.
23830 (cortexa57_tunings): Likewise.
23831 (cortexa72_tunings): Likewise.
23832 (cortexa73_tunings): Likewise.
23834 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
23836 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
23837 Set loop alignment to 8.
23839 2017-05-04 Martin Sebor <msebor@redhat.com>
23841 PR translation/80280
23842 * builtins.c (expand_builtin_object_size): Add missing quoting to
23843 %D and like directives.
23844 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
23845 (hsa_type_for_tree_type): Same.
23846 (verify_function_arguments): Same.
23847 * symtab.c (symbol_table::change_decl_assembler_name): Same.
23848 * varasm.c (get_section): Same.
23851 2017-05-04 Martin Sebor <msebor@redhat.com>
23853 PR translation/80280
23854 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
23856 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
23858 * config/aarch64/aarch64.c (generic_addrcost_table):
23859 Change HI/TI mode setting.
23861 2017-05-04 Martin Jambor <mjambor@suse.cz>
23863 PR tree-optimization/80622
23864 * tree-sra.c (comes_initialized_p): New function.
23865 (build_accesses_from_assign): Only set write lazily when
23866 comes_initialized_p is false.
23867 (analyze_access_subtree): Use comes_initialized_p.
23868 (propagate_subaccesses_across_link): Assert !comes_initialized_p
23869 instead of testing for PARM_DECL.
23871 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23873 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
23874 constraint on operand 0 to allow more general addressing modes.
23875 Adjust output template.
23876 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
23878 * config/aarch64/aarch64-protos.h
23879 (aarch64_address_valid_for_prefetch_p): Declare prototype.
23880 * config/aarch64/constraints.md (Dp): New address constraint.
23881 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
23884 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
23886 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
23887 update use of estimate_ipcp_clone_size_and_time.
23888 (estimate_local_effects): Update use of
23889 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
23890 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
23891 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
23892 Return nonspecialized time.
23894 2017-05-04 Richard Biener <rguenther@suse.de>
23896 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
23897 for the last VUSE which def dominates the PHI. Directly call
23899 (get_continuation_for_phi_1): Remove.
23901 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
23903 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
23904 to explain the use of truncating division. Cap the number of
23905 iterations to the maximum given by nb_iterations_upper_bound,
23908 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
23910 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
23911 * configure: Regenerate.
23912 * config.in: Regenerate.
23913 * config/i386/driver-mingw32.c: new file.
23914 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
23915 * config.host: Link driver-mingw32.o on MinGW host.
23916 * doc/install.texi: Document new --enable-mingw-wildcard configure
23919 2017-05-04 Marek Polacek <polacek@redhat.com>
23921 PR tree-optimization/80612
23922 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
23924 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23925 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
23927 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
23928 (movt splitter): Likewise.
23929 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
23930 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
23931 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
23932 block for Thumb-1 with MOVT.
23933 (thumb2_legitimate_address_p): Move code block ...
23934 (can_avoid_literal_pool_for_label_p): ... into this new function.
23935 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
23937 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
23938 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
23939 "M-profile targets with the MOVT instruction".
23941 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23943 * config/arm/arm-builtins.c (arm_init_builtins): Rename
23944 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
23945 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
23947 2017-05-04 Martin Liska <mliska@suse.cz>
23949 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
23950 variable cond_code.
23952 2017-05-04 Richard Biener <rguenther@suse.de>
23954 * tree.c (array_at_struct_end_p): Handle arrays at struct
23955 end with flexarrays more conservatively. Refactor and treat
23956 arrays of arrays or aggregates more strict. Fix
23957 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
23958 * tree.c (array_at_struct_end_p): Adjust prototype.
23959 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
23960 * gimple-fold.c (get_range_strlen): Likewise.
23961 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
23963 2017-05-04 Richard Biener <rguenther@suse.de>
23965 PR tree-optimization/31130
23966 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
23968 (supports_overflow_infinity): Likewise.
23969 (is_negative_overflow_infinity): Likewise.
23970 (is_positive_overflow_infinity): Likewise.
23971 (is_overflow_infinity): Likewise.
23972 (stmt_overflow_infinity): Likewise.
23973 (overflow_infinity_range_p): Likewise.
23974 (usable_range_p): Remove as always returning true.
23975 (make_overflow_infinity): Remove.
23976 (negative_overflow_infinity): Likewise.
23977 (positive_overflow_infinity): Likewise.
23978 (avoid_overflow_infinity): Likewise.
23979 (set_value_range): Adjust accordingly.
23980 (set_value_range_to_nonnegative): Likewise, remove now unused
23981 overflow_infinity arg.
23982 (vrp_operand_equal_p): Adjust.
23983 (update_value_range): Likewise.
23984 (range_int_cst_singleton_p): Likewise.
23985 (operand_less_p): Likewise.
23986 (compare_values_warnv): Likewise.
23987 (extract_range_for_var_from_comparison_expr): Likewise.
23988 (vrp_int_const_binop): Likewise.
23989 (zero_nonzero_bits_from_vr): Likewise.
23990 (extract_range_from_multiplicative_op_1): Likewise.
23991 (extract_range_from_binary_expr_1): Likewise.
23992 (extract_range_from_unary_expr): Likewise.
23993 (extract_range_from_comparison): Likewise.
23994 (extract_range_basic): Likewise.
23995 (adjust_range_with_scev): Likewise.
23996 (compare_ranges): Likewise.
23997 (compare_range_with_value): Likewise.
23998 (dump_value_range): Likewise.
23999 (test_for_singularity): Likewise, remove strict_overflow_p parameter
24001 (simplify_cond_using_ranges): Adjust.
24003 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
24005 * brig-builtins.def: Added a builtin for class_f64.
24006 * builtin-types.def: Added a builtin type needed by class_f64.
24008 2017-05-03 Jason Merrill <jason@redhat.com>
24010 * timevar.def: Add TV_CONSTEXPR.
24012 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24014 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
24016 2017-05-03 Martin Jambor <mjambor@suse.cz>
24018 * ipa-prop.c (ipa_update_after_lto_read): Removed.
24019 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
24020 * ipa-cp.c (ipcp_propagate_stage): Do not call
24021 ipa_update_after_lto_read.
24022 * ipa-inline.c (ipa_inline): Likewise.
24024 2017-05-03 Martin Jambor <mjambor@suse.cz>
24026 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
24027 tag. Added a default constructor and a destructor.
24028 (ipa_edge_args_sum_t): New class;
24029 (ipa_edge_args_sum): Declare.
24030 (ipa_edge_args_vector): Remove declaration.
24031 (IPA_EDGE_REF): Use ipa_edge_args_sum.
24032 (ipa_free_edge_args_substructures): Remove declaration.
24033 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
24034 (ipa_edge_args_info_available_for_edge_p): Likewise.
24035 * ipa-prop.c (ipa_edge_args_vector): Removed.
24036 (edge_removal_hook_holder): Likewise.
24037 (edge_duplication_hook_holder): Likewise.
24038 (ipa_edge_args_sum): New variable.
24039 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
24040 ipa_edge_args_vector.
24041 (ipa_free_edge_args_substructures): Likewise.
24042 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
24043 ipa_edge_args_vector.
24044 (ipa_edge_removal_hook): Turned into method
24045 ipa_edge_args_sum_t::remove.
24046 (ipa_edge_duplication_hook): Turned into method
24047 ipa_edge_args_sum_t::duplicate.
24048 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
24049 registering edge hooks.
24050 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
24051 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
24052 ipa_edge_args_sum instead of ipa_edge_args_vector.
24053 * ipa-profile.c (ipa_profile): Likewise.
24055 2017-05-03 Martin Jambor <mjambor@suse.cz>
24057 * symbol-summary.h (function_summary): New method exists.
24058 (function_summary::symtab_removal): Deallocate through release.
24059 (call_summary): New class.
24060 (gt_ggc_mx): New overload.
24061 (gt_pch_nx): Likewise.
24062 (gt_pch_nx): Likewise.
24064 2017-05-03 Jeff Law <law@redhat.com>
24066 PR tree-optimization/78496
24067 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
24068 from simplify_cond_using_ranges. Split off code to walk
24069 backwards through casts into ...
24070 (simplify_cond_using_ranges_2): New function.
24071 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
24072 (execute_vrp): After identifying jump threads, call
24073 simplify_cond_using_ranges_2.
24075 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
24078 * ipa-inline.h (inline_summary): Add ctor.
24079 (create_ggc): Do not use ggc_cleared_alloc.
24081 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
24082 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24084 * gcc.c (handle_braces): Support escaping in switch matching
24086 * doc/invoke.texi (Spec Files): Document it.
24087 Remove superfluous @code markup in items.
24089 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24091 * diagnostic-show-locus.c (struct column_range): New struct.
24092 (get_affected_columns): New function.
24093 (get_printed_columns): New function.
24094 (struct correction): New struct.
24095 (correction::ensure_capacity): New function.
24096 (correction::ensure_terminated): New function.
24097 (struct line_corrections): New struct.
24098 (line_corrections::~line_corrections): New dtor.
24099 (line_corrections::add_hint): New function.
24100 (layout::print_trailing_fixits): Reimplement in terms of the new
24102 (selftest::test_overlapped_fixit_printing): New function.
24103 (selftest::diagnostic_show_locus_c_tests): Call it.
24105 2017-05-03 Nathan Sidwell <nathan@acm.org>
24107 Canonicalize canonical type hashing
24108 * tree.h (type_hash_canon_hash): Declare.
24109 * tree.c (type_hash_list, attribute_hash_list): Move into
24110 type_hash_canon_hash.
24111 (build_type_attribute_qual_variant): Break out hash code calc into
24112 type_hash_canon_hash.
24113 (type_hash_canon_hash): New. Generic type hash computation.
24114 (build_range_type_1, build_array_type_1, build_function_type,
24115 build_method_type_directly, build_offset_type, build_complex_type,
24116 make_vector_type): Call it.
24118 2017-05-03 Richard Biener <rguenther@suse.de>
24120 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
24121 When all DRs have unknown misaligned do not always peel
24122 when there is a store but apply the same costing model as if
24123 there were only loads.
24125 2017-05-03 Richard Biener <rguenther@suse.de>
24128 PR tree-optimization/80492
24129 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
24130 compare_base_decls returning dont-know properly.
24132 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
24134 * config/arm/iterators.md (CCSI): New mode iterator.
24135 (arch): New mode attribute.
24136 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
24137 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
24138 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
24139 code iterator for success result mode.
24140 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
24141 the corresponding new insn generators.
24143 2017-05-03 Bin Cheng <bin.cheng@arm.com>
24146 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24147 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24149 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
24151 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
24152 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
24153 (DDR_A): Wrap DDR argument in brackets.
24154 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
24155 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
24156 (DDR_REVERSED_P): Likewise.
24158 2017-05-03 Jakub Jelinek <jakub@redhat.com>
24160 PR tree-optimization/79472
24161 * tree-switch-conversion.c (struct switch_conv_info): Add
24162 contiguous_range and default_case_nonstandard fields.
24163 (collect_switch_conv_info): Compute contiguous_range and
24164 default_case_nonstandard fields, don't clear final_bb if
24165 contiguous_range and only the default case doesn't have the required
24167 (check_all_empty_except_final): Set default_case_nonstandard instead
24168 of failing if contiguous_range and the default case doesn't have empty
24170 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
24171 and only the default case doesn't have the required constants. Skip
24173 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
24174 if default_case_nonstandard.
24175 (build_constructors): Build constant 1 just once. Assert that default
24176 values aren't inserted in between cases if contiguous_range. Skip
24178 (build_arrays): Skip virtual phis.
24179 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
24180 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
24181 Handle virtual phis.
24182 (gen_inbound_check): Handle default_case_nonstandard case.
24183 (process_switch): Adjust check_final_bb caller. Call
24184 gather_default_values with the first non-default case instead of
24185 default case if default_case_nonstandard.
24187 2017-05-02 Nathan Sidwell <nathan@acm.org>
24189 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
24190 check. Fix formatting.
24192 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
24194 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
24195 errors when comparing specialized and unspecialized times.
24197 2017-05-02 David Malcolm <dmalcolm@redhat.com>
24199 * diagnostic-show-locus.c
24200 (layout::should_print_annotation_line_p): Make private.
24201 (layout::print_annotation_line): Make private.
24202 (layout::annotation_line_showed_range_p): Make private.
24203 (layout::show_ruler): Make private.
24204 (layout::print_source_line): Make private. Pass in line and
24205 line_width, rather than calling location_get_source_line. Drop
24207 (layout::print_leading_fixits): New method.
24208 (layout::print_any_fixits): Rename to...
24209 (layout::print_trailing_fixits): ...this, and make private.
24210 Don't print newline fixits.
24211 (diagnostic_show_locus): Move logic for printing one row into...
24212 (layout::print_line): ...this new function. Move the
24213 location_get_source_line call and error-handling from
24214 print_source_line to here. Call print_leading_fixits, and rename
24215 print_any_fixits to print_trailing_fixits.
24216 (selftest::test_fixit_insert_containing_newline): Update now that
24217 newlines are partially supported.
24218 (selftest::test_fixit_insert_containing_newline_2): New test.
24219 (selftest::test_fixit_replace_containing_newline): Update comments.
24220 (selftest::diagnostic_show_locus_c_tests): Call the new test.
24221 * edit-context.c (class added_line): New class.
24222 (class edited_line): Describe newline handling in comment.
24223 (edited_line::actually_edited_p): New method.
24224 (edited_line::print_content): Delete redundant decl.
24225 (edited_line::m_predecessors): New field.
24226 (edited_file::print_content): Call edited_line::print_content.
24227 (edited_file::print_diff): Update to support newlines.
24228 (edited_file::print_diff_hunk): Likewise.
24229 (edited_file::print_run_of_changed_lines): New function.
24230 (edited_file::print_diff_line): Convert to...
24231 (print_diff_line): ...this.
24232 (edited_file::get_effective_line_count): New function.
24233 (edited_line::edited_line): Initialize new field m_predecessors.
24234 (edited_line::~edited_line): Clean up m_predecessors.
24235 (edited_line::apply_fixit): Handle newlines.
24236 (edited_line::get_effective_line_count): New function.
24237 (edited_line::print_content): New function.
24238 (edited_line::print_diff_lines): New function.
24239 (selftest::test_applying_fixits_insert_containing_newline): New
24241 (selftest::test_applying_fixits_replace_containing_newline): New
24243 (selftest::insert_line): New function.
24244 (selftest::test_applying_fixits_multiple_lines): Add example of
24246 (selftest::edit_context_c_tests): Call the new tests.
24248 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24250 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
24251 parameter cand. Update dump information.
24252 (get_computation_cost): Update uses.
24254 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24256 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
24257 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
24258 (get_computation_at, rewrite_use_address): Update use of
24259 get_computation_aff.
24261 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24263 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
24264 (get_computation): Delete.
24265 (get_computation_cost): Implement like get_computation_cost_at.
24266 Use get_computation_at.
24267 (get_computation_cost_at): Delete.
24268 (rewrite_use_nonlinear_expr): Use get_computation_at.
24269 (rewrite_use_compare, remove_unused_ivs): Ditto.
24271 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24273 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
24275 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24277 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
24278 (ivopts_global_cost_for_size): Rename parameter and update uses.
24279 (iv_ca_recount_cost): Update uses.
24280 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
24281 candidates seperately in n_invs and n_cands.
24282 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
24284 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24286 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
24287 (find_inv_vars_cb): New.
24288 (find_depends): Renamed to ...
24289 (find_inv_vars): ... this.
24290 (add_candidate_1, force_var_cost): Call find_inv_vars.
24291 (split_address_cost, determine_group_iv_cost_cond): Ditto.
24293 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24295 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
24296 inv_vars. Add inv_exprs.
24297 (struct iv_cand): Rename depends_on to inv_vars.
24298 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
24299 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
24300 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
24301 (dump_cand): Dump inv_vars.
24302 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
24303 (record_invariant, find_depends, add_candidate_1): Ditto.
24304 (set_group_iv_cost, force_var_cost): Ditto.
24305 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
24306 (get_computation_cost_at, get_computation_cost): Ditto.
24307 (determine_group_iv_cost_generic): Ditto.
24308 (determine_group_iv_cost_address): Ditto.
24309 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
24310 (determine_group_iv_costs): Ditto.
24311 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
24312 (iv_ca_set_remove_invariants): Renamed to ...
24313 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
24314 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
24315 (iv_ca_set_add_invariants): Renamed to ...
24316 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
24317 (iv_ca_set_cp): Use iv_ca_set_add_invs.
24318 (iv_ca_has_deps): Support inv_vars and inv_exprs.
24319 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
24320 (create_new_ivs): Remove useless dump.
24322 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24324 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
24326 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
24327 (iv_ca_set_no_cp, create_new_iv): Ditto.
24329 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24331 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24333 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
24335 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
24336 function tree_check2.
24338 2017-05-02 Martin Liska <mliska@suse.cz>
24340 * doc/gcov.texi: Add missing preposition.
24341 * gcov.c (function_info::function_info): Properly fill up
24342 all member variables.
24344 2017-05-02 Tamar Christina <tamar.christina@arm.com>
24346 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
24348 2017-05-02 Tamar Christina <tamar.christina@arm.com>
24350 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
24352 2017-05-02 Martin Liska <mliska@suse.cz>
24355 * lto-streamer-in.c (lto_read_tree_1): Remove
24356 LTO_STREAMER_DEBUG.
24357 * lto-streamer.c (struct tree_hash_entry): Likewise.
24358 (struct tree_entry_hasher): Likewise.
24359 (tree_entry_hasher::hash): Likewise.
24360 (tree_entry_hasher::equal): Likewise.
24361 (lto_streamer_init): Likewise.
24362 (lto_orig_address_map): Likewise.
24363 (lto_orig_address_get): Likewise.
24364 (lto_orig_address_remove): Likewise.
24365 * lto-streamer.h: Likewise.
24366 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
24367 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
24369 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
24371 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
24372 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
24373 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
24374 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
24375 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
24376 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
24377 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
24378 (mm_maskz_sub_ss): New intrinsics.
24379 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24380 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
24381 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
24382 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
24383 (__builtin_ia32_subss_mask_round): New builtins.
24384 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24385 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24386 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
24388 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
24389 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
24391 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
24394 2017-05-02 Martin Jambor <mjambor@suse.cz>
24396 PR tree-optimization/78687
24397 * tree-sra.c (access): New field parent.
24398 (process_subtree_disqualification): New function.
24399 (disqualify_candidate): Call it.
24400 (build_accesses_from_assign): Reset write flag if creating an
24402 (build_access_subtree): Fill in parent field and also prpagate
24403 down grp_write flag.
24404 (create_artificial_child_access): New parameter set_grp_write, set
24405 grp_write to its value.
24406 (propagate_subaccesses_across_link): Also propagate grp_write flag
24408 (propagate_all_subaccesses): Push the closest parent back to work
24409 queue if add_access_to_work_queue returned true.
24411 2017-05-02 Richard Biener <rguenther@suse.de>
24413 * common.opt (fstrict-overflow): Alias negative to fwrapv.
24414 * doc/invoke.texi (fstrict-overflow): Remove all traces of
24415 -fstrict-overflow documentation.
24416 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
24417 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
24418 flag_strict_overflow.
24419 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
24420 * lto-opts.c (lto_write_options): Do not stream it.
24421 * lto-wrapper.c (merge_and_complain): Do not handle it.
24422 * opts.c (default_options_table): Do not set -fstrict-overflow.
24423 (finish_options): Likewise do not clear it when sanitizing.
24424 * simplify-rtx.c (simplify_const_relational_operation): Do not
24425 test flag_strict_overflow.
24427 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
24429 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
24430 using enabled attribute.
24431 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
24432 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
24433 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
24434 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
24435 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
24436 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
24437 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
24438 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
24439 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
24440 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
24442 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
24444 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
24446 2017-05-02 Richard Biener <rguenther@suse.de>
24448 PR tree-optimization/80591
24450 2017-04-10 Richard Biener <rguenther@suse.de>
24452 * tree-ssa-structalias.c (find_func_aliases): Properly handle
24455 2017-05-02 Richard Biener <rguenther@suse.de>
24457 PR tree-optimization/80549
24458 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
24459 (cleanup_tree_cfg_noloop): Create forwarders to known loop
24460 headers if they do not have a preheader.
24462 2017-05-02 Martin Liska <mliska@suse.cz>
24465 * common.opt: Fix typo.
24466 * doc/invoke.texi: Likewise.
24468 2017-05-01 Jan Beulich <jbeulich@suse.com>
24470 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
24471 swapping, add (x,x,m,x,n) alternative.
24473 2017-05-01 Nathan Sidwell <nathan@acm.org>
24475 * calls.c (combine_pending_stack_adjustment_and_call): Remove
24476 unnecessary unadjusted_alignment check.
24478 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
24481 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
24483 * gimplify.c (gimplify_cilk_detach): New function.
24484 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
24485 * tree-core.h: Document EXPR_CILK_SPAWN.
24486 * tree.h (EXPR_CILK_SPAWN): Define.
24488 2017-05-01 David Malcolm <dmalcolm@redhat.com>
24490 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
24491 to use new fixit_hint representation, using the "replace" logic.
24492 (get_line_span_for_fixit_hint): Likewise.
24493 (layout::print_any_fixits): Likewise.
24494 (selftest::test_one_liner_many_fixits): Rename to...
24495 (selftest::test_one_liner_many_fixits_1): ...this, and update
24496 comment and expected output to reflect that the multiple fix-it
24497 hints are now consolidated into one insertion.
24498 (selftest::test_one_liner_many_fixits_2): New test.
24499 (selftest::test_diagnostic_show_locus_one_liner): Update for
24501 (selftest::test_fixit_consolidation): Update for fix-it API
24503 * diagnostic.c (print_parseable_fixits): Likewise.
24504 * edit-context.c (edited_line::m_line_events): Convert from
24505 auto_vec <line_event *> to auto_vec <line_event>.
24506 (class line_event): Convert from abstract base class to a concrete
24507 class, taking over the role of replace_event.
24508 (class insert_event): Delete.
24509 (class replace_event): Rename to class line_event. Convert to
24511 (edit_context::add_fixits): Reimplement.
24512 (edit_context::apply_insert): Delete.
24513 (edit_context::apply_replace): Rename to...
24514 (edit_context::apply_fixit): ...this. Convert to half-open range.
24515 (edited_file::apply_insert): Delete.
24516 (edited_file::apply_replace): Rename to...
24517 (edited_file::apply_fixit): ...this.
24518 (edited_line::~edited_line): Drop deletion of events.
24519 (edited_line::apply_insert): Delete.
24520 (edited_line::apply_replace): Rename to...
24521 (edited_line::apply_fixit): ...this. Convert to half-open range.
24522 Update for change to type of m_line_events.
24523 * edit-context.h (edit_context::apply_insert): Delete.
24524 (edit_context::apply_replace): Rename to...
24525 (edit_context::apply_fixit): ...this.
24527 2017-05-01 Martin Sebor <msebor@redhat.com>
24529 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
24532 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
24535 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
24536 __get_cpuid_max returns 0.
24537 (__get_cpuid_count): Ditto.
24539 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
24541 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
24542 replacement expression is another instance of one of its arguments.
24544 2017-05-01 Jakub Jelinek <jakub@redhat.com>
24547 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
24548 check for stack push/pop autoinc.
24549 * config/i386/i386.c (ix86_agi_dependent): Return false
24550 if the only reason why modified_in_p returned true is that
24551 addr is SP based and set_insn is a push or pop.
24553 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
24555 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
24558 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
24561 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
24562 (account_size_time): Use two predicates - exec_pred and
24564 (evaluate_conditions_for_known_args): Compute both clause and
24566 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
24567 (inline_summary_t::duplicate): Update.
24568 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
24570 (compute_inline_parameters): Likewise.
24571 (estimate_edge_size_and_time): Update caluclation of time.
24572 (estimate_node_size_and_time): Compute both time and nonspecialized
24574 (estimate_ipcp_clone_size_and_time): Update.
24575 (inline_merge_summary): Update.
24576 (do_estimate_edge_time): Update.
24577 (do_estimate_edge_size): Update.
24578 (do_estimate_edge_hints): Update.
24579 (inline_read_section, inline_write_summary): Stream both new predicates.
24580 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
24582 (compute_inlined_call_time): Cleanup.
24583 (big_speedup_p): Update.
24584 (edge_badness): Update.
24585 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
24586 (size_time_entry): Replace predicate by exec_predicate and
24587 nonconst_predicate.
24588 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
24589 (estimate_edge_time): Return also nonspec_time.
24590 (reset_edge_growth_cache): Update.
24592 2017-04-29 Jakub Jelinek <jakub@redhat.com>
24594 PR rtl-optimization/80491
24595 * ifcvt.c (noce_process_if_block): When looking for x setter
24596 with missing else_bb, don't check only the insn right before
24597 cond_earliest, but look for the last insn that x is modified in
24598 within the same bb.
24600 PR rtl-optimization/80491
24601 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
24603 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
24605 PR tree-optimization/80487
24606 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
24608 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24610 PR tree-optimization/79697
24611 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
24612 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
24613 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
24615 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
24616 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
24618 2017-04-28 Martin Sebor <msebor@redhat.com>
24620 PR tree-optimization/80523
24621 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
24622 (init_target_to_host_charmap, target_to_host, target_strtol10): New
24624 (maybe_warn, format_directive, parse_directive): Use new functions.
24625 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
24627 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
24629 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
24631 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
24633 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
24634 target_header_dir): Set correctly.
24635 * configure: Regenerated.
24636 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
24637 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
24638 instead of SYSTEM_HEADER_DIR.
24640 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
24642 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
24643 (estimate_local_effects): Likewise.
24644 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
24645 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
24646 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
24647 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
24648 do_estimate_edge_time, estimate_edge_time): Likewise.
24649 * ipa-inline-analysis.c (estimate_node_size_and_time,
24650 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
24651 (estimate_time_after_inlining): Remove.
24653 2017-04-28 Martin Liska <mliska@suse.cz>
24655 * doc/gcov.texi: Enhance documentation of gcov.
24657 2017-04-28 Martin Liska <mliska@suse.cz>
24659 * doc/gcov.texi: Sort options in alphabetic order.
24660 * doc/gcov-dump.texi: Likewise.
24661 * doc/gcov-tool.texi: Likewise.
24662 * gcov.c (print_usage): Likewise.
24663 * gcov-dump.c (print_usage): Likewise.
24664 * gcov-tool.c (print_merge_usage_message): Likewise.
24665 (print_rewrite_usage_message): Likewise.
24666 (print_overlap_usage_message): Likewise.
24668 2017-04-28 Martin Liska <mliska@suse.cz>
24670 PR gcov-profile/53915
24671 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
24673 2017-04-28 Martin Liska <mliska@suse.cz>
24675 PR gcov-profile/79891
24676 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
24677 is marked by compiler as living on a line.
24678 (get_cycles_count): Remove usage of the union.
24679 (output_intermediate_file): Likewise.
24680 (find_source): Fix GNU coding style.
24681 (accumulate_line_counts): Remove old non-all block mode.
24682 (output_lines): Remove usage of the union.
24683 * profile.c (output_location): Include all BBs, even if
24684 belonging to a same line (and file) as a previous BB.
24686 2017-04-28 Martin Liska <mliska@suse.cz>
24688 * gcov.c (process_args): Handle new argument 'w'.
24689 (read_graph_file): Assign ID to BBs.
24690 (output_branch_count): Display BB # if verbose flag is set.
24691 (output_lines): Likewise for arcs.
24692 (print_usage): Add '--verbose' option help.
24693 * doc/gcov.texi: Document --verbose (-w) option.
24695 2017-04-28 Martin Liska <mliska@suse.cz>
24697 * gcov.c (struct block_location_info): New struct.
24698 (process_file): Fill up the new structure.
24699 (read_graph_file): Replace usage of encoding by the newly added
24701 (add_line_counts): Likewise.
24702 (accumulate_line_counts): Remove usage of the union.
24703 (function_info::function_info): New function.
24704 (function_info::~function_info): Likewise.
24705 (process_file): Call delete instead of release_function.
24706 (release_function): Release the function.
24707 (release_structures): Call delete instead of release_function.
24708 (solve_flow_graph): Replace usage of num_blocks.
24709 (find_exception_blocks): Likewise.
24710 (output_lines): Fix GNU coding style.
24712 2017-04-28 Martin Liska <mliska@suse.cz>
24715 * coverage.c (coverage_remove_note_file): New function.
24716 * coverage.h: Declare the function.
24717 * toplev.c (finalize): Clean if an error has been seen.
24719 2017-04-28 Martin Liska <mliska@suse.cz>
24721 PR gcov-profile/80031
24722 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
24723 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
24724 * gcov.c (read_graph_file): Read just number of blocks.
24725 * profile.c (branch_prob): Do not stream 0 flags per a basic
24728 2017-04-28 Martin Liska <mliska@suse.cz>
24730 * gcov-dump.c (tag_*): Add new argument to declarations.
24731 (dump_gcov_file): Likewise.
24732 (tag_blocks): Add and use new argument depth.
24733 (tag_arcs): Likewise.
24734 (tag_lines): Likewise.
24735 (tag_counters): Likewise.
24736 (tag_summary): Likewise.
24737 (dump_working_sets): Use depth to do a proper indentation.
24739 2017-04-28 Jakub Jelinek <jakub@redhat.com>
24742 * cgraph.h (symtab_node::debug_symtab): No longer inline.
24743 * symtab.c (symtab_node::debug_symtab): Move definition here.
24745 2017-04-28 Richard Biener <rguenther@suse.de>
24747 * lto-streamer.h (LTO_major_version): Bump to 7.
24749 2017-04-28 Richard Biener <rguenther@suse.de>
24751 * tree-vrp.c (assert_info): New struct.
24752 (add_assert_info): New helper.
24753 (register_edge_assert_for_2): Refactor to add asserts to a vector
24755 (register_edge_assert_for_1): Likewise.
24756 (register_edge_assert_for): Likewise.
24757 (finish_register_edge_assert_for): New helper actually registering
24758 asserts where live on edge.
24759 (find_conditional_asserts): Adjust.
24760 (find_switch_asserts): Likewise.
24761 (evrp_dom_walker::try_find_new_range): Generalize.
24762 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
24764 2017-04-27 Marek Polacek <polacek@redhat.com>
24767 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
24768 arg10 and arg11 to itype.
24770 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
24772 * doc/extend.texi (Object Size Checking): Improve grammar.
24774 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
24777 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
24778 that the logic for permitting reciprocal estimates matches that
24781 2017-04-27 Jakub Jelinek <jakub@redhat.com>
24784 * tree.c (type_cache_hasher::equal): Only compare
24785 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
24786 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
24787 non-aggregate element types.
24788 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
24789 about the flag on ARRAY_TYPEs in the comment, formatting fix.
24791 2017-04-27 Richard Biener <rguenther@suse.de>
24793 PR middle-end/80533
24794 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
24795 stripping ARRAY_REFs from MEM_EXPR make sure we're not
24796 keeping a reference to a trailing array.
24798 2017-04-27 Richard Biener <rguenther@suse.de>
24800 PR middle-end/80539
24801 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
24802 being in loop-closed SSA form conservatively.
24803 (chrec_fold_multiply_poly_poly): Likewise.
24805 2017-04-27 Tamar Christina <tamar.christina@arm.com>
24807 PR middle-end/79665
24808 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
24809 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
24811 2017-04-27 Jakub Jelinek <jakub@redhat.com>
24814 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
24815 (aarch64_function_arg_alignment): Return unsigned int again, but still
24816 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
24817 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
24818 Don't emit -Wpsabi note.
24819 (aarch64_function_arg_boundary): Likewise.
24820 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
24823 2017-04-26 Nathan Sidwell <nathan@acm.org>
24825 * tree.h (crc32_unsigned_n): Declare.
24826 (crc32_unsigned, crc32_unsigned): Make inline.
24827 * tree.c (crc32_unsigned_bits): Replace with ...
24828 (crc32_unsigned_n): ... this.
24829 (crc32_unsigned, crc32_byte): Remove.
24830 (crc32_string): Remove unnecessary braces.
24832 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
24834 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
24835 * ipa-inline-analysis.c (MAX_TIME): Remove.
24836 (account_size_time): Use sreal for time.
24837 (dump_inline_summary): Update.
24838 (estimate_function_body_sizes): Update.
24839 (estimate_edge_size_and_time): Update.
24840 (estimate_calls_size_and_time): Update.
24841 (estimate_node_size_and_time): Update.
24842 (inline_merge_summary): Update.
24843 (inline_update_overall_summary): Update.
24844 (estimate_time_after_inlining): Update.
24845 (inline_read_section): Update.
24846 (inline_write_summary): Update.
24847 * ipa-inline.c (compute_uninlined_call_time): Update.
24848 (compute_inlined_call_time): Update.
24849 (recursive_inlining): Update.
24850 (inline_small_functions): Update.
24851 (dump_overall_stats): Update.
24852 * ipa-inline.h: Include sreal.h.
24853 (size_time_entry): Turn time to sreal.
24854 (inline_summary): Turn self_time nad time to sreal.
24856 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
24858 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
24860 (sreal::stream_out, sreal::stream_in): New.
24861 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
24863 2017-04-25 Jakub Jelinek <jakub@redhat.com>
24865 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
24868 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
24871 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
24873 (dimode_scalar_chain::compute_convert_gain): Ditto.
24874 (dimode_scalar_chain::make_vector_copies): Ditto.
24875 (dimode_scalar_chain::convert_reg): Ditto.
24876 (dimode_scalar_chain::convert_insn): Ditto.
24877 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
24878 (VI248_AVX512BW_1): New mode iterator.
24879 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
24880 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
24883 2017-04-25 Martin Sebor <msebor@redhat.com>
24885 PR tree-optimization/80497
24886 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
24887 constants are representable in HOST_WIDE_INT.
24888 (parse_directive): Ditto.
24890 2017-04-25 Martin Sebor <msebor@redhat.com>
24893 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
24894 (new_zero_array): Adjust signature.
24895 (dom_info::dom_init): Used unsigned rather that size_t.
24896 (dom_info::dom_info): Same.
24898 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24899 Jakub Jelinek <jakub@redhat.com>
24902 * config/arm/arm.c: Include gimple.h.
24903 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
24904 returns negative, increment ncrn only if it returned positive.
24905 (arm_needs_doubleword_align): Return int instead of bool,
24906 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
24907 members, but if there is any such non-FIELD_DECL
24908 > PARM_BOUNDARY aligned decl, return -1 instead of false.
24909 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
24910 returns negative, increment nregs only if it returned positive.
24911 (arm_setup_incoming_varargs): Likewise.
24912 (arm_function_arg_boundary): Emit -Wpsabi note if
24913 arm_needs_doubleword_align returns negative, return
24914 DOUBLEWORD_ALIGNMENT only if it returned positive.
24916 2017-04-25 Marek Polacek <polacek@redhat.com>
24919 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
24920 first argument to type.
24922 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
24925 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
24926 type checks to test for compatibility instead of equality.
24928 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24929 Jakub Jelinek <jakub@redhat.com>
24932 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
24934 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
24935 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
24936 the alignment computation, but return their maximum in warn_alignment.
24937 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
24938 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
24940 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
24941 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
24944 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24946 * config/arc/simdext.md (dmpyh): Fix typo.
24948 2017-04-25 Richard Biener <rguenther@suse.de>
24950 PR tree-optimization/80492
24951 * alias.c (compare_base_decls): Handle registers with asm
24952 specification conservatively.
24953 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
24954 compare_base_decls returning dont-know properly.
24956 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24958 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
24959 (legitimate_offset_address_p): New function.
24960 (arc_legitimate_address_p): Use above function.
24962 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24964 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
24966 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24968 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
24969 ACCH registers whenever they are available.
24971 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24973 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
24974 double regs fix when not used.
24976 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24978 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
24980 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
24981 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
24983 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24985 * config/arc/arc.c (arc_output_addsi): Check for h-register class
24986 when emitting short ADD instructions.
24988 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24990 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
24992 (cmpsi_cc_c_insn): Likewise.
24993 (cbranchsi4_scratch): Compute proper instruction length using
24994 compact_hreg_operand.
24995 * config/arc/predicates.md (compact_hreg_operand): New predicate.
24997 2017-04-25 Richard Biener <rguenther@suse.de>
24999 PR middle-end/80509
25000 * passes.c (pass_manager::pass_manager): Initialize
25001 m_name_to_pass_map.
25003 2017-04-25 Richard Biener <rguenther@suse.de>
25005 PR tree-optimization/79201
25006 * tree-ssa-sink.c (statement_sink_location): Handle calls.
25008 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25011 * config/s390/vector.md: Split MEM->GPR vector moves for
25012 non-s_operand addresses.
25014 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25017 * config/s390/predicates.md (reload_const_wide_int_operand): New
25019 * config/s390/s390.md ("movti"): Remove d/P alternative.
25020 ("movti_bigconst"): New pattern definition.
25022 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25025 * s390-protos.h (s390_expand_cs_hqi): Removed.
25026 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
25027 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
25028 modes as well as CCZ1mode and CCZmode.
25029 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
25030 signature of s390_emit_compare_and_swap.
25031 (s390_expand_cs_hqi): Likewise, make static.
25032 (s390_expand_cs_tdsi): Generate an explicit compare before trying
25033 compare-and-swap, in some cases.
25034 (s390_expand_cs): Wrapper function.
25035 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
25037 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
25038 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
25039 patterns for small and large integers. Forbid symref memory operands.
25040 Move expander to s390.c. Require cc register.
25041 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
25042 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
25043 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
25044 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
25045 symref memory operands. Remove CC mode and call s390_match_ccmode
25047 ("atomic_exchange<mode>"): Allow and implement all integer modes.
25049 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25051 * config/s390/s390.md (define_peephole2): New peephole to help
25052 combining the load-and-test pattern with volatile memory.
25054 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25056 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
25057 with CCZmode for TARGET_Z196.
25059 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25061 PR rtl-optimization/80501
25062 * combine.c (make_compound_operation_int): Set subreg_code to SET
25063 even for AND with mask of the sign bit of mode.
25065 PR rtl-optimization/80500
25066 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
25067 sum's initial value.
25069 2017-04-25 Julian Brown <julian@codesourcery.com>
25070 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25072 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
25074 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
25076 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
25078 2017-04-25 Julian Brown <julian@codesourcery.com>
25079 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25081 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
25082 (thunderx2t99_sha): New Reservation.
25084 2017-04-25 Julian Brown <julian@codesourcery.com>
25085 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25087 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
25088 type for 1-element load.
25090 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
25092 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
25094 2017-04-24 Martin Jambor <mjambor@suse.cz>
25096 PR tree-optimization/80293
25097 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
25098 char arrays not totally scalarizable if it is false.
25099 (analyze_all_variable_accesses): Pass correct value in the new
25100 parameter. Add a statistics counter.
25102 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
25104 PR middle-end/79931
25105 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
25107 2017-04-24 Richard Biener <rguenther@suse.de>
25109 PR tree-optimization/80494
25110 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
25111 out for complex types.
25113 2017-04-24 Richard Biener <rguenther@suse.de>
25115 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
25116 * tree-ssa-sccvn.c (print_scc): Print SCC size.
25117 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
25118 (DFS): Adjust and never fail.
25119 (sccvn_dom_walker::fail): Remove.
25120 (sccvn_dom_walker::before_dom_children): Adjust.
25121 (run_scc_vn): Likewise and never fail.
25122 * tree-ssa-pre.c (pass_pre::execute): Adjust.
25123 (pass_fre::execute): Likewise.
25125 2017-04-24 Richard Biener <rguenther@suse.de>
25127 PR tree-optimization/79725
25128 * tree-ssa-sink.c (statement_sink_location): Return whether
25129 failure reason was zero uses. Move that check later.
25130 (sink_code_in_bb): Deal with zero uses by removing the stmt
25133 2017-04-24 Richard Biener <rguenther@suse.de>
25136 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
25137 pointer-based references.
25139 2017-04-24 Richard Biener <rguenther@suse.de>
25142 * pass_manager.h (pass_manager::operator new): Remove.
25143 (pass_manager::operator delete): Likewise.
25144 * passes.c (pass_manager::operator new): Remove.
25145 (pass_manager::operator delete): Likewise.
25146 (pass_manager::pass_manager): Zero individual pass members.
25148 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
25151 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
25152 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
25153 Check "XEXP (src, 1)" operand here.
25154 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
25155 Check "XEXP (src, 1)" operand here.
25156 (dimode_scalar_chain::make_vector_copies): Detect count register
25157 of a shift instruction. Zero extend count register from QImode
25158 to DImode to satisfy vector shift pattern count operand predicate.
25159 Substitute vector shift count operand with a DImode copy.
25160 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
25163 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
25165 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
25166 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
25167 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
25168 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
25169 (UNSPEC_NOREX_MEM): Remove definition.
25171 2017-04-21 Richard Biener <rguenther@suse.de>
25173 PR tree-optimization/79547
25174 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
25175 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
25176 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
25177 without any constraints.
25179 2017-04-21 Richard Biener <rguenther@suse.de>
25181 PR tree-optimization/78847
25182 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
25184 2017-04-21 Richard Biener <rguenther@suse.de>
25186 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
25187 (build_distinct_type_copy): Likewise.
25188 (build_variant_type_copy): Likewise.
25189 * tree.c (build_qualified_type): Pass down mem-stat info.
25190 (build_distinct_type_copy): Likewise.
25191 (build_variant_type_copy): Likewise.
25193 2017-04-21 Richard Biener <rguenther@suse.de>
25195 PR tree-optimization/80237
25196 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
25198 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
25199 for a simplified result.
25201 2016-04-21 Richard Biener <rguenther@suse.de>
25203 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
25204 sth as strict as a simple_iv but a chrec without symbols and an
25205 operand defined in the loop we are peeling (and not some subloop).
25206 (propagate_constants_for_unrolling): Propagate all constants.
25208 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25211 * config/i386/i386.c (print_reg): Remove assert for disalowed
25212 regno values, call output_operand_lossage instead.
25214 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25217 * config/i386/constraints.md (Yc): New register constraint.
25218 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
25219 Use Yc constraint for alternative 2 of operand 0. Remove
25220 preferred_for_speed attribute.
25222 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
25224 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
25225 lastprivate clauses in SIMT case.
25227 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
25229 * doc/invoke.texi (-Wextra-semi): Document new warning option.
25231 2017-04-20 Richard Biener <rguenther@suse.de>
25233 PR tree-optimization/57796
25234 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
25235 as N scalar stores.
25236 (vect_model_load_cost): Cost gathers as N scalar loads.
25238 2017-04-20 Richard Biener <rguenther@suse.de>
25240 * ggc-page.c (ggc_allocated_p): Rename to ...
25241 (safe_lookup_page_table_entry): ... this and return the lookup
25243 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
25245 2017-04-20 Richard Biener <rguenther@suse.de>
25247 PR tree-optimization/80453
25248 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
25249 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
25250 from the conditions.
25251 (vn_phi_eq): Pass them down.
25252 (vn_phi_lookup): Record them.
25253 (vn_phi_insert): Likewise.
25255 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
25257 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
25258 uninitialized variable warning to avoid buffer overrun.
25260 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
25263 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
25264 is suppressed for '{ 0 }' in C.
25266 2017-04-20 Jakub Jelinek <jakub@redhat.com>
25268 * BASE-VER: Set to 8.0.0.
25270 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
25272 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
25273 priority .init_array and .fini_array section with SECTION_NOTYPE
25276 2017-04-20 Jakub Jelinek <jakub@redhat.com>
25278 PR middle-end/80423
25279 * tree.h (build_array_type): Add typeless_storage default argument.
25280 * tree.c (type_cache_hasher::equal): Also compare
25281 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
25282 (build_array_type): Add typeless_storage argument, set
25283 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
25285 (build_nonshared_array_type): Adjust build_array_type_1 caller.
25286 (build_array_type): Likewise. Add typeless_storage argument.
25288 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
25289 Jakub Jelinek <jakub@redhat.com>
25291 PR tree-optimization/80426
25292 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
25293 operation on symbolic operands, also compute the overflow for the
25294 invariant part when the operation degenerates into a negation.
25296 2017-04-19 Jakub Jelinek <jakub@redhat.com>
25299 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
25300 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
25303 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
25305 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
25308 * config/avr/avr.c (tree.h): Include it.
25309 (cgraph.h): Include it.
25310 (avr_encode_section_info): Don't warn for uninitialized progmem
25311 variable if it's just an alias.
25313 2017-04-19 Richard Biener <rguenther@suse.de>
25316 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
25317 when needed by AutoPGO.
25319 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
25322 * doc/lto.texi: Remove an extra 'that'.
25324 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
25326 PR rtl-optimization/80429
25327 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
25328 are only used in debug insns.
25330 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
25331 Vladimir Makarov <vmakarov@redhat.com>
25333 * config/sparc/predicates.md (input_operand): Add comment. Return
25334 true for any memory operand when LRA is in progress.
25335 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
25337 2017-04-18 Jeff Law <law@redhat.com>
25340 * mips.md ({return,simple_return}_internal): Do not overwrite
25343 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25345 PR tree-optimization/80443
25346 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
25347 instead of adding 1, subtract -1 and similarly instead of subtracting
25350 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
25352 PR rtl-optimization/80357
25353 * haifa-sched.c (tmp_bitmap): New variable.
25354 (model_recompute): Handle duplicate use records.
25355 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
25356 (free_global_sched_pressure_data): Free it.
25358 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
25361 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
25362 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25363 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25364 instead of SYSTEM_HEADER_DIR.
25366 2017-04-18 Jeff Law <law@redhat.com>
25368 PR middle-end/80422
25369 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
25370 predecessors after walking up the insn chain.
25372 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25375 * dwarf2out.c (modified_type_die): Try harder not to emit internal
25376 sizetype type into debug info.
25378 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
25381 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
25382 unneeded test for TARGET_UPPER_REGS_SF.
25383 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
25385 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25388 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
25389 instead of gsi_after_labels.
25391 2017-04-18 Jeff Law <law@redhat.com>
25393 * regcprop.c (maybe_mode_change): Avoid creating copies of the
25397 2017-04-13 Jeff Law <law@redhat.com>
25398 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
25399 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
25401 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
25404 * config/avr/avr.c (intl.h): Include it.
25405 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
25407 2017-04-18 Martin Liska <mliska@suse.cz>
25409 PR gcov-profile/78783
25410 * gcov-tool.c (gcov_output_files): Validate that destination
25411 file is either removed by the tool or by a user.
25413 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
25414 Guy Benyei <guybe@mellanox.com>
25416 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
25417 block, and do not negate it, the stored id is already negative.
25419 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
25421 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
25423 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
25426 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
25427 masks of options that should be turned off if the VSX vector
25428 options are turned off.
25429 (OTHER_P8_VECTOR_MASKS): Likewise.
25430 (OTHER_VSX_VECTOR_MASKS): Likewise.
25431 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
25432 rs6000_disable_incompatible_switches to validate no type switches
25434 (rs6000_incompatible_switch): New function to disallow turning on
25435 other vector options if -mno-vsx, -mno-power8-vector, or
25436 -mno-power9-vector are specified.
25438 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25440 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
25442 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25444 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
25445 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
25446 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
25447 (ARG_POINTER_CFA_OFFSET): Likewise.
25449 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25451 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
25452 conditions to take advantage of various optimizations.
25454 2017-04-13 Jeff Law <law@redhat.com>
25456 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
25457 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
25458 (zero_extendsidi2_dext): Likewise.
25460 2017-04-13 Jakub Jelinek <jakub@redhat.com>
25463 * fold-const.c (fold_ternary_loc): Revert
25464 use op0 instead of fold_convert_loc (loc, type, arg0) part of
25467 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
25469 PR rtl-optimization/80343
25470 * lra-remat.c (update_scratch_ops): Assign original hard reg to
25471 new scratch pseudo.
25473 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
25476 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
25477 to ubsan_encode_value.
25479 2017-04-13 Jeff Law <law@redhat.com>
25481 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
25482 appearing in DEBUG_INSNs.
25484 2017-04-13 Martin Liska <mliska@suse.cz>
25486 PR gcov-profile/80413
25487 * gcov-io.c (gcov_write_string): Copy to buffer just when
25488 allocated size is greater than zero.
25490 2017-04-13 Jakub Jelinek <jakub@redhat.com>
25493 * dwarf2out.c (decls_for_scope): Ignore declarations of
25494 current_function_decl in BLOCK_NONLOCALIZED_VARS.
25496 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
25499 * ipa-visibility.c (non_local_p): Fix typos.
25500 (localize_node): When localizing symbol in same comdat group,
25501 dissolve the group only when we know external symbols are going
25503 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
25505 2017-04-12 Jakub Jelinek <jakub@redhat.com>
25507 PR tree-optimization/79390
25508 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
25509 order does not result in usable sequence, retry with reversed operand
25515 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
25516 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
25517 op0 instead of fold_convert_loc (loc, type, arg0).
25519 2017-04-12 Jeff Law <law@redhat.com>
25521 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
25522 has a delay slot in the generated code.
25524 * config/cris/cris.md (cris_preferred_reload_class): Return
25525 GENNONACR_REGS rather than GENERAL_REGS.
25527 2017-04-12 Jakub Jelinek <jakub@redhat.com>
25530 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
25531 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
25532 signedness of the result type.
25534 2017-04-12 Richard Biener <rguenther@suse.de>
25535 Jeff Law <law@redhat.com>
25537 PR tree-optimization/80359
25538 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
25539 trim stores to TARGET_MEM_REFs.
25541 2017-04-12 Richard Biener <rguenther@suse.de>
25543 PR tree-optimization/79390
25544 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
25545 threading case even more.
25547 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
25550 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
25551 for quad_address_p for TImode, instead of just not indexed_address.
25553 2017-04-12 Richard Biener <rguenther@suse.de>
25554 Bernd Edlinger <bernd.edlinger@hotmail.de>
25556 PR middle-end/79671
25557 * alias.c (component_uses_parent_alias_set_from): Handle
25558 TYPE_TYPELESS_STORAGE.
25559 (get_alias_set): Likewise.
25560 * tree-core.h (tree_type_common): Add typeless_storage flag.
25561 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
25562 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
25563 for types containing members with TYPE_TYPELESS_STORAGE.
25564 (place_field): Likewise.
25565 (layout_type): Likewise for ARRAY_TYPE.
25566 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
25567 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
25568 TYPE_TYPELESS_STORAGE.
25569 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
25571 2017-04-12 Jakub Jelinek <jakub@redhat.com>
25574 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
25575 first argument to type.
25577 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25581 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
25582 CONST0_RTX (mode) rather than const0_rtx where appropriate.
25583 (rs6000_expand_binop_builtin): Likewise.
25584 (rs6000_expand_ternop_builtin): Likewise; also add missing
25585 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
25586 vshasigma built-ins.
25587 * doc/extend.texi: Document that vec_xxpermdi's third argument
25588 must be a constant.
25590 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
25592 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
25593 Use shift_const cost parameter when calculating gain of STV shifts.
25595 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
25597 PR rtl-optimization/70478
25598 * lra-constraints.c (process_alt_operands): Check memory for
25599 disfavoring memory insn operand.
25601 2017-04-11 Jakub Jelinek <jakub@redhat.com>
25603 PR middle-end/80100
25604 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
25605 left shift in unsigned HOST_WIDE_INT type.
25607 PR rtl-optimization/80385
25608 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
25609 (not (neg X)) into (plus X -1) for complex or non-integral modes.
25612 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
25613 if they have any depend clauses.
25615 2017-04-11 Martin Liska <mliska@suse.cz>
25618 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
25619 * ipa-split.c (split_function): Create a local comdat symbol
25620 if caller is in a comdat group.
25622 2017-04-11 Martin Liska <mliska@suse.cz>
25625 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
25628 2017-04-11 Martin Sebor <msebor@redhat.com>
25630 PR middle-end/80364
25631 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
25632 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
25634 (directive::set_width, directive::set_precision, format_character):
25636 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
25639 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
25642 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
25643 conflict, set target->arch_name instead of target->cpu_name.
25645 2017-04-11 Richard Biener <rguenther@suse.de>
25647 PR tree-optimization/80374
25648 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
25649 build_zero_cst, remove fold_convertible_p check again.
25651 2017-04-11 Martin Liska <mliska@suse.cz>
25654 * ubsan.c (instrument_object_size): Do not instrument register
25657 2017-04-11 Jakub Jelinek <jakub@redhat.com>
25660 * config/i386/i386-builtin-types.def
25661 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
25662 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
25663 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
25664 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
25665 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
25666 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
25667 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
25668 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
25669 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
25670 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
25671 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
25672 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
25673 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
25674 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
25675 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
25676 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
25677 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
25678 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
25679 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
25680 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
25681 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
25682 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
25683 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
25684 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
25685 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
25686 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
25687 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
25688 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
25689 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
25690 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
25691 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
25692 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
25693 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
25694 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
25695 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
25696 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
25697 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
25698 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
25699 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
25700 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
25701 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
25702 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
25703 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
25704 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
25705 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
25706 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
25708 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
25709 flag to second_arg_count, handle 4 argument function type _COUNT
25710 aliases, handle second_arg_count on second argument rather than last.
25712 2017-04-10 Jeff Law <law@redhat.com>
25714 PR tree-optimization/80374
25715 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
25716 record anything if we can not convert integer_zero_node to the
25719 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
25722 * config/rs6000/rs6000.c (rs6000_option_override_internal):
25723 Enhance special handling given to the TARGET_P9_MINMAX option in
25724 relation to certain other options.
25726 2017-04-10 Bin Cheng <bin.cheng@arm.com>
25728 PR tree-optimization/80153
25729 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
25730 remove POINTER_PLUS_EXPR's base part directly, rather than through
25733 2017-04-10 Richard Biener <rguenther@suse.de>
25734 Bin Cheng <bin.cheng@arm.com>
25736 PR tree-optimization/80153
25737 * tree-affine.c (aff_combination_to_tree): Get base pointer from
25738 the first element of pointer type aff_tree. Build result expr in
25740 (add_elt_to_tree): Convert to type unconditionally. Remove other
25741 fold_convert calls.
25742 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
25743 (rewrite_use_nonlinear_expr): Check invariant using iv information.
25745 2017-04-10 Richard Biener <rguenther@suse.de>
25747 * tree-ssa-structalias.c (find_func_aliases): Properly handle
25750 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
25752 PR rtl-optimization/70478
25753 * lra-constraints.c (curr_small_class_check): New.
25754 (update_and_check_small_class_inputs): New.
25755 (process_alt_operands): Update curr_small_class_check. Disfavor
25756 alternative insn memory operands. Check available regs for small
25759 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
25762 * config/mips/mips.opt (-mvirt): Update description.
25763 * doc/invoke.texi (-mvirt): Likewise.
25765 2017-04-10 Richard Biener <rguenther@suse.de>
25767 PR middle-end/80362
25768 * fold-const.c (fold_binary_loc): Look at unstripped ops when
25769 looking for NEGATE_EXPR in -A / -B to A / B folding.
25771 2017-04-10 Martin Liska <mliska@suse.cz>
25773 PR gcov-profile/80224
25774 * gcov.c (print_usage): Fix usage string.
25775 (get_gcov_intermediate_filename): Remove.
25776 (output_gcov_file): Use both for normal and intermediate format.
25777 (generate_results): Do not initialize special file for
25778 intermediate format.
25780 2017-04-10 Richard Biener <rguenther@suse.de>
25782 PR tree-optimization/80304
25783 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
25786 2017-04-10 Nathan Sidwell <nathan@acm.org>
25789 * config/rs6000/rs6000.c (rs6000_vector_type): New.
25790 (rs6000_init_builtins): Use it.
25792 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25794 * config/arm/arm.md (<mrc>): Add mode to SET source.
25795 (<mrrc>): Likewise.
25797 2017-04-10 Richard Biener <rguenther@suse.de>
25799 PR middle-end/80344
25800 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
25802 2017-04-10 Jakub Jelinek <jakub@redhat.com>
25805 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
25806 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
25807 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
25808 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
25809 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
25810 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
25811 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
25812 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
25813 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
25814 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
25815 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
25816 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
25817 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
25818 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
25819 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
25820 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
25821 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
25822 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
25823 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
25824 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
25825 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
25826 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
25827 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
25829 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
25831 PR rtl-optimization/70478
25832 * lra-constraints.c: Reverse the last patch.
25834 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
25836 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
25837 Add comment for WCHAR_T.
25839 2017-04-08 Martin Liska <mliska@suse.cz>
25842 2017-04-07 Martin Liska <mliska@suse.cz>
25845 * ipa-split.c (split_function): Add function part to a same comdat
25848 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25851 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
25853 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
25855 * rs6000/rs6000.c (vec_load_pendulum): Rename...
25856 (vec_pairing): ...to this.
25857 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
25858 (rs6000_sched_init): Adjust for name change.
25859 (struct rs6000_sched_context): Likewise.
25860 (rs6000_init_sched_context): Likewise.
25861 (rs6000_set_sched_context): Likewise.
25863 2017-04-07 Jakub Jelinek <jakub@redhat.com>
25869 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
25871 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
25872 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
25873 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
25875 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
25877 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
25879 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
25881 PR rtl-optimization/70703
25882 * ira-color.c (update_conflict_hard_regno_costs): Use
25883 int64_t instead of HOST_WIDE_INT.
25885 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
25887 PR rtl-optimization/70478
25888 * lra-constraints.c (process_alt_operands): Disfavor alternative
25889 insn memory operands.
25891 2017-04-07 Jeff Law <law@redhat.com>
25893 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
25894 CALL and NOTE_INSN_CALL_ARG_LOCATION.
25896 2017-04-07 Martin Liska <mliska@suse.cz>
25899 * config/aarch64/aarch64.c (aarch64_process_target_attr):
25900 Show error message instead of an ICE.
25902 2017-04-07 Martin Liska <mliska@suse.cz>
25905 * ipa-split.c (split_function): Add function part to a same comdat
25908 2017-04-07 Richard Biener <rguenther@suse.de>
25910 PR middle-end/80341
25911 * tree.c (get_unwidened): Also handle ! for_type case for
25913 * convert.c (do_narrow): Split out from ...
25914 (convert_to_integer_1): ... here. Do not pass final truncation
25915 type to get_unwidened for TRUNC_DIV_EXPR.
25917 2017-04-07 Richard Biener <rguenther@suse.de>
25919 * tree-affine.c (wide_int_ext_for_comb): Take type rather
25921 (aff_combination_const): Adjust.
25922 (aff_combination_scale): Likewise.
25923 (aff_combination_add_elt): Likewise.
25924 (aff_combination_add_cst): Likewise.
25925 (aff_combination_convert): Likewise.
25926 (add_elt_to_tree): Likewise. Remove unused argument.
25927 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
25929 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
25931 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
25933 * config/arm/arm.c (arm_default_short_enums): Use
25934 ARM_DEFAULT_SHORT_ENUMS.
25935 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
25937 2017-04-06 Jakub Jelinek <jakub@redhat.com>
25940 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
25941 members with redundant out-of-class redeclaration.
25943 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
25946 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
25947 * config/i386/i386.md (*zero_extendsidi2):
25948 Add (?*x,*x) and (?*v,*v) alternatives.
25950 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
25953 * config/i386/i386.c (ix86_expand_builtin)
25954 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
25955 mode from insn data. Convert operands to insn operand mode.
25956 Copy operands that don't satisfy insn predicate to a register.
25958 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
25960 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
25963 2017-04-06 Richard Biener <rguenther@suse.de>
25965 PR tree-optimization/80334
25966 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
25967 preserve alignment of accesses.
25969 2017-04-06 Richard Biener <rguenther@suse.de>
25971 PR tree-optimization/80262
25972 * tree-sra.c (build_ref_for_offset): Preserve address-space
25974 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
25975 Drop useless address-space information on MEM_REF offsets.
25977 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
25979 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
25981 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
25983 PR rtl-optimization/70703
25984 * ira-color.c (update_conflict_hard_regno_costs): Use
25985 HOST_WIDE_INT instead of long.
25987 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
25990 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
25991 not defined for x86_64 target. Add -mmmx target option when __SSE2__
25993 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
25994 for x86_64 target. Handle -m3dnowa option.
25996 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
25998 PR rtl-optimization/70703
25999 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
26000 (update_conflict_hard_regno_costs): Use long instead of unsigned
26001 arithmetic for cost calculation.
26003 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26004 Bernd Edlinger <bernd.edlinger@hotmail.de>
26007 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
26010 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
26013 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
26014 ptr_mode with Pmode throughout.
26015 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
26016 into probe_stack_range and use DImode.
26018 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
26021 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
26022 call_eh_return is true.
26024 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26026 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
26027 Initialize last_match_fntype_index.
26029 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26032 * tree-nvr.c: Include internal-fn.h.
26033 (pass_return_slot::execute): Ignore internal calls without
26036 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26037 Richard Biener <rguenther@suse.de>
26040 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
26041 captures used multiple times, except for the last use.
26042 * generic-match-head.c: Include gimplify.h.
26044 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26046 PR tree-optimization/79390
26047 * target.h (struct noce_if_info): Declare.
26048 * targhooks.h (default_noce_conversion_profitable_p): Declare.
26049 * target.def (noce_conversion_profitable_p): New target hook.
26050 * ifcvt.h (struct noce_if_info): New type, moved from ...
26051 * ifcvt.c (struct noce_if_info): ... here.
26052 (noce_conversion_profitable_p): Renamed to ...
26053 (default_noce_conversion_profitable_p): ... this. No longer
26055 (noce_try_store_flag_constants, noce_try_addcc,
26056 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
26057 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
26058 instead of noce_conversion_profitable_p.
26059 * config/i386/i386.c: Include ifcvt.h.
26060 (ix86_option_override_internal): Don't override
26061 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
26062 (ix86_noce_conversion_profitable_p): New function.
26063 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
26064 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
26065 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
26066 * doc/tm.texi: Regenerated.
26068 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26070 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
26073 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26076 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
26077 instructions for small multiply cores.
26079 2017-04-04 Jeff Law <law@redhat.com>
26081 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
26083 (mips_expand_vec_perm_const): Initialize elements in orig_perm
26084 that are not set by the loop over the elements.
26086 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26089 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
26090 int mode, convert_modes it to mode as unsigned, otherwise use
26091 lowpart_subreg to mode rather than SImode.
26092 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
26093 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
26094 Use DImode instead of SImode for the shift count operand.
26095 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
26098 2017-04-04 Richard Biener <rguenther@suse.de>
26100 PR middle-end/80281
26101 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
26102 arithmetic done for the negate or the plus. Simplify.
26103 (A - (-B) -> A + B): Likewise.
26104 * fold-const.c (split_tree): Make sure to not negate pointers.
26106 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
26108 PR rtl-optimization/60818
26109 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
26110 a compare of comparisons with the thing compared if this results
26111 in a different machine mode.
26113 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
26115 * alias.c (base_alias_check): Fix typo in comment.
26116 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
26117 * cgraphunit.c (symbol_table::compile): Likewise.
26118 * collect2.c (maybe_run_lto_and_relink): Likewise.
26119 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
26120 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
26121 * config/avr/avr.c (avr_map_op_t): Likewise.
26122 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
26123 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
26124 * config/epiphany/epiphany.md (movcc): Likewise.
26125 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
26126 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
26128 * config/mips/mips.c (mips_save_restore_reg): Likewise.
26129 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
26130 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
26131 * config/sh/sh.c (sh_rtx_costs): Likewise.
26132 * fold-const.c (fold_truth_andor): Likewise.
26133 * genautomata.c (collapse_flag): Likewise.
26134 * gengtype.h (struct type::u::s): Likewise.
26135 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
26136 * input.c (FORMAT_AMOUNT): Likewise.
26137 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
26138 (known_aggs_to_agg_replacement_list): Likewise.
26139 * ipa-inline-analysis.c: Likewise.
26140 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
26141 * ipa-polymorphic-call.c
26142 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
26143 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
26144 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
26146 * modulo-sched.c (apply_reg_moves): Likewise.
26147 * omp-expand.c (build_omp_regions_1): Likewise.
26148 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
26149 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
26150 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
26151 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
26152 * value-prof.c: Likewise.
26153 * var-tracking.c (val_reset): Likewise.
26155 2017-04-03 Richard Biener <rguenther@suse.de>
26157 PR tree-optimization/80275
26158 * fold-const.c (split_address_to_core_and_offset): Handle
26161 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
26163 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
26164 descriptors is at least equal to that of functions.
26166 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
26168 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
26170 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
26173 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
26174 (mov<IMOD4:mode>): New expander.
26175 (*mov<IMOD4:mode>_internal): New insn and split pattern.
26177 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
26179 PR rtl-optimization/79405
26180 * fwprop.c (propagations_left): New variable.
26181 (forward_propagate_into): Decrement it.
26182 (fwprop_init): Initialize it.
26183 (fw_prop): If the variable has reached zero, stop propagating.
26184 (fwprop_addr): Ditto.
26186 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26189 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
26190 a FUNCTION_DECL, pass it as decl instead of origin to
26193 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
26195 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
26198 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
26201 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
26202 TARGET_VSX_SMALL_INTEGER.
26204 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26206 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
26207 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
26209 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26211 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
26212 extraction from odd-numbered MSA register.
26214 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26216 PR middle-end/80173
26217 * expmed.c (store_bit_field_1): Don't attempt to create
26218 a word subreg out of hard registers wider than word if they
26219 have HARD_REGNO_NREGS of 1 for their mode.
26221 PR middle-end/80163
26222 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
26223 conversions to integer types wider than word and pointer.
26226 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
26227 (rtx_equal_for_cselib_p): Pass 0 to it.
26228 * cselib.c (cselib_hasher::equal): Likewise.
26229 (rtx_equal_for_cselib_1): Add depth argument. If depth
26230 is 128, don't look up VALUE locs and punt. Increment
26231 depth in recursive calls when walking VALUE locs.
26233 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
26235 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
26236 (make_gcov_file_name): Use the canonical path name for generating
26238 (read_line): Fix handling of files with ascii null bytes.
26240 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
26242 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
26243 to initialise a vector register instead
26244 of using a const_int.
26246 2017-03-30 Jakub Jelinek <jakub@redhat.com>
26248 PR translation/80189
26249 * gimplify.c (omp_default_clause): Use %qs instead of %s in
26250 diagnostic messages.
26252 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
26255 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
26256 (dfp_diex_<mode>): Update mode of operand 1.
26257 * doc/extend.texi (dxex, dxexq): Document change to return type.
26258 (diex, diexq): Document change to argument type.
26260 2017-03-30 Martin Jambor <mjambor@suse.cz>
26263 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
26264 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
26265 it reflects the signature changes performed at the callee side.
26266 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
26267 to cgraph_build_function_type_skip_args.
26268 (build_function_decl_skip_args): Adjust call to the above function.
26270 2017-03-30 Jakub Jelinek <jakub@redhat.com>
26273 * config/i386/sse.md
26274 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
26275 register as dest whenever it is a MEM not rtx_equal_p to the
26276 corresponding dup operand, and when forcing into reg move the
26277 reg into the memory afterwards.
26278 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
26279 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
26280 for the force_reg mode.
26281 (avx512vl_vextractf128<mode>): Use register as dest either
26282 always when a MEM, or when it is a MEM not rtx_equal_p to the
26283 corresponding dup operand, or even not when it is a CONST_VECTOR
26284 depending on the mode and lo vs. hi.
26285 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
26287 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
26288 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
26289 Likewise. Require that operands[2] is even.
26290 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
26291 Remove extraneous parens. Require that operands[2] is a multiple
26293 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
26294 operands[0] is a MEM if <mask_applied>, the predicates/constraints
26295 disallow memory then.
26297 2017-03-30 Richard Biener <rguenther@suse.de>
26299 PR tree-optimization/77498
26300 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
26301 to non-constants over backedges.
26303 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
26305 PR rtl-optimization/80233
26306 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
26307 as last_combined_insn. Do not test for BARRIER_P separately.
26309 2017-03-29 Andreas Schwab <schwab@suse.de>
26312 * calls.c (prepare_call_address): Convert funexp to Pmode before
26313 copying to temp reg.
26315 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26317 PR tree-optimization/80158
26318 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
26319 Handle possible future case of more than one alternate
26321 (replace_rhs_if_not_dup): Likewise.
26322 (replace_one_candidate): Likewise.
26324 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
26326 PR rtl-optimization/80193
26327 * ira.c (ira): Do not check allocation for LRA.
26329 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
26331 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
26332 (nvptx_output_simt_exit): Declare.
26333 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
26334 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
26335 (init_softstack_frame): Move initialization of crtl->is_leaf to...
26336 (nvptx_declare_function_name): ...here. Emit declaration of local
26337 memory space buffer for omp_simt_enter insn.
26338 (nvptx_output_unisimt_switch): New.
26339 (nvptx_output_softstack_switch): New.
26340 (nvptx_output_simt_enter): New.
26341 (nvptx_output_simt_exit): New.
26342 * config/nvptx/nvptx.h (struct machine_function): New fields
26343 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
26344 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
26345 (UNSPECV_SIMT_EXIT): Ditto.
26346 (omp_simt_enter_insn): New insn.
26347 (omp_simt_enter): New expansion.
26348 (omp_simt_exit): New insn.
26349 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
26351 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
26352 (expand_GOMP_SIMT_ENTER_ALLOC): New.
26353 (expand_GOMP_SIMT_EXIT): New.
26354 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
26355 (GOMP_SIMT_ENTER_ALLOC): Ditto.
26356 (GOMP_SIMT_EXIT): Ditto.
26357 * target-insns.def (omp_simt_enter): New insn.
26358 (omp_simt_exit): Ditto.
26359 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
26361 (lower_rec_simd_input_clauses): Implement SIMT privatization.
26362 (lower_rec_input_clauses): Likewise.
26363 (lower_lastprivate_clauses): Handle SIMT privatization.
26365 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
26366 (ompdevlow_adjust_simt_enter): New.
26367 (find_simtpriv_var_op): New.
26368 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
26369 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
26371 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
26372 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
26373 (copy_decl_for_dup_finish): Ditto.
26375 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
26377 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
26380 * config/i386/i386.c (ix86_option_override_internal): Always
26381 allow -mpreferred-stack-boundary=3 for 64-bit targets.
26383 2017-03-28 Bin Cheng <bin.cheng@arm.com>
26385 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
26387 2017-03-28 Bin Cheng <bin.cheng@arm.com>
26389 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
26390 mark new edge's irreducible flag accordign to it.
26391 (vect_do_peeling): Check loop preheader edge's irreducible flag
26392 and pass it to function slpeel_add_loop_guard.
26394 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
26396 PR tree-optimization/80218
26397 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
26398 Update block frequencies and counts.
26400 2017-03-28 Richard Biener <rguenther@suse.de>
26402 PR tree-optimization/78644
26403 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
26404 of a simplification result we may not use it at all.
26406 2017-03-28 Richard Biener <rguenther@suse.de>
26409 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
26410 without arguments, generate default definition of a SSA name.
26412 2017-03-28 Richard Biener <rguenther@suse.de>
26414 PR middle-end/80222
26415 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
26416 TYPE_REF_CAN_ALIAS_ALL references.
26417 * fold-const.c (fold_indirect_ref_1): Likewise.
26419 2017-03-28 Martin Liska <mliska@suse.cz>
26422 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
26423 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
26425 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
26426 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
26428 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
26429 (EXTRA_SPECS): Define.
26430 (SUBTARGET_EXTRA_SPECS): Likewise.
26431 (SUBTARGET_CPP_SPEC): Likewise.
26432 * config/arc/elf.h (EXTRA_SPECS): Renamed to
26433 SUBTARGET_EXTRA_SPECS.
26434 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
26436 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
26438 * config/arc/simdext.md (vst64_insn): Update pattern.
26439 (vld32wh_insn): Likewise.
26440 (vld32wl_insn): Likewise.
26441 (vld64_insn): Likewise.
26442 (vld32_insn): Likewise.
26444 2017-03-28 Marek Polacek <polacek@redhat.com>
26447 * fold-const.c (fold_comparison): Use protected_set_expr_location
26448 instead of SET_EXPR_LOCATION.
26450 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
26452 * tree.c (add_expr): Avoid name lookup warning.
26454 2017-03-27 Jeff Law <law@redhat.com>
26456 PR tree-optimization/80216
26457 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
26458 function name. Limit recursion depth.
26459 (record_temporary_equivalences): Corresponding changes.
26461 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
26463 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
26466 2017-03-27 Jakub Jelinek <jakub@redhat.com>
26469 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
26471 * cfgcleanup.c (reg_note_cfa_p): New array.
26472 (insns_have_identical_cfa_notes): New function.
26473 (old_insns_match_p): Don't cross-jump in between /f
26474 and non-/f instructions. If both i1 and i2 are frame related,
26475 verify all CFA notes, their order and content.
26477 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
26480 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
26481 HImode and SImode with zero extend to DImode to one insn.
26482 (bswap<mode>2_extenddi): Likewise.
26483 (bswapsi2_extenddi): Likewise.
26484 (bswaphi2_extendsi): Likewise.
26485 (bswaphi2): Combine bswap HImode and SImode into one insn.
26486 Separate memory insns from swapping register.
26487 (bswapsi2): Likewise.
26488 (bswap<mode>2): Likewise.
26489 (bswaphi2_internal): Delete, no longer used.
26490 (bswapsi2_internal): Likewise.
26491 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
26492 store, and gpr<-gpr swap insns.
26493 (bswap<mode>2_store): Likewise.
26494 (bswaphi2_reg): Register only splitter, combine with the splitter.
26495 (bswaphi2 splitter): Likewise.
26496 (bswapsi2_reg): Likewise.
26497 (bswapsi2 splitter): Likewise.
26498 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
26499 the insns into load, store, and register/register insns.
26500 (bswapdi2_ldbrx): Likewise.
26501 (bswapdi2_load): Likewise.
26502 (bswapdi2_store): Likewise.
26503 (bswapdi2_reg): Likewise.
26505 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
26507 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
26508 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
26510 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
26513 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
26515 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
26516 special handling for target option conflicts between dform
26517 options (-mpower9-dform, -mpower9-dform-vector,
26518 -mpower9-dform-scalar) and -mno-direct-move.
26520 2017-03-27 Richard Biener <rguenther@suse.de>
26522 PR tree-optimization/80181
26523 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
26525 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26527 * config/arc/predicates.md (move_double_src_operand): Replace the
26528 call to move_double_src_operand with a call to address_operand.
26530 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26532 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
26533 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
26534 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
26536 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26538 * config/arc/predicates.md (long_immediate_loadstore_operand):
26539 Consider scaled addresses cases.
26541 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26543 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
26544 restored when in interrupt.
26545 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
26546 doesn't have delay slot.
26548 2017-03-27 Richard Biener <rguenther@suse.de>
26551 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
26552 inlined thunk clones.
26554 2017-03-27 Jakub Jelinek <jakub@redhat.com>
26557 * asan.c (instrument_derefs): Copy over last operand from
26558 original COMPONENT_REF to the new COMPONENT_REF with
26559 DECL_BIT_FIELD_REPRESENTATIVE.
26560 * ubsan.c (instrument_object_size): Likewise.
26562 2017-03-27 Richard Biener <rguenther@suse.de>
26564 PR tree-optimization/80170
26565 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
26566 sure DR/SCEV didnt fold in constants we do not see when looking
26567 at the reference base alignment.
26569 2017-03-27 Richard Biener <rguenther@suse.de>
26571 PR middle-end/80171
26572 * gimple-fold.c (fold_ctor_reference): Properly guard against
26573 NULL return value from canonicalize_constructor_val.
26575 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
26578 * config/i386/i386.c (ix86_expand_builtin)
26579 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
26580 flags reg setting and flags reg using instructions.
26581 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
26582 clobbering instructions to zero extend op2.
26584 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
26586 * doc/install.texi (Configuration) <--with-aix-soname>:
26587 Update link to AIX ld.
26589 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
26591 PR rtl-optimization/80160
26592 PR rtl-optimization/80159
26593 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
26594 reg_alternate_class into account.
26596 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
26599 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
26600 to consider in curr_insn_transform.
26602 2017-03-24 Jakub Jelinek <jakub@redhat.com>
26604 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
26605 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
26606 and emit_mode_inner.
26608 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26610 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
26611 argument to the overloaded builtin variants. Use the new flag to
26612 deprecate certain builtin variants.
26613 * config/s390/s390-builtin-types.def: Add new builtin types.
26614 * config/s390/s390-builtins.h: Support new flags field for
26615 overloaded builtins.
26616 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
26617 (s390_macro_to_expand): Enable vector float data type.
26618 (s390_cpu_cpp_builtins_internal): Indicate support of the new
26619 builtins by incrementing the __VEC__ version number.
26620 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
26622 (s390_resolve_overloaded_builtin): Emit error messages depending
26623 on the builtin flags.
26624 * config/s390/s390.c (s390_expand_builtin): Support additional
26625 flags argument. Change error message to match the messages
26626 emitted in s390-c.c.
26627 * config/s390/s390.md: New UNSPEC_* constants.
26628 (op_type): Add new instruction types.
26629 * config/s390/vecintrin.h: Add new builtins and test data class
26631 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
26632 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
26633 (VEC_INEXACT, VEC_NOINEXACT): New constants.
26634 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
26635 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
26636 ("vec_mergel<mode>"): V_HW -> VEC_HW.
26638 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
26639 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
26640 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
26641 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
26643 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
26644 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
26645 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
26646 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
26648 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
26649 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
26650 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
26651 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
26652 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
26653 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
26654 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
26656 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
26657 ("vec_scatter_element<V_HW_4:mode>_DI")
26658 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
26659 ("vec_fpint<mode>", "vflls")
26660 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
26661 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
26662 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
26663 ("*vec_cmphe<mode>_cc"): ... these.
26665 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
26666 mode constant instead of magic value.
26668 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26670 * config/s390/s390.c (s390_expand_vec_compare): Support other
26671 vector floating point modes than just V2DF.
26672 (s390_expand_vcond): Likewise.
26673 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
26674 (s390_cannot_change_mode_class): Prevent mode changes between TF
26675 and V1TF in vector registers.
26676 * config/s390/s390.md (DF, SF): New mode attributes.
26677 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
26678 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
26679 SFmode support for VRs.
26680 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
26682 (VFT, VF_HW): New mode iterators.
26683 (vw, sdx): New mode attributes.
26684 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
26685 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
26686 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
26687 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
26688 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
26689 also the new vector floating point modes. Renaming to ...
26691 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
26692 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
26693 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
26694 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
26695 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
26696 ("vec_unordered<mode>"): ... these.
26698 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
26699 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
26700 ("*vec_extendv2df"): New insn definitions.
26702 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26704 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
26705 ("mulditi3_2", "*muldi3_sign"): New patterns.
26706 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
26707 rename the pattern definition.
26709 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26711 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
26713 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
26715 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26717 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
26718 instruction if possible.
26719 * config/s390/vector.md (vec_halfnumelts): New mode
26721 ("*vec_vllezlf<mode>"): New pattern.
26723 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26725 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
26726 ("popcountv4si2", "popcountv2di2"): Rename to ...
26727 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
26728 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
26730 ("popcount<mode>2_vxe"): New pattern.
26732 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26734 * common/config/s390/s390-common.c (processor_flags_table): Add
26736 * config.gcc: Add arch12.
26737 * config/s390/driver-native.c (s390_host_detect_local_cpu):
26738 Default to arch12 for unknown CPU model numbers.
26739 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
26740 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
26741 PROCESSOR_max sanity check.
26742 * config/s390/s390-opts.h (enum processor_type): Add
26744 * config/s390/s390.c (processor_table): Add arch12.
26745 (s390_expand_builtin): Add check for B_VXE flag.
26746 (s390_issue_rate): Add PROCESSOR_ARCH12.
26747 (s390_get_sched_attrmask): Likewise.
26748 (s390_get_unit_mask): Likewise.
26749 (s390_sched_score): Enable z13 scheduling for arch12.
26750 (s390_sched_reorder): Likewise.
26751 (s390_sched_variable_issue): Likewise.
26752 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
26754 (s390_tune_attr): Use z13 scheduling also for arch12.
26755 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
26756 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
26757 (TARGET_VXE_P): New macros.
26758 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
26759 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
26760 * config/s390/s390.opt: Add arch12 as processor_type.
26762 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26764 * config/s390/s390.md
26765 ("fixuns_truncdddi2", "fixuns_trunctddi2")
26766 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
26767 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
26769 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
26770 Rename expanders to ...
26772 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
26773 ("fixuns_truncdddi2_emu"): ... these.
26775 ("fixuns_trunc<mode>si2_emu"): New expander.
26777 ("*fixuns_truncdfdi2_z13"): Rename to ...
26778 ("*fixuns_truncdfdi2_vx"): ... this.
26780 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26782 * config/s390/2964.md: Remove the single element vector compare
26783 instructions which are no longer used.
26784 * config/s390/s390.c (s390_select_ccmode): Remove handling of
26786 (s390_canonicalize_comparison): Remove handling of DFmode
26788 (s390_expand_vec_compare_scalar): Remove function.
26789 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
26790 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
26792 ("*cmp<mode>_ccs"): Add wfcdb instruction.
26794 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26796 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
26798 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
26799 will anyway by matched by mov<mode>_64dfp.
26801 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26803 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
26804 vlef/vstef. Add missing operand to vleif.
26806 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26808 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
26809 pair for all vector types with 64 bit elements.
26810 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
26811 * config/s390/vector.md (V_HW_64): ... here.
26812 (V_128_NOSINGLE): New mode iterator.
26813 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
26814 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
26815 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
26816 ("*vec_load_pairv2di"): Change to ...
26817 ("*vec_load_pair<mode>"): ... this one.
26819 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26821 * config/s390/constraints.md: Add comments.
26822 (jKK): Reject element sizes > 8 bytes.
26823 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
26825 * config/s390/s390.md: Add the s_operand checks formerly in
26826 s390_split_ok_p to various splitters where they are still
26828 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
26829 for 128 bit vectors. Plus two splitters.
26831 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26833 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
26836 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26839 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
26840 error if the boundary argument is not constant.
26842 2017-03-24 Jakub Jelinek <jakub@redhat.com>
26844 PR rtl-optimization/80112
26845 * loop-doloop.c (doloop_condition_get): Don't check condition
26846 if cmp isn't SET with IF_THEN_ELSE src.
26848 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26850 PR tree-optimization/80158
26851 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
26852 replacing a candidate statement, also replace it for the
26853 candidate's alternate interpretation.
26854 (replace_rhs_if_not_dup): Likewise.
26855 (replace_one_candidate): Likewise.
26857 2017-03-24 Richard Biener <rguenther@suse.de>
26859 PR tree-optimization/80167
26860 * graphite-isl-ast-to-gimple.c
26861 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
26863 (translate_isl_ast_to_gimple::get_rename): Likewise.
26865 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
26867 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
26868 handling of certain combinations of target options, including the
26869 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
26870 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
26872 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26875 * config/arm/arm.md (*load_multiple): Add reload_completed to
26876 matching condition.
26878 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26879 Richard Biener <rguenth@suse.de>
26881 PR tree-optimization/79908
26882 PR tree-optimization/80136
26883 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
26884 been cast away, gimplify_and_add suffices.
26886 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
26888 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
26890 2017-03-23 Richard Biener <rguenther@suse.de>
26892 PR tree-optimization/80032
26893 * gimplify.c (gimple_push_cleanup): Forced unconditional
26894 cleanups still have to go to the conditional_cleanups
26897 2017-03-22 Jakub Jelinek <jakub@redhat.com>
26899 PR tree-optimization/80072
26900 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
26902 (next_operand_entry_id): Change type to unsigned int.
26903 (sort_by_operand_rank): Make sure to return the right return value
26904 even if unsigned fields are bigger than INT_MAX.
26905 (struct oecount): Change cnt and id type to unsigned int.
26906 (oecount_hasher::equal): Formatting fix.
26907 (oecount_cmp): Make sure to return the right return value
26908 even if unsigned fields are bigger than INT_MAX.
26909 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
26912 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
26913 TREE_READONLY on result if writing it more than once.
26916 * doc/invoke.texi (-fsanitize=thread): Document that with
26917 -fnon-call-exceptions atomics are not able to throw
26921 * tsan.c: Include tree-eh.h.
26922 (instrument_builtin_call): Call maybe_clean_eh_stmt or
26923 maybe_clean_or_replace_eh_stmt where needed.
26924 (instrument_memory_accesses): Add cfg_changed argument.
26925 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
26926 if it returned true.
26927 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
26929 PR rtl-optimization/63191
26930 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
26931 wrapper function, moved the whole old content into ...
26932 (ix86_delegitimize_address_1): ... this. New inline function.
26933 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
26934 true as last argument instead of ix86_delegitimize_address.
26936 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
26938 * config/aarch64/aarch64.c (generic_branch_cost): Copy
26939 cortexa57_branch_cost.
26941 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
26943 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
26945 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26948 * doc/md.texi (Constraints): Document wA constraint.
26949 * config/rs6000/constraints.md (wA): New.
26950 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
26951 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
26952 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
26953 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
26955 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
26958 * gimplify.c (is_oacc_declared): New function.
26959 (oacc_default_clause): Use it to set default flags for acc declared
26960 variables inside parallel regions.
26961 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
26962 declared variables.
26963 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
26964 declare attribute to any decl as necessary.
26966 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
26969 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
26970 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
26971 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
26972 (arm_arch_lpae): This.
26973 * config/arm/arm.c (arm_arch7ve): Rename into ...
26974 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
26975 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
26978 2017-03-22 Martin Liska <mliska@suse.cz>
26981 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
26982 error message instead of an ICE.
26984 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26986 * doc/extend.texi (6.11 Additional Floating Types): Revise.
26988 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
26990 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
26992 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
26995 2017-03-21 Martin Sebor <msebor@redhat.com>
26997 * doc/extend.texi: Use "cannot" instead of "can't."
26998 * doc/hostconfig.texi: Same.
26999 * doc/install.texi: Same.
27000 * doc/invoke.texi: Same.
27001 * doc/loop.texi: Same.
27002 * doc/md.texi: Same.
27003 * doc/objc.texi: Same.
27004 * doc/rtl.texi: Same.
27005 * doc/tm.texi: Same.
27006 * doc/tm.texi.in: Same.
27007 * doc/trouble.texi: Same.
27009 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
27012 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
27013 (collect_checksum_attributes): Set it.
27014 (die_checksum_ordered): Use it.
27016 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27018 PR tree-optimization/79908
27019 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
27020 change: For a VA_ARG whose LHS has been cast away, use
27021 force_gimple_operand to construct the side effects.
27023 2017-03-21 David Malcolm <dmalcolm@redhat.com>
27025 PR translation/80001
27026 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
27027 more amenable to translation.
27028 (oacc_loop_auto_partitions): Likewise.
27030 2017-03-21 Marek Polacek <polacek@redhat.com>
27031 Martin Sebor <msebor@redhat.com>
27033 PR tree-optimization/80109
27034 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
27035 on INTEGRAL_TYPE_P.
27037 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27038 Segher Boessenkool <segher@kernel.crashing.org>
27041 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
27042 check reg_used_between_p between insn and one of succ or succ2
27043 depending on if succ is artificial insn not inserted into insn
27046 2017-03-21 Martin Liska <mliska@suse.cz>
27048 PR gcov-profile/80081
27049 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
27050 * doc/gcc.texi: Include gcov-dump stuff.
27051 * doc/gcov-dump.texi: New file.
27053 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
27055 PR rtl-optimization/79150
27056 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
27057 conditional jump, if the jump is the last insn of the loop.
27059 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27060 Richard Biener <rguenth@suse.de>
27062 PR tree-optimization/79908
27063 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27064 been cast away, use force_gimple_operand to construct the side
27067 2017-03-21 Martin Liska <mliska@suse.cz>
27069 PR libfortran/79956
27070 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
27073 2017-03-21 Brad Spengler <spender@grsecurity.net>
27076 * plugin.c (htab_hash_plugin): New function.
27077 (add_new_plugin): Use it and adjust.
27078 (parse_plugin_arg_opt): Adjust.
27079 (init_one_plugin): Likewise.
27081 2017-03-21 Richard Biener <rguenther@suse.de>
27083 PR tree-optimization/80032
27084 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
27085 if set force the cleanup to happen unconditionally.
27086 (gimplify_target_expr): Push inserted clobbers with force_uncond
27087 to avoid them being removed by control-dependent DCE.
27089 2017-03-21 Richard Biener <rguenther@suse.de>
27091 PR tree-optimization/80122
27092 * tree-inline.c (copy_bb): Do not expans va-arg packs or
27093 va_arg_pack_len when the inlined call stmt requires pack
27095 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
27097 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27100 * tsan.c (instrument_builtin_call): If the memory model argument
27101 is not a constant, assume it is valid.
27104 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
27107 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27109 PR rtl-optimization/79910
27110 * combine.c (can_combine_p): Do not allow combining an I0 or I1
27111 if its dest is used by an insn before I2 (other than the combined
27112 insns themselves, which are properly handled already).
27114 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27117 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27119 * combine.c (record_used_regs): New static function.
27120 (try_combine): Handle situations where there is an additional
27121 instruction between I2 and I3 which needs to have a LOG_LINK
27125 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27127 * combine.c (try_combine): Delete redundant i1 test. Call
27128 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27130 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27133 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
27134 alternatives 13/14.
27136 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27138 PR tree-optimization/80054
27139 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
27140 the optimization if a PHI or any of its arguments is not dominated
27141 by the candidate's basis. Use gphi* rather than gimple* as
27143 (replace_profitable_candidates): Clean up a gimple* variable that
27144 should be a gphi* variable.
27146 2017-03-20 Martin Sebor <msebor@redhat.com>
27149 * doc/extend.texi (attribute constructor): Document present limitation.
27151 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
27154 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
27155 __POWER9_VECTOR__ #ifdef control, change template definition to
27156 use Power9-specific built-in function.
27157 (vec_any_eq): Likewise.
27158 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
27159 to control outcomes from this test.
27160 (vector_ae_<mode>p): For VEC_F modes, likewise.
27162 2017-03-20 Ian Lance Taylor <iant@google.com>
27164 * config/i386/i386.c (ix86_function_regparm): Save an extra
27165 register for -fsplit-stack with DECL_STATIC_CHAIN.
27167 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
27170 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
27171 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
27173 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
27175 * config/riscv/riscv.c (riscv_print_operand): Use "fence
27177 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
27180 2017-03-20 Marek Polacek <polacek@redhat.com>
27183 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
27185 2017-03-20 Richard Biener <rguenther@suse.de>
27187 PR tree-optimization/80113
27188 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
27189 allocate extra SSA name for PHI def.
27190 (add_close_phis_to_outer_loops): Likewise.
27191 (add_close_phis_to_merge_points): Likewise.
27192 (copy_loop_close_phi_args): Likewise.
27193 (copy_cond_phi_nodes): Likewise.
27195 2017-03-20 Martin Liska <mliska@suse.cz>
27197 PR middle-end/79753
27198 * tree-chkp.c (chkp_build_returned_bound): Do not build
27199 returned bounds for a LHS that's not a BOUNDED_P type.
27201 2017-03-20 Martin Liska <mliska@suse.cz>
27205 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
27206 COMPLEX_CST and VECTOR_CST.
27208 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27211 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
27212 target operand. A new splitter adds the clobber statement in case
27213 the target operand is dead anyway.
27215 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
27217 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
27218 to age-old versions of binutils and glibc.
27220 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
27222 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
27224 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27226 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
27228 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27230 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
27231 requirement for binutils 2.13.
27233 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27235 * combine.c (try_combine): Delete redundant i1 test. Call
27236 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27238 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
27240 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
27241 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
27243 <riscv64-*-elf>: Re-arrange section
27244 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
27245 <riscv32-*-linux>: Likewise.
27246 <riscv64-*-elf>: Likewise
27247 <riscv64-*-linux>: Likewise.
27249 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
27252 * aarch64.opt(verbose-cost-dump): Fix typo.
27254 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
27257 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
27258 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
27260 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27262 * reload.c (find_reloads): When reloading a nonoffsettable address,
27263 use RELOAD_OTHER for it and its address reloads.
27265 PR rtl-optimization/79910
27266 * combine.c (record_used_regs): New static function.
27267 (try_combine): Handle situations where there is an additional
27268 instruction between I2 and I3 which needs to have a LOG_LINK
27271 2017-03-17 Jeff Law <law@redhat.com>
27273 PR tree-optimization/71437
27274 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
27275 conditional in the hash table first.
27276 (vrp_dom_walker::before_dom_children): Extract condition from
27277 ASSERT_EXPR. Record condition, its inverion and any implied
27278 conditions as well.
27280 2017-03-17 Marek Polacek <polacek@redhat.com>
27281 Markus Trippelsdorf <markus@trippelsdorf.de>
27283 PR tree-optimization/80079
27284 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
27287 2017-03-17 Richard Biener <rguenther@suse.de>
27289 PR middle-end/80075
27290 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
27291 Properly verify the LHS before the RHS possibly claims to be
27293 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
27296 2017-03-17 Martin Jambor <mjambor@suse.cz>
27298 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
27299 (List of -O2 options): Likewise.
27300 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
27303 2017-03-17 Tom de Vries <tom@codesourcery.com>
27305 * gcov-dump.c (print_usage): Print bug_report_url.
27307 2017-03-17 Richard Biener <rguenther@suse.de>
27309 PR middle-end/80050
27310 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
27311 (parser::peek): Likewise.
27313 2017-03-17 Richard Biener <rguenther@suse.de>
27315 PR tree-optimization/80048
27316 * sese.c (free_sese_info): Properly release rename_map and
27317 copied_bb_map elements.
27319 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
27321 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
27322 Add linked-list forward and backlinks. Insert on
27323 construction, remove on destruction.
27324 (class pass_store_merging): Add m_stores_head field.
27325 (pass_store_merging::terminate_and_process_all_chains):
27326 Iterate over m_stores_head list.
27327 (pass_store_merging::terminate_all_aliasing_chains):
27329 (pass_store_merging::execute): Check for debug stmts first.
27330 Push new chains onto the m_stores_head stack.
27332 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
27335 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
27336 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
27337 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
27339 2017-03-16 Jeff Law <law@redhat.com>
27341 PR tree-optimization/71437
27342 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
27343 member function. Implementation moved into after_dom_children
27344 member function and into the threader's thread_outgoing_edges
27346 (dom_opt_dom_walker::after_dom_children): Simplify by moving
27347 some code into new thread_outgoing_edges.
27348 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
27349 definition. Simplify marker handling (do it here). Assume we always
27350 have the available expression and the const/copies tables.
27351 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
27353 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
27354 * tree-vrp.c (equiv_stack): No longer file scoped.
27355 (vrp_dom_walker): New class.
27356 (vrp_dom_walker::before_dom_children): New member function.
27357 (vrp_dom_walker::after_dom_children): Likewise.
27358 (identify_jump_threads): Setup domwalker. Use it rather than
27359 walking edges in a random order by hand. Simplify setup/finalization.
27360 (finalize_jump_threads): Remove.
27361 (vrp_finalize): Do not call identify_jump_threads here.
27362 (execute_vrp): Do it here instead and call thread_through_all_blocks
27365 PR tree-optimization/71437
27366 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
27368 (simplify_stmt_for_jump_threading): Add basic_block argument. All
27370 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
27371 (dom_opt_dom_walker::thread_across_edge): Remove
27372 handle_dominating_asserts argument. All callers changed.
27373 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
27374 changes. Remove calls to lhs_of_dominating_assert. Other
27375 uses of handle_dominating_asserts turn into unconditional code
27376 (simplify_control_stmt_condition_1): Likewise.
27377 (simplify_control_stmt_condition): Likewise.
27378 (thread_through_normal_block, thread_across_edge): Likewise.
27379 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
27380 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
27381 object if it is not an SSA_NAME.
27382 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
27383 before calling into the VRP specific simplifiers.
27384 (identify_jump_threads): Remove handle_dominating_asserts
27387 2017-03-16 Jakub Jelinek <jakub@redhat.com>
27390 * tree-diagnostic.c (default_tree_printer): No longer static.
27391 * tree-diagnostic.h (default_tree_printer): New prototype.
27393 2017-03-16 Tamar Christina <tamar.christina@arm.com>
27395 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
27396 Change ins into fmov.
27398 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27400 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
27401 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
27402 Use h_con constraint for operand 1.
27403 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
27404 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
27406 2017-03-15 Jeff Law <law@redhat.com>
27408 PR tree-optimization/71437
27409 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
27410 (record_temporary_equivalences): Use it.
27412 PR tree-optimization/71437
27413 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
27414 tree-ssa-scopedtables.
27415 (lookup_avail_expr, build_and_record_new_cond): Likewise.
27416 (record_conditions, record_cond, vuse_eq): Likewise.
27417 (record_edge_info): Adjust to API tweak of record_conditions.
27418 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
27419 (record_temporary_equivalences, optimize_stmt): Likewise.
27420 (eliminate_redundant_computations): Likewise.
27421 (record_equivalences_from_stmt): Likewise.
27422 * tree-ssa-scopedtables.c: Include options.h and params.h.
27423 (vuse_eq): New function, moved from tree-ssa-dom.c
27424 (build_and_record_new_cond): Likewise.
27425 (record_conditions): Likewise. Accept vector of conditions rather
27426 than edge_equivalence structure for first argument.
27427 for the first argument.
27428 (avail_exprs_stack::lookup_avail_expr): New member function, moved
27429 from tree-ssa-dom.c.
27430 (avail_exprs_stack::record_cond): Likewise.
27431 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
27432 from tree-ssa-dom.c.
27433 (avail_exprs_stack): Add new member functions lookup_avail_expr
27435 (record_conditions): Declare.
27437 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
27440 * lra-constraints.c (process_alt_operands): Increase reject for
27441 reloading an input/output operand.
27443 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
27446 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
27447 insns to convert from signed/unsigned char/short to IEEE 128-bit
27449 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
27451 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
27454 * config/i386/i386.c (ix86_vector_duplicate_value): Create
27455 subreg of inner mode for values already in registers.
27457 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
27459 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
27460 iteration reg is used after the loop.
27462 2017-03-14 Martin Sebor <msebor@redhat.com>
27464 PR tree-optimization/79800
27465 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
27466 precision in negative-positive range.
27467 (format_floating): Call non-const overload with adjusted precision.
27469 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
27472 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
27475 2017-03-14 Martin Sebor <msebor@redhat.com>
27477 PR middle-end/80020
27478 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
27479 * builtins.def (aligned_alloc): Use it.
27482 * Makefile.in (GTFILES): Add calls.c.
27483 * calls.c: Include "gt-calls.h".
27485 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
27487 PR rtl-optimization/79728
27488 * regs.h (struct target_regs): New field
27489 x_contains_allocatable_regs_of_mode.
27490 (contains_allocatable_regs_of_mode): New macro.
27491 * reginfo.c (init_reg_sets_1): Initialize it, and change
27492 contains_reg_of_mode so it includes global regs as well.
27493 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
27494 rather than contains_regs_of_mode.
27496 2017-03-14 Martin Liska <mliska@suse.cz>
27498 * doc/invoke.texi: Document options that can't be combined with
27499 -fcheck-pointer-bounds.
27501 2017-03-14 Martin Liska <mliska@suse.cz>
27503 PR middle-end/79831
27504 * doc/invoke.texi (-Wchkp): Document the option.
27506 2017-03-14 Martin Liska <mliska@suse.cz>
27508 * Makefile.in: Install gcov-dump.
27510 2017-03-14 Martin Liska <mliska@suse.cz>
27512 * multiple_target.c (expand_target_clones): Bail out for
27513 an invalid attribute.
27515 2017-03-14 Richard Biener <rguenther@suse.de>
27517 * alias.c (struct alias_set_entry): Pack properly.
27518 * cfgloop.h (struct loop): Likewise.
27519 * cse.c (struct set): Likewise.
27520 * ipa-utils.c (struct searchc_env): Likewise.
27521 * loop-invariant.c (struct invariant): Likewise.
27522 * lra-remat.c (struct cand): Likewise.
27523 * recog.c (struct change_t): Likewise.
27524 * rtl.h (struct address_info): Likewise.
27525 * symbol-summary.h (function_summary): Likewise.
27526 * tree-loop-distribution.c (struct partition): Likewise.
27527 * tree-object-size.c (struct object_size_info): Likewise.
27528 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
27529 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
27530 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
27531 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
27532 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
27533 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
27534 (struct _stmt_vec_info): Likewise.
27536 2017-03-14 Martin Liska <mliska@suse.cz>
27539 * multiple_target.c (create_dispatcher_calls): Check that
27540 a target can create a function dispatcher.
27542 2017-03-14 Martin Liska <mliska@suse.cz>
27545 * multiple_target.c (expand_target_clones): Drop local.local
27546 flag for default implementation.
27548 2017-03-14 Richard Biener <rguenther@suse.de>
27550 PR tree-optimization/80030
27551 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
27553 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
27555 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
27556 gcc_fallthrough() instead of __attribute__((fallthrough));
27558 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
27560 * doc/gcc.texi: Remove "up" link to (DIR).
27561 * doc/gccint.texi: Ditto.
27563 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
27565 * doc/install.texi (Specific) <avr>: Remove reference to
27568 2017-03-13 Jeff Law <law@redhat.com>
27570 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
27571 attribute rather than comments.
27573 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
27574 match_scratch operand is highest.
27576 2017-03-13 Martin Liska <mliska@suse.cz>
27578 PR middle-end/78339
27579 * ipa-pure-const.c (warn_function_noreturn): If the declarations
27580 is a CHKP clone, use original declaration.
27582 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27584 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
27585 (arc_conditional_register_usage): Use a different allocation order
27586 when optimizing for size.
27587 * common/config/arc/arc-common.c (arc_option_optimization_table):
27588 Section anchors default on when optimizing for size.
27590 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27592 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
27594 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27596 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
27597 * config/arc/arc.md (cpu_facility): Add cd variant.
27598 (*movqi_insn): Add code density variant.
27599 (*movhi_insn): Likewise.
27600 (*movqi_insn): Likewise.
27601 (*addsi3_mixed): Likewise.
27602 (subsi3_insn): Likewise.
27604 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27606 * config/arc/arc.md (movsi_cond_exec): Update constraint.
27608 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27610 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
27611 expressions with MINUS and UNARY ops.
27613 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27616 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
27618 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
27619 between vec_select and vector argument.
27620 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
27621 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
27622 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
27623 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
27624 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
27625 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
27627 2017-03-13 Richard Biener <rguenther@suse.de>
27630 * params.def (vect-max-peeling-for-alignment): Fix typo.
27632 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
27634 * doc/install.texi (Specific) <mips-*-*>: Remove description of
27635 issue that only occurred with binutils below 2.18.
27637 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
27639 * doc/install.texi (Specific) <cris-axis-elf>: No longer
27640 refer to binutils 2.11/2.12 minimum.
27642 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
27644 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
27645 ftp.kernel.org and simplify binutils requirement.
27647 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
27649 * doc/invoke.texi (Warning Options): Fix spelling of link-time
27651 (Optimize Options): Ditto. Also remove redundancy.
27653 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27655 PR translation/79848
27656 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
27658 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
27659 to G_ to avoid double translation.
27661 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27663 PR translation/79923
27664 * auto-profile.c (get_combined_location): Convert leading
27665 character of diagnostics to lower case and remove trailing period.
27666 (read_profile): Likewise for various diagnostics.
27667 * config/arm/arm.c (arm_option_override): Remove trailing period
27668 from various diagnostics.
27669 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
27670 (msp430_expand_delay_cycles): Likewise.
27672 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27675 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
27676 full command-line argument, rather than just "str".
27677 (aarch64_validate_march): Likewise.
27678 (aarch64_validate_mtune): Likewise.
27680 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
27682 PR rtl-optimization/78911
27683 * lra-assigns.c (must_not_spill_p): New function.
27684 (spill_for): Use it.
27686 2017-03-10 Jakub Jelinek <jakub@redhat.com>
27688 PR tree-optimization/79981
27689 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
27690 ATOMIC_COMPARE_EXCHANGE ifn result.
27691 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
27692 IFN_ATOMIC_COMPARE_EXCHANGE.
27694 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27697 * opts.c (parse_sanitizer_options): Add missing question mark to
27698 "did you mean" message.
27700 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27702 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
27704 (VMULEUH_UNS): Likewise.
27705 (VMULOUB_UNS): Likewise.
27706 (VMULOUH_UNS): Likewise.
27707 * config/rs6000/rs6000.c (builtin_function_type): Remove
27708 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
27710 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27713 * read-rtl-function.c (function_reader::read_rtx_operand): Update
27714 x with result of extra_parsing_for_operand_code_0.
27715 (function_reader::extra_parsing_for_operand_code_0): Convert
27716 return type from void to rtx, returning x. When reading
27717 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
27718 larger size containing struct block_symbol.
27720 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
27722 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
27723 -mfloat128-hardware without -m64.
27725 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
27728 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
27729 entries to the case statement that marks unsigned arguments to
27730 overloaded functions.
27732 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
27734 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
27735 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
27737 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
27740 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
27741 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
27743 2017-03-10 Martin Liska <mliska@suse.cz>
27747 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
27748 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
27751 2017-03-10 Olivier Hainque <hainque@adacore.com>
27753 * tree-switch-conversion (array_value_type): Start by resetting
27754 candidate type to it's main variant.
27756 2017-03-10 Jakub Jelinek <jakub@redhat.com>
27758 PR rtl-optimization/79909
27759 * combine.c (try_combine): Use simplify_replace_rtx on individual
27760 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
27761 of the whole CALL_INSN_FUNCTION_USAGE.
27763 PR tree-optimization/79972
27764 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
27765 get_range_info on SSA_NAMEs. Formatting fixes.
27767 2017-03-10 Richard Biener <rguenther@suse.de>
27768 Jakub Jelinek <jakub@redhat.com>
27770 PR tree-optimization/77975
27771 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
27772 edge to be constant.
27773 (get_val_for): For constant x return it. Formatting fix.
27774 (loop_niter_by_eval): Avoid pointless looping if the next iteration
27775 would use the same bases as the current one.
27777 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27779 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
27780 instead of vec_select for V1TImode.
27781 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
27783 (VSX_LE_128): Add V1TI to this mode iterator.
27784 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
27785 (*vsx_le_perm_store_<mode>): Likewise.
27786 (pre-reload splitter for VSX stores): Likewise.
27787 (post-reload splitter for VSX stores): Likewise.
27788 (*vsx_xxpermdi2_le_<mode>): Likewise.
27789 (*vsx_lxvd2x2_le_<mode>): Likewise.
27790 (*vsx_stxvd2x2_le_<mode>): Likewise.
27792 2017-03-09 Michael Eager <eager@eagercon.com>
27794 Correct failures with --enable-checking=yes,rtl.
27796 * config/microblaze/microblaze.c (microblaze_expand_shift):
27797 Replace GET_CODE test with CONST_INT_P and INTVAL test with
27798 test for const0_rtx.
27799 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
27800 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
27802 2017-03-09 Richard Biener <rguenther@suse.de>
27804 PR tree-optimization/79977
27805 * graphite-scop-detection.c (scop_detection::merge_sese):
27806 Handle the case of extra exits to blocks dominating the entry.
27808 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
27810 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
27813 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
27815 PR rtl-optimization/79949
27816 * lra-constraints.c (process_alt_operands): Check memory when
27817 trying to predict a cycle. Print about the overall increase.
27819 2017-03-09 Richard Biener <rguenther@suse.de>
27821 PR middle-end/79971
27822 * gimple-expr.c (useless_type_conversion_p): Preserve
27823 TYPE_SATURATING for fixed-point types.
27825 2017-03-09 Richard Biener <rguenther@suse.de>
27828 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
27829 alignment of BLKmode params.
27831 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27834 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
27835 (VALL_NO_V2Q): Likewise.
27837 * config/aarch64/aarch64-simd.md
27838 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
27840 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
27841 VALL_NO_V2Q mode iterator.
27842 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
27844 2017-03-09 Martin Liska <mliska@suse.cz>
27846 PR tree-optimization/79631
27847 * tree-chkp-opt.c (chkp_is_constant_addr): Call
27848 tree_int_cst_sign_bit just for INTEGER constants.
27850 2017-03-09 Martin Liska <mliska@suse.cz>
27854 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
27857 2017-03-09 Marek Polacek <polacek@redhat.com>
27860 * tree.c (inchash::add_expr): Handle TREE_VEC.
27862 2017-03-09 Martin Liska <mliska@suse.cz>
27865 (chkp_narrow_size_and_offset): New function.
27866 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
27867 (void chkp_parse_bit_field_ref): New function.
27868 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
27869 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
27871 2017-03-09 Martin Liska <mliska@suse.cz>
27874 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
27875 (chkp_find_bounds_1): Remove gcc_unreachable.
27877 2017-03-09 Jakub Jelinek <jakub@redhat.com>
27880 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
27881 BUILT_IN_SYNC*, determine the access type from the size suffix and
27882 always build a MEM_REF with that type. Handle forgotten
27883 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
27886 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
27887 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
27888 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
27889 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
27890 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
27891 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
27892 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
27893 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
27894 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
27895 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
27896 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
27897 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
27898 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
27899 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
27900 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
27901 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
27902 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
27903 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
27904 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
27905 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
27906 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
27907 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
27908 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
27909 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
27910 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
27911 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
27912 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
27913 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
27914 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
27915 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
27916 definitions outside of __OPTIMIZE__ guarded section.
27919 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
27920 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
27921 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
27922 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
27925 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27927 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
27928 ("vfenez<mode>"): Add missing constraints.
27930 2017-03-08 Martin Sebor <msebor@redhat.com>
27933 * config/nds32/nds32.c (nds32_option_override):
27934 Fix misspelled diagnostic.
27936 2017-03-08 Jakub Jelinek <jakub@redhat.com>
27939 * gimplify.c (gimplify_omp_for): Replace index var in outer
27940 taskloop statement with an artificial variable and add
27941 OMP_CLAUSE_PRIVATE clause for it.
27943 2017-03-08 Richard Biener <rguenther@suse.de>
27945 PR tree-optimization/79955
27946 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
27947 for accesses that are completely outside of the variable.
27949 2017-03-08 Andrew Haley <aph@redhat.com>
27951 PR tree-optimization/79943
27952 * tree-ssa-loop-split.c (compute_new_first_bound): When
27953 calculating the new upper bound, (END-BEG) should be added, not
27956 2017-03-08 Jakub Jelinek <jakub@redhat.com>
27958 * config/avr/avr.md (setmemhi): Make sure match_dup
27959 operand number comes before match_scratch.
27961 2017-03-08 Richard Biener <rguenther@suse.de>
27963 PR tree-optimization/79920
27964 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
27965 with ncopies == 1 to ...
27966 (vect_transform_slp_perm_load): ... here. Properly compute
27967 all element loads by iterating VF times over the group. Do
27968 not handle ncopies (computed in a broken way) in
27969 vect_create_mask_and_perm.
27971 2017-03-08 Jakub Jelinek <jakub@redhat.com>
27974 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
27975 is a uniform vector, use uniform_vector_p return value instead of
27976 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
27978 2017-03-07 Marek Polacek <polacek@redhat.com>
27980 PR middle-end/79809
27981 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
27982 (alloca_call_type): Likewise.
27984 2017-03-07 Martin Liska <mliska@suse.cz>
27986 * gcov.c (process_args): Put comment to correct location.
27988 2017-03-07 Martin Liska <mliska@suse.cz>
27990 PR middle-end/68270
27991 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
27992 Use array_at_struct_end_p instead of DECL_CHAIN (field).
27993 (chkp_narrow_bounds_for_field): Likewise.
27994 (chkp_parse_array_and_component_ref): Pass one more argument to
27997 2017-03-07 Richard Biener <rguenther@suse.de>
27999 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
28002 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
28004 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
28005 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
28007 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28010 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
28011 to end of description.
28012 (PARAM_MAX_STORES_TO_MERGE): Likewise.
28014 2017-03-07 Jakub Jelinek <jakub@redhat.com>
28016 PR rtl-optimization/79901
28017 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
28019 (*avx512f_<code><mode>3<mask_name>): ... this.
28020 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
28021 iterator instead of VI8_AVX2_AVX512BW.
28023 PR rtl-optimization/79901
28024 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
28025 min/max expander, expand it using expand_vec_cond_expr.
28028 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
28031 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28034 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
28035 to void * for PCH reasons.
28036 * dwarf2out.c (output_loc_operands, output_die): Cast
28037 v.val_vec.array to unsigned char *.
28039 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
28042 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
28045 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
28047 PR rtl-optimization/79571
28048 * lra-constraints.c (process_alt_operands): Calculate static
28049 reject and subtract it from overall when only addresses will be
28052 2017-03-06 Julia Koval <julia.koval@intel.com>
28055 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
28056 incoming stack boundary to 128 for 64-bit targets.
28058 2017-03-06 Richard Biener <rguenther@suse.de>
28060 PR tree-optimization/79894
28061 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
28062 to NULL after folding it.
28064 2017-03-06 Richard Biener <rguenther@suse.de>
28066 PR tree-optimization/79824
28067 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
28068 check disabling peeling for gaps.
28070 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
28072 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28073 attributes): Document gettimeofday.
28075 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28077 * config/s390/s390.c (s390_option_override_internal): Set
28078 PARAM_MIN_VECT_LOOP_BOUND
28080 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28082 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
28083 * config/s390/s390.md: Likewise.
28085 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28088 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
28089 (<avx2_avx512>_perm<mode>): Rename to ...
28090 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
28092 (<avx512>_perm<mode>_mask): Rename to ...
28093 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
28095 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
28096 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
28097 instead of VI8F_256_512.
28098 (avx512f_perm<mode>): New define_expand.
28099 (avx512f_perm<mode>_mask): Likewise.
28100 (avx512f_perm<mode>_1<mask_name>): New define_insn.
28101 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
28103 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28105 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
28106 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
28108 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
28110 2017-03-06 Martin Liska <mliska@suse.cz>
28113 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
28114 when having a SSA NAME w/o VAR_DECL assigned to it.
28116 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28118 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
28119 msa_dpsub_<su>_d): Fix MODE for vec_select.
28121 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28123 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
28125 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
28127 2017-03-06 Richard Biener <rguenther@suse.de>
28129 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
28130 * plugin.c (register_plugin_info): Likewise.
28131 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
28133 2017-03-05 Jakub Jelinek <jakub@redhat.com>
28135 * config/i386/sse.md (sse_storehps, sse_storelps,
28136 avx_<castmode><avxsizesuffix>_<castmode>,
28137 avx512f_<castmode><avxsizesuffix>_<castmode>,
28138 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
28139 in condition that at least one operand is not a MEM.
28141 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28143 PR middle-end/79805
28144 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
28145 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
28147 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
28148 gimple_call_nothrow_p flag based on whether original builtin can throw.
28149 If it can, emit following stmts on the fallthrough edge.
28150 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
28151 don't create new bb if inserting just debug stmts on the edge, try to
28152 insert them on the fallthru bb or just reset debug stmts.
28154 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
28157 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
28158 restore recog_data (including the operand rtxes inside it) around
28159 the call to get_insn_template.
28161 2017-03-03 Martin Sebor <msebor@redhat.com>
28163 PR tree-optimization/79699
28164 * context.c (context::~context): Free MPFR caches to avoid
28165 a memory leak on program exit.
28167 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28169 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
28170 Use wide_int::ulow () instead of .elt (0).
28172 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28174 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
28175 (*pushxf): Limit oF constraint to 32bit targets and add oC
28176 constraint for 64bit targets.
28177 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
28178 (*pushdf): Change rmF constraint to rmC.
28180 2017-03-03 Martin Liska <mliska@suse.cz>
28182 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
28183 Remove unused variable.
28185 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28188 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
28189 is a memory operand, increase num_memory.
28190 (ix86_expand_args_builtin): Likewise.
28192 2017-03-03 Jan Hubicka <jh@suse.cz>
28195 * ipa-devirt.c (maybe_record_node): Properly handle
28196 __cxa_pure_virtual visibility.
28198 2017-03-03 Martin Liska <mliska@suse.cz>
28200 PR tree-optimization/79803
28201 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
28203 (pass_loop_prefetch::execute): Disabled optimization if an
28204 assumption about L1 cache size is not met.
28206 2017-03-03 Martin Liska <mliska@suse.cz>
28208 PR rtl-optimization/79574
28209 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
28210 (hash_scan_set): Likewise.
28211 (dump_hash_table): Likewise.
28212 (hoist_code): Likewise.
28214 2017-03-03 Richard Biener <rguenther@suse.de>
28216 * fixed-value.c (fixed_from_string): Restore use of elt (1)
28217 in place of uhigh ().
28218 (fixed_convert_from_real): Likewise.
28220 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28223 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
28225 2017-03-03 Richard Biener <rguenther@suse.de>
28227 PR middle-end/79818
28228 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
28229 TYPE_OVERFLOW_UNDEFINED check.
28231 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28233 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
28235 (vector_ae_<mode>_p): Likewise.
28236 (vector_nez_<mode>_p): Likewise.
28237 (vector_ne_v2di_p): Likewise.
28238 (vector_ae_v2di_p): Likewise.
28239 (vector_ne_<mode>_p): Likewise.
28240 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
28242 (vsx_tsqrt<mode>2_fe): Likewise.
28244 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
28247 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
28249 2017-03-02 Jakub Jelinek <jakub@redhat.com>
28251 PR rtl-optimization/79780
28252 * cprop.c (one_cprop_pass): When second and further conditional trap
28253 in a single basic block is turned into an unconditional trap, turn it
28254 into a deleted note to avoid RTL verification failures.
28256 2017-03-02 Richard Biener <rguenther@suse.de>
28258 * fold-const.c (const_binop): Use ulow () instead of elt (0).
28260 2017-03-02 Richard Biener <rguenther@suse.de>
28262 PR tree-optimization/79345
28264 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
28265 param and abort the walk, returning -1 if it is hit.
28266 (walk_aliased_vdefs): Take a limit param and pass it on.
28267 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
28268 defaulting to 0 and return a signed int.
28269 * tree-ssa-uninit.c (struct check_defs_data): New struct.
28270 (check_defs): New helper.
28271 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
28272 about uninitialized memory.
28273 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
28274 bogus uninitialized warning.
28275 (fixed_convert_from_real): Likewise.
28277 2017-03-02 Bin Cheng <bin.cheng@arm.com>
28279 PR tree-optimization/66768
28280 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
28281 iv_use if base object can't be determined.
28283 2017-03-02 Jakub Jelinek <jakub@redhat.com>
28285 PR tree-optimization/79345
28286 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
28287 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
28288 (get_pattern_stats): Initialize it.
28289 * genemit.c (gen_expand): Verify match_scratch numbers come after
28290 match_operand/match_dup numbers.
28291 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
28292 match_scratch numbers.
28293 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
28295 * config/s390/s390.md (trunctdsd2): Likewise.
28297 2017-03-02 Richard Biener <rguenther@suse.de>
28299 * wide-int.h (wide_int_storage::operator=): Implement in terms
28302 2017-03-02 Richard Biener <rguenther@suse.de>
28304 PR tree-optimization/79777
28305 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
28306 the to insert expression to sth existing.
28308 2017-03-01 Martin Sebor <msebor@redhat.com>
28310 PR middle-end/79692
28311 * gimple-ssa-sprintf.c
28312 (directive::known_width_and_precision): New function.
28313 (format_integer): Use it.
28314 (get_mpfr_format_length): Consider the full range of precision
28315 when computing %g output with the # flag. Set the likely byte
28316 count to 3 rather than 1 when precision is indeterminate.
28317 (format_floating): Correct the lower bound of precision.
28319 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28321 * doc/invoke.texi: Document default code model for 64-bit Linux.
28323 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28326 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
28327 udiv rather than div since input pattern is unsigned.
28329 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
28331 * config/i386/i386.c (print_reg): Warn for values of
28332 unsupported size in integer register.
28334 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
28337 * config/rs6000/predicates.md (current_file_function_operand): Do
28338 not allow self calls to be local if the function is replaceable.
28340 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
28343 * config/rs6000/altivec.h (vec_ctz and others): Change the
28344 preprocessor macro that controls conditional compilation from
28345 _ARCH_PWR9 to __POWER9_VECTOR__.
28346 (vec_all_ne): Change parameterization of __altivec_scalar_pred
28347 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
28348 control (instead of _ARCH_PWR9 control) so that template
28349 definition uses power9-specific function.
28350 (vec_any_eq): Likewise.
28351 (vec_all_ne): Change macro definition to use a power9-specific
28352 expansion under #ifdef __POWER9_VECTOR__ control (instead of
28353 _ARCH_PWR9 control).
28354 (vec_any_eq) Likewise.
28355 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
28356 expansion for CMPNEF to remove support for xvcmpnesp instruction.
28357 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
28358 support for xvcmpnedp instruction.
28359 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
28360 macro expansion so that Power9 implementation of vec_all_ne does
28361 not use the AltiVec predicate framework.
28362 (VCMPNEH_P): Likewise.
28363 (VCMPNEW_P): Likewise.
28364 (VCMPNED_P): Likewise.
28365 (VCMPNEFP_P): Likewise.
28366 (VCMPNEDP_P): Likewise.
28367 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
28368 implementation of vec_any_eq to not use AltiVec predicate
28370 (VCMPAEH_P): Likewise.
28371 (VCMPAEW_P): Likewise.
28372 (VCMPAED_P): Likewise.
28373 (VCMPAEFP_P): Likewise.
28374 (VCMPAEDP_P): Likewise.
28375 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
28376 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
28377 not use the AltiVec predicate framework.
28378 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
28379 of vec_any_eq to not use AltiVec predicate framework.
28380 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
28381 support for predefined __POWER9_VECTOR__ macro to indicate that
28382 Power9 instruction selection is enabled.
28383 (altivec_overloaded_builtins): Remove extraneous
28384 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
28385 function argument types RS6000_BTI_bool_V16QI and
28386 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
28387 entry for overloaded function argument types RS6000_BTI_bool_V4SI
28388 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
28389 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
28390 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
28391 Power9 for implementations of vec_cmpne. Change the signature for
28392 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
28393 (representing vec_all_ne) to remove the previously described first
28394 argument of type RS6000_BTI_INTSI, as this was an artifact of
28395 reliance on the AltiVec predicate framework, which is no longer
28396 used in the implementation of these functions. Add
28397 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
28398 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
28399 since, unlike the AltiVec predicate framework implementation, we
28400 do not share function descriptors between vec_alle and vec_anyeq.
28401 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
28402 set of modes that receive special treatment even when
28403 TARGET_P9_VECTOR is true. The special treatment emits code that
28404 does not depend on Power9 instructions.
28405 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
28406 define_expand to not rely on AltiVec predicate framework.
28407 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
28409 (vector_ne_v2di_p): Change this define_expand to not rely on
28410 AltiVec predicate framework.
28411 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
28413 (vector_ne_<mode>_p): Change this define_expand to not rely on
28414 AltiVec predicate framework.
28415 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
28417 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
28418 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
28419 define_insn pattern.
28420 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
28421 define_insn pattern because the xvcmpne<VSs>. instruction is not
28423 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
28424 instruction is not supported.
28426 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28428 * config/nvptx/nvptx.c: Include intl.h.
28430 2017-03-01 Martin Jambor <mjambor@suse.cz>
28433 * ipa-prop.h (ipa_bits): Removed field known.
28434 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
28435 to pointers. Adjusted their comments to warn about their sharing.
28436 (ipcp_transformation_summary): Change bits to a vector of pointers.
28437 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
28438 (ipa_get_ipa_bits_for_value): Declare.
28439 * tree-vrp.h (value_range): Mark as GTY((for_user)).
28440 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
28441 (ipa_bits_hash_table): Likewise.
28442 (ipa_vr_ggc_hash_traits): Likewise.
28443 (ipa_vr_hash_table): Likewise.
28444 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
28445 being pointers and vr_known being removed.
28446 (ipa_set_jf_unknown): Likewise.
28447 (ipa_get_ipa_bits_for_value): New function.
28448 (ipa_set_jfunc_bits): Likewise.
28449 (ipa_get_value_range): New overloaded functions.
28450 (ipa_set_jfunc_vr): Likewise.
28451 (ipa_compute_jump_functions_for_edge): Use the above functions to
28452 construct bits and vr parts of jump functions.
28453 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
28454 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
28456 (ipcp_grow_transformations_if_necessary): Also allocate
28457 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
28459 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
28460 them. Fix too long lines.
28461 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
28462 vr_known being removed.
28463 (ipa_read_jump_function): Use new setter functions to construct bits
28464 and vr parts of jump functions or set them to NULL.
28465 (write_ipcp_transformation_info): Adjust for bits being pointers.
28466 (read_ipcp_transformation_info): Likewise.
28467 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
28469 Include gt-ipa-prop.h.
28470 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
28472 (ipcp_store_bits_results): Likewise.
28473 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
28474 Do not write to existing jump functions but use a temporary instead.
28476 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28479 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
28480 attempt to use its first operand as BIT_FIELD_REF base.
28482 2017-03-01 Richard Biener <rguenther@suse.de>
28484 PR middle-end/79721
28485 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
28486 interpolating formula in wrapping arithmetic.
28487 (chrec_apply): Convert chrec_evaluate return value to wanted type.
28489 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28491 PR tree-optimization/79734
28492 * tree-vect-generic.c (expand_vector_condition): Optimize
28493 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
28494 Handle VEC_COND_EXPR where comparison has different inner width from
28495 type's inner width.
28497 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
28499 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
28500 markup, and similar issues. Remove @opindex entries for things
28501 that aren't options. Add missing -mmpy-option entries.
28503 2017-02-28 Jakub Jelinek <jakub@redhat.com>
28505 PR tree-optimization/79737
28506 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
28507 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
28508 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
28509 instead of byte_size. Formatting fix.
28510 (shift_bytes_in_array_right): Formatting fix.
28512 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
28515 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
28516 condition on optimize for the leaf function test.
28518 2017-02-28 Martin Liska <mliska@suse.cz>
28521 * read-rtl-function.c (function_reader::handle_unknown_directive):
28522 Bail out when one uses -flto.
28524 2017-02-28 Martin Liska <mliska@suse.cz>
28526 * common.opt: Replace space with tabular for options of <number>
28528 * config/i386/i386.opt: Show <number> value for
28529 -mlarge-data-threshold.
28530 * opts.c (print_filtered_help): Do not display number in hexadecimal
28533 2017-02-28 Martin Liska <mliska@suse.cz>
28535 * common.opt: Fix --help=option -Q for options which are of
28538 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
28540 * config/i386/i386.c (print_reg): Error out for values
28541 of 8-bit size in invalid integer register.
28543 2017-02-28 Martin Sebor <msebor@redhat.com>
28545 PR tree-optimization/79691
28546 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
28548 2017-02-28 Jakub Jelinek <jakub@redhat.com>
28551 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
28552 gcc_unreachable with output_operand_lossage.
28554 2017-02-28 Richard Biener <rguenther@suse.de>
28556 PR tree-optimization/79740
28557 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
28559 (visit_nary_op): Insert the nary into the hashtable if we
28560 pattern-matched sth.
28561 * tree-ssa-pre.c (eliminate_insert): Robustify.
28563 2017-02-28 Richard Biener <rguenther@suse.de>
28565 PR middle-end/79731
28566 * fold-const.c (decode_field_reference): Reject out-of-bound
28569 2017-02-28 Jakub Jelinek <jakub@redhat.com>
28571 * config/i386/i386.c: Include intl.h.
28572 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
28573 instead of just cond ? "..." : "...".
28574 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
28575 * coverage.c (read_counts_file): Likewise.
28576 * omp-offload.c: Include intl.h.
28577 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
28578 of just cond ? "..." : "...".
28579 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
28580 of just cond ? "..." : "...".
28582 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
28585 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
28587 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
28588 'tune for' CPU name.
28589 * config/arm/arm-cpu-data.h: Regenerated.
28591 2017-02-28 Richard Biener <rguenther@suse.de>
28593 PR tree-optimization/79732
28594 * tree-inline.c (expand_call_inline): Do not shadow var.
28596 2017-02-28 Richard Biener <rguenther@suse.de>
28598 PR tree-optimization/79723
28599 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
28600 address-space properly.
28602 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
28604 * doc/optinfo.texi (Optimization groups): Fix option used for
28606 * doc/invoke.texi (-fopt-info): Document "omp".
28607 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
28608 (OPTGROUP_ALL): Add OPTGROUP_OMP.
28609 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
28610 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
28611 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
28613 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
28615 * dumpfile.c (optgroup_options): Instead of "openmp", associate
28616 OPTGROUP_OMP with "omp".
28618 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
28621 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
28622 for arithmetic shift of unsigned V2DI.
28624 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
28626 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
28627 arc/linux.h headers.
28628 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
28629 (LINK_SPEC): Likewise.
28630 (ARC_TLS_EXTRA_START_SPEC): Likewise.
28631 (EXTRA_SPECS): Likewise.
28632 (STARTFILE_SPEC): Likewise.
28633 (ENDFILE_SPEC): Likewise.
28634 (LIB_SPEC): Likewise.
28635 (TARGET_SDATA_DEFAULT): Likewise.
28636 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
28637 (MULTILIB_DEFAULTS): Likewise.
28638 (DWARF2_UNWIND_INFO): Likewise.
28639 * config/arc/big.h: New file.
28640 * config/arc/elf.h: Likewise.
28641 * config/arc/linux.h: Likewise.
28642 * config/arc/t-uClibc: Remove.
28644 2017-02-27 Bin Cheng <bin.cheng@arm.com>
28646 PR tree-optimization/77536
28647 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
28648 (tree_transform_and_unroll_loop): Use above function to compute the
28649 estimated niter of unrolled loop and use it when scaling profile.
28650 Also use count info rather than frequency if it's non-zero.
28651 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
28652 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
28653 (vect_transform_loop): Call above function.
28655 2017-02-27 Richard Biener <rguenther@suse.de>
28657 PR tree-optimization/45397
28658 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
28659 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
28660 (visit_nary_op): Add pattern matching for CSEing sign-changed
28661 or truncated operations with wider ones.
28663 2017-02-27 Richard Biener <rguenther@suse.de>
28665 PR tree-optimization/79690
28666 * tree-vect-stmts.c (vectorizable_store): Use vector type
28667 built from the DR with address-space.
28669 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
28671 * doc/invoke.texi (Optimize Options): Refine the description
28672 of asan-use-after-return.
28674 2017-02-25 Alan Modra <amodra@gmail.com>
28676 PR rtl-optimization/79584
28677 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
28678 base, not ad->base_term, the reg within base. Remove assertion
28679 that ad->base == ad->base_term. Replace gen_int_mode using
28680 bogus mode with const0_rtx.
28682 2017-02-25 Jakub Jelinek <jakub@redhat.com>
28684 PR middle-end/79396
28685 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
28686 FMA_EXPR like tcc_binary or tcc_unary.
28688 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
28691 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
28693 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
28694 (output_loc_operands): Handle DW_OP_call_ref and
28695 DW_OP_GNU_variable_value.
28696 (struct variable_value_struct): New type.
28697 (struct variable_value_hasher): Likewise.
28698 (variable_value_hash): New variable.
28699 (string_types): Remove.
28700 (copy_loc_descr): New function.
28701 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
28702 (prepend_loc_descr_to_each): New function.
28703 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
28704 instead of add_loc_descr_to_each if the first argument is single
28705 location list and the second has multiple.
28706 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
28707 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
28708 when looking for variable value which doesn't have other location info.
28709 (loc_list_from_tree): Formatting fix.
28710 (gen_array_type_die): Simplify DW_AT_string_length handling.
28711 (adjust_string_types): Remove.
28712 (gen_subprogram_die): Don't call adjust_string_types nor test/set
28713 string_types. Call resolve_variable_values.
28714 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
28715 (resolve_addr_in_expr): Likewise. Add A argument.
28716 (copy_deref_exprloc): Remove deref argument. Adjust for the
28717 original expression being DW_OP_GNU_variable_value with optionally
28718 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
28719 optionally after it.
28720 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
28721 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
28722 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
28723 (variable_value_hasher::hash, variable_value_hasher::equal): New
28725 (resolve_variable_value_in_expr, resolve_variable_value,
28726 resolve_variable_values, note_variable_value_in_expr,
28727 note_variable_value): New functions.
28728 (dwarf2out_early_finish): Call note_variable_value on all toplevel
28731 2017-02-24 Jakub Jelinek <jakub@redhat.com>
28734 * opts.h (handle_generated_option): Add GENERATED_P argument.
28735 * opts-common.c (handle_option): Adjust function comment.
28736 (handle_generated_option): Add GENERATED_P argument, pass it to
28738 (control_warning_option): Pass false to handle_generated_option
28740 * opts.c (maybe_default_option): Pass true to handle_generated_option
28742 * optc-gen.awk: Likewise.
28744 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28746 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
28747 a REG, look at the REG it is a SUBREG of.
28748 (splitter for cmpeqsi_t): Ditto.
28750 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28752 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
28753 the special USEs with the pattern of the insn, not the insn itself.
28755 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
28758 * doc/invoke.texi: Document -mload-store-pairs.
28760 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28761 Sandra Loosemore <sandra@codesourcery.com>
28763 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
28764 argument isn't a CONST_INT.
28765 (nios2_alternate_compare_const): Assert op is a CONST_INT.
28766 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
28767 (nios2_validate_compare): Bypass alternate compare logic if *op2
28768 is not a CONST_INT.
28769 (ldstwm_operation_p): Return false if first_base is not a REG or
28770 if first_offset is not a CONST_INT.
28772 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28774 * config/cris/cris.md: Use correct operand in a define_peephole2.
28776 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28778 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
28780 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28782 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
28783 this_insn if it is an INSN or JUMP_INSN.
28784 (force_offsettable): Look at base, not at addr.
28785 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
28786 on things that aren't necessarily CONST_INTs.
28788 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
28790 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
28791 -mfpmath=sse is the default also for x86-32 targets with SSE2
28792 instruction set when @option{-ffast-math} is enabled
28794 2017-02-24 Jeff Law <law@redhat.com>
28796 PR rtl-optimizatoin/79286
28797 * ira.c (update_equiv_regs): Drop may_trap_p exception to
28800 2017-02-24 Richard Biener <rguenther@suse.de>
28802 PR tree-optimization/79389
28803 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
28806 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
28808 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
28809 function comment to reflect reality.
28810 (loop_exits_before_overflow): Fix typo in function description.
28812 2017-02-24 Richard Biener <rguenther@suse.de>
28814 PR tree-optimization/79389
28815 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
28816 properly that a threading opportunity exists. Detect conditional
28817 copy/constant propagation opportunities.
28819 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
28821 * config/visium/visium.md (type): Add trap.
28822 (b): New mode attribute.
28823 (*btst): Rename into...
28824 (*btst<mode>): ...this and adjust.
28825 (*cbranchsi4_btst_insn): Rename into...
28826 (*cbranch<mode>4_btst_insn): ...this and adjust.
28827 (trap): New define_insn.
28829 2017-02-23 Jakub Jelinek <jakub@redhat.com>
28831 PR tree-optimization/79389
28832 * ifcvt.c (struct noce_if_info): Add rev_cond field.
28833 (noce_reversed_cond_code): New function.
28834 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
28835 reversed_comparison_code. Formatting fix.
28836 (noce_try_store_flag): Test rev_cond != NULL in addition to
28837 reversed_comparison_code.
28838 (noce_try_store_flag_constants): Likewise.
28839 (noce_try_store_flag_mask): Likewise.
28840 (noce_try_addcc): Use rev_cond if non-NULL instead of
28841 reversed_comparison_code.
28842 (noce_try_cmove_arith): Likewise. Formatting fixes.
28843 (noce_try_minmax, noce_try_abs): Clear rev_cond.
28844 (noce_find_if_block): Initialize rev_cond.
28845 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
28846 instead of false as last argument never attempt to reverse it
28849 2017-02-23 Bin Cheng <bin.cheng@arm.com>
28851 PR tree-optimization/79663
28852 * tree-predcom.c (combine_chains): Process refs in reverse order
28853 only for ZERO length chains, and add explaining comment.
28855 2017-02-23 Jeff Law <law@redhat.com>
28857 PR tree-optimization/79578
28858 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
28859 in call to operand_equal_p.
28861 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
28864 * config/i386/cpuid.h: Fix another undefined behavior.
28866 2017-02-23 Richard Biener <rguenther@suse.de>
28868 PR tree-optimization/79683
28869 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
28870 vector types for data-refs.
28872 2017-02-23 Martin Liska <mliska@suse.cz>
28874 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
28876 2017-02-23 Jakub Jelinek <jakub@redhat.com>
28878 PR middle-end/79665
28879 * internal-fn.c (get_range_pos_neg): Moved to ...
28880 * tree.c (get_range_pos_neg): ... here. No longer static.
28881 * tree.h (get_range_pos_neg): New prototype.
28882 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
28883 are known to be in between 0 and signed maximum inclusive, try to
28884 expand both unsigned and signed divmod and use the cheaper one from
28887 2017-02-22 Jeff Law <law@redhat.com>
28889 PR tree-optimization/79578
28890 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
28891 to compare base operands.
28893 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
28896 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
28897 gpc_reg_operand instead of fpr_reg_operand.
28899 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
28901 * config/mips/mips.c (mips_return_in_memory): Force FP
28902 vector types to be returned in memory for o32 ABI.
28904 2017-02-22 Jakub Jelinek <jakub@redhat.com>
28906 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
28907 instead of DW_TAG_member for static data member declarations and don't
28908 set no_linkage_name for static inline data members.
28909 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
28912 2017-02-22 Martin Liska <mliska@suse.cz>
28914 * doc/invoke.texi: Replace inequality signs with square brackets
28917 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
28920 * lra-constraints.c (simplify_operand_subreg): Handle
28921 WORD_REGISTER_OPERATIONS targets.
28923 2017-02-22 Jakub Jelinek <jakub@redhat.com>
28926 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
28927 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
28928 elimination by swapping fld*.
28930 2017-02-22 Richard Biener <rguenther@suse.de>
28932 PR tree-optimization/79673
28933 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
28934 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
28935 irrelevant address-space qualifiers and avoiding a
28936 ADDR_SPACE_CONVERT_EXPR from fold_convert.
28938 2017-02-22 Richard Biener <rguenther@suse.de>
28940 PR tree-optimization/79666
28941 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
28942 to not symbolically negate if that may introduce undefined
28945 2017-02-22 Martin Liska <mliska@suse.cz>
28948 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
28949 * data-streamer-out.c (streamer_write_gcov_count_stream):
28951 * value-prof.c (stream_out_histogram_value): Make assert more
28952 precise based on type of counter.
28954 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
28957 * config/i386/i386.md (standard_x87sse_constant_load splitter):
28958 Use nonimmediate_operand instead of memory_operand for operand 1.
28959 (float-extend standard_x87sse_constant_load splitter): Ditto.
28961 2017-02-21 Jeff Law <law@redhat.com>
28963 PR tree-optimization/79621
28964 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
28965 blocks with edges to themselves.
28967 2017-02-21 Jakub Jelinek <jakub@redhat.com>
28970 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
28971 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
28972 Use gimple_call_builtin_p.
28975 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
28976 on temporarily removed DEBUG_INSNs.
28978 PR tree-optimization/79649
28979 * tree-loop-distribution.c (classify_partition): Give up on
28980 non-generic address space loads/stores.
28982 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
28984 * doc/loop.texi (Loop manipulation): Remove nonexistent
28985 tree_ssa_loop_version from the documentation.
28986 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
28988 2017-02-21 Jakub Jelinek <jakub@redhat.com>
28991 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
28992 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
28993 * config/rs6000/rs6000.c: Include except.h.
28994 (rs6000_expand_split_stack_prologue): Call
28995 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
28997 2017-02-21 Martin Jambor <mjambor@suse.cz>
29000 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
29001 have been analyzed.
29003 2017-02-21 Martin Jambor <mjambor@suse.cz>
29005 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
29006 for backward compatibility only.
29007 * doc/invoke.texi (Option Summary): Remove all references to
29008 -fipa-cp-alignment.
29010 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
29014 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29016 * lra-constraints.c (curr_insn_transform): Handle
29017 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29019 2017-02-21 Martin Liska <mliska@suse.cz>
29021 * config/i386/i386.opt: Replace -masm-dialect with -masm.
29023 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
29025 PR translation/79638
29026 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
29028 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
29031 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
29032 (arm_function_ok_for_sibcall): Return false for an indirect call by
29033 descriptor if all the argument registers are used.
29034 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
29035 alignment of the function.
29037 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29039 PR tree-optimization/61441
29040 * simplify-rtx.c (simplify_const_unary_operation): For
29041 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
29042 the sNaN unmodified.
29044 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
29046 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29047 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29048 instead of SYSTEM_HEADER_DIR.
29050 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
29051 Martin Liška <mliska@suse.cz>
29053 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
29054 Fix typos and grammar, use active voice, and clarify.
29056 2017-02-20 Marek Polacek <polacek@redhat.com>
29058 PR middle-end/79537
29059 * gimplify.c (gimplify_expr): Handle unused *&&L;.
29062 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
29064 2017-02-20 Jakub Jelinek <jakub@redhat.com>
29067 * config/i386/i386.c (ix86_expand_builtin): Handle
29068 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
29069 ix86_builtins_isa[fcode].isa as a requirement of those
29070 flags and any other flag in the bitmask.
29071 (ix86_init_mmx_sse_builtins): Use 0 instead of
29072 ~OPTION_MASK_ISA_64BIT as mask.
29073 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
29074 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
29075 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
29076 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
29078 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29081 * lra-constraints.c (split_reg): Check requested split mode
29082 is supported by the register.
29084 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29086 * lra-constraints.c (simplify_operand_subreg): Remove early
29089 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29092 * lra-constraints.c (curr_insn_transform): Tighten condition
29093 for converting SUBREG reloads from OP_OUT to OP_INOUT.
29095 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29098 * lra-constraints.c (curr_insn_transform): Handle
29099 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29101 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
29104 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
29106 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
29108 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
29111 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
29114 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29116 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
29119 2017-02-18 Jakub Jelinek <jakub@redhat.com>
29121 * final.c (last_columnnum, override_columnnum): New variables.
29122 (final_start_function): Set last_columnnum, pass it to begin_prologue
29123 hook and pass 0 to dwarf2out_begin_prologue.
29124 (final_scan_insn): Update override_columnnum. Pass last_columnnum
29125 to source_line debug hook.
29126 (notice_source_line): Compute last_columnnum and for debug_column_info
29127 return true on column changes.
29128 * debug.h (struct gcc_debug_hooks): Add column argument to
29129 source_line and begin_prologue hooks.
29130 (debug_nothing_int_charstar_int_bool): Remove prototype.
29131 (debug_nothing_int_int_charstar,
29132 debug_nothing_int_int_charstar_int_bool): New prototypes.
29133 (dwarf2out_begin_prologue): Add column argument.
29134 * debug.c (do_nothing_debug_hooks): Adjust source_line and
29135 begin_prologue hooks.
29136 (debug_nothing_int_charstar_int_bool): Remove.
29137 (debug_nothing_int_int_charstar,
29138 debug_nothing_int_int_charstar_int_bool): New functions.
29139 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
29140 through to dwarf2out_source_line.
29141 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
29142 (dwarf2out_source_line): Add column argument, emit it if requested.
29143 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
29145 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29146 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29147 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
29148 through to dwarf2out_begin_prologue.
29149 (vmsdbgout_source_line): Add column argument, pass it through to
29150 dwarf2out_source_line.
29151 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
29152 dbxout_source_line caller.
29153 (dbxout_source_line): Add column argument.
29155 * common.opt (gno-column-info, gcolumn-info): New options.
29156 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
29157 (check_die): Also test for multiple DW_AT_decl_column attributes.
29158 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
29159 DW_AT_decl_column if requested.
29160 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
29162 (gen_variable_die): Likewise.
29163 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
29164 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
29167 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
29168 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
29169 (ix86_handle_option): Handle OPT_m3dnowa.
29170 * doc/invoke.texi (-m3dnowa): Document.
29171 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
29172 -m3dnowa instead of -m3dnow -march=athlon.
29175 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
29176 instead of gcc_assert for K, r and R code checks. Formatting fixes.
29178 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29181 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
29182 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
29183 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
29184 generator for vsx_xxpermdi_<mode>_be.
29185 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
29186 force big-endian semantics.
29187 (vsx_xxpermdi_<mode>_be): New define_expand with same
29188 implementation as previous version of vsx_xxpermdi_<mode>.
29190 2017-02-17 Jakub Jelinek <jakub@redhat.com>
29192 PR tree-optimization/79327
29193 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
29194 variable, its initialization and use.
29196 2017-02-17 Julia Koval <julia.koval@intel.com>
29198 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
29199 (OPTION_MASK_ISA_PKU_UNSET): New.
29200 (ix86_handle_option): Handle -mrdpid.
29201 * config/i386/cpuid.h (bit_RDPID): New.
29202 * config/i386/driver-i386.c (host_detect_local_cpu):
29203 Detect RDPID feature.
29204 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
29205 * config/i386/i386-c.c (ix86_target_macros_internal):
29207 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
29208 (ix86_valid_target_attribute_inner_p): Add "rdpid".
29209 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
29210 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
29211 * config/i386/i386.md (define_insn "rdpid"): New.
29212 * config/i386/i386.opt Add -mrdpid.
29213 * config/i386/immintrin.h (_rdpid_u32): New.
29215 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
29217 PR rtl-optimization/79541
29218 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
29219 instead of transforming it into USE.
29221 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
29223 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
29224 If HONOR_SNANS (SFmode) force the input to a register.
29225 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
29226 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
29227 an frsp or similar insn.
29229 2017-02-17 Martin Liska <mliska@suse.cz>
29231 PR rtl-optimization/79577
29232 * params.def (selsched-max-sched-times): Increase minimum to 1.
29234 2017-02-17 Martin Liska <mliska@suse.cz>
29236 PR rtl-optimization/79574
29237 * gcse.c (want_to_gcse_p): Prevent integer overflow.
29239 2017-02-17 Martin Liska <mliska@suse.cz>
29241 PR tree-optimization/79529
29242 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
29243 ssa_defined_default_def_p to handle cases which are implicitly
29245 * tree-ssa.c (ssa_defined_default_def_p): New function.
29246 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
29247 which are implicitly defined.
29248 * tree-ssa.h (ssa_defined_default_def_p): Declare.
29250 2017-02-17 Richard Biener <rguenther@suse.de>
29252 PR middle-end/79576
29253 * params.def (max-ssa-name-query-depth): Limit to 10.
29255 2017-02-17 Richard Biener <rguenther@suse.de>
29257 PR tree-optimization/79552
29258 * tree-ssa-structalias.c (visit_loadstore): Properly verify
29261 2017-02-17 Richard Biener <rguenther@suse.de>
29264 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
29266 2017-02-17 Marek Polacek <polacek@redhat.com>
29268 PR middle-end/79536
29269 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
29270 (fold_negate_expr): New wrapper.
29272 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
29274 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
29275 Correct terminology and de-emphasize pre-standard behavior.
29277 2017-02-16 Alan Modra <amodra@gmail.com>
29279 PR rtl-optimization/79286
29280 * ira.c (def_dominates_uses): New function.
29281 (update_equiv_regs): Don't create an equivalence for insns that
29282 may trap where the register def does not dominate the use.
29284 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
29286 PR rtl-optimization/78127
29287 * lra.c (lra): Call lra_eliminate before finish the loop after
29290 2017-02-16 Richard Biener <rguenther@suse.de>
29292 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
29294 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
29295 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
29296 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
29297 (isl_val_int_from_wi): New function.
29298 (extract_affine_gmp): Rename to ...
29299 (extract_affine_wi): ... this, take a widest_int.
29300 (extract_affine_int): Just wrap extract_affine_wi.
29301 (add_param_constraints): Use isl_val_int_from_wi.
29302 (add_loop_constraints): Likewise, and extract_affine_wi.
29304 2017-02-15 Jeff Law <law@redhat.com>
29306 PR middle-end/79521
29307 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
29308 ira_init_register_move_cost_if_necessary.
29310 2017-02-15 Martin Sebor <msebor@redhat.com>
29312 PR middle-end/32003
29313 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
29314 removed in a prior commit.
29316 2017-02-15 Bin Cheng <bin.cheng@arm.com>
29318 PR tree-optimization/79347
29319 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
29320 counters during peeling.
29322 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
29324 * Makefile.in (site.exp): Remove "set ISLVER".
29326 2017-02-15 Jakub Jelinek <jakub@redhat.com>
29329 * real.c (real_from_integer): Call real_convert even for decimal.
29331 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
29334 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
29336 2017-02-14 Andrew Pinski <apinski@cavium.com>
29338 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
29339 cores and change the partno/implementer to be correct.
29340 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
29341 the 'B" as the implementer.
29342 * config/aarch64/aarch64-tune.md: Regenerate.
29344 2017-02-14 Carl Love <cel@us.ibm.com>
29346 * config/rs6000/rs6000.c: Add case statement entry to make the
29347 xvcvuxdsp built-in argument unsigned.
29348 * config/rs6000/vsx.md: Fix the source and return operand types so they
29349 match the instruction definitions from the ISA document. Fix typo
29350 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
29353 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
29356 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
29357 member early_clobber_alts.
29358 * lra-lives.c (reg_early_clobber_p): New.
29359 (process_bb_lives): Use it.
29360 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
29361 (debug_operand_data): Initialize early_clobber_alts.
29362 (setup_operand_alternative): Set up early_clobber_alts.
29363 (collect_non_operand_hard_regs): Ditto. Pass early clobber
29364 alternatives to new_insn_reg.
29365 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
29367 (lra_update_insn_regno_info): Pass the new arg.
29369 2017-02-14 Jakub Jelinek <jakub@redhat.com>
29371 PR middle-end/79505
29372 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
29373 (new_oacc_loop_raw): Don't clear already cleared fields.
29376 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
29377 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
29378 _mm512_prefetch_i64gather_ps): New inline functions and macros.
29380 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
29383 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
29385 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
29388 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
29389 the extra instruction to the right place to store 128-bit constant
29392 2017-02-14 Martin Sebor <msebor@redhat.com>
29394 PR middle-end/79448
29395 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
29396 warning for strings of unknown length.
29398 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
29400 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
29402 2017-02-14 Jeff Law <law@redhat.com>
29405 * ira-costs.c (scan_one_insn): Initialize register move costs
29406 for pseudos seen in USE/CLOBBER insns.
29408 PR tree-optimization/79095
29409 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
29410 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
29411 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
29412 if the operands are known to be not equal, then the resulting range
29414 (intersect_ranges): If the new range is ~[0,0] and the old range is
29415 wide, then prefer ~[0,0].
29416 * tree-vrp.c (overflow_comparison_p_1): New function.
29417 (overflow_comparison_p): New function.
29418 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
29419 if NAME is used in an overflow test.
29420 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
29421 overflow check that can be expressed as an equality test, then adjust
29422 ops to be that equality test.
29424 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29426 * config/s390/s390-builtin-types.def: Remove flags argument.
29427 * config/s390/s390.c (s390_init_builtins): Likewise.
29429 2017-02-14 Martin Liska <mliska@suse.cz>
29431 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
29432 vector. Fix trailing white spaces.
29434 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
29436 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
29439 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29441 PR rtl-optimization/68664
29442 * config/arm/arm.c (arm_sched_can_speculate_insn):
29443 New function. Declare prototype.
29444 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
29446 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29448 PR rtl-optimization/68664
29449 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
29451 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
29453 2017-02-14 Amit Pawar <amit.pawar@amd.com>
29455 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
29456 max skip bytes for function, loop and jump.
29458 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29460 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
29461 ABS_EXPR for gimple dump.
29463 2017-02-14 Jakub Jelinek <jakub@redhat.com>
29466 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
29468 PR tree-optimization/79408
29469 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
29470 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
29471 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
29472 also if rhs1 is INTEGER_CST.
29474 2017-02-14 Richard Biener <rguenther@suse.de>
29476 PR middle-end/79432
29477 * tree-into-ssa.c (insert_phi_nodes): When the function can
29478 have abnormal edges rewrite SSA names with broken use-def
29479 dominance out of SSA and register them for PHI insertion.
29481 2017-02-13 Martin Sebor <msebor@redhat.com>
29483 PR middle-end/79496
29484 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
29485 clearing info.nowrite flag when snprintf size argument is a range.
29487 2017-02-13 Jakub Jelinek <jakub@redhat.com>
29489 * cprop.c (cprop_jump): Add missing space in string literal.
29490 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
29491 (get_constraint_for_component_ref): Likewise.
29492 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
29493 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
29494 * lra-constraints.c (process_alt_operands): Likewise.
29495 * ipa-inline.c (inline_small_functions): Likewise.
29496 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
29497 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
29498 * trans-mem.c (diagnose_tm_1_op): Likewise.
29499 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
29500 (grid_parallel_clauses_gridifiable): Likewise.
29502 * config/nvptx/mkoffload.c (process): Add space in between
29505 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
29506 "MOD4_SSE_REGS" and "ALL_REGS".
29508 * spellcheck.c (test_data): Add , in between "foo" and "food".
29510 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29513 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
29514 boundary crossing check and subsequent code generation agree.
29516 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29518 * config/aarch64/aarch64.c (has_memory_op): Delete.
29519 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
29522 2017-02-13 Jakub Jelinek <jakub@redhat.com>
29524 PR rtl-optimization/79388
29525 PR rtl-optimization/79450
29526 * combine.c (distribute_notes): When removing TEM_INSN for which
29527 corresponding dest has last value recorded, invalidate that last
29530 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29532 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
29533 of explicit '@'. Add missing assembly comment marker on branch costs
29536 2017-02-13 Nathan Sidwell <nathan@acm.org>
29538 * gengtype-lex.l (<in_struct>): Add '/'.
29540 2017-02-13 Martin Liska <mliska@suse.cz>
29543 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
29545 2017-02-13 Richard Biener <rguenther@suse.de>
29547 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
29549 * configure: Re-generate.
29550 * config.in: Likewise.
29551 * graphite-dependences.c: Simplify as if
29552 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
29553 * graphite-isl-ast-to-gimple.c: Likewise.
29554 * graphite-optimize-isl.c: Likewise.
29555 * graphite-poly.c: Likewise.
29556 * graphite-sese-to-poly.c: Likewise.
29557 * graphite.h: Likewise.
29558 * toplev.c: Include isl/version.h and use isl_version () for
29559 printing the ISL version.
29560 * doc/install.texi: Update ISL requirement.
29562 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
29564 * doc/standards.texi (Standards): Update reference to
29567 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
29569 * doc/extend.texi (Named Address Spaces): sourceware.org now
29571 * doc/install.texi (Binaries): Ditto.
29574 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
29576 * doc/cpp.texi: Replace "stringify"/"stringification" with C
29577 standard terminology "stringize"/"stringizing" throughout.
29578 * doc/cppinternals.texi: Likewise.
29580 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
29582 * doc/extend.texi: Fix some spelling mistakes and typos.
29583 * doc/invoke.texi: Likewise.
29585 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
29588 * params.def (inline-min-speedup) Change from 10 to 8.
29590 2017-02-11 Jakub Jelinek <jakub@redhat.com>
29592 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
29595 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
29598 * ipa-inline-analysis.c (get_minimal_bb): New function.
29599 (record_modified): Use it.
29600 (remap_edge_change_prob): Handle also ancestor functions.
29602 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
29604 * doc/contrib.texi (Contributors): Remove broken link into
29605 the Mauve CVS repository.
29607 2017-02-11 Jakub Jelinek <jakub@redhat.com>
29609 PR middle-end/79454
29610 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
29611 result computation whenever lhs doesn't have vector mode, not
29612 just when it has BLKmode.
29614 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
29616 * doc/makefile.texi (profiledbootstrap): Refer to the
29617 installation instructions only in textual form.
29619 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29622 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
29624 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
29626 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
29627 (Specific): Update mingw-w64 reference.
29629 (Specific): Remove broken link to Renesas RX processor.
29631 2017-02-10 Richard Biener <rguenther@suse.de>
29633 * toplev.c (process_options): Do not mention obsolete graphite
29634 options when printing sorry message about missing graphite support.
29635 Mention -floop-nest-optimize.
29637 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
29639 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
29640 (vtst_p16): Likewise.
29641 (vtstq_p8): Likewise.
29642 (vtstq_p16): Likewise.
29644 (vtstq_p64): Likewise.
29645 * config/arm/arm_neon.h (vgetq_lane_p64): New.
29646 (vset_lane_p64): New.
29647 (vsetq_lane_p64): New.
29649 2017-02-10 Jakub Jelinek <jakub@redhat.com>
29651 PR tree-optimization/79411
29652 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
29653 stmt operands are SSA_NAMEs used in abnormal phis.
29654 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
29657 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
29660 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
29663 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
29665 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
29667 2017-02-09 Jakub Jelinek <jakub@redhat.com>
29669 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
29673 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
29674 not arbitrary TREE_CONSTANT.
29677 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
29678 "omp declare target link" attribute unless is_global_var.
29679 * omp-offload.c (find_link_var_op): Likewise.
29681 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
29682 Chung-Lin Tang <cltang@codesourcery.com>
29684 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
29686 (gimplify_adjust_omp_clauses): Don't delete TILE.
29687 (gimplify_omp_for): Deal with TILE.
29688 * internal-fn.c (expand_GOACC_TILE): New function.
29689 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
29691 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
29692 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
29694 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
29695 avoid DIV for outermost collapse var.
29696 (expand_oacc_for): Insert tile element loop as needed. Adjust.
29697 Remove out of date comments, fix whitespace.
29698 * omp-general.c (omp_extract_for_data): Deal with tiling.
29699 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
29700 adjust OLF_DIM_BASE value.
29701 (struct omp_for_data): Add tiling field.
29702 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
29703 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
29704 for auto loops. Remove default auto determining, moved to
29705 oacc_loop_fixed_partitions.
29706 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
29707 stmts, add e_mask field.
29708 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
29709 (oacc_thread_numbers): Use oacc_dim_call.
29710 (oacc_xform_tile): New.
29711 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
29712 (finish_oacc_loop): Adjust for ifns vector.
29713 (oacc_loop_discover_walk): Append loop abstraction sites to list,
29714 add case for GOACC_TILE fns.
29715 (oacc_loop_xform_loop): Delete.
29716 (oacc_loop_process): Iterate over call list directly, and add
29717 handling for GOACC_TILE fns.
29718 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
29720 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
29721 vector partitioning to outer loops. Assign 2 partitions to loops
29722 when available. Add TILE handling.
29723 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
29724 (execite_oacc_device_lower): Process GOACC_TILE fns,
29725 ignore unknown specs.
29726 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
29727 * tree.c (omp_clause_num_ops): Adjust TILE ops.
29728 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
29730 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
29732 * configure.ac (ACX_BUGURL): Update.
29733 * configure: Regenerate.
29735 2017-02-09 Richard Biener <rguenther@suse.de>
29737 PR tree-optimization/69823
29738 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
29739 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
29741 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
29743 * config/arc/arc-c.def: Add __NPS400__ definition.
29744 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
29745 (TARGET_NPS400): Define.
29747 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
29749 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
29751 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
29752 pointer, arch_info.
29753 (arc_cpu_types): Fill the arch_info field with a pointer into the
29754 arc_arch_types table.
29755 (arc_selected_cpu): Declare.
29756 * config/arc/arc.c (arc_selected_cpu): Make global.
29757 (arc_selected_arch): Delete.
29758 (arc_base_cpu): Delete.
29759 (arc_override_options): Remove references to deleted variables,
29760 update access to arch information.
29761 (ARC_OPT): Update access to arch information.
29762 (ARC_OPTX): Likewise.
29763 * config/arc/arc.h (arc_base_cpu): Remove declaration.
29764 (TARGET_ARC600): Update access to arch information.
29765 (TARGET_ARC601): Likewise.
29766 (TARGET_ARC700): Likewise.
29767 (TARGET_EM): Likewise.
29768 (TARGET_HS): Likewise.
29769 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
29772 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
29775 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
29776 condition/operands for integer GE/LE/GEU/LEU operations.
29778 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
29780 PR translation/79397
29781 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
29784 2017-02-08 Martin Jambor <mjambor@suse.cz>
29787 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
29788 whether allocation happened.
29789 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
29790 nothing was allocated.
29792 2017-02-08 Jakub Jelinek <jakub@redhat.com>
29794 PR tree-optimization/79408
29795 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
29796 constant, but SSA_NAME with a known integer range, use the minimum
29797 of that range instead of op1 to determine if modulo can be replaced
29798 with its first operand.
29800 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29802 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
29804 2017-02-08 Richard Biener <rguenther@suse.de>
29806 PR tree-optimization/71824
29807 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
29808 Check all loops contained in the merged region.
29810 2017-02-07 Andrew Pinski <apinski@cavium.com>
29812 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
29814 2017-02-07 Andrew Pinski <apinski@cavium.com>
29816 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
29817 (thunderxt88): Likewise.
29818 (thunderxt81): Disable LSE and change v8.1 to v8.
29819 (thunderxt83): Likewise.
29821 2017-02-07 Jakub Jelinek <jakub@redhat.com>
29822 Richard Biener <rguenther@suse.de>
29824 PR middle-end/79399
29825 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
29826 type from int to size_t.
29827 * ira-costs.c (struct_costs_size): Change type from int to size_t.
29829 2017-02-07 Jakub Jelinek <jakub@redhat.com>
29831 PR rtl-optimization/79386
29832 * cprop.c (bypass_conditional_jumps): Initialize
29833 bypass_last_basic_block already before splitting bbs after
29834 unconditional traps...
29835 (bypass_conditional_jumps): ... rather than here.
29838 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
29839 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
29840 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
29841 fix -masm=intel patterns.
29843 2017-02-07 Richard Biener <rguenther@suse.de>
29845 PR tree-optimization/79256
29846 PR middle-end/79278
29847 * builtins.c (get_object_alignment_2): Use min_align_of_type
29848 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
29849 and ADJUST_FIELD_ALIGN.
29851 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
29853 * doc/tm.texi: Regenerate.
29854 * stor-layout.c (layout_decl): Adjust.
29855 (update_alignment_for_field): Likewise.
29856 (place_field): Likewise.
29857 (min_align_of_type): Likewise.
29858 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
29859 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
29860 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
29861 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
29862 * config/frv/frv.c (frv_adjust_field_align): Likewise.
29863 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
29864 * config/i386/i386.c (x86_field_alignment): Likewise.
29865 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
29866 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
29867 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
29868 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
29869 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
29870 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
29874 2017-01-30 Richard Biener <rguenther@suse.de>
29876 PR tree-optimization/79256
29877 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
29878 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
29881 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
29883 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
29884 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
29885 builtins to SImode and emit a zero-extend, if necessary.
29887 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
29889 * docs/invoke.texi (RISC-V Options): Alphabetize.
29891 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
29893 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
29896 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
29898 * config/riscv/riscv.c: New file.
29899 * common/config/riscv/riscv-common.c: Likewise.
29900 * config.gcc: Likewise.
29901 * config/riscv/constraints.md: Likewise.
29902 * config/riscv/elf.h: Likewise.
29903 * config/riscv/generic.md: Likewise.
29904 * config/riscv/linux.h: Likewise.
29905 * config/riscv/multilib-generator: Likewise.
29906 * config/riscv/peephole.md: Likewise.
29907 * config/riscv/pic.md: Likewise.
29908 * config/riscv/predicates.md: Likewise.
29909 * config/riscv/riscv-builtins.c: Likewise.
29910 * config/riscv/riscv-c.c: Likewise.
29911 * config/riscv/riscv-ftypes.def: Likewise.
29912 * config/riscv/riscv-modes.def: Likewise.
29913 * config/riscv/riscv-opts.h: Likewise.
29914 * config/riscv/riscv-protos.h: Likewise.
29915 * config/riscv/riscv.h: Likewise.
29916 * config/riscv/riscv.md: Likewise.
29917 * config/riscv/riscv.opt: Likewise.
29918 * config/riscv/sync.md: Likewise.
29919 * config/riscv/t-elf-multilib: Likewise.
29920 * config/riscv/t-linux: Likewise.
29921 * config/riscv/t-linux-multilib: Likewise.
29922 * config/riscv/t-riscv: Likewise.
29923 * configure.ac: Likewise.
29924 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
29925 Waterman as RISC-V maintainers.
29926 * doc/install.texi: Add RISC-V entries.
29927 * doc/invoke.texi: Add RISC-V options section.
29928 * doc/md.texi: Add RISC-V constraints section.
29929 * configure: Regenerated.
29931 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
29934 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
29935 false values to be constant vectors with all 0 or all 1 bits set.
29936 (vcondu<mode><mode>): Likewise.
29937 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
29939 (fpmask_comparison_operator): Update comment.
29940 (vecint_comparison_operator): New predicate.
29941 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
29942 vector conditionals when the true and false values are constant
29943 vectors with all 0 bits or all 1 bits set.
29945 2017-02-06 Martin Sebor <msebor@redhat.com>
29947 PR tree-optimization/79376
29948 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
29950 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
29952 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
29953 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
29954 to simplify split condition.
29956 2017-02-06 Jakub Jelinek <jakub@redhat.com>
29958 * omp-expand.c (oxpand_omp_atomic_fetch_op,
29959 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
29962 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
29964 PR rtl-optimization/68664
29965 * target.def (can_speculate_insn): New hook.
29966 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
29967 * doc/tm.texi: Regenerate.
29968 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
29969 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
29970 (rs6000_sched_can_speculate_insn): New function.
29972 2017-02-06 Jakub Jelinek <jakub@redhat.com>
29974 PR tree-optimization/79284
29975 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
29976 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
29977 vectorizable_mask_load_store, vectorizable_operation,
29978 vect_is_simple_cond, get_same_sized_vectype): Use it instead
29979 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
29980 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
29981 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
29982 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
29983 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
29984 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
29985 is_gimple_assign (stmt). Replace another such test with
29986 is_gimple_assign (stmt).
29988 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
29991 * config/avr/avr.c (rtl-iter.h): Include it.
29992 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
29993 (avr_legitimate_combined_insn): ...and implementation.
29995 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
29997 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
29998 * config/s390/s390.c (s390_const_operand_ok)
29999 (s390_canonicalize_comparison, s390_extract_part)
30000 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
30001 (s390_contiguous_bitmask_p, s390_rtx_costs)
30002 (legitimize_pic_address): Likewise.
30003 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
30004 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
30005 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
30006 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
30007 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
30009 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30011 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
30012 REGNO($0) == REGNO($1).
30014 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30016 * config/s390/linux.h(SIZE_TYPE): Add comment.
30018 2017-02-06 Julian Brown <julian@codesourcery.com>
30019 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30020 Virendra Pathak <virendra.pathak@broadcom.com>
30022 * config/aarch64/aarch64-cores.def: Change the scheduler
30024 * config/aarch64/aarch64.md: Include thunderx2t99.md.
30025 * config/aarch64/thunderx2t99.md: New file.
30027 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30029 * doc/standards.texi (Go Language): Update link to language
30032 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30034 * tree-eh.c (lower_resx): Sanitize profile.
30035 (cleanup_empty_eh_move_lp): Likewise.
30037 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30040 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
30042 * cfgloopmanip.h (loop_version): Update prototype.
30043 * modulo-sched.c (sms_schedule): Update call of loop_version.
30044 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
30045 * tree-parloops.c (gen_parallel_loop): Likewise.
30046 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
30047 * tree-ssa-loop-split.c (split_loop): Likewise.
30048 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
30049 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
30051 2017-02-05 Martin Liska <mliska@suse.cz>
30054 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
30056 (print_operand_address): Initialize a struct to zero.
30058 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30060 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
30061 garbage collector only in textual form.
30063 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30065 * doc/extend.texi (x86 specific memory model extensions for
30066 transactional memory): Simplify a phrase.
30068 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
30071 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
30072 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
30073 (atomic_storedi_1): Likewise.
30075 2017-02-04 Jakub Jelinek <jakub@redhat.com>
30077 PR tree-optimization/79338
30078 * tree-parloops.c (gather_scalar_reductions): Don't call
30079 vect_analyze_loop_form for loop->inner before destroying loop's
30082 2017-02-03 Martin Sebor <msebor@redhat.com>
30084 PR tree-optimization/79327
30085 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
30086 when precision has resulted in leading zeros.
30087 (format_integer): Adjust the likely counter to assume an unknown
30088 argument that may be zero is non-zero.
30090 2017-02-03 Jason Merrill <jason@redhat.com>
30093 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
30094 avoid copying non-taken branch.
30096 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30098 PR tree-optimization/79340
30099 * tree-vect-loop.c (vectorizable_reduction): Release
30100 vec_defs elements after safe_splicing them into other vectors.
30103 PR tree-optimization/79327
30104 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
30105 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
30107 (format_integer): Use wide_int_to_tree instead of build_int_cst
30108 + to_?hwi. If argmin is NULL, just set argmin and argmax to
30109 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
30110 of shortest and longest sequence.
30112 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
30114 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
30115 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
30117 2017-02-03 Walter Lee <walt@tilera.com>
30120 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
30121 after initial stackframe link reg save.
30122 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
30124 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30127 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
30128 wu for stxssp alternative.
30130 2017-02-03 Martin Sebor <msebor@redhat.com>
30132 PR tree-optimization/79352
30133 * gimple-fold.c (get_range_strlen): Add argument.
30134 (get_range_strlen): Change return type to bool.
30135 (get_maxval_strlen): Pass in a dummy argument.
30136 * gimple-fold.h (get_range_strlen): Change return type to bool.
30137 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
30138 * tree.h (array_at_struct_end_p): Add argument.
30139 * tree.c (array_at_struct_end_p): Handle it.
30141 2017-02-03 Martin Liska <mliska@suse.cz>
30144 * multiple_target.c (create_dispatcher_calls): Redirect edge
30145 from a caller of a dispatcher.
30146 (expand_target_clones): Make the clones local.
30147 (ipa_target_clone): Do both target clones and resolvers.
30148 (ipa_dispatcher_calls): Remove the pass.
30149 (pass_dispatcher_calls::gate): Likewise.
30150 (make_pass_dispatcher_calls): Likewise.
30151 * passes.def (pass_target_clone): Put as very first IPA early
30154 2017-02-03 Martin Liska <mliska@suse.cz>
30156 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
30157 in case of a function with ifunc attribute.
30159 2017-02-03 Martin Liska <mliska@suse.cz>
30161 * cgraph.c (cgraph_node::dump): Dump function version info.
30162 * symtab.c (symtab_node::dump_base): Add missing new line.
30164 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30166 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
30167 (ifcombine_ifandif): Use it.
30169 2017-02-03 Martin Liska <mliska@suse.cz>
30171 * doc/invoke.texi: Document default value for
30172 use-after-scope-direct-emission-threshold.
30174 2017-02-03 Martin Liska <mliska@suse.cz>
30176 PR tree-optimization/79339
30177 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
30178 (format_floating): Likewise.
30180 2017-02-03 Martin Liska <mliska@suse.cz>
30183 * ipa-prop.c (ipa_node_params_t::insert): Remove current
30185 (ipa_node_params_t::remove): Likewise.
30186 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
30187 initialization from removed ipa_node_params_t::insert.
30188 (ipa_node_params::~ipa_node_params): Move from removed
30189 ipa_node_params_t::release.
30190 * symbol-summary.h (symbol_summary::m_released): New member.
30191 Do not release a summary twice. Do not allow to call finalizer
30192 for types of a summary that live in GGC memory.
30194 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30196 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
30199 2017-02-02 Martin Sebor <msebor@redhat.com>
30201 PR middle-end/79275
30202 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
30203 (format_string): Tighten up the range of output for non-constant
30204 strings and correct the expected range for wide non-constant strings.
30206 2017-02-02 Martin Sebor <msebor@redhat.com>
30208 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
30210 PR middle-end/32003
30211 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
30213 (-fdump-tree-@var): Add to index and document how to come up
30214 with pass-specific option and dump file names.
30215 (-fdump-passes): Clarify where to look for output.
30217 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30219 PR middle-end/77445
30220 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
30221 statistics of the analyzed path; allow threading for speed when
30222 any of BBs along the path are optimized for speed.
30224 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
30226 PR middle-end/78468
30227 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
30228 settings of the virtual registers.
30231 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
30233 * explow.c (get_dynamic_stack_size): Take known alignment of stack
30234 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
30237 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30239 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
30240 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
30242 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30244 * config/s390/s390.md: Add missing comments with the expanded
30246 * config/s390/vector.md: Likewise.
30247 * config/s390/vx-builtins.md: Likewise.
30249 2017-02-02 Jakub Jelinek <jakub@redhat.com>
30252 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
30253 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
30254 conditions on a single line.
30256 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30258 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
30259 __S390_VX__ to __VX__.
30261 2017-02-01 Andrew Pinski <apinski@cavium.com>
30263 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
30264 stmt_info to record_stmt_cost.
30265 (vect_get_known_peeling_cost): Pass stmt_info if known to
30267 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
30268 cpu_vector_cost field into
30269 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
30270 field into vec_int_stmt_cost and vec_fp_stmt_cost.
30271 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
30272 splitting of scalar_stmt_cost and vec_stmt_cost.
30273 (thunderx_vector_cost): Likewise.
30274 (cortexa57_vector_cost): LIkewise.
30275 (exynosm1_vector_cost): Likewise.
30276 (xgene1_vector_cost): Likewise.
30277 (thunderx2t99_vector_cost): Improve after the splitting of the two
30279 (aarch64_builtin_vectorization_cost): Update for the splitting of
30280 scalar_stmt_cost and vec_stmt_cost.
30282 2017-02-01 Torvald Riegel <triegel@redhat.com>
30283 Richard Henderson <rth@redhat.com>
30285 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
30286 conditional on existance of a fast atomic load.
30287 * optabs-query.c (can_atomic_load_p): New function.
30288 * optabs-query.h (can_atomic_load_p): Declare it.
30289 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
30290 no fast atomic load is available for the particular size of access.
30291 (expand_atomic_compare_and_swap): Likewise.
30292 (expand_atomic_load): Likewise.
30293 (expand_atomic_store): Likewise.
30294 (expand_atomic_fetch_op): Likewise.
30295 * testsuite/lib/target-supports.exp
30296 (check_effective_target_sync_int_128): Remove x86 because it provides
30297 no fast atomic load.
30298 (check_effective_target_sync_int_128_runtime): Likewise.
30300 2017-02-01 Richard Biener <rguenther@suse.de>
30302 * graphite.c: Include tree-vectorizer.h for find_loop_location.
30303 (graphite_transform_loops): Provide opt-info for optimized nests.
30304 * tree-parloop.c (parallelize_loops): Provide opt-info for
30305 parallelized loops.
30307 2017-02-01 Richard Biener <rguenther@suse.de>
30309 PR middle-end/79315
30310 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
30311 was not set before.
30313 2017-02-01 Richard Biener <rguenther@suse.de>
30315 PR tree-optimization/71824
30316 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
30317 Verify the loops are valid in the merged SESE region.
30318 (scop_detection::can_represent_loop_1): Check analyzing the
30319 evolution of the number of iterations in the region succeeds.
30321 2017-01-31 Ian Lance Taylor <iant@golang.org>
30323 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
30324 REG_ARGS_SIZE note to 32-bit push insns and call insn.
30326 2017-01-31 David Malcolm <dmalcolm@redhat.com>
30328 PR preprocessor/79210
30329 * input.c (get_substring_ranges_for_loc): Replace line_width
30330 assertion with error-handling.
30332 2017-01-31 Richard Biener <rguenther@suse.de>
30334 PR tree-optimization/77318
30335 * graphite-sese-to-poly.c (extract_affine): Fix assert.
30336 (create_pw_aff_from_tree): Take loop parameter.
30337 (add_condition_to_pbb): Pass loop of the condition to
30338 create_pw_aff_from_tree.
30340 2017-01-31 Jakub Jelinek <jakub@redhat.com>
30342 * config/s390/s390.c (s390_asan_shadow_offset): New function.
30343 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
30345 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
30349 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
30351 (convert_int_to_float128): Likewise.
30352 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
30353 (convert_int_to_float128): Likewise.
30354 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
30355 (UNSPEC_IEEE128_CONVERT): Likewise.
30356 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
30357 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
30358 Use local variables for IBM extended format.
30359 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
30360 (fix_trunc<mode>si2_fprs): Likewise.
30361 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
30362 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
30363 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
30364 to know that we can now have integers of all sizes in vector
30366 (fix<uns>_<mode>di2_hw): Likewise.
30367 (float<uns>_<mode>si2_hw): Likewise.
30368 (fix_<mode>si2_hw): Likewise.
30369 (fixuns_<mode>si2_hw): Likewise.
30370 (float<uns>_<mode>di2_hw): Likewise.
30371 (float_<mode>di2_hw): Likewise.
30372 (float_<mode>si2_hw): Likewise.
30373 (floatuns_<mode>di2_hw): Likewise.
30374 (floatuns_<mode>si2_hw): Likewise.
30375 (xscvqp<su>wz_<mode>): Delete, no longer used.
30376 (xscvqp<su>dz_<mode>): Likewise.
30377 (xscv<su>dqp_<mode>): Likewise.
30378 (ieee128_mfvsrd_64bit): Likewise.
30379 (ieee128_mfvsrd_32bit): Likewise.
30380 (ieee128_mfvsrwz): Likewise.
30381 (ieee128_mtvsrw): Likewise.
30382 (ieee128_mtvsrd_64bit): Likewise.
30383 (ieee128_mtvsrd_32bit): Likewise.
30385 2017-01-31 Martin Liska <mliska@suse.cz>
30388 * ipa-prop.c (ipa_free_all_node_params): Call release method
30389 instead of ~sumbol_summary to not to trigger double times
30392 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
30394 PR tree-optimization/71691
30395 * bitmap.h (class auto_bitmap): New.
30396 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
30397 is_maybe_undefined instead of ssa_undefined_value_p.
30399 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30401 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
30402 __S390_ARCH_LEVEL__ to __ARCH__.
30404 2017-01-31 Jakub Jelinek <jakub@redhat.com>
30406 PR tree-optimization/79267
30407 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
30408 if should_remove_lhs_p is true.
30410 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
30413 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
30414 (add_alignment_attribute): New.
30415 (base_type_die): Add alignment attribute.
30416 (subrange_type_die): Likewise.
30417 (modified_type_die): Likewise.
30418 (gen_array_type_die): Likewise.
30419 (gen_descr_array_type_die: Likewise.
30420 (gen_enumeration_type_die): Likewise.
30421 (gen_subprogram_die): Likewise.
30422 (gen_variable_die): Likewise.
30423 (gen_field_die): Likewise.
30424 (gen_ptr_to_mbr_type_die): Likewise.
30425 (gen_struct_or_union_type_die): Likewise.
30426 (gen_subroutine_type_die): Likewise.
30427 (gen_typedef_die): Likewise.
30428 (base_type_cmp): Compare alignment attribute.
30430 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30433 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
30434 (setb_unsigned) New pattern for setb with CCUNS.
30435 * config/rs6000/rs6000.c (expand_block_compare): Use a different
30436 subfc./subfe sequence to avoid overflow problems. Generate a
30437 shorter sequence with cmpld/setb for power9.
30438 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
30439 for generating subfc. instruction.
30440 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
30441 now uses this instruction.
30443 2017-01-30 Ian Lance Taylor <iant@google.com>
30446 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
30447 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
30449 2017-01-30 Martin Sebor <msebor@redhat.com>
30451 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
30452 Move constant to the right of a relational operator.
30453 (get_mpfr_format_length, format_character, format_string): Ditto.
30454 (should_warn_p, maybe_warn): Same.
30456 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
30458 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
30461 * asan.c (get_translation_unit_decl): Remove function.
30462 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
30464 2017-01-30 Martin Liska <mliska@suse.cz>
30466 PR gcov-profile/79259
30467 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
30468 -fprofile-generate.
30470 2017-01-30 Martin Liska <mliska@suse.cz>
30473 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
30474 Initialize variables with NULL value.
30476 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
30479 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
30481 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
30483 2017-01-30 Richard Biener <rguenther@suse.de>
30485 PR tree-optimization/79276
30486 * tree-vrp.c (process_assert_insertions): Properly adjust common
30487 when removing a duplicate.
30489 2017-01-30 Richard Biener <rguenther@suse.de>
30491 PR tree-optimization/79256
30492 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
30493 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
30495 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
30497 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
30500 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
30501 ("*r<noxa>sbg_<mode>_sll_bitmask")
30502 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
30503 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
30504 Use contiguous_bitmask_nowrap_operand.
30506 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30509 * config/rs6000/altivec.h (vec_xl): Revise #define.
30510 (vec_xst): Likewise.
30512 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
30514 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
30516 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
30518 PR rtl-optimization/79194
30519 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
30520 traps before call to bypass_conditional_jumps.
30522 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
30524 PR tree-optimization/71374
30525 * lra-constraints.c (check_conflict_input_operands): New.
30526 (match_reload): Use it.
30528 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
30531 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
30532 account to calculate conflict_set.
30534 2017-01-27 Bin Cheng <bin.cheng@arm.com>
30536 PR rtl-optimization/78559
30537 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
30538 other_insn in combine.
30540 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
30542 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
30543 uint16_type_node for BT_UINT16.
30545 2017-01-27 David Malcolm <dmalcolm@redhat.com>
30547 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
30548 "RTL Tests" to menu.
30549 (GIMPLE Tests): New node.
30550 (RTL Tests): New node.
30552 2017-01-27 Richard Biener <rguenther@suse.de>
30554 PR tree-optimization/79245
30555 * tree-loop-distribution.c (distribute_loop): Apply cost
30556 modeling also to detected patterns.
30558 2017-01-27 Richard Biener <rguenther@suse.de>
30560 PR tree-optimization/71433
30561 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
30562 (compare_assert_loc): New function.
30563 (process_assert_insertions): Sort and optimize assert locations
30564 to remove duplicates and push down identical assertions on
30565 edges to their destination block.
30567 2017-01-27 Richard Biener <rguenther@suse.de>
30569 PR tree-optimization/79244
30570 * tree-vrp.c (remove_range_assertions): Forcefully propagate
30571 out SSA names even if abnormal.
30573 2017-01-27 Jakub Jelinek <jakub@redhat.com>
30575 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
30576 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
30577 instead of MPFR_RNDN.
30579 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
30582 * arm.c (arm_option_override): Don't call build_target_option_node
30583 until after doing all option overrides.
30584 (arm_valid_target_attribute_tree): Likewise.
30586 2017-01-27 Martin Liska <mliska@suse.cz>
30588 * doc/invoke.texi (-fprofile-arcs): Document profiling support
30589 for {cd}tors and C++ {cd}tors.
30591 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
30593 * config/s390/s390.md ("*setmem_long_and")
30594 ("*setmem_long_and_31z"): Use zero_extend instead of and.
30596 2017-01-26 Martin Sebor <msebor@redhat.com>
30598 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
30601 2017-01-26 Martin Sebor <msebor@redhat.com>
30603 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
30604 HAVE_DFmode before using XFmode or DFmode.
30605 (parse_directive): Avoid using the z length modifier to avoid
30606 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
30608 PR middle-end/78703
30609 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
30610 to accept adjustment as an array.
30611 (get_int_range): New function.
30612 (struct directive): Make width and prec arrays.
30613 (directive::set_width, directive::set_precision): Call get_int_range.
30614 (format_integer, format_floating): Handle width and precision ranges.
30615 (format_string, parse_directive): Same.
30617 2017-01-26 Jakub Jelinek <jakub@redhat.com>
30620 * dwarf2out.c (generate_skeleton_bottom_up): For children with
30621 comdat_type_p set, just clone them, but keep the children in the
30625 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
30626 which have direct callers with -fvar-tracking-assignments enabled
30628 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
30629 inside of type units.
30631 2017-01-26 Martin Sebor <msebor@redhat.com>
30633 PR middle-end/78703
30634 * gimple-ssa-sprintf.c (struct result_range): Add likely and
30636 (struct format_result): Replace number_chars, number_chars_min,
30637 and number_chars_max with a single member of struct result_range.
30639 (format_result::operator+=): Adjust.
30640 (struct fmtresult): Remove bounded. Handle likely and unlikely
30642 (fmtresult::adjust_for_width_or_precision): New function.
30643 (fmtresult:type_max_digits): New function.
30644 (bytes_remaining): Handle likely and unlikely counters.
30645 (min_bytes_remaining): Remove.
30646 (format_percent): Simplify.
30647 (format_integer, format_floating): Set likely and unlikely counters.
30648 (get_string_length, format_character, format_string): Same.
30649 (format_plain, should_warn_p): New function.
30650 (maybe_warn): Call should_warn_p. Update diagnostic messages
30651 and handle those for all directives, including plain strings.
30652 (format_directive): Handle likely and unlikely counters.
30653 Remove unnecessary quoting from diagnostics. Add an informational
30655 (add_bytes): Remove.
30656 (pass_sprintf_length::compute_format_length): Simplify.
30657 (try_substitute_return_value): Handle likely and unlikely counters.
30659 2017-01-26 Carl Love <cel@us.ibm.com>
30661 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
30662 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
30664 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
30667 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
30668 endianess for subregs into account.
30669 * lra-constraints.c (lra_constraints): Do risky transformations
30670 always on the first iteration.
30671 * lra-lives.c (check_pseudos_live_through_calls): Add arg
30672 last_call_used_reg_set.
30673 (process_bb_lives): Define and use last_call_used_reg_set.
30674 * lra.c (lra): Always continue after lra_constraints on the first
30677 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
30679 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
30681 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
30683 2017-01-26 Jakub Jelinek <jakub@redhat.com>
30685 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
30686 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
30687 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
30688 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
30689 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
30690 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
30691 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
30692 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
30693 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
30695 2017-01-26 Marek Polacek <polacek@redhat.com>
30698 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
30699 for the third operand.
30701 2017-01-26 Jakub Jelinek <jakub@redhat.com>
30703 PR middle-end/79236
30704 * omp-low.c (struct omp_context): Add simt_stmt field.
30705 (scan_omp_for): Return omp_context *.
30706 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
30707 context to the _simt_ SIMD stmt.
30708 (lower_omp_for): For combined SIMD with sibling _simt_
30709 SIMD, make sure to use the same decls in _looptemp_
30710 clauses as in the sibling.
30712 2017-01-26 David Sherwood <david.sherwood@arm.com>
30714 PR middle-end/79212
30715 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
30718 2017-01-26 Jakub Jelinek <jakub@redhat.com>
30721 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
30722 emit fld b; fld a; if possible.
30724 * brig-builtins.def: Update copyright years.
30725 * config/arm/arm_acle_builtins.def: Update copyright years.
30727 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
30730 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
30731 constraint instead of o for the stxsd instruction.
30733 2017-01-25 Carl Love <cel@us.ibm.com>
30735 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
30736 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
30738 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
30740 * doc/invoke.texi (C++ Dialect Options): Fix typo.
30742 2017-01-25 Richard Biener <rguenther@suse.de>
30744 PR tree-optimization/69264
30745 * target.def (vector_alignment_reachable): Improve documentation.
30746 * doc/tm.texi: Regenerate.
30747 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
30749 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
30750 earlier changes with respect to TYPE_USER_ALIGN.
30751 (vector_alignment_reachable_p): Likewise. Improve dumping.
30753 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30756 * config/arm/arm.md (xordi3): Force constant operand into a register
30759 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30761 * doc/invoke.texi (-fstore-merging): Correct default optimization
30762 levels at which it is enabled.
30763 (-O): Move -fstore-merging from list to...
30766 2017-01-25 Richard Biener <rguenther@suse.de>
30769 * omp-expand.c: Include debug.h.
30770 (expand_omp_taskreg): Make sure to generate early debug before
30771 outlining anything from a function.
30772 (expand_omp_target): Likewise.
30773 (grid_expand_target_grid_body): Likewise.
30775 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
30778 * asan.c (get_translation_unit_decl): New function.
30779 (asan_add_global): Extract modules file name from globals
30780 TRANSLATION_UNIT_DECL name.
30782 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
30785 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
30786 for long calls with APCS frame and VFP.
30788 2017-01-24 David Malcolm <dmalcolm@redhat.com>
30790 * cfg.c (original_copy_tables_initialized_p): New function.
30791 * cfg.h (original_copy_tables_initialized_p): New decl.
30792 * cfgrtl.c (relink_block_chain): Guard the call to
30793 free_original_copy_tables with a call to
30794 original_copy_tables_initialized_p.
30795 * cgraph.h (symtab_node::native_rtl_p): New decl.
30796 * cgraphunit.c (symtab_node::native_rtl_p): New function.
30797 (symtab_node::needed_p): Don't assert for early assembly output
30798 for __RTL functions.
30799 (cgraph_node::finalize_function): Set "force_output" for __RTL
30801 (cgraph_node::analyze): Bail out early for __RTL functions.
30802 (analyze_functions): Update assertion to support __RTL functions.
30803 (cgraph_node::expand): Bail out early for __RTL functions.
30804 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
30806 * function.h (struct function): Update comment for field
30808 * gimple-expr.c: Include "tree-pass.h".
30809 (gimple_has_body_p): Return false for __RTL functions.
30810 * Makefile.in (OBJS): Add run-rtl-passes.o.
30811 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
30813 (gcc::pass_manager::get_clean_slate): New accessor.
30814 * passes.c: Include "insn-addr.h".
30815 (should_skip_pass_p): Add logging. Update logic for running
30816 "expand" to be compatible with both __GIMPLE and __RTL. Guard
30817 property-provider override so it is only done for gimple passes.
30819 (skip_pass): New function.
30820 (execute_one_pass): Call skip_pass when skipping passes.
30821 * read-md.c (md_reader::read_char): Support filtering
30822 the input to a subset of line numbers.
30823 (md_reader::md_reader): Initialize fields
30824 m_first_line and m_last_line.
30825 (md_reader::read_file_fragment): New function.
30826 * read-md.h (md_reader::read_file_fragment): New decl.
30827 (md_reader::m_first_line): New field.
30828 (md_reader::m_last_line): New field.
30829 * read-rtl-function.c (function_reader::create_function): Only
30830 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
30831 curr_properties. Set DECL_INITIAL to a dummy block.
30832 (read_rtl_function_body_from_file_range): New function.
30833 * read-rtl-function.h (read_rtl_function_body_from_file_range):
30835 * run-rtl-passes.c: New file.
30836 * run-rtl-passes.h: New file.
30838 2017-01-24 Jeff Law <law@redhat.com>
30840 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
30843 2017-01-24 Bin Cheng <bin.cheng@arm.com>
30845 PR tree-optimization/79159
30846 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
30847 (record_nonwrapping_iv): Improve boundary using above function if no
30848 value range information.
30850 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
30851 Martin Jambor <mjambor@suse.cz>
30853 * brig-builtins.def: New file.
30854 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
30855 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
30856 (DEF_HSAIL_SAT_BUILTIN): Likewise.
30857 (DEF_HSAIL_INTR_BUILTIN): Likewise.
30858 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
30859 * builtin-types.def (BT_INT8): New.
30860 (BT_INT16): Likewise.
30861 (BT_UINT8): Likewise.
30862 (BT_UINT16): Likewise.
30863 (BT_FN_ULONG): Likewise.
30864 (BT_FN_UINT_INT): Likewise.
30865 (BT_FN_UINT_ULONG): Likewise.
30866 (BT_FN_UINT_LONG): Likewise.
30867 (BT_FN_UINT_PTR): Likewise.
30868 (BT_FN_ULONG_PTR): Likewise.
30869 (BT_FN_INT8_FLOAT): Likewise.
30870 (BT_FN_INT16_FLOAT): Likewise.
30871 (BT_FN_UINT32_FLOAT): Likewise.
30872 (BT_FN_UINT16_FLOAT): Likewise.
30873 (BT_FN_UINT8_FLOAT): Likewise.
30874 (BT_FN_UINT64_FLOAT): Likewise.
30875 (BT_FN_UINT16_UINT32): Likewise.
30876 (BT_FN_UINT32_UINT16): Likewise.
30877 (BT_FN_UINT16_UINT16_UINT16): Likewise.
30878 (BT_FN_INT_PTR_INT): Likewise.
30879 (BT_FN_UINT_PTR_UINT): Likewise.
30880 (BT_FN_LONG_PTR_LONG): Likewise.
30881 (BT_FN_ULONG_PTR_ULONG): Likewise.
30882 (BT_FN_VOID_UINT64_UINT64): Likewise.
30883 (BT_FN_UINT8_UINT8_UINT8): Likewise.
30884 (BT_FN_INT8_INT8_INT8): Likewise.
30885 (BT_FN_INT16_INT16_INT16): Likewise.
30886 (BT_FN_INT_INT_INT): Likewise.
30887 (BT_FN_UINT_FLOAT_UINT): Likewise.
30888 (BT_FN_FLOAT_UINT_UINT): Likewise.
30889 (BT_FN_ULONG_UINT_UINT): Likewise.
30890 (BT_FN_ULONG_UINT_PTR): Likewise.
30891 (BT_FN_ULONG_ULONG_ULONG): Likewise.
30892 (BT_FN_UINT_UINT_UINT): Likewise.
30893 (BT_FN_VOID_UINT_PTR): Likewise.
30894 (BT_FN_UINT_UINT_PTR: Likewise.
30895 (BT_FN_UINT32_UINT64_PTR): Likewise.
30896 (BT_FN_INT_INT_UINT_UINT): Likewise.
30897 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
30898 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
30899 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
30900 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
30901 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
30902 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
30903 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
30904 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
30905 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
30906 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
30907 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
30908 * doc/frontends.texi: List BRIG FE.
30909 * doc/install.texi (Testing): Add BRIG tesring requirements.
30910 * doc/invoke.texi (Overall Options): Mention BRIG.
30911 * doc/standards.texi (Standards): Doucment BRIG HSA version.
30913 2017-01-24 Richard Biener <rguenther@suse.de>
30915 PR translation/79208
30916 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
30918 2017-01-24 Martin Jambor <mjambor@suse.cz>
30921 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
30922 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
30923 and known_contexts.
30925 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
30927 PR middle-end/79123
30928 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
30929 casts from signed to unsigned really don't have a range.
30931 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
30933 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
30934 GMP_RNDx for compatiblity.
30936 2017-01-24 Martin Liska <mliska@suse.cz>
30939 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
30940 that would prevent us to call alloca with -1 as argument.
30942 2017-01-24 Jakub Jelinek <jakub@redhat.com>
30944 * dwarf2out.c (output_compilation_unit_header, output_file_names):
30945 Avoid -Wformat-security warning.
30947 2017-01-23 Andrew Pinski <apinski@cavium.com>
30949 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
30952 2017-01-23 Martin Sebor <msebor@redhat.com>
30954 PR middle-end/78703
30955 * gimple-ssa-sprintf.c (warn_level): New global.
30956 (format_integer): Use it here and throughout the rest of the file.
30957 Use the same switch to compute sign as base.
30958 (maybe_warn): New function.
30959 (format_directive): Factor out warnings into maybe_warn.
30960 Add debugging output. Use warn_level.
30961 (add_bytes): Use warn_level.
30962 (pass_sprintf_length::compute_format_length): Add debugging output.
30963 (try_substitute_return_value): Same.
30964 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
30966 PR middle-end/78703
30967 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
30968 (struct fmtresult, format_integer, format_floating): Adjust.
30969 (fmtresult::fmtresult): Set max correctly in two argument ctor.
30970 (get_string_length, format_string,format_directive): Same.
30971 (pass_sprintf_length::compute_format_length): Same.
30972 (try_substitute_return_value): Simplify slightly.
30974 PR middle-end/78703
30975 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
30976 (fmtresult::operator+=): Outlined.
30977 (struct fmtresult): Add ctors.
30978 (struct conversion_spec): Rename...
30979 (struct directive): ...to this. Add and remove data members.
30980 (directive::set_width, directive::set_precision): New functions.
30981 (format_percent): Use fmtresult ctor.
30982 (get_width_and_precision): Remove.
30983 (format_integer): Make naming changes. Avoid computing width and
30985 (format_floating): Same. Adjust indentation.
30986 (format_character, format_none): New functions.
30987 (format_string): Moved character handling to format_character.
30988 (format_directive): Remove arguments, change return type.
30989 (parse_directive): New function.
30990 (pass_sprintf_length::compute_format_length): Move directive
30991 parsing to parse_directive.
30993 2017-01-23 Jakub Jelinek <jakub@redhat.com>
30995 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
30996 (assign_assembler_name_if_needed): ... this.
30997 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
30998 (assign_assembler_name_if_needed): ... this.
30999 (free_lang_data_in_cgraph): Adjust callers.
31000 * cgraphunit.c (cgraph_node::analyze): Likewise.
31001 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
31004 2017-01-23 Richard Biener <rguenther@suse.de>
31006 PR tree-optimization/79088
31007 PR tree-optimization/79188
31008 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
31009 resetting loop bounds after last path deletion. Reset loop
31010 bounds of the target loop, make code match the comments.
31011 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
31012 Make sure loops need no fixups.
31014 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
31016 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
31017 exponent support with double type for first argument.
31018 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
31019 type returned by __builtin_vec_extract_sig,
31020 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
31021 functions from "vector int" to "vector unsigned int" or from
31022 "vector long long int" to "vector unsigned long long int".
31023 Changed type returned by __builtin_vec_extract_exp,
31024 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
31025 functions from "vector int" to "vector unsigned int" or from
31026 "vector long long int" to "vector unsigned long long int".
31027 Changed return type of __builtin_vec_test_data_class,
31028 __builtin_vec_test_data_class_sp, and
31029 __builtin_vec_test_data_class_dp from "vector int" to
31030 "vector bool int" or from "vector long long int" to "vector bool
31031 long long int" and changed second argument type from "unsigned
31032 int" to "int". Added new overloaded function forms "vector float
31033 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
31034 "vector float __builtin_vec_insert_exp_sp (vector float, vector
31035 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
31036 double, vector unsigned long long int)" and "vector double
31037 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
31038 long int)". Changed return type of
31039 __builtin_scalar_test_data_class and
31040 __builtin_scalar_test_data_class_sp and
31041 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
31042 int" and changed second argument from "unsigned int" to "int".
31043 Changed type returned by __builtin_scalar_test_neg,
31044 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
31045 from "int" to "bool int". Added new overloaded function form
31046 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
31047 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
31048 exponent double-precision with floating point first argument.
31049 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
31050 documentation of scalar_test_data_class, scalar_test_neg,
31051 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
31052 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
31053 vec_test_data_class built-in functions to reflect refinements in
31054 their type signatures.
31056 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
31058 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
31060 (aarch64_elf_asm_destructor): Likewise.
31062 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
31064 PR rtl-optimization/78634
31065 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
31066 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
31067 * ifcvt.c (noce_try_cmove): Add missing cost check.
31069 PR rtl-optimization/71724
31070 * combine.c (if_then_else_cond): Look for situations where it is
31071 beneficial to undo the work of one of the recursive calls.
31073 2017-01-23 Bin Cheng <bin.cheng@arm.com>
31075 PR tree-optimization/70754
31076 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
31077 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
31078 combined stmt before it if not NULL.
31079 (combine_chains): Process refs reversely and compute dominance point
31082 2017-01-23 Martin Liska <mliska@suse.cz>
31084 PR tree-optimization/79196
31085 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
31086 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
31088 (strlen_optimize_stmt): Call the renamed function.
31090 2017-01-23 Michael Matz <matz@suse.de>
31092 PR tree-optimization/78384
31093 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
31095 2017-01-23 Richard Biener <rguenther@suse.de>
31097 PR tree-optimization/79186
31098 * tree-vrp.c (register_new_assert_for): Make sure we've seen
31099 both incoming edges before moving an assert.
31101 2017-01-23 Martin Jambor <mjambor@suse.cz>
31103 * ipa-prop.c (load_from_param_1): Removed.
31104 (load_from_unmodified_param): Bits from load_from_param_1 put back
31106 (load_from_param): Removed.
31107 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
31108 with stmt. Reverted back to use of load_from_unmodified_param.
31110 2017-01-23 Martin Jambor <mjambor@suse.cz>
31113 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
31114 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
31115 field a pointer to garbage collected vector, mark lattices and
31116 ipcp_orig_node with GTY((skip)).
31117 (ipa_get_param_count): Adjust to descriptors being a pointer.
31118 (ipa_get_param): Likewise.
31119 (ipa_get_type): Likewise.
31120 (ipa_get_param_move_cost): Likewise.
31121 (ipa_set_param_used): Likewise.
31122 (ipa_get_controlled_uses): Likewise.
31123 (ipa_set_controlled_uses): Likewise.
31124 (ipa_is_param_used): Likewise.
31125 (ipa_node_params_t): Move into garbage collector. New methods insert
31127 (ipa_node_params_sum): Annotate wth GTY(()).
31128 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
31130 (ipa_load_from_parm_agg): Adjust declaration.
31131 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
31132 * ipa-profile.c (ipa_profile): Likewise.
31133 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
31134 (ipa_populate_param_decls): Make descriptors parameter garbage
31136 (ipa_dump_param): Adjust to descriptors being a pointer.
31137 (ipa_alloc_node_params): Likewise.
31138 (ipa_initialize_node_params): Likewise.
31139 (load_from_param_1): Make descriptors parameter garbage collected.
31140 (load_from_unmodified_param): Likewise.
31141 (load_from_param): Likewise.
31142 (ipa_load_from_parm_agg): Likewise.
31143 (ipa_node_params::~ipa_node_params): Removed.
31144 (ipa_free_all_node_params): Remove call to delete operator.
31145 (ipa_node_params_t::insert): New.
31146 (ipa_node_params_t::remove): Likewise.
31147 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
31148 copy known_csts and known_contexts vectors.
31149 (ipa_read_node_info): Adjust to descriptors being a pointer.
31150 (ipcp_modif_dom_walker): Make m_descriptors field garbage
31152 (ipcp_transform_function): Make descriptors variable garbage
31155 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
31157 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
31158 * config/i386/avx512dqintrin.h: Ditto.
31159 * config/i386/avx512fintrin.h: Ditto.
31160 * config/i386/i386.c: Handle new builtins.
31161 * config/i386/i386-builtin.def: Add new builtins.
31162 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
31163 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
31165 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31166 Martin Liska <mliska@suse.cz>
31168 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
31169 * asan.c (asan_expand_poison_ifn): Support stores and use
31170 appropriate ASAN report function.
31171 * internal-fn.c (expand_ASAN_POISON_USE): New function.
31172 * internal-fn.def (ASAN_POISON_USE): Declare.
31173 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
31174 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
31175 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
31176 ASAN_POISON calls w/o LHS.
31177 * tree-ssa.c (execute_update_addresses_taken): Create clobber
31178 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
31179 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
31180 * gimplify.c (asan_poison_variables): Add attribute
31181 use_after_scope_memory to variables that really needs to live
31183 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
31184 having the attribute.
31186 2017-01-23 Martin Liska <mliska@suse.cz>
31188 * asan.c (create_asan_shadow_var): New function.
31189 (asan_expand_poison_ifn): Likewise.
31190 * asan.h (asan_expand_poison_ifn): New declaration.
31191 * internal-fn.c (expand_ASAN_POISON): Likewise.
31192 * internal-fn.def (ASAN_POISON): New builtin.
31193 * sanopt.c (pass_sanopt::execute): Expand
31194 asan_expand_poison_ifn.
31195 * tree-inline.c (copy_decl_for_dup_finish): Make function
31197 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
31198 * tree-ssa.c (is_asan_mark_p): New function.
31199 (execute_update_addresses_taken): Rewrite local variables
31200 (identified just by use-after-scope as addressable) into SSA.
31202 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31204 * doc/install.texi (Specific): opensource.apple.com uses https
31205 now. Remove trailing slash.
31207 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31209 * README.Portability: Remove note on an Irix compatibility issue.
31211 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
31213 * gcov.c (INCLUDE_ALGORITHM): Define.
31214 (INCLUDE_VECTOR): Define.
31215 No longer include <vector> and <algorithm> directly.
31217 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
31219 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
31221 * doc/invoke.texi (Code Gen Options): Ditto.
31223 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
31226 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
31228 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
31230 rtl-optimization/79125
31231 * cprop.c (local_cprop_pass): Handle cases where we make an
31232 unconditional trap.
31234 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
31238 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
31239 read from, for big endian.
31241 2017-01-20 Jiong Wang <jiong.wang@arm.com>
31243 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
31244 register pauth builtins for LP64 only.
31246 2017-01-20 Marek Polacek <polacek@redhat.com>
31249 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
31252 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
31254 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
31256 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
31257 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
31258 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
31260 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31263 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
31264 in big-endian mode when they are not a single duplicated value.
31266 2017-01-20 Richard Biener <rguenther@suse.de>
31268 * BASE-VER: Bump to 7.0.1.
31270 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
31272 * omp-low.c (omplow_simd_context): New struct. Use it...
31273 (lower_rec_simd_input_clauses): ...here and...
31274 (lower_rec_input_clauses): ...here to hold common data. Adjust all
31275 references to idx, lane, max_vf, is_simt.
31277 2017-01-20 Graham Markall <graham.markall@embecosm.com>
31279 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
31282 2017-01-20 Martin Jambor <mjambor@suse.cz>
31284 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
31285 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
31287 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
31288 (GTFILES): Rename hsa.c to hsa-common.c.
31289 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
31290 * hsa-dump.c: Likewise.
31291 * hsa-gen.c: Likewise.
31292 * hsa-regalloc.c: Likewise.
31293 * ipa-hsa.c: Likewise.
31294 * omp-expand.c: Likewise.
31295 * omp-low.c: Likewise.
31296 * toplev.c: Likewise.
31298 2017-01-20 Marek Polacek <polacek@redhat.com>
31301 * doc/invoke.texi: Document -Wduplicated-branches.
31302 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
31303 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
31304 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
31305 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
31306 return 0 only when not OEP_LEXICOGRAPHIC.
31307 (fold_build_cleanup_point_expr): Use the expression
31308 location when building CLEANUP_POINT_EXPR.
31309 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
31310 * tree.c (add_expr): Handle error_mark_node.
31312 2017-01-20 Martin Liska <mliska@suse.cz>
31315 * tree-profile.c (init_ic_make_global_vars): Do not call
31317 (gimple_init_gcov_profiler): Likewise.
31319 2017-01-20 Martin Liska <mliska@suse.cz>
31322 * cgraph.h (maybe_create_reference): Remove argument and
31324 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
31326 * ipa-cp.c (create_specialized_node): Likewise.
31327 * symtab.c (symtab_node::maybe_create_reference): Handle
31328 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
31330 2017-01-20 Martin Liska <mliska@suse.cz>
31332 * read-rtl-function.c (function_reader::create_function): Use
31333 build_decl instread of build_decl_stat.
31335 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
31337 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
31338 * config/i386/avx512dqintrin.h: Ditto.
31339 * config/i386/avx512fintrin.h: Ditto.
31340 * config/i386/i386-builtin-types.def: Add new types.
31341 * config/i386/i386.c: Handle new types.
31342 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
31343 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
31344 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
31345 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
31346 (__builtin_ia32_kshiftridi): New.
31347 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
31349 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
31353 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
31354 define to rs6000_init_stack_protect_guard.
31355 (rs6000_init_stack_protect_guard): New function.
31357 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
31358 Yunqiang Su <yunqiang.su@imgtec.com>
31360 * config.gcc (supported_defaults): Add madd4.
31361 (with_madd4): Add validation.
31362 (all_defaults): Add madd4.
31363 * config/mips/mips.opt (mmadd4): New option.
31364 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
31366 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
31368 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
31369 (ISA_HAS_FUSED_MADD4): Likewise.
31370 * doc/invoke.texi (-mmadd4): Document the new option.
31371 * doc/install.texi (--with-madd4): Document the new option.
31373 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31375 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
31376 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
31377 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
31378 (aarch64_init_pauth_hint_builtins): New.
31379 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
31380 (aarch64_expand_builtin): Expand new builtins.
31382 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31384 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
31385 * combine-stack-adj.c (no_unhandled_cfa): Handle
31386 REG_CFA_TOGGLE_RA_MANGLE.
31387 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
31388 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
31389 info for return address signing.
31390 (aarch64_expand_epilogue): Likewise.
31392 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31394 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
31395 * config/aarch64/aarch64-protos.h
31396 (aarch64_return_address_signing_enabled): New declaration.
31397 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
31399 (aarch64_expand_prologue): Sign return address before it's pushed onto
31401 (aarch64_expand_epilogue): Authenticate return address fetched from
31403 (aarch64_override_options): Sanity check for ILP32 and ISA level.
31404 (aarch64_attributes): New function attributes for "sign-return-address".
31405 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
31406 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
31407 ("*do_return"): Generate combined instructions according to key index.
31408 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
31409 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
31411 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
31412 * config/aarch64/aarch64.opt (msign-return-address=): New.
31413 * doc/extend.texi (AArch64 Function Attributes): Documents
31414 "sign-return-address=".
31415 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
31417 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
31419 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
31420 overall option summary.
31422 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31424 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
31425 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
31426 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
31427 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
31429 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
31431 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
31432 -mpower9-minmax by default for -mcpu=power9.
31433 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
31434 128-bit floating point.
31436 2017-01-20 Alan Modra <amodra@gmail.com>
31438 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
31439 optimizing for size.
31441 2017-01-20 Alan Modra <amodra@gmail.com>
31444 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
31445 for strcmp and strncmp from corresponding builtin decl.
31447 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
31449 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
31450 instead of i386/rtems-64.h.
31451 * config/i386/rtems-64.h: Remove.
31453 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
31457 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
31459 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
31461 2017-01-19 Tamar Christina <tamar.christina@arm.com>
31463 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
31464 Change int to HOST_WIDE_INT.
31465 * config/aarch64/aarch64-protos.h
31466 (aarch64_simd_gen_const_vector_dup): Likewise.
31467 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
31469 2017-01-19 David Malcolm <dmalcolm@redhat.com>
31471 * langhooks-def.h (lhd_type_for_size): New decl.
31472 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
31473 * langhooks.c (lhd_type_for_size): New function, taken from
31476 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
31478 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
31479 define_bypass for CR latency.
31480 (power9-cracked-alu): Update bypass latency and remove power9-branch.
31481 (power9-alu2): Add define_bypass for CR latency.
31483 (power9-mul): Update insn latency.
31484 (power9-mul-compare): Update insn latency, bypass latency and remove
31487 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31489 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
31491 * config/aarch64/aarch64.md
31492 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
31493 aarch64_nopcrelative_literal_loads.
31494 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
31496 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
31498 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
31499 TARGET_LOONGSON_3A.
31500 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
31502 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
31505 * config.gcc (supported_defaults): Add lxc1-sxc1.
31506 (with_lxc1_sxc1): Add validation.
31507 (all_defaults): Add lxc1-sxc1.
31508 * config/mips/mips.opt (mlxc1-sxc1): New option.
31509 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
31511 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
31512 __mips_no_lxc1_sxc1.
31513 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
31514 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
31515 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
31517 2017-01-19 Richard Biener <rguenther@suse.de>
31519 PR tree-optimization/72488
31520 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
31521 sure to restore SSA info.
31522 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
31524 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
31526 PR rtl-optimization/79121
31527 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
31528 of the inner type when shifting an extended value.
31530 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
31533 * symtab.c (symtab_node::equal_address_to): Fix comparing of
31534 interposable aliases.
31536 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
31539 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
31540 Use the evmergelohi instruction.
31541 (mov_si<mode>_e500_subreg4_2_le): Likewise.
31542 (mov_sitf_e500_subreg8_2_be): Likewise.
31543 (mov_sitf_e500_subreg12_2_le): Likewise.
31544 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
31545 (mov_si<mode>_e500_subreg4_2_be): Likewise.
31546 (mov_sitf_e500_subreg8_2_le): Likewise.
31547 (mov_sitf_e500_subreg12_2_be): Likewise.
31549 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31551 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
31552 attribute from vecsimple to vecperm.
31553 (altivec_vbpermq2): Likewise.
31555 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31558 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
31560 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31561 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
31562 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
31563 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
31564 case where N arg is SIZE_MAX.
31565 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
31566 (cmpstrsi): Add pattern.
31568 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
31570 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
31571 __builtin_vec_revb builtins.
31572 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
31573 built-in functions to support generation of the ISA 3.0 XXBR<x>
31574 vector byte reverse instructions.
31575 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
31576 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
31577 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
31578 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
31579 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
31580 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
31581 (P9V_BUILTIN_VEC_REVB): Likewise.
31582 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
31583 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
31584 (p9_xxbrq_v16qi): Likewise.
31585 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
31586 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
31587 (p9_xxbrh_v8hi): Likewise.
31588 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
31589 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
31590 vec_revb built-in functions.
31592 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
31594 PR rtl-optimization/78952
31595 * config/i386/i386.md (any_extract): New code iterator.
31596 (*insvqi_2): Use any_extract for source operand.
31597 (*insvqi_3): Use any_shiftrt for source operand.
31599 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
31601 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
31603 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
31605 2017-01-18 Matthias Klose <doko@ubuntu.com>
31607 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
31609 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31611 * config/rs6000/altivec.h (vec_bperm): Change #define.
31612 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
31613 (altivec_vbpermq2): New define_insn.
31614 (altivec_vbpermd): Likewise.
31615 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
31616 function interface.
31617 (VBPERMD): Likewise.
31618 (VBPERM): New polymorphic function interface.
31619 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
31620 Add entries for P9V_BUILTIN_VEC_VBPERM.
31621 * doc/extend.texi: Add interfaces for vec_bperm.
31623 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31625 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
31626 first letter of error messages.
31627 (s390_resolve_overloaded_builtin): Likewise.
31628 * config/s390/s390.c (s390_expand_builtin): Likewise.
31629 (s390_invalid_arg_for_unprototyped_fn): Likewise.
31630 (s390_valid_target_attribute_inner_p): Likewise.
31631 * config/s390/s390.md ("tabort"): Likewise.
31633 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
31635 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
31636 (ISA_AVOID_DIV_HILO): New macro.
31637 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
31638 (ISA_HAS_DDIV): Likewise.
31640 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
31642 * doc/invoke.texi (fabi-version): Correct number of occurrences.
31644 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
31646 * doc/invoke.texi (fabi-version): Spelling fix.
31648 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
31651 * doc/invoke.texi (fabi-version): Mention mangling fix for
31654 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
31657 * doc/invoke.texi (fabi-version): Document discriminator mangling.
31659 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
31662 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
31663 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
31665 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
31667 (stack_protect_test): Ditto.
31668 * config/rs6000/rs6000.opt (mstack-protector-guard=,
31669 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
31671 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
31672 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
31673 -mstack-protector-guard-offset=.
31674 (RS/6000 and PowerPC Options): Ditto.
31676 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
31678 * config/i386/i386.h (MASK_CLASS_P): New define.
31679 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
31680 there are no registers from different register sets also when
31681 mask registers are used. Update function comment.
31682 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
31683 to (*k/*r) and (*k/*km) alternatives.
31685 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
31687 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
31688 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
31689 (EH_RETURN_HANDLER_RTX): New define.
31690 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
31691 Force frame pointer in EH return functions.
31692 (aarch64_expand_epilogue): Add barrier for eh_return.
31693 (aarch64_final_eh_return_addr): Remove.
31694 (aarch64_eh_return_handler_rtx): New function.
31695 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
31697 (aarch64_eh_return_handler_rtx): New prototype.
31699 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31701 * config/rs6000/altivec.h (vec_rlmi): New #define.
31702 (vec_vrlnm): Likewise.
31703 (vec_rlnm): Likewise.
31704 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
31705 (UNSPEC_VRLNM): Likewise.
31706 (VIlong): New mode iterator.
31707 (altivec_vrl<VI_char>mi): New define_insn.
31708 (altivec_vrl<VI_char>nm): Likewise.
31709 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
31711 (VRLDNM): Likewise.
31712 (RLNM): New polymorphic function entry.
31713 (VRLWMI): New monomorphic function entry.
31714 (VRLDMI): Likewise.
31715 (RLMI): New polymorphic function entry.
31716 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
31717 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
31718 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
31721 2017-01-17 Jakub Jelinek <jakub@redhat.com>
31724 * dwarf2out.c (field_byte_offset): Restore the
31725 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
31726 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
31727 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
31730 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
31733 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
31735 2017-01-17 Jakub Jelinek <jakub@redhat.com>
31738 * dwarf2out.c (add_data_member_location_attribute): For constant
31739 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
31740 instead of DW_AT_data_member_location, DW_AT_bit_offset and
31741 DW_AT_byte_size attributes.
31743 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
31745 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
31746 after forcing to constant memory when the code model is medium.
31748 2017-01-17 Julia Koval <julia.koval@intel.com>
31751 * config/i386/avx512fintrin.h
31752 (_mm512_i32gather_ps): Change __addr type to void const*.
31753 (_mm512_mask_i32gather_ps): Ditto.
31754 (_mm512_i32gather_pd): Ditto.
31755 (_mm512_mask_i32gather_pd): Ditto.
31756 (_mm512_i64gather_ps): Ditto.
31757 (_mm512_mask_i64gather_ps): Ditto.
31758 (_mm512_i64gather_pd): Ditto.
31759 (_mm512_mask_i64gather_pd): Ditto.
31760 (_mm512_i32gather_epi32): Ditto.
31761 (_mm512_mask_i32gather_epi32): Ditto.
31762 (_mm512_i32gather_epi64): Ditto.
31763 (_mm512_mask_i32gather_epi64): Ditto.
31764 (_mm512_i64gather_epi32): Ditto.
31765 (_mm512_mask_i64gather_epi32): Ditto.
31766 (_mm512_i64gather_epi64): Ditto.
31767 (_mm512_mask_i64gather_epi64): Ditto.
31768 (_mm512_i32scatter_ps): Change __addr type to void*.
31769 (_mm512_mask_i32scatter_ps): Ditto.
31770 (_mm512_i32scatter_pd): Ditto.
31771 (_mm512_mask_i32scatter_pd): Ditto.
31772 (_mm512_i64scatter_ps): Ditto.
31773 (_mm512_mask_i64scatter_ps): Ditto.
31774 (_mm512_i64scatter_pd): Ditto.
31775 (_mm512_mask_i64scatter_pd): Ditto.
31776 (_mm512_i32scatter_epi32): Ditto.
31777 (_mm512_mask_i32scatter_epi32): Ditto.
31778 (_mm512_i32scatter_epi64): Ditto.
31779 (_mm512_mask_i32scatter_epi64): Ditto.
31780 (_mm512_i64scatter_epi32): Ditto.
31781 (_mm512_mask_i64scatter_epi32): Ditto.
31782 (_mm512_i64scatter_epi64): Ditto.
31783 (_mm512_mask_i64scatter_epi64): Ditto.
31784 * config/i386/avx512pfintrin.h
31785 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
31786 (_mm512_mask_prefetch_i32gather_ps): Ditto.
31787 (_mm512_mask_prefetch_i64gather_pd): Ditto.
31788 (_mm512_mask_prefetch_i64gather_ps): Ditto.
31789 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
31790 (_mm512_prefetch_i32scatter_ps): Ditto.
31791 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
31792 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
31793 (_mm512_prefetch_i64scatter_pd): Ditto.
31794 (_mm512_prefetch_i64scatter_ps): Ditto.
31795 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
31796 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
31797 * config/i386/avx512vlintrin.h
31798 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
31799 (_mm_mmask_i32gather_ps): Ditto.
31800 (_mm256_mmask_i32gather_pd): Ditto.
31801 (_mm_mmask_i32gather_pd): Ditto.
31802 (_mm256_mmask_i64gather_ps): Ditto.
31803 (_mm_mmask_i64gather_ps): Ditto.
31804 (_mm256_mmask_i64gather_pd): Ditto.
31805 (_mm_mmask_i64gather_pd): Ditto.
31806 (_mm256_mmask_i32gather_epi32): Ditto.
31807 (_mm_mmask_i32gather_epi32): Ditto.
31808 (_mm256_mmask_i32gather_epi64): Ditto.
31809 (_mm_mmask_i32gather_epi64): Ditto.
31810 (_mm256_mmask_i64gather_epi32): Ditto.
31811 (_mm_mmask_i64gather_epi32): Ditto.
31812 (_mm256_mmask_i64gather_epi64): Ditto.
31813 (_mm_mmask_i64gather_epi64): Ditto.
31814 (_mm256_i32scatter_ps): Change __addr type to void*.
31815 (_mm256_mask_i32scatter_ps): Ditto.
31816 (_mm_i32scatter_ps): Ditto.
31817 (_mm_mask_i32scatter_ps): Ditto.
31818 (_mm256_i32scatter_pd): Ditto.
31819 (_mm256_mask_i32scatter_pd): Ditto.
31820 (_mm_i32scatter_pd): Ditto.
31821 (_mm_mask_i32scatter_pd): Ditto.
31822 (_mm256_i64scatter_ps): Ditto.
31823 (_mm256_mask_i64scatter_ps): Ditto.
31824 (_mm_i64scatter_ps): Ditto.
31825 (_mm_mask_i64scatter_ps): Ditto.
31826 (_mm256_i64scatter_pd): Ditto.
31827 (_mm256_mask_i64scatter_pd): Ditto.
31828 (_mm_i64scatter_pd): Ditto.
31829 (_mm_mask_i64scatter_pd): Ditto.
31830 (_mm256_i32scatter_epi32): Ditto.
31831 (_mm256_mask_i32scatter_epi32): Ditto.
31832 (_mm_i32scatter_epi32): Ditto.
31833 (_mm_mask_i32scatter_epi32): Ditto.
31834 (_mm256_i32scatter_epi64): Ditto.
31835 (_mm256_mask_i32scatter_epi64): Ditto.
31836 (_mm_i32scatter_epi64): Ditto.
31837 (_mm_mask_i32scatter_epi64): Ditto.
31838 (_mm256_i64scatter_epi32): Ditto.
31839 (_mm256_mask_i64scatter_epi32): Ditto.
31840 (_mm_i64scatter_epi32): Ditto.
31841 (_mm_mask_i64scatter_epi32): Ditto.
31842 (_mm256_i64scatter_epi64): Ditto.
31843 (_mm256_mask_i64scatter_epi64): Ditto.
31844 (_mm_i64scatter_epi64): Ditto.
31845 (_mm_mask_i64scatter_epi64): Ditto.
31846 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
31847 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
31848 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
31849 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
31850 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
31851 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
31852 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
31853 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
31854 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
31855 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
31856 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
31857 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
31858 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
31859 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
31860 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
31861 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
31862 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
31863 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
31864 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
31865 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
31866 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
31867 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
31868 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
31869 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
31870 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
31871 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
31872 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
31873 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
31874 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
31875 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
31876 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
31877 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
31878 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
31879 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
31880 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
31881 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
31882 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
31883 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
31884 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
31885 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
31886 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
31887 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
31888 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
31889 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
31890 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
31891 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
31892 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
31893 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
31894 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
31895 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
31896 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
31897 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
31898 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
31899 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
31900 definitions accordingly.
31902 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
31903 Kuan-Lin Chen <kuanlinchentw@gmail.com>
31906 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
31909 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
31912 * ira-conflicts.c (ira_build_conflicts): Update total conflict
31913 hard regs for inner regno.
31915 2017-01-17 Martin Liska <mliska@suse.cz>
31918 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
31919 assumption and add comment.
31921 2017-01-17 Nathan Sidwell <nathan@acm.org>
31923 * ipa-visibility.c (localize_node): New function, broken out of ...
31924 (function_and_variable_visibility): ... here. Call it.
31926 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
31928 PR middle-end/77445
31929 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
31930 correctly set frequency of oudgoing edge.
31931 (duplicate_thread_path): Fix profile updating.
31933 2017-01-17 Jakub Jelinek <jakub@redhat.com>
31936 * configure.ac: Add GCC_BASE_VER.
31937 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
31938 version from BASE-VER file.
31939 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
31940 (gcc.o): Depend on $(BASEVER).
31941 * common.opt (dumpfullversion): New option.
31942 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
31943 * doc/invoke.texi: Document -dumpfullversion.
31944 * doc/install.texi: Document --with-gcc-major-version-only.
31945 * configure: Regenerated.
31947 2017-01-17 Richard Biener <rguenther@suse.de>
31949 PR tree-optimization/71433
31950 * tree-vrp.c (register_new_assert_for): Merge same asserts
31951 on all incoming edges.
31952 (process_assert_insertions_for): Handle insertions at the
31955 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
31957 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
31958 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
31960 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
31963 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
31966 2017-01-17 Alan Modra <amodra@gmail.com>
31969 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
31970 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
31971 symbolic stack limit when pic.
31973 2017-01-16 Martin Sebor <msebor@redhat.com>
31975 PR tree-optimization/78608
31976 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
31978 2017-01-16 Jeff Law <law@redhat.com>
31981 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
31982 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
31983 for several include directories that may be relative to sysroot.
31984 * config/i386/x-mingw32 (gplus_includedir): Define.
31985 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
31986 (native_system_includedir): Likewise.
31987 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
31988 override if TARGET_SYSTEM_ROOT is defined.
31989 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
31991 PR tree-optimization/79090
31992 PR tree-optimization/33562
31993 PR tree-optimization/61912
31994 PR tree-optimization/77485
31995 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
31996 and computed trims into the dump file.
31998 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32000 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
32002 2017-01-16 Jakub Jelinek <jakub@redhat.com>
32005 * gimplify.c (gimplify_init_constructor): If want_value and
32006 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
32010 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
32011 sequence. Formatting fixes.
32012 (doloop_optimize): Formatting fixes.
32015 * gcc.c (debug_level_greater_than_spec_func): New function.
32016 (static_spec_functions): Add debug-level-gt spec function.
32017 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
32019 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32020 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32021 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
32022 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
32023 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
32024 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
32026 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
32028 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
32029 QImode fixups to general and mask registers only.
32031 2017-01-16 Carl Love <cel@us.ibm.com>
32033 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
32034 for built-in functions
32035 vector signed char vec_nabs (vector signed char)
32036 vector signed short vec_nabs (vector signed short)
32037 vector signed int vec_nabs (vector signed int)
32038 vector signed long long vec_nabs (vector signed long long)
32039 vector float vec_nabs (vector float)
32040 vector double vec_nabs (vector double)
32041 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
32043 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
32044 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
32045 * doc/extend.texi: Update the documentation file for the new built-in
32048 2017-01-16 Martin Sebor <msebor@redhat.com>
32050 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
32053 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32055 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
32056 UNSPEC_VSX__XXSPLTD to require special splat handling.
32058 2017-01-16 David Malcolm <dmalcolm@redhat.com>
32061 * system.h: Poison strndup.
32063 2017-01-16 Alan Modra <amodra@gmail.com>
32066 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
32069 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
32071 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
32073 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
32075 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
32076 call recog here. Assert that INSN_CODE (insn) is non-negative.
32078 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32081 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
32083 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
32084 in the currently scheduled RTL fragment.
32086 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32088 PR rtl-optimization/78751
32089 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
32092 2017-01-14 Jeff Law <law@redhat.com>
32094 PR tree-optimization/79090
32095 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
32096 variable length stores.
32097 (compute_trims): Delete dead assignment to *trim_tail.
32098 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
32101 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
32103 PR rtl-optimization/78626
32104 PR rtl-optimization/78727
32105 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
32106 of a block, and split such blocks after everything else is finished.
32108 2017-01-14 Alan Modra <amodra@gmail.com>
32111 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
32112 target legitimate_combined_insn.
32113 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
32114 (rs6000_legitimate_combined_insn): New function.
32115 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
32117 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
32118 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
32119 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
32121 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
32123 * doc/frontends.texi (G++ and GCC): Remove references to Java.
32125 2017-01-13 Jeff Law <law@redhat.com>
32127 PR tree-optimization/33562
32128 PR tree-optimization/61912
32129 PR tree-optimization/77485
32130 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
32132 (delete_dead_assignment): Likewise.
32133 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
32134 statement to delete_dead_call and delete_dead_assignment.
32136 2017-01-13 David Malcolm <dmalcolm@redhat.com>
32139 * substring-locations.c (format_warning_va): Strengthen case 1 so
32140 that both endpoints of the substring must be within the format
32141 range for just the substring to be printed.
32143 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
32145 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
32146 * config/i386/i386.c (ix86_target_string): Add missing options
32147 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
32148 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
32149 flags_other and ix86_target_other to flags2_other. Display unknown
32151 (ix86_valid_target_attribute_inner_p): Add missing options and
32152 reorder options by implied ISAs, as in ix86_target_string.
32154 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32156 * hash-table.h (hash_table::too_empty_p): New function.
32157 (hash_table::expand): Use it.
32158 (hash_table::traverse): Likewise.
32159 (hash_table::empty_slot): Use sizeof (value_type) instead of
32160 sizeof (PTR) to convert bytes to elements. Shrink the table
32161 if the current size is excessive for the current number of
32164 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32166 * ira-costs.c (record_reg_classes): Break from the inner loop
32167 early once alt_fail is known to be true. Update outer loop
32168 handling accordingly.
32170 2017-01-13 Jeff Law <law@redhat.com>
32172 * tree-ssa-dse.c (decrement_count): New function.
32173 (increment_start_addr, maybe_trim_memstar_call): Likewise.
32174 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
32175 when we know the partially dead statement is a mem* function.
32177 PR tree-optimization/61912
32178 PR tree-optimization/77485
32179 * tree-ssa-dse.c: Include expr.h.
32180 (maybe_trim_constructor_store): New function.
32181 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
32183 PR tree-optimization/33562
32184 PR tree-optimization/61912
32185 PR tree-optimization/77485
32186 * doc/invoke.texi: Document new dse-max-object-size param.
32187 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
32188 * tree-ssa-dse.c: Include params.h.
32189 (dse_store_status): New enum.
32190 (initialize_ao_ref_for_dse): New, partially extracted from
32192 (valid_ao_ref_for_dse, normalize_ref): New.
32193 (setup_live_bytes_from_ref, compute_trims): Likewise.
32194 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
32195 (maybe_trim_partially_dead_store): Likewise.
32196 (maybe_trim_complex_store): Likewise.
32197 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
32198 Track what bytes live from the original store. Return tri-state
32199 for dead, partially dead or live.
32200 (dse_dom_walker): Add constructor, destructor and new private members.
32201 (delete_dead_call, delete_dead_assignment): New extracted from
32203 (dse_optimize_stmt): Make a member of dse_dom_walker.
32204 Use initialize_ao_ref_for_dse.
32206 PR tree-optimization/33562
32207 PR tree-optimization/61912
32208 PR tree-optimization/77485
32209 * sbitmap.h (bitmap_count_bits): Prototype.
32210 (bitmap_clear_range, bitmap_set_range): Likewise.
32211 * sbitmap.c (bitmap_clear_range): New function.
32212 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
32214 2017-01-13 Martin Liska <mliska@suse.cz>
32217 * function.c (set_cfun): Add new argument force.
32218 * function.h (set_cfun): Likewise.
32219 * ipa-inline-transform.c (inline_call): Use the function when
32220 strict alising from is dropped for function we inline to.
32222 2017-01-13 Richard Biener <rguenther@suse.de>
32224 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
32225 for dumping GIMPLE INTEGER_CSTs.
32227 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32229 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
32230 to 201112L since C++17.
32232 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
32235 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
32236 if -fsanitize=kernel-address is present.
32238 2017-01-13 Richard Biener <rguenther@suse.de>
32240 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
32241 as _Literal ( type ) number in case usual suffixes do not
32242 preserve all information.
32244 2017-01-13 Richard Biener <rguenther@suse.de>
32246 PR tree-optimization/77283
32247 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
32248 and ssa-iterators.h.
32249 (is_feasible_trace): Implement a cost model based on joiner
32252 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
32255 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
32256 char or short to __float128/_Float128 directly.
32258 2017-01-12 Martin Sebor <msebor@redhat.com>
32260 to -Wformat-overflow.
32261 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
32262 (min_bytes_remaining): Same.
32263 (get_string_length): Same.
32264 (format_string): Same.
32265 (format_directive): Same.
32267 (pass_sprintf_length::handle_gimple_call): Same.
32269 2017-01-12 Jakub Jelinek <jakub@redhat.com>
32271 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
32272 info.nowrite calls with no lhs that can't throw. Return bool
32273 whether gsi_remove has been called or not.
32274 (pass_sprintf_length::handle_gimple_call): Return bool whether
32275 try_substitute_return_value called gsi_remove. Formatting fix.
32276 (pass_sprintf_length::execute): Don't use gsi_remove if
32277 handle_gimple_call returned true.
32280 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
32281 be removed due to side-effects, don't remove following barrier nor
32282 turn the successor edge into fallthru edge.
32284 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32287 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
32288 element-reversing loads and stores as not swappable.
32290 2017-01-12 Nathan Sidwell <nathan@acm.org>
32291 Nicolai Stange <nicstange@gmail.com>
32293 * combine.c (try_combine): Don't ignore result of overlap checking
32294 loop. Combine overlap & asm check into single loop.
32296 2017-01-12 Richard Biener <rguenther@suse.de>
32298 * tree-pretty-print.c (dump_generic_node): Provide -gimple
32299 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
32301 2017-01-12 Richard Biener <rguenther@suse.de>
32303 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
32304 and TS_TARGET_OPTION directly derive from TS_BASE.
32305 * tree-core.h (tree_optimization_option): Derive from tree_base.
32306 (tree_target_option): Likewise.
32308 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
32310 * config/i386/i386.c (memory_address_length): Increase len
32311 only when rip_relative_addr_p returns false.
32313 2017-01-11 Julia Koval <julia.koval@intel.com>
32315 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
32316 (OPTION_MASK_ISA_SGX_SET): New.
32317 (ix86_handle_option): Handle OPT_msgx.
32318 * config.gcc: Added sgxintrin.h.
32319 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
32320 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
32321 * config/i386/i386.c (ix86_target_string): Add -msgx.
32323 (ix86_option_override_internal): Handle new options.
32324 (ix86_valid_target_attribute_inner_p): Add sgx.
32325 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
32326 * config/i386/i386.opt: Add msgx.
32327 * config/i386/sgxintrin.h: New file.
32328 * config/i386/x86intrin.h: Add sgxintrin.h.
32330 2017-01-11 Jakub Jelinek <jakub@redhat.com>
32333 * fold-const.c (maybe_nonzero_address): Return 1 for function
32335 (tree_single_nonzero_warnv_p): Don't handle function local objects
32339 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
32342 2017-01-11 David Malcolm <dmalcolm@redhat.com>
32345 * opts.c: Include "spellcheck.h"
32346 (struct string_fragment): New struct.
32347 (struct edit_distance_traits<const string_fragment &>): New
32349 (get_closest_sanitizer_option): New function.
32350 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
32352 2017-01-11 Jakub Jelinek <jakub@redhat.com>
32354 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
32356 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
32357 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
32358 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
32359 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
32360 for initial die_offset if dwarf_split_debug_info.
32361 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
32362 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
32364 (output_skeleton_debug_sections): Formatting fix. Use
32365 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
32366 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
32368 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
32370 * config/arm/cortex-a53.md: Add bypasses for
32371 cortex_a53_r2f_cvt.
32372 (cortex_a53_r2f): Only use for transfers.
32373 (cortex_a53_f2r): Likewise.
32374 (cortex_a53_r2f_cvt): Add reservation for conversions.
32375 (cortex_a53_f2r_cvt): Likewise.
32377 2017-01-11 Tamar Christina <tamar.christina@arm.com>
32379 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
32380 to all inlined functions, change static to extern.
32382 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
32385 * config/arm/arm.c (legitimize_pic_address): Handle reference to
32387 (arm_assemble_integer): Likewise.
32389 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
32391 * config.gcc: Use new awk script to check CPU, FPU and architecture
32392 parameters for --with-... options.
32393 * config/arm/parsecpu.awk: New file
32394 * config/arm/arm-cpus.in: New file.
32395 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
32397 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
32399 * config/arm/t-arm: Update dependency rules.
32400 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
32401 of processing .def files.
32402 * config/arm/genopt.sh: Deleted.
32403 * config/arm/gentune.sh: Deleted.
32404 * config/arm/arm-cores.def: Deleted.
32405 * config/arm/arm-arches.def: Deleted.
32406 * config/arm/arm-fpus.def: Deleted.
32407 * config/arm/arm-tune.md: Regenerated.
32408 * config/arm/arm-tables.opt: Regenerated.
32409 * config/arm/arm-cpu.h: New generated file.
32410 * config/arm/arm-cpu-data.h: New generated file.
32411 * config/arm/arm-cpu-cdata.h: New generated file.
32413 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
32416 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
32418 (input_varpool_node): Unpack dynamically_initialized bit.
32420 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
32422 PR rtl-optimization/79032
32423 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
32424 the alignment of the adjusted memory reference against that of MODE,
32425 instead of the alignment of the original memory reference.
32427 2017-01-11 Martin Jambor <mjambor@suse.cz>
32429 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
32431 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
32432 decorated functions.
32434 2017-01-11 Richard Biener <rguenther@suse.de>
32436 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
32437 set range/nonnull info for PHI results. Do not set it on
32438 stmts marked for removal.
32440 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
32442 * expr.c (store_field): In the bitfield case, fetch the return value
32443 from the registers before applying a single big-endian adjustment.
32444 Always do a final load for a BLKmode value not larger than a word.
32446 2017-01-10 David Malcolm <dmalcolm@redhat.com>
32449 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
32450 that we correctly handle column numbers greater than
32451 LINE_MAP_MAX_COLUMN_NUMBER.
32453 2017-01-10 Martin Sebor <msebor@redhat.com>
32455 PR middle-end/78245
32456 * gimple-ssa-sprintf.c (get_destination_size): Call
32457 {init,fini}object_sizes.
32458 * tree-object-size.c (addr_object_size): Adjust.
32459 (pass_through_call): Adjust.
32460 (pass_object_sizes::execute): Adjust.
32461 * tree-object-size.h (fini_object_sizes): Declare.
32463 2017-01-10 Martin Sebor <msebor@redhat.com>
32465 PR tree-optimization/78775
32466 * builtins.c (get_size_range): Move...
32467 * calls.c: ...to here.
32468 (alloc_max_size): Accept zero argument.
32469 (operand_signed_p): Remove.
32470 (maybe_warn_alloc_args_overflow): Call get_size_range.
32471 * calls.h (get_size_range): Declare.
32473 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
32475 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
32476 from TI's devices.csv file as of September 2016.
32477 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
32479 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
32481 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
32482 * doc/invoke.texi: Likewise.
32483 * doc/md.texi: Likewise.
32484 * doc/objc.texi: Likewise.
32486 2017-01-10 Joshua Conner <joshconner@google.com>
32488 * config/arm/fuchsia-elf.h: New file.
32489 * config/fuchsia.h: New file.
32490 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
32491 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
32493 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
32495 2016-01-10 Richard Biener <rguenther@suse.de>
32497 PR tree-optimization/79034
32498 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
32499 Propagate out degenerate PHIs in the joiner.
32501 2017-01-10 Martin Liska <mliska@suse.cz>
32503 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
32504 (sort_congruence_classes_by_decl_uid): Likewise.
32505 (sort_congruence_class_groups_by_decl_uid): Likewise.
32506 (sem_item_optimizer::merge_classes): Sort class, groups in these
32507 classes and members in the groups by DECL_UID of declarations.
32508 This would make merge operations stable.
32510 2017-01-10 Martin Liska <mliska@suse.cz>
32512 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
32513 usage of m_classes_vec.
32514 (sem_item_optimizer::~sem_item_optimizer): Likewise.
32515 (sem_item_optimizer::get_group_by_hash): Likewise.
32516 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
32517 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
32518 (sem_item_optimizer::verify_classes): Likewise.
32519 (sem_item_optimizer::process_cong_reduction): Likewise.
32520 (sem_item_optimizer::dump_cong_classes): Likewise.
32521 (sem_item_optimizer::merge_classes): Likewise.
32522 * ipa-icf.h (congruence_class_hash): Rename from
32523 congruence_class_group_hash. Remove declaration of m_classes_vec.
32525 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
32527 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
32528 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
32529 * config.gcc: Add avx512vpopcntdqintrin.h.
32530 * config/i386/avx512vpopcntdqintrin.h: New.
32531 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
32532 * config/i386/i386-builtin-types.def: Add new types.
32533 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
32534 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
32535 __builtin_ia32_vpopcountq_v8di_mask): New.
32536 * config/i386/i386-c.c (ix86_target_macros_internal): Define
32537 __AVX512VPOPCNTDQ__.
32538 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
32539 (PTA_AVX512VPOPCNTDQ): Define.
32540 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
32541 TARGET_AVX512VPOPCNTDQ_P): Define.
32542 * config/i386/i386.opt: Add mavx512vpopcntdq.
32543 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
32544 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
32546 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32548 PR middle-end/77484
32549 * predict.def (PRED_CALL): Set to 67.
32551 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
32553 * expr.c (store_field): In the bitfield case, if the value comes from
32554 a function call and is of an aggregate type returned in registers, do
32555 not modify the field mode; extract the value in all cases if the mode
32556 is BLKmode and the size is not larger than a word.
32558 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
32561 * config/i386/cpuid.h: Fix undefined behavior.
32563 2017-01-04 Jeff Law <law@redhat.com>
32565 PR tree-optimization/79007
32566 PR tree-optimization/67955
32567 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
32568 conservative for pt.null when flag_non_call_exceptions is on.
32570 2017-01-09 Jakub Jelinek <jakub@redhat.com>
32572 PR translation/79019
32573 PR translation/79020
32574 * params.def (PARAM_INLINE_MIN_SPEEDUP,
32575 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
32576 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
32578 * config/avr/avr.opt (maccumulate-args): Likewise.
32579 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
32580 * common.opt (freport-bug): Likewise.
32581 * cif-code.def (CIF_FINAL_ERROR): Likewise.
32582 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
32583 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
32584 translatable string.
32585 * config/i386/i386.c (function_value_32): Likewise.
32586 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
32587 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
32589 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
32590 * common/config/msp430/msp430-common.c (msp430_handle_option):
32592 * symtab.c (symtab_node::verify_base): Likewise.
32593 * opts.c (set_debug_level): Likewise.
32594 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
32595 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
32596 missing whitespace to translatable strings.
32597 * config/avr/avr.md (bswapsi2): Fix typo in comment.
32598 * config/sh/superh.h: Likewise.
32599 * config/i386/xopintrin.h: Likewise.
32600 * config/i386/znver1.md: Likewise.
32601 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
32602 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
32603 * double-int.h (struct double_int): Likewise.
32604 * double-int.c (div_and_round_double): Likewise.
32605 * wide-int.cc: Likewise.
32606 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
32607 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
32608 * cfgcleanup.c (crossjumps_occured): Renamed to ...
32609 (crossjumps_occurred): ... this.
32610 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
32613 PR tree-optimization/78899
32614 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
32615 returning bool return struct loop *, NULL for failure and the new
32617 (versionable_outer_loop_p): Don't version outer loop if it has
32618 dont_vectorized bit set.
32619 (tree_if_conversion): When versioning outer loop, ensure
32620 tree_if_conversion is performed also on the inner loop of the
32621 non-vectorizable outer loop copy.
32622 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
32623 LOOP_VECTORIZED in inner loop of the scalar outer loop and
32624 prevent vectorization of it.
32625 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
32626 the outer loop vectorization of the non-scalar version is attempted
32627 before vectorization of the inner loop in scalar version. If
32628 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
32629 vectorization of its inner loop.
32630 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
32631 has 2 inner loops, rename also on edges from bb whose single pred
32632 is outer_loop->header. Fix typo in function comment.
32634 2017-01-09 Martin Sebor <msebor@redhat.com>
32637 * asan.c (asan_emit_stack_protection): Increase local buffer size
32638 to avoid snprintf truncation warning.
32640 2017-01-09 Andrew Pinski <apinski@cavium.com>
32642 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
32643 to reference thunderx2t99 for the tuning structure
32644 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
32646 (thunderx2t99_extra_costs): This.
32647 * config/aarch64/aarch64-tune.md: Regenerate.
32648 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
32649 (vulcan_addrcost_table): This.
32650 (vulcan_regmove_cost): Rename to ...
32651 (thunderx2t99_regmove_cost): This.
32652 (vulcan_vector_cost): Rename to ...
32653 (thunderx2t99_vector_cost): this.
32654 (vulcan_branch_cost): Rename to ...
32655 (thunderx2t99_branch_cost): This.
32656 (vulcan_tunings): Rename to ...
32657 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
32658 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
32660 2017-01-09 Martin Jambor <mjambor@suse.cz>
32664 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
32665 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
32666 (propagate_vr_accross_jump_function): Use the above function for all
32667 value range computations for pass-through jump functions and type
32668 converasion from explicit value range values.
32669 (ipcp_propagate_stage): Do not attempt to deduce types of formal
32670 parameters from TYPE_ARG_TYPES.
32671 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
32672 (ipa_write_node_info): Stream type of the actual argument.
32673 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
32675 2017-01-09 Martin Liska <mliska@suse.cz>
32678 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
32679 (lookup_compiler): Do not show error message with have_E.
32681 2017-01-09 Jakub Jelinek <jakub@redhat.com>
32683 PR tree-optimization/78938
32684 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
32685 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
32686 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
32687 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
32690 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32692 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
32695 2017-01-09 Richard Biener <rguenther@suse.de>
32697 PR tree-optimization/78997
32698 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
32699 name condition properly.
32701 2017-01-09 Richard Biener <rguenther@suse.de>
32704 * dwarf2out.c (is_cxx): New overload with context.
32705 (is_naming_typedef_decl): Use it.
32707 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
32709 * invoke.texi (Option Summary): Correct spacing in option lists
32710 and add line breaks to fix over-long lines.
32712 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
32714 PR middle-end/17660
32716 * extend.texi (Common Variable Attributes): Add xref to GCC
32717 Internals manual to explain mode attribute keywords.
32719 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
32722 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
32723 and Preprocessor Options.
32724 (Options for Linking): Document -pthread here....
32725 (RS/6000 and PowerPC Options): ...not here.
32726 (Solaris 2 Options): ...or here.
32727 * doc/cppopts.texi: Document -pthread.
32729 2017-01-08 Martin Sebor <msebor@redhat.com>
32731 PR middle-end/77708
32732 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
32733 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
32734 New member functions.
32735 (format_directive): Used them.
32737 (pass_sprintf_length::handle_gimple_call): Same.
32738 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
32739 to avoid truncation for any argument.
32740 (extract_affine_mul): Same.
32741 * tree.c (get_file_function_name): Same.
32743 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32745 PR middle-end/77484
32746 * predict.def (PRED_INDIR_CALL): Set to 86.
32748 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
32750 PR preprocessor/54124
32751 * doc/cppopts.texi: Reformat -d subtable to list the full name
32752 of the options. Add cross-reference to the docs for the general
32753 compiler -d options.
32754 * doc/invoke.texi (Developer Options): Add cross-reference to the
32755 preprocessor-specific -d option documentation.
32757 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
32759 PR preprocessor/13498
32760 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
32761 redudant material, and reflect new command-line options.
32762 (System Headers): Likewise.
32764 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
32766 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
32767 -isystem, and -idirafter. Copy-edit.
32768 * doc/cppopts.texi: Copy-edit. Remove contradiction about
32769 default for -ftrack-macro-expansion. Delete obsolete and
32770 badly-formatted implementation details about -fdebug-cpp output.
32771 * doc/cppwarnopts.texi: Copy-edit.
32773 2017-01-07 David Malcolm <dmalcolm@redhat.com>
32776 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
32777 that the transition from a max line width >= 1<<10 to narrower
32778 lines works correctly.
32780 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
32782 * doc/options.texi (PerFunction): New.
32783 * opt-functions.awk (switch_flags): Map both Optimization and
32784 PerFunction to CL_OPTIMIZATION.
32785 * opth-gen.awk: Test for PerFunction flag along with
32787 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
32788 it only when the latter is present. Skip those that don't in
32789 the hash function generator.
32790 * common.opt (fvar-tracking): Mark as PerFunction instead of
32792 (fvar-tracking-assignments): Likewise.
32793 (fvar-tracking-assignments-toggle): Likewise.
32794 (fvar-tracking-uninit): Likewise.
32796 2017-01-07 Jakub Jelinek <jakub@redhat.com>
32798 PR translation/79018
32799 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
32802 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
32805 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
32806 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
32807 TARGET_LONG_JUMP_TABLE_OFFSETS.
32808 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
32809 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
32810 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
32811 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
32812 * config/m68k/m68k.md (tablejump expander): Likewise.
32813 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
32814 TARGET_LONG_JUMP_TABLE_OFFSETS.
32815 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
32816 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
32818 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
32819 David Holsgrove <david.holsgrove@xilinx.com>
32821 * common/config/microblaze/microblaze-common.c
32822 (TARGET_EXCEPT_UNWIND_INFO): Remove.
32823 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
32825 * config/microblaze/microblaze.c (microblaze_must_save_register)
32826 (microblaze_expand_epilogue, microblaze_return_addr): Handle
32828 (microblaze_eh_return): New function.
32829 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
32830 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
32831 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
32832 * config/microblaze/microblaze.md (eh_return): New pattern.
32834 2017-01-06 Jakub Jelinek <jakub@redhat.com>
32836 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
32837 GCC_DIAGNOSTIC_STRINGIFY): Define.
32839 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
32841 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32843 * config/arm/arm.md (<mcrr>): New.
32845 * config/arm/arm.c (arm_arch5te): New.
32846 (arm_option_override): Set arm_arch5te.
32847 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
32849 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
32850 (arm_mcrr_qualifiers): ... this. New.
32851 (MRRC_QUALIFIERS): Define to...
32852 (arm_mrrc_qualifiers): ... this. New.
32853 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
32855 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
32856 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
32857 (MRRCI, mrrc, MRRC): New.
32858 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
32859 VUNSPEC_MRRC2): New.
32861 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32863 * config/arm/arm.md (<mcr>): New.
32865 * config/arm/arm.c (arm_coproc_builtin_available): Add
32866 support for mcr, mrc, mcr2 and mrc2.
32867 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
32868 (arm_mcr_qualifiers): ... this. New.
32869 (MRC_QUALIFIERS): Define to ...
32870 (arm_mrc_qualifiers): ... this. New.
32871 (MCR_QUALIFIERS): Define to ...
32872 (arm_mcr_qualifiers): ... this. New.
32873 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
32875 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
32876 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
32877 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
32878 VUNSPEC_MRC2): New.
32880 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32882 * config/arm/arm.md (*ldc): New.
32886 * config/arm/arm.c (arm_coproc_builtin_available): Add
32887 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
32888 (arm_coproc_ldc_stc_legitimate_address): New.
32889 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
32890 'qualifier_const_pointer'.
32891 (LDC_QUALIFIERS): Define to...
32892 (arm_ldc_qualifiers): ... this. New.
32893 (STC_QUALIFIERS): Define to...
32894 (arm_stc_qualifiers): ... this. New.
32895 * config/arm/arm-protos.h
32896 (arm_coproc_ldc_stc_legitimate_address): New.
32897 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
32898 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
32899 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
32900 stc2, stcl, stc2l): New.
32901 * config/arm/constraints.md (Uz): New.
32902 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
32903 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
32904 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
32905 VUNSPEC_STC2L): New.
32907 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32909 * config/arm/arm.md (<cdp>): New.
32910 * config/arm/arm.c (neon_const_bounds): Rename this ...
32911 (arm_const_bounds): ... this.
32912 (arm_coproc_builtin_available): New.
32913 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
32914 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
32915 (CDP_QUALIFIERS): Define to...
32916 (arm_cdp_qualifiers): ... this. New.
32918 (arm_expand_builtin_args): Add case for 6 arguments.
32919 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
32920 (arm_const_bounds): ... this.
32921 (arm_coproc_builtin_available): New.
32922 * config/arm/arm_acle.h (__arm_cdp): New.
32924 * config/arm/arm_acle_builtins.def (cdp): New.
32926 * config/arm/iterators.md (CDPI,CDP,cdp): New.
32927 * config/arm/neon.md: Rename all 'neon_const_bounds' to
32928 'arm_const_bounds'.
32929 * config/arm/types.md (coproc): New.
32930 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
32931 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
32932 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
32933 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
32935 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32937 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
32938 (UBINOP_QUALIFIERS): New.
32940 (acle_builtin_data): New. Change comment.
32941 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
32942 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
32943 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
32944 arm_acle_builtins.def.
32945 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
32946 (arm_init_acle_builtins): New.
32947 (CRC32_BUILTIN): Remove.
32948 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
32949 crc32cb, crc32ch and crc32cw.
32950 (arm_init_crc32_builtins): Remove.
32951 (arm_init_builtins): Use arm_init_acle_builtins rather
32952 than arm_init_crc32_builtins.
32953 (arm_expand_acle_builtin): New.
32954 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
32955 * config/arm/arm_acle_builtins.def: New.
32957 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32959 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
32960 (arm_builtin_datum): ... this.
32961 (arm_init_neon_builtin): Rename to ...
32962 (arm_init_builtin): ... this. Add a new parameters PREFIX
32963 and USE_SIG_IN_NAME.
32964 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
32965 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
32966 'arm_builtin_datum'.
32967 (arm_init_vfp_builtins): Likewise.
32968 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
32969 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
32970 (arm_expand_neon_args): Rename to ...
32971 (arm_expand_builtin_args): ... this. Rename builtin_arg
32972 enum values and differentiate between ARG_BUILTIN_MEMORY
32973 and ARG_BUILTIN_NEON_MEMORY.
32974 (arm_expand_neon_builtin_1): Rename to ...
32975 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
32976 values, arm_expand_builtin_args and add bool parameter NEON.
32977 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
32978 (arm_expand_vfp_builtin): Likewise.
32979 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
32981 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32983 PR middle-end/77484
32984 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
32985 * predict.c (tree_estimate_probability_bb): Reverse direction of
32986 polymorphic call predictor.
32988 2017-01-06 David Malcolm <dmalcolm@redhat.com>
32990 * passes.c (execute_one_pass): Split out pass-skipping logic into...
32991 (determine_pass_name_match): ...this new function and...
32992 (should_skip_pass_p): ...this new function.
32994 2017-01-06 Nathan Sidwell <nathan@acm.org>
32996 * ipa-visibility.c (function_and_variable_visibility): Reformat
32997 comments and long lines. Remove extrneous if.
32998 * symtab.c (symtab_node::make_decl_local): Fix code format.
32999 (symtab_node::set_section_for_node): Fix comment typo.
33001 2017-01-06 Martin Liska <mliska@suse.cz>
33004 * lra-constraints.c: Rename invariant to lra_invariant.
33005 * predict.c (set_even_probabilities): Initialize e to NULL.
33007 2017-01-05 Martin Sebor <msebor@redhat.com>
33009 PR tree-optimization/78910
33010 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
33011 (format_integer): Correct off-by-one error in the handling
33012 of precision with negative numbers in signed conversions..
33014 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
33016 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
33018 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33020 PR tree-optimization/71016
33021 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
33022 factor_out_conditional_conversion. Formatting fix.
33023 (factor_out_conditional_conversion): Add cond_stmt argument.
33024 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
33025 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
33028 2017-01-05 David Malcolm <dmalcolm@redhat.com>
33030 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
33031 read-rtl-function.o, and selftest-rtl.o.
33032 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
33033 (selftest::aarch64_test_loading_full_dump): New function.
33034 (selftest::aarch64_run_selftests): New function.
33035 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
33036 selftest::aarch64_run_selftests.
33037 * config/i386/i386.c
33038 (selftest::ix86_test_loading_dump_fragment_1): New function.
33039 (selftest::ix86_test_loading_call_insn): New function.
33040 (selftest::ix86_test_loading_full_dump): New function.
33041 (selftest::ix86_test_loading_unspec): New function.
33042 (selftest::ix86_run_selftests): Call the new functions.
33043 * emit-rtl.c (maybe_set_max_label_num): New function.
33044 * emit-rtl.h (maybe_set_max_label_num): New decl.
33045 * function.c (instantiate_decls): Guard call to
33046 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
33047 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
33049 * gensupport.c (gen_reader::gen_reader): Pass "false"
33050 for new "compact" param of rtx_reader.
33051 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
33052 rather than an empty string for NULL strings.
33053 * read-md.c: Potentially include config.h rather than bconfig.h.
33054 Wrap include of errors.h with #ifdef GENERATOR_FILE.
33055 (have_error): New global, copied from errors.c.
33056 (md_reader::read_name): Rename to...
33057 (md_reader::read_name_1): ...this, adding "out_loc" param,
33058 and converting "missing name or number" to returning false, rather
33060 (md_reader::read_name): Reimplement in terms of read_name_1.
33061 (md_reader::read_name_or_nil): New function.
33062 (md_reader::read_string): Handle "(nil)" by returning NULL.
33063 (md_reader::md_reader): Add new param "compact".
33064 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
33065 (md_reader::read_file): New method.
33066 * read-md.h (md_reader::md_reader): Add new param "compact".
33067 (md_reader::read_file): New method.
33068 (md_reader::is_compact): New accessor.
33069 (md_reader::read_name): Convert return type from void to file_location.
33070 (md_reader::read_name_or_nil): New decl.
33071 (md_reader::read_name_1): New decl.
33072 (md_reader::m_compact): New field.
33073 (noop_reader::noop_reader): Pass "false" for new "compact" param
33075 (rtx_reader::rtx_reader): Add new "compact" param.
33076 (rtx_reader::read_rtx_operand): Make virtual and convert return
33077 type from void to rtx.
33078 (rtx_reader::read_until): New decl.
33079 (rtx_reader::handle_any_trailing_information): New virtual function.
33080 (rtx_reader::postprocess): New virtual function.
33081 (rtx_reader::finalize_string): New virtual function.
33082 (rtx_reader::m_in_call_function_usage): New field.
33083 (rtx_reader::m_reuse_rtx_by_id): New field.
33084 * read-rtl-function.c: New file.
33085 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
33086 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
33087 (selftest::verify_three_block_rtl_cfg): New decl.
33088 * read-rtl-function.h: New file.
33089 * read-rtl.c: Potentially include config.h rather than bconfig.h.
33090 For host, include function.h, memmodel.h, and emit-rtl.h.
33091 (one_time_initialization): New function.
33092 (struct compact_insn_name): New struct.
33093 (compact_insn_names): New array.
33094 (find_code): Handle insn codes in compact dumps.
33095 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
33096 (bind_subst_iter_and_attr): Likewise.
33097 (add_condition_to_string): Likewise.
33098 (add_condition_to_rtx): Likewise.
33099 (apply_attribute_uses): Likewise.
33100 (add_current_iterators): Likewise.
33101 (apply_iterators): Likewise.
33102 (initialize_iterators): Guard usage of apply_subst_iterator with
33103 #ifdef GENERATOR_FILE.
33104 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
33105 (md_reader::read_mapping): Likewise.
33106 (add_define_attr_for_define_subst): Likewise.
33107 (add_define_subst_attr): Likewise.
33108 (read_subst_mapping): Likewise.
33109 (check_code_iterator): Likewise.
33110 (rtx_reader::read_rtx): Likewise. Move one-time initialization
33112 (one_time_initialization): New function.
33113 (rtx_reader::read_until): New method.
33114 (read_flags): New function.
33115 (parse_reg_note_name): New function.
33116 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
33117 Handle reuse_rtx ids.
33118 Wrap iterator lookup within #ifdef GENERATOR_FILE.
33119 Add parsing support for RTL dumps, mirroring the special-cases in
33120 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
33121 values, and calling handle_any_trailing_information.
33122 (rtx_reader::read_rtx_operand): Convert return type from void
33123 to rtx, returning return_rtx. Handle case 'e'. Call
33124 finalize_string on XSTR and XTMPL fields.
33125 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
33126 "(nil)" values were omitted. Call the postprocess vfunc on the
33128 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
33129 class ctor. Initialize m_in_call_function_usage. Call
33130 one_time_initialization.
33131 * rtl-tests.c (selftest::test_uncond_jump): Call
33132 set_new_first_and_last_insn.
33133 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
33134 * selftest-rtl.c: New file.
33135 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
33136 (selftest::get_insn_by_uid): New decl.
33137 * selftest-run-tests.c (selftest::run_tests): Call
33138 read_rtl_function_c_tests.
33139 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
33140 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
33143 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
33145 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
33146 operands in a special way. Assert that pos+len <= mode precision.
33148 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33150 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
33151 3 argument Alias with unlimited for the negative form.
33152 (fno-vect-cost-model): Removed.
33154 2017-01-05 Martin Liska <mliska@suse.cz>
33156 * hsa-gen.c (gen_hsa_divmod): New function.
33157 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
33159 2017-01-05 Martin Liska <mliska@suse.cz>
33162 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
33165 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33167 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
33168 small constant length operands.
33170 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33172 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
33173 between loop iterations.
33175 2017-01-05 Martin Liska <mliska@suse.cz>
33178 * gimplify.c (gimplify_decl_expr): Compare to
33179 asan_poisoned_variables instread of checking flags.
33180 (gimplify_target_expr): Likewise.
33181 (gimplify_expr): Likewise.
33182 (gimplify_function_tree): Conditionally initialize
33183 asan_poisoned_variables.
33185 2017-01-04 Jeff Law <law@redhat.com>
33187 PR tree-optimizatin/78812
33188 * rtl.h (contains_mem_rtx_p): Prototype.
33189 * ifcvt.c (containts_mem_rtx_p): Move from here to...
33190 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
33191 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
33192 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
33193 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
33195 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33197 * input.c (assert_char_at_range): Default-initialize actual_range.
33199 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33201 * df-scan.c (df_ref_create_structure): Make regno unsigned,
33202 to match the caller.
33204 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33206 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
33207 insns after final jump in test to emit dummy move.
33209 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33211 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
33212 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
33214 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33216 * multiple_target.c (create_dispatcher_calls): Init e_next.
33217 * tree-ssa-loop-split.c (split_loop): Init border.
33218 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
33221 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
33226 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
33227 (altivec_register_operand): Do not return true if the operand
33228 contains a SUBREG mixing SImode and SFmode.
33229 (vsx_register_operand): Likewise.
33230 (vsx_reg_sfsubreg_ok): New predicate.
33231 (vfloat_operand): Do not return true if the operand contains a
33232 SUBREG mixing SImode and SFmode.
33233 (vint_operand): Likewise.
33234 (vlogical_operand): Likewise.
33235 (gpc_reg_operand): Likewise.
33236 (int_reg_operand): Likewise.
33237 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
33238 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
33239 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
33241 (rs6000_emit_move_si_sf_subreg): New helper function.
33242 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
33243 fixup SUBREGs involving SImode and SFmode.
33244 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
33245 numbers for the new peephole2 optimization.
33246 (peephole2 for SFmode unions): New peephole2 to optimize cases in
33247 the GLIBC math library that do AND/IOR/XOR operations on single
33248 precision floating point.
33249 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
33250 target macros to say whether we need to avoid SUBREGs mixing
33252 (TARGET_ALLOW_SF_SUBREG): Likewise.
33253 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
33254 (UNSPEC_SI_FROM_SF): Likewise.
33255 (iorxor): Change spacing.
33256 (and_ior_xor): New iterator for AND, IOR, and XOR.
33257 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
33258 (movdi_from_sf_zero_ext): Likewise.
33259 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
33260 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
33261 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
33262 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
33263 (fms<mode>4): Likewise.
33264 (fnma<mode>4): Likewise.
33265 (fnms<mode>4): Likewise.
33266 (nfma<mode>4): Likewise.
33267 (nfms<mode>4): Likewise.
33269 2017-01-04 Marek Polacek <polacek@redhat.com>
33272 * doc/invoke.texi: Document -Wpointer-compare.
33274 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33276 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
33279 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
33280 descriptions for -gdwarf-5 and emit them as uleb128 instead of
33283 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
33286 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
33287 documentation of the powerpc_popcntb_ok attribute.
33288 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
33289 code to issue warning messages if a requested CPU configuration is
33290 not supported by the binary (assembler and loader) toolchain.
33291 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
33292 made to define a built-in function that has been disabled.
33293 (paired_init_builtins): Add assertion to prevent ICE if attempt is
33294 made to define a built-in function that has been disabled.
33295 (altivec_init_builtins): Add comment explaining why definition
33296 of the DST built-in functions is not preceded by an assertion
33297 check. Add assertions to prevent ICE if attempts are made to
33298 define an altivec predicate or an abs* built-in function that has
33300 (htm_init_builtins): Add comment explaining why definition of the
33301 htm built-in functions is not preceded by an assertion check.
33303 2017-01-04 Jeff Law <law@redhat.com>
33305 PR tree-optimizatin/67955
33306 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
33307 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
33308 the points-to solution does not include pt_null. Use DECL_PT_UID
33311 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
33313 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
33314 Use gen_int_mode instead of gen_lopwart for const_int operands.
33316 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33318 PR tree-optimization/71563
33319 * match.pd: Simplify X << Y into X if Y is known to be 0 or
33320 out of range value - has low bits known to be zero.
33322 2017-01-04 Alan Modra <amodra@gmail.com>
33324 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
33325 * configure: Regenerate.
33326 * config.in: Regenerate.
33328 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33331 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
33332 a substring of the message, but strcmp with the whole message. Ifdef
33333 ENABLE_NLS, translate the message first using dgettext.
33335 2017-01-03 Jeff Law <law@redhat.com>
33337 PR tree-optimizatin/78856
33338 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
33339 (mark_threaded_blocks): Remove code to truncate thread paths that
33340 cross multiple loop headers. Instead invalidate the cached loop
33341 iteration information and handle case of a thread path walking
33342 into an irreducible region.
33344 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
33347 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
33348 assertions. Add support for doing the signbit if the IEEE 128-bit
33349 floating point value is in a GPR.
33350 * config/rs6000/rs6000.md (Fsignbit): Delete.
33351 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
33352 Update the length attribute if the value is in a GPR.
33353 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
33354 the sign or zero extension instruction, since the value is always 0/1.
33355 (signbit<mode>2_dm2): Delete using <Fsignbit>.
33358 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
33359 extracting SImode to a GPR register so that we can generate a
33360 store, limit the vector to be in a traditional Altivec register
33361 for the vextuwrx instruction.
33363 2017-01-03 Ian Lance Taylor <iant@google.com>
33365 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
33367 2017-01-03 Martin Sebor <msebor@redhat.com>
33369 PR tree-optimization/78696
33370 * gimple-ssa-sprintf.c (format_floating): Correct handling of
33371 precision. Use MPFR for %f for greater fidelity. Correct handling
33373 (pass_sprintf_length::compute_format_length): Set width and precision
33374 specified by asrerisk to void_node for vararg functions.
33375 (try_substitute_return_value): Adjust dump output.
33377 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
33379 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
33381 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
33383 * doc/invoke.texi (SPARC options): Document -mlra as the default.
33384 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
33385 -mlra/-mno-lra was passed to the compiler.
33387 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
33389 PR rtl-optimization/65618
33390 * emit-rtl.c (try_split): Move initialization of "before" and
33391 "after" to just before the call to emit_insn_after_setloc.
33393 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
33395 * doc/md.texi (Standard Names): Remove reference to Java frontend.
33397 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
33399 * dwarf2out.c (gen_enumeration_type_die): When
33400 -gno-strict-dwarf, add a DW_AT_encoding attribute.
33402 2017-01-03 Jakub Jelinek <jakub@redhat.com>
33404 PR tree-optimization/78965
33405 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
33406 Change first argument from const call_info & to call_info &. For %n
33407 set info.nowrite to false.
33409 PR middle-end/78901
33410 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
33411 possibly throwing calls.
33413 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
33414 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
33415 and fns handling, rather than in a separate case SSA_NAME.
33417 2017-01-02 Jeff Law <law@redhat.com>
33419 * config/darwin-driver.c (darwin_driver_init): Const-correctness
33420 fixes for first_period and second_period variables.
33422 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
33425 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
33426 (*insvqi_1): New insn pattern.
33427 (*insvqi_1_mem_rex64): Ditto.
33428 (*insvqi_2): Ditto.
33429 (*insvqi_3): Rename from *insvqi.
33431 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
33433 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
33435 * doc/cfg.texi (Edges): Remove reference to Java.
33436 (Maintaining the CFG): Ditto.
33438 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33440 PR middle-end/77674
33441 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
33442 transparent aliases.
33444 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33446 PR middle-end/77484
33447 * predict.def (PRED_CALL): Update hitrate.
33448 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
33449 * predict.c (tree_estimate_probability_bb): Split CALL predictor
33450 into direct/indirect/polymorphic variants.
33452 2017-01-01 Jakub Jelinek <jakub@redhat.com>
33454 Update copyright years.
33456 * gcc.c (process_command): Update copyright notice dates.
33457 * gcov-dump.c (print_version): Ditto.
33458 * gcov.c (print_version): Ditto.
33459 * gcov-tool.c (print_version): Ditto.
33460 * gengtype.c (create_file): Ditto.
33461 * doc/cpp.texi: Bump @copying's copyright year.
33462 * doc/cppinternals.texi: Ditto.
33463 * doc/gcc.texi: Ditto.
33464 * doc/gccint.texi: Ditto.
33465 * doc/gcov.texi: Ditto.
33466 * doc/install.texi: Ditto.
33467 * doc/invoke.texi: Ditto.
33469 Copyright (C) 2017 Free Software Foundation, Inc.
33471 Copying and distribution of this file, with or without modification,
33472 are permitted in any medium without royalty provided the copyright
33473 notice and this notice are preserved.